Periphrasis, Replacement and Renewal: Studies in English Historical Linguistics

Periphrasis, Replacement and Renewal: Studies in English Historical Linguistics

Periphrasis, Replacement and Renewal: Studies in English Historical Linguistics

Edited by

Irén HegedĦs and Dóra PĘdör

Periphrasis, Replacement and Renewal: Studies in English Historical Linguistics, Edited by Irén HegedĦs and Dóra PĘdör This book first published 2013 Cambridge Scholars Publishing 12 Back Chapman Street, Newcastle upon Tyne, NE6 2XX, UK

British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library

Copyright © 2013 by Irén HegedĦs and Dóra PĘdör and contributors All rights for this book reserved. No part of this book may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior permission of the copyright owner. ISBN (10): 1-4438-4423-3, ISBN (13): 978-1-4438-4423-9

TABLE OF CONTENTS

Foreword and Acknowledgements ............................................................ vii Introduction Periphrasis, Replacement and Renewal: Lessons for and from the History of English Irén HegedĦs and Dóra PĘdör .................................................................... 1 Part I: Patterns and Models of Replacement What are ‘Grammatical Periphrases’? John Anderson .......................................................................................... 14 Morphological Productivity as a Trigger of the þurfan / need Replacement Lucía Loureiro-Porto ................................................................................ 34 Aspectual Loss and Renewal: onginnan, beginnan, start Lynn D. Sims ............................................................................................ 57 Litteral Substitution Sets in MSS of The Man of Law’s Tale Justyna Rogos ........................................................................................... 83 Part II: Syntactic and Semantic Variation and Change Syntactic and Semantic Variation: Middle English Causatives and Complement Types Brian Lowrey ............................................................................................ 92 V-1 Main Declarative Clauses in Old English and Old High German Biblical Translations Anna Cichosz .......................................................................................... 125 The Grammaticalization of Old English whether in the Left Periphery Víctor Parra-Guinaldo ............................................................................ 152

vi

Table of Contents

The Loss of Lexical Case in the History of English: What is Behind Transitivation? Fuyo Osawa ............................................................................................ 167 Modal Adjectives in the History of English Octav Eugen DeLazero ........................................................................... 187 Part III: Adverbials and Particles Adverbials in the History of English: The Case of before Artur Bartnik .......................................................................................... 212 The Historical Development of no sooner … than and its Semantic Change Daisuke Suzuki ....................................................................................... 228 A Turn of the Screw in the Semantics of Phrasal Verbs: Phrasal Verbs with up as a Test Case Paula Rodríguez-Puente ......................................................................... 243 Part IV: Functional and Regional Variation in Discourse and Vocabulary Discourse Strategies in Margery Kempe’s Descriptions of Her Pilgrimages Fumiko Yoshikawa ................................................................................. 268 The Rise of Disciplinary Identity Lilo Moessner ......................................................................................... 286 White Kennett’s Etymological Collections of English Words and Provincial Expressions: An Unprinted Source of Regional Vocabulary Javier Ruano-García ............................................................................... 310 Contributors ............................................................................................ 333 Index........................................................................................................ 339

FOREWORD AND ACKNOWLEDGEMENTS This volume carries the peer-reviewed, revised version of selected papers from the 16th International Conference on English Historical Linguistics held in Pécs, Hungary in 2010. That year was special both for the city and for its university because Pécs enjoyed the title and status of “European Capital of Culture”, and to celebrate this, the University of Pécs hosted numerous academic events, among them the 16th ICEHL. Since this conference was integrated into the academic programme specially designed by the University of Pécs to mark the year of “European Capital of Culture”, organizing the conference was greatly facilitated by the generous financial grant from the University of Pécs, which is gratefully acknowledged here. Furthermore, the organizers wish to express their gratitude to Professor Ferenc Fischer, Dean of the Faculty of Humanities, University of Pécs for the moral and logistic support provided in the phase of conference preparations. There were ca. 130 papers presented at the 16th ICEHL; the abstracts are available at the Pécs website of the 16th ICEHL (http://www.icehl16.pte.hu). Only 15 papers appear in the present selection but other papers have also been published; their publication details are also available at the above website. The editors wish to thank the following experts for their participation in the arduous process of paper selection: József Andor, Michael Benskin, Charlotte Brewer, Tamás Eitler, Alexandra Fodor, Leena Kahlas-Tarkka, László Kristó, Ursula Lenker, Bettelou Los, Gabriella Mazzon, and Stefan Thim. Their suggestions, corrections have helped not only the editors’ tasks but were also appreciated by the contributors to the present volume. Special thanks go to our colleague, Alexandra Fodor (Semmelweis University, Budapest) for all the work she has put into managing the lengthy process of paper review and selection preceding the actual editing of this volume. The editors are indebted to the contributors not only for their kind cooperation but also for a variety of comments and corrections that they made in connection with the introductory chapter written by the editors. Last but not least we want to thank Carol Koulikourdi, Amanda Millar and Emily Surrey at Cambridge Scholars Publishing for the assistance they gave us during the preparation of the manuscript for publishing. Pécs, October 2012

The Editors

INTRODUCTION PERIPHRASIS, REPLACEMENT AND RENEWAL: LESSONS FOR AND FROM THE HISTORY OF ENGLISH IRÉN HEGEDĥS AND DÓRA PėDÖR

In the infinite process of language change various factors can lead to the loss of lexical elements or grammatical features. Parallel with this, the solutions to remedy such losses and to renew the affected part of the linguistic system can also be multifarious. Synchronic theory and diachronic investigations smoothly blend in research inquiring into the question of how the language system is renewed after loss or how a defective paradigm is amended. A major device for gap-filling in the structure of languages is periphrasis, a mechanism that has both morphological and syntactic properties. This Janus-faced nature of periphrasis poses a problem for linguists working with any linguistic model when they try to formulate the criteria for labelling a certain construction as an instance of periphrasis. Attempts have been made to achieve a better understanding of periphrasis in the framework of various grammatical theories, as for example in a lexical grammatical approach (Kiparsky 2005), or in canonical typology (Brown et al. 2012), to mention only a couple. Growing interest in periphrasis has motivated the publication of in-depth studies of this phenomenon: a recent collection of studies (Chumakina & Corbett 2012) focuses on typological variation of periphrastic constructions observable in languages of diverse language families, while John Anderson (2011), in his trilogy on the substance of language, devoted the second half of volume 2 to the discussion of periphrasis. Analytic constructions, thus periphrasis, are known to have gained increasing significance in the history of the English language. With the decline of inflectional endings in the Late Old English period it is not an unexpected development that the subjunctive, which was marked by

2

Introduction

suffixes, gradually lost its importance against periphrastic forms conveying modal distinctions: the optative subjunctive was replaced by mayperiphrasis, while the hortative subjunctive by the let-construction. Replacement of the hortative seems to have spread faster, so much so that by the end of the 16th century it became more frequent than the subjunctive but optative periphrasis did not develop that fast (see Rissanen 1999: 229). As opposed to mood, aspect in Old English was expressed with the help of prefixes (e.g. Old English perfective ge- or be-), so the spread of periphrastic replacement – as, for example, in the case of the perfect – is connected not only with the formal and functional weakening of prefixes but also with the gradual intrusion of the once exclusively possessive have into the domain of the perfect aspect. This typologically widespread expansion and grammaticalization of have may have to do with Scandinavian influence (see McWhorter 2002: 236f). The phenomenon of linguistic replacement is closely connected to periphrasis; in the examples cited above, a periphrastic structure eventually replaced a synthetic one. One of the most exciting tasks of historical linguists is to discover and uncover the possible reasons and causes behind the development of periphrastic constructions and the replacement of older linguistic forms by these. As some of the papers in this volume show, replacement is always preceded by synchronic variation, and for the study of this phenomenon the use of digitized corpora provides great help. This relatively recent aid for historical linguistic research has been exploited by most authors in the present volume. The replacement of an older linguistic item (be it a morpheme, a lexical element or a more complex structure) will often affect some of the sub-systems, or sometimes even the whole linguistic system of a language, thus it may lead to the renewal and restructuring of some of these (to mention just one well-known example, the gradual loss of final inflectional endings in Middle English triggered off significant changes in syntax). One can thus view the phenomena of periphrasis, replacement and renewal as constantly recurring processes in the cycle of language change.

Patterns and Models of Replacement Periphrasis is a phenomenon often neglected or incorrectly treated by handbooks and textbooks. In this volume John Anderson points out unfortunate aspects in the terminological tradition and significantly refines the theoretical distinction between the notions of ‘grammatical periphrasis’ and ‘lexico-grammatical periphrasis’, which is equally relevant for general

Irén HegedĦs & Dóra PĘdör

3

linguistics and for historical linguistic considerations. Taking the expression of the Latin perfect passive as a prototype for the central notion of the term ‘grammatical periphrasis’, the author examines several potential cases of grammatical periphrasis in English – such as the progressive, perfect, and passive constructions, as well as expressions with used (to). He finds that modal constructions that interact with the morphological subjunctives are closer to the prototype than other instances of modals, which turn out not to qualify as instances of periphrasis. The loss and replacement of the Old English preterite-present verb þurfan – an unsolved mystery in the history of English – is investigated by Lucía Loureiro-Porto. In an earlier work (Loureiro-Porto 2010), she already discussed the competition between verbo-nominal constructions with the nouns þearf and need, and rejected the earlier assumptions ascribing the loss of þurfan to phonological confusion with durren in Middle English (see Visser 1963í1973, Molencki 2005) because the noun neod had replaced the noun þearf more than two centuries earlier. Loureiro-Porto’s study in the present volume approaches the question why neod ousted þearf from a psycholinguistic perspective and evaluates the role of morphological productivity in a quantitative analysis. The author concludes that the Old English form neod had a high resting activation, and the circumstance that it was easily retrievable from the mental lexicon in the process of speech production secured its productivity, which led to the substitution of þearf by neod in the contexts where originally they used to be in free variation. Aspectual loss and renewal in the history of some ingressive markers is investigated by Lynn Sims, who analyses how the increasing degree of grammaticalization of Old English onginnan contributes to its decline and replacement by Old English beginnan. In the Middle English period the occurrence of {begin + bare infinitive} gradually disappeared, and a new -ing complement appeared. But then the distribution pattern of the {begin + -ing} structure was challenged and – by the beginning of the 20th century – suppressed by a new competitor: {start + -ing}. The introduction of the new ingressive marker start in early Middle English may have been facilitated by some semantic restrictions on the use of begin. The evolution of ingressive markers and their complements exemplify how history sometimes repeats itself because the cyclic mechanism of loss and renewal by replacement tends to recur. Justyna Rogos investigates the intricate network of “multivocal relationships in the mapping of sound to symbol” (Laing & Lass 2009: 2) in seven manuscripts of Geoffrey Chaucer’s The Man of Law’s Tale. Based on the views of Laing & Lass 2003, her paper makes use of the

4

Introduction

medieval doctrine of the littera, which conceives of the letter as a unit of both writing and speech, in order to show the multivalent relation between spelling and sound in the manuscripts examined. She sets out several Litteral Substitution Sets (LSS) – that is, when a number of litterae can represent in variation the same potestas (sound value) – used in the manuscripts. She comes to the conclusion that the LSS attested in the manuscripts under examination indicate that the scribes in question worked within the framework of conventionalized spelling, which admitted little variability. She also argues that an examination of Potestatic Substitution Sets (PSS) is necessary for a thorough analysis of the spelling systems in the 15th-century manuscripts of the Canterbury Tales.

Syntactic and Semantic Variation and Change The loss and replacement of some causative verbs is investigated by Brian Lowrey, who looks at syntactic and semantic variation in Middle English causatives and complement types. He argues that up to now too little attention has been paid to their variation, and that is why the explanations for the distribution of Middle English complement types in causative structures have only been partially successful. He examines five different complement structures in his corpus within the framework of a dynamic model of variation and change proposed by Smith (1996). On the basis of his findings concerning the distribution of causers/causees by thematic role in each of the texts examined, the author argues that the agentivity of the causer and that of the causee are very important factors to be considered, as they affect the variational spaces (i.e. all the contexts in which a given linguistic item may appear) of both causative verbs and complement types. He also discusses the case of make, do, gar, let and cause in detail and demonstrates that the shift in their respective variational spaces in Middle English was heavily influenced by the factor of agentivity. Syntactic variation is examined by Anna Cichosz from another vantage point. In a comparative study of Old English and Old High German translations of Latin texts she investigates how far the Latin source may have syntactically influenced these translations, especially with respect to the relative position of the finite verb and its subject in main and subordinate clauses. Her results show that the verb-first pattern of Latin is closely followed in the Old High German translation, while in the Old English translation the verb-first structure rarely occurs, and even if it does, the original Latin word order is slightly modified. When the subject is expressed in the Latin text, Old High German faithfully reflects this

Irén HegedĦs & Dóra PĘdör

5

order but the Old English translation consistently moves the verb from initial position (V-1) to V-2, V-3 or even to final position. So Old English seems to have been less dependent on the syntactic structure of the source, while Old High German closely follows the pattern of the Latin original. When word order in the translations is the same as that in the Latin text, it is difficult to decide whether the structure used in the target languages can be considered a Latin syntactic calque or a native structure, i.e. the structural correspondence may only be coincidental. The most recent trend in generative grammar, the Minimalist Program is used as a theoretical framework in Víctor Parra-Guinaldo’s study, which examines the grammaticalization of Old English whether. In the first part of his article the author sets out the basic principles of the Minimalist Program, laying emphasis on the concept of economy, and arguing for the relevance of van Gelderen’s Complementizer Phrase Cycle model based on her two economy principles – Head Preference and Late Merge (Gelderen 2004: 18, 28) – for examining and interpreting the grammaticalization of Old English whether. The article discusses the different functions of Old English whether (including, among others, whether as a declinable and an indeclinable pronoun, and as a conjunction in various contexts) based on Ukaji 1997, and provides illustrative examples for each function. The author argues that the descriptive work done by Ukaji on whether should be revised applying the Minimalist Program and van Gelderen’s economy principles. The Minimalist Program, however, cannot provide an explanation for all aspects of the history of English, as case in Present-Day English is considered uninterpretable in the Minimalist framework. In the present volume Fuyo Osawa proposes a different viewpoint of case interpretability on the basis of investigating transitivation in earlier periods of the history of English. She shows that uninterpretability in English emerged as a result of the separation of case from thematic roles, thus new syntactic constructions could emerge and ‘deviant’ transitive constructions (i.e. constructions that have few transitive properties) could become productive. Transitivity in Old English was easily expressed by internal vowel contrast or by prefixation, since a form–meaning correspondence obtained to some extent in Old English. But later, due to phonological coalescence and morphological simplification, the situation changed, and this had an effect on transitivity as well. The author purports that a shift from a lexical-thematic stage to a functional one took place in the history of English and she proposes a new case hierarchy in which the instrumental case is ordered below the genitive, while the genitive is

6

Introduction

below the dative (as opposed to earlier interpretations expressed, for example, in Calabrese 1998 or Blake 2001). Octav Eugen DeLazero uses the theoretical framework of formal semantics – among others, the views put forward by Zucchi (1993), Larson (1998) and Gentner (2005) – to make several valuable observations in connection with the development of modal adjectives (e.g. likely, probable, possible, virtual, potential) in the history of English. The author points out that all of these adjectives are borrowings in English, almost exclusively from French or Latin. These adjectives in English have certain selectional restrictions that were inherited from the languages from which they were borrowed: namely that originally they could only stand with nouns which described situations, and they started to be used with nouns denoting physical entities only in the 18th–19th centuries. The reader is introduced to the concept of modal adjectives and some diachronic data are provided about them; furthermore, a cross-linguistic analysis of this class is given with examples from Ancient Greek, Latin, Old Icelandic, Old Church Slavonic, Gothic and Sanskrit. DeLazero also gives an outline of the status of these adjectives in Present-Day English, and then proceeds to analyze the denotations of the modal adjectives and those of the nouns they can combine with. The author’s conclusion is that ‘proper’ modal adjectives are only those that can combine both with situational and relational nouns, and he also argues that languages that do not allow such combinations (of which English was one up to the end of the 18th century) are not able to express modality at determiner phrase level.

Adverbials and Particles The methods of corpus linguistics have facilitated research into the smaller building blocks of language like adverbials, particles and conjunctions. Thus the changes that occurred in these areas can be traced fairly accurately, and detailed analyses can be provided about the phenomena connected to these developments. Artur Bartnik examines the development and distribution of adverbial before in Old and Middle English. In order to provide a clearer picture, he also carries out a parallel investigation of Old English ær, Middle English er (‘before’), and demonstrates that there are a number of differences in the use of before and ær. One of these is a quantitative difference, as their number of occurrences in the examined corpora shows considerable variation. These differences can also be semantic, as these adverbials do not quite carry the same meanings (while before can have both temporal and locative meaning, aer carries only temporal meaning). Furthermore,

Irén HegedĦs & Dóra PĘdör

7

the use of before and ær can also be grammatically different, as they do not always occur with the same type of modification (see also Molencki 2007). Apart from discussing these differences, Bartnik also shows what changes occurred in the use of these two adverbials with respect to the differences mentioned above during the Old and Middle English periods; moreover, he also gives a comprehensive list of all the possible Middle English variant forms of before. The author convincingly illustrates all of his points with examples from his corpora. The loss of the adverbial status of the no sooner … than construction is examined by Daisuke Suzuki. Although in Present-Day English this construction correlates with than, earlier it used to combine with but as well. Following a period of competitive coexistence between the two correlatives, the use of than became established. On the basis of corpus evidence the author shows that the no sooner construction gradually lost its status as an adverb in a process of grammaticalization, and became a conjunction. Suzuki also investigates other factors connected to the development of this construction. One of these is the frequency of the occurrence of inversion and the position of no sooner in the construction. Another factor is the use of different tenses in the no sooner clause. The third factor under scrutiny is the co-occurrence of no sooner with synonymous expressions like immediately, instantly, at once, presently. The author also examines the semantic change which led to the presentday meaning of no sooner, and he argues that this change was heavily influenced by the co-existence of but and than. Furthermore, he also points out that it is justified to use the framework of grammaticalization when analyzing the development of the no sooner construction. English phrasal verbs or verb-particle combinations have always received considerable attention from scholars. They are generally grouped into five semantic categories: literal, reiterative, aspectual/aktionsart, figurative, and non-compositional. A relatively large number of phrasal verbs or verb-particle combinations, however, cannot be easily fitted into these categories. Paula Rodríguez-Puente, in a study focusing on the combinations of verbs with the particle up, offers to refine this categorization by distinguishing a sixth group, that of emphatic phrasal verbs, which are likely to have emerged in the Middle English period. The corpus evidence she examined allows for a tentative explanation: Latinate verbs started to be combined with the particle up by way of analogy with native verbs in reiterative combinations. This process, probably also fuelled by the increasing analyticity in the history of English, must have facilitated the naturalization of the foreign elements.

8

Introduction

Functional and Regional Variation in Discourse and Vocabulary Discussing discourse strategies in early texts has enjoyed a strengthening popularity over the past two decades. In her study of Early Modern English travelogues, Virtanen (1995) introduced three discourse strategies, namely temporal, locative, and participant-/topic-oriented, and since then these strategies have served as bases for subsequent work dealing with historical discourse analysis. The use of these three discourse strategies in Margery Kempe and in Julian of Norwich’s Revelations of Divine Love were compared by Fumiko Yoshikawa (2008). Her paper in this volume extends the earlier investigation and offers a comparative study of the discourse strategies used in The Book of Margery Kempe (the chapters describing Margery Kempe’s pilgrimage) with the discourse strategies applied in Mandeville’s Travels and The Stacions of Rome. The comparative analysis focuses on examining adverbials of time and place in sentence-/clause-initial position, as well as references to the author or to participants in the above-mentioned texts. The analysis reveals that discourse strategies used in travelogues differ from those in pilgrimage stories inasmuch as the latter tend to combine locative and participantoriented strategies. Illuminating the differences between Mandeville’s Travels and Margery Kempe, Yoshikawa argues that while the former can be considered descriptive, the latter is rather narrative. This can be concluded from Yoshikawa’s observation that the way coherence is maintained in the two texts is not alike: in Margery Kempe coherence is achieved by temporal discourse strategy, while in Mandeville’s Travels by locative discourse strategy. The author also calls attention to the use of participant-oriented strategy, which is consistently used in Margery Kempe but not in Mandeville’s Travels. Earlier discourse models, however, are found inadequate for explaining a discrepancy revealed in a corpus study carried out by Lilo Moessner, in which she used a corpus of ca. 90,000 words from the fields of medicine and natural science from the period ranging between the middle of the 17th and the middle of the 18th century. The author applied a modified version of multidimensional analysis for highlighting characteristic register properties of these early texts, and she found that – contrary to earlier long-term discourse studies – the registers of medicine and science began to diverge with respect to the properties informativeness, narrative concerns, overt expression of persuasion, and degree of abstractness in the hundred-year period examined by her. To explain the discrepancy between earlier studies and her findings, she proposes a new and more powerful

Irén HegedĦs & Dóra PĘdör

9

discourse model that integrates extralinguistic aspects (such as format, purpose, audience, etc.) in the linguistic investigation. A so far undeservedly neglected source for the study of regional vocabulary in 17th-century England is analyzed by Javier Ruano-García. He surveys the unpublished manuscript of Bishop White Kennett’s Etymological Collections of English Words and Provincial Expressions, which was compiled most likely in the late 1690s and the early 18th century. Ruano-García provides an extensive lexical survey of the unique material contained in this manuscript: he discusses the Northern words, the Midland words and the Southern words, and emphasizes the importance of the fact the Kennett very often provides a precise location for the lexeme listed: thus we get information about the vocabulary used in 38 counties in England, and some further information concerning the Isle of Man, Wales, Ireland and Scotland. Although information provided about the Northern counties of England is the most extensive, still, in comparison with other contemporary lexicographical works, one of the greatest merits of Kennett’s work is that it documents the lexical history of the Midland and southern counties, and thus provides valuable information for the study of hitherto unknown aspects of historical regional variation. No matter whether the processes of periphrasis, replacement or renewal are at work, the linguistic changes analyzed by the studies collected in this volume testify to the intricate relationship between the cognitive needs of the speakers and their concrete representations at the level of language, as well as to the immense human creativity in the changes affecting the linguistic system. Several papers combine various aspects of theoretical linguistics with historical linguistic research, and they testify to the beneficial results of ‘cross-fertilization’ between these two areas. Thus the papers in this volume also show that there are still important lessons to be learnt both for and from the history of English.

Works Cited Anderson, John M. 2011. Morphology, Paradigms, and Periphrases. (The Substance of Language. Volume II.) Oxford: Oxford University Press. Blake, Barry, J. 2001. Case. (2nd ed.) Cambridge: Cambridge University Press. Brown, Dunstan, Marina Chumakina, Greville Corbett, Gergana Popova & Andrew Spencer. 2012. “Defining ‘Periphrasis’: Key Notions”. Morphology 22:2.233-275. Calabrese, Andrea. 1998. “Some Remarks on the Latin Case System and Its Development in Romance”. Theoretical Analyses on Romance

10

Introduction

Languages ed. by José Lema & Esthela Treviño, 71-126. Amsterdam & Philadelphia: John Benjamins. Chumakina, Marina & Greville G. Corbett (eds.) 2012. Periphrasis. The Role of Syntax and Morphology in Paradigms. (Proceedings of the British Academy, 180.) Oxford: Oxford University Press & British Academy. Gelderen, Elly van. 2004. Grammaticalization as Economy. Amsterdam: John Benjamins. Gentner, Dedre. 2005. “The Development of Relational Category Knowledge”. Building Object Categories in Developmental Time ed. by Lisa Gershkoff-Stow & David H. Rakison, 245-275. Hillsdale, NJ: Lawrence Earlbaum. Kiparsky, Paul. 2005. “Blocking and Periphrasis in Inflectional Paradigms”. Yearbook of Morphology 2004 ed. by Geert Booij & Jaap van Marle, 113-135. Dordrecht: Springer. Laing, Margaret & Roger Lass. 2003. “Tales of the 1001 Nists: the Phonological Implications of Litteral Substitution Sets in Some Thirteenth-Century South-West Midland Texts”. English Language and Linguistics 7.257-278. —. 2009. “Shape-Shifting, Sound Change and the Genesis of Prodigal Writing Systems”. English Language and Linguistics 13.1-31. Larson, Richard. 1998. “Events and Modification in Nominals”. Proceedings from Semantics and Linguistic Theory (SALT) VIII ed. by Devon Strolovitch & Aaron Lawson, 1-27. Ithaca, NY: Cornell University Press. Loureiro-Porto, Lucía. 2010. “Verbo-nominal Constructions of Necessity with þearf n. and need n.: Competition and Grammaticalization from OE to eModE”. English Language and Linguistics 14:3.373-397. McWhorter, John H. 2002. “What happened to English?” Diachronica 19: 217-272. Molencki, Rafaá. 2005. “The confusion between thurven and durren in Middle English”. Rethinking Middle English: Linguistic and Literary Approaches ed. by Nikolas Ritt & Herbert Schendl, 147-160. Frankfurt am Main: Peter Lang. —. 2007. “On the Rise of the Temporal Preposition/Conjunction Before”. To Make his Englissh Sweete upon his Tonge. Medieval English Mirror (3rd ed.) by Marcin Krygier & Liliana Sikorska, 37-54. Frankfurt am Main: Peter Lang. Rissanen, Matti. 1999. “Syntax”. The Cambridge History of the English Language, vol. III: 1476–1776 ed. by Roger Lass, 187-331. Cambridge: Cambridge University Press.

Irén HegedĦs & Dóra PĘdör

11

Smith, Jeremy J. 1996. An Historical Study of English: Function, Form and Change. London & New York: Routledge. Ukaji, Masatomo. 1997. “A History of Whether”. Studies in English Linguistics: A Festschrift for Akira Ota on the Occasion of His Eightieth Birthday ed. by Masatomo Ukaji, Toshio Nakao, Masaru Kajita & Shuji Chiba, 1236-1261. Tokyo: Taishukan. Virtanen, Tuija. 1995. “‫ލ‬Then I Saw to Antique Heddes’: Discourse Strategies in Early Modern English Travelogues”. Historical Pragmatics: Pragmatic Developments in the History of English ed. by Andreas H. Jucker, 499-513. Amsterdam & Philadelphia: John Benjamins. Visser, Fredericus T. 1963–1973. An Historical Syntax of the English Language (4 vols.). Leiden: Brill. Yoshikawa, Fumiko. 2008. “Discourse Strategies in Late Middle English Women’s Mystical Writing”. Historical Englishes in Varieties of Texts and Contexts (= Studies in English Medieval Language and Literature, 22.) ed. by Masachiyo Amano, Michiko Ogura & Masayuki Ohkado, 233-244. Frankfurt am Main: Peter Lang. Zucchi, Alessandro. 1993. “Incomplete Events, Intensionality, and Imperfective Aspect”. Natural Language Semantics 7.179-215.

PART I PATTERNS AND MODELS OF REPLACEMENT

WHAT ARE ‘GRAMMATICAL PERIPHRASES’? JOHN ANDERSON

For some little time I’ve been interested in the development of what I’ve thought of to myself as ‘subjunctive periphrases’ (see e.g. Anderson 2004). What I have in mind are constructions such as those in the second clauses of each of (1) and (2): (1) If it rained, I would be happy (2) I wish that it would rain Historically, the two-word verbal construction involving would in the apodosis of (1) has replaced the synthetic subjunctive that is retained in the protasis, in the form of rained. And the would-rain-construction in the subordinate clause in (2) is apparently in contrast with the synthetic subjunctive rained that we find in (3): (3) I wish that it rained occasionally The expression in (2) normally involves future reference; (3) does not. My use here of the term ‘subjunctive periphrasis’, or more generally, ‘grammatical periphrasis’, appeals to a traditional usage in grammatical studies. A more recent exemplar of the Greek tradition is provided in Triantafyllidis et al. (1941/1991: §812, ȕ), for example. But it is also a term that is not uncontroversial in how precisely it is to be interpreted, and even as to its usefulness. Does it denote some consistent significant grammatical phenomenon distinct from, say, ‘analytic expression’? It is a term inherited in my case from one tradition that I was familiar with. The tradition is illustrated by Mustanoja’s (1960: 453) remark on developments in Middle English: “the subjunctive mood begins to be indicated periphrastically by means of modal auxiliaries”. Warner (1993: 171-172) comments that “the equivalence implied here is, however, misleading”. And he suggests (1993: 172) “rather, some of the particular functions discharged by the inflectional subjunctive begin to be discharged by modal verbs, without any simple replacement or equivalence”. Also, Yawamoto (2010), for instance, observes that any bleaching of the Old

John Anderson

15

English modals did not always lead to subjunctive periphrases. And in Modern English it seems clear that only some modal constructions might be described as ‘periphrastic’, and only in some of their bleached senses; and these senses often co-exist with use of the subjunctive for related senses. Nevertheless, the interaction in (1)-(3) between modal+verb and the subjunctive of the same verb invites inquiry into whether this interaction might legitimately be described as involving ‘grammatical periphrasis’. If it does, my usage in describing the phenomena discussed in Anderson (2004) as “the development of subjunctive periphrases” is appropriate. But are these constructions ‘grammatical periphrases’? In interpreting such statements as Mustanoja’s I had always thought that ‘periphrasis’ was not just a label for a diachronic development, but that it was synchronically relevant to our understanding of an important element in the relationship between morphology and syntax. This is why I want to spend some time here trying to establish what might be a reasonable characterization of ‘grammatical periphrasis’, given established usages, and with a view to trying to make more explicit this particular interaction between morphology and syntax. This is not merely a terminological question – if anything is ever ‘merely terminological’. It’s a question of clarifying what kind of link between morphology and syntax is implied by the use of the term ‘grammatical periphrasis’. Understanding what periphrasis involves is crucial to understanding how it develops. My impression is that, as historical grammarians, we too often fail to examine closely enough the concepts that are crucial to understanding the developments that we’re interested in. Here I’d at least like to try to clarify how my understanding relates to others. Then I can maybe start to get back to worrying about developments. Let’s see if we can get a sense of what people have meant by ‘grammatical periphrasis’. The ‘ordinary-language’ sense of periphrasis is “a roundabout way of speaking” or “a roundabout expression”, as the Macquarie Dictionary of 1981 has it. And that’s not too remote from the etymology. But that dictionary also has a definition of “periphrastic” as used in grammar: “denoting a construction of two or more words with a class meaning which in other languages or in other forms of the same language is expressed by the inflectional modification of a single word”. This seems to reduce ‘grammatical periphrasis’ to the analytic term in an analytic vs. synthetic distinction between different language systems (if “other forms of the same language” is taken to denote ‘other varieties’). And this interlinguistic sense doesn’t seem to adequately reflect my impression of current usage among grammarians – and in particular my

16

What are ‘Grammatical Periphrases’?

own. As implied above, I take ‘grammatical periphrasis’ to involve a relationship with inflection that is true of a particular language system or sub-system. Let me spell this out before examining the usage of others. What is it, in such terms, that distinguishes a periphrasis as ‘grammatical’? In relation to individual language systems we can begin crudely to differentiate between ‘lexical’ and ‘grammatical periphrasis’ as in (4) vs. (5): (4) LEXICAL PERIPHRASIS: a sequence of words ‘equivalent to’ a single word or lexeme), as in take/have a bath vs. bathe (5) GRAMMATICAL PERIPHRASIS: a sequence of words ‘equivalent to’ a word form We have different interpretations of ‘equivalence’ here: in (4) the equivalence is a rough equivalence in meaning; in (5) we have an equivalence in function. Specifically, the sequence referred to in (5) has a paradigmatic function equivalent to that of an inflected form. And typically the sequence, or periphrasis, is extending the paradigm in some way. We can be more specific still, but before I plunge into that, I want to compare what I’m suggesting with the usage of grammars. A survey of familiar grammars and histories of English reveals a variety of usage, and much non-usage, concerning the term ‘grammatical periphrasis’. And in general, where the term is invoked at all, there is not much explicitness about its nature. I’ll come back to the kind of definition that does emerge from those works that attempt one. However, a rather different picture emerges, perhaps unsurprisingly, from typological studies devoted, at least in part, to phenomena related to some notion of ‘grammatical periphrasis’. ‘Periphrasis’ does seem to have been regarded as a useful distinctive term in talking about the relation between syntax and morphology. But again there is often an absence of explicit characterization of the term. For instance, several contributors to Harris & Ramat (1987) employ the term. But it seems from them that the reader is assumed to understand this term without definition, or even prompting. Thus, in his important paper in this 1987 volume, Dik makes the claim “Copula Auxiliarization always originates in periphrastic constructions with an aspectual meaning” (Dik 1987: 59) without making it clear how “periphrastic constructions” is to be understood. I come back to an exception in this volume to the neglect of clarification. This impression of the grammatical relevance of a concept, or at least a term, ‘periphrasis’, is confirmed in works that focus on the term – even more unsurprisingly. However, these also reveal the need to clarify usage.

John Anderson

17

For from the literature we don’t gain an impression of a homogeneous concept, but at best a chain of related concepts conforming to different ‘criteria’. This emerges very plainly from Haspelmath (2000), which surveys some of the general linguistic literature. What is involved in the works he considers seems to be more specific than simply ‘analytic construction’ (to the extent that that is simple), but it still remains uncertain exactly in what way, despite Haspelmath’s efforts. Haspelmath does identify three main usages concerning the term ‘grammatical periphrasis’. And among these he distinguishes two recurrent concepts that he suggests agree with Hockett’s (1958) definition. According to this latter, grammatical periphrasis “can be recognized only where there is a clear gap in the inflectional patterns, which the phrases serve to fill” (1958: 212). And this is not unlike what is suggested by a number of other grammarians. One of Haspelmath’s conceptions that corresponds to Hockett’s definition, what he calls “paradigm symmetry” (2000: §2), is also close to my own implicit idea of ‘grammatical periphrasis’. Let’s look at “paradigm symmetry”. The most commonly discussed and agreed example of the achievement of ‘grammatical periphrasis’ in the sense associated with its role in “paradigm symmetry” is the Latin perfect passive construction of (6). (6) Auditus est ‘S/he/it was heard’ (perfect passive) (7) Audiebatur ‘S/he/it was/used to be heard’ (imperfect passive) (8)

Audivit ‘S/he/it heard’ (perfect active)

(9)

Audiebat ‘S/he/it heard/used to hear’ (imperfect active)

(7) to (9) illustrate combinations within a single word form of a term from the tense-aspect category with a term from the voice category. The sequence in (6) signifying ‘perfect passive’ seems to be in contrast with the synthetic imperfect passive of (7) and the synthetic perfect active and imperfect active of (8) and (9). And the corresponding forms in the present are all synthetic. The periphrasis in (6) is functionally equivalent to an inflected form. It ‘fills a gap’ where certain combinations of terms of finite verbal categories fail to have a morphological exponent – here the combination of perfect and passive. Compare Vincent (1987: 251) on such periphrases: ... certain constructions may exhibit properties of linear order, surface separability and perhaps even deleteability which suggest that they are

18

What are ‘Grammatical Periphrases’? syntactic, while at the same time entering into contrastive opposition with elements more normally regarded as morphological.

The other conception that Haspelmath singles out as involving what he calls “periphrasis in the narrower sense” (2000: 656) is what he calls “inflectional generality” (2000: §3). Of this he says (ibid. 657): If a certain inflectional pattern is not applicable to some members of the word class, a periphrasis may fill this gap. An example of this type is the English periphrastic comparative ..., which allows adjectives that lack the bound comparative (*beautifuller) to have a comparative form (more beautiful).

This other conception of periphrasis seems to me a rather different kind of animal from what is illustrated in (6) to (9). I would prefer to think of it as involving what one might call ‘lexico-grammatical periphrasis’. It is like the “paradigm symmetry” illustrated by the Latin forms in ‘filling a gap’, but the gap is associated with particular lexical items, not with the general absence in the morphology of a particular combination of terms of categories. And the relationship between more and the comparative seems to be rather different from that between the function verb in (6) and the inflected forms in (7)-(9): not only does (6) contrast with these inflected forms, unlike the ‘comparative periphrasis’ and the comparative form, but more itself is a comparative form. I am going to pursue the former conception of periphrasis here since, on a personal basis, it seems best adapted to helping me to sort out what I might have meant by calling the modal constructions in (1) to (3) “subjunctive periphrases”. But it is also the sense that has most occupied work in linguistics that is specifically concerned with grammatical periphrasis, work in which the Latin example has played an important role. But before proceeding with this I must first acknowledge the other main usage of ‘periphrasis’ that Haspelmath identifies – though it doesn’t involve “periphrasis in the narrower sense” (Haspelmath 2000: 656). It is, however, relevant to evaluating to what extent the constructions in (1)-(2) that I’ve been concerned with are themselves “periphrases in the narrower sense”, and specifically examples of “paradigm symmetry”. Haspelmath calls this third type “categorial periphrasis”. And he comments (ibid. 660): Examples of categorial periphrasis are the English have-perfect ..., the French aller-future ..., and the Spanish estar-progressive ... There exists no monolectal form of any of these categories in the languages in which they occur, so these forms do not fill a gap defined by a system of monolectal forms, i.e. “real” inflectional forms.

John Anderson

19

Haspelmath concludes that these do not seem to be “‘circumlocutions’ for anything” (ibid). I suggest that we can say, however, that what he describes are analytic expressions for a category that is expressed in another language synthetically. Analysis and synthesis are comparative terms. And this resolves in principle another problem Haspelmath sees with this third kind of ‘periphrasis’. He raises the question of how we decide when such a potential periphrasis is expressing a grammatical meaning: what is a ‘grammatical meaning’? This is resolved comparatively. A multi-word expression may be said to be analytic if the meaning it expresses is elsewhere expressed synthetically. Sure, there can be problems in crossidentifying meanings. But often a cross-linguistic analytic vs. synthetic parallel is establishable. So what I’m suggesting here is that we do not have to do with periphrases at all in the last situation described by Haspelmath, but simply analytic expression. But I shall also suggest, later, that not all the expressions he cites in the above quotation necessarily conform, even in his own terms, to what he describes as “categorial periphrasis” (i.e. analytic forms as I interpret this notion), but that they are much closer to “periphrasis in the narrower sense”. It is Haspelmath’s last, least specific, of the three applications of the term ‘periphrasis’ that is pervasive in those works on the history of English that attempt to characterize and not just invoke – or, more commonly ignore – the term. Consider, for example, the definition offered by Brinton & Arnovick (2006: 499): “A construction employing function words in place of inflectional endings to express grammatical meaning”. This does not explicitly distinguish between constructions and inflections in the same language system and an analytic construction in one language or epoch and a synthetic one elsewhere. It seems to me that this terminological tradition – in so far as there is one – is unfortunate, particularly given that the analytic vs. synthetic distinction is wellestablished in designating the latter situation. And it is at odds with much general-linguistic work on ‘grammatical periphrasis’. All this is why I decided to concentrate here on the first kind of ‘grammatical periphrasis’ distinguished by Haspelmath, the one illustrated by (6). It’s not just that it seems most relevant to understanding the constructions in (1) and (2) that started me off. This conception also seems to me, as I’ve said, an attempt to capture an important relationship between syntax and morphology, involving the interaction of the two. It recognizes the paradigmaticity of certain syntactic constructions, in their filling a gap in what is otherwise realized as a set of systematically

20

What are ‘Grammatical Periphrases’?

contrasting inflections. And it offers the most articulated and detailed concept of periphrasis. I’m going to take the construction in (6) as a prototype, and its various properties (e.g. finiteness, functional head, paradigmaticity) as prototypical for a grammatical periphrasis. This reflects the centrality of this Latin construction in theoretical discussions of grammatical periphrasis. But this choice also embodies an anticipation that we might be able to range potential periphrases in English in terms of which and how many of the properties of (6) are missing in each case. It offers a helpful template in talking about the extent to which potential periphrases conform to the complex concept exemplified by (6), and a useful taxonomy might eventually emerge from such a procedure. As a basis for what is minimally necessary for a grammatical periphrasis, I propose that, at the very least, a verbal periphrasis, to qualify as such, should enhance the resources of the finite paradigm. But there may be intermediate cases differing from (6) in various ways – sometimes extensively – but nevertheless contributing to paradigmaticity. For the historical linguist, such comparisons and such a potential taxonomy should also eventually throw some light on how grammatical periphrases develop. Given my startingpoint, i.e. the analytic constructions in (1) and (2), I have decided to examine various potential ‘grammatical periphrases’ in English apart from these, to determine to what extent they each conform to this prototype, to what extent they merit the label ‘periphrastic’. What I report on are thus some preliminary results of research on periphrases in general, before finally taking up (1) and (2) for the same examination. Another somewhat fuller attempt at progress in this area is made in Anderson (2011: chapters 3-4), which also addresses some of the issues neglected here that are mentioned at the end. Firstly, let me list what I see as relevant properties of the prototype of specifically verbal periphrasis. This is presented in what follows: The Prototypical Verbal Periphrasis a) it consists of a finite function verb plus a non-finite lexical verb b) the function verb governs the lexical verb c) the construction enhances the paradigmatic resources of verbs, particularly the finite paradigm; specifically: i) the function verb requires its complement to express certain terms of morphological categories ii) the combination of terms in (i) is one missing from the potential maximal paradigm of the finite lexical verb d) the function verb is otherwise categorially empty.

John Anderson

21

In terms of (a) and (b), the function verb takes the lexical verb as complement. In the case of (6), the missing combination alluded to in (c(ii)) is that of perfect and passive. This combination is manifested by the form of the non-finite verb, the perfect participle, that, as in (c(i)), complements the function verb. This form occurs independently of the periphrasis in (10), with the ablative absolute construction: (10) Xerxe victo

‘Xerxes having been defeated’

In (10) the name and the participle are both in the ablative and the construction functions as equivalent to an adverbial clause or phrase. In (8) and (9) active combines with perfect and imperfect respectively, and in (7) passive combines with imperfect, but among finites there is no synthetic expression combining perfect and passive. The terms perfect and passive are associated with the non-finite participle in (6), not the finite function verb. The specifications in (c) are crucial in our characterization, as it is they that introduce paradigmaticity. As regards (d), in content the est in (6) is categorially minimal compared with other verbs. Concerning this Latin periphrasis Kiparsky (2005) argues that it is a “last resort”, appealed to on account of the gap in the paradigm. Synthetic expression is preferred to periphrastic on grounds of “economy”. He also suggests that this gap in the Latin finite conjugation is associated with the complexity of the Latin perfect. Specifically, the perfect can be interpreted as either a perfective past – contrasting with the imperfect – or a relative past, like the English perfect. But I won’t pursue this here, since I want to, finally, address our attention to English. The confronting of my own terminological usage becomes more urgent in the light of denials of periphrastic status to any English verbal construction. There are certainly no two-word verbal sequences in English that exactly conform to this prototype. But some come very close. And some resemble the constructions rejected in the above quotation, from Haspelmath, as periphrases “in the narrower sense”. Consider first of all the progressive in (11): (11) She was leaving Here we have a function verb, apparently empty of content except the requirement that it takes as a complement a non-finite verb form that signals progressiveness. Progressive seems to be a feature of this nonfinite form. The progressive form appears independently of the function verb in (11), as in (12):

What are ‘Grammatical Periphrases’?

22

(12) I saw her leaving (12) is in contrast with the non-progressive in (13): (13) I saw her leave We find a similar contrast with (11) vs. (14): (14) She left This is beginning to look rather like the Latin situation in (6)-(9). The contrast in (12) vs. (13) is of course not to be confused with what’s happening in a pair like (15) and (16): (15) Leaving the house is forbidden (16) To leave the house is forbidden The first word in (15) is a simple nominalization. Whatever the distinction between it and the to-infinitive in (16) might be, it does not involve progressiveness and its absence. The historical basis for the ambivalence of -ing is well-known, if not yet, in my impression, well understood. Example (11) seems to conform to all of the properties associated with the prototype – except in one respect: there is only one category involved. As in specification (a), it consists of a finite function verb plus a non-finite lexical verb, and, as in (b), the function verb governs the lexical verb. (11) conforms to (c(i)) in that the function verb takes a complement which bears certain terms of morphological categories – except that there is only one term involved. And the same is true of (c(ii)) which requires that the combination of terms in (c(i)) is the set missing from the potential maximal paradigm of the finite lexical verb. And, indeed, (11) satisfies (d): the function verb is otherwise categorially empty. The Latin paradigm in (6) to (9) involves more than one dimension: it involves tense-aspect and voice – and, indeed, the former can be split into tense and aspect (perfective vs. imperfective). The English non-finite paradigm in (12) vs. (13) is unidimensional and is expressed privatively: progressive vs. non-progressive. The same is true of (11) vs. (14). So that the introduction of the analytic form in (11) might be said to ‘create’ the finite paradigm rather than filling a gap in an existing one. Haspelmath seems to regard this as crucial in relegating similar expressions to merely “categorial periphrases”. I suggested indeed that in his description of them these were simply ‘analytic expressions’; in his terms they did not warrant the name ‘periphrasis’. But Haspelmath is wrong concerning some of the

John Anderson

23

expressions he cites when he claims, as quoted above, that “[t]here exists no monolectal form of any of these categories in the languages in which they occur”. This is not true of the English perfect that he cites, at least – just as it is not true of the progressive. They are closer to the prototype than being simply ‘analytic expressions’. Consider first the progressive we’ve just looked at. Just as in Latin, the contrast is present in the non-finite paradigm, as illustrated by (12) vs. (13). The progressive construction may not ‘fill a gap’ in the finite paradigm, but it can be said to extend the non-finite paradigm into the finite domain. The contrast already exists in the morphology of the language concerned. And this is also true of the perfect in English, despite what Haspelmath says. The difference from the prototype is a result of the unidimensionality of the paradigm. So that the progressive, for instance, is the closest that a morphologically impoverished language like English can come to a prototypical grammatical periphrasis. Let’s turn now to the perfect. The relative past non-finite form occurs in the attributive use of (17) and (18), independently of any periphrasis: (17) That is the (recently) escaped parrot (18) That is the (recently) deposed president These are non-finite forms of a verb that is respectively intransitive and transitive. The periphrasis in (19) again enables the non-finite form to participate in a finite paradigm: (19) The parrot has escaped The construction in (19) shows most of the same properties as the progressive. But this perfect periphrasis is a little further from the prototype than the progressive. It has content that prevents the construction in (20), while (21) is unexceptionable: (20) *The parrot is having escaped (21) The parrot has been escaping The head of the perfect construction has a ‘stative’ feature that prevents it complementing the be of the progressive. The non-finite form in (19) has the same kind of history as the Latin in (6). But the Latin form has remained multidimensional; it is perfect passive. Whereas these two properties are associated with different

24

What are ‘Grammatical Periphrases’?

periphrases in English – the perfect we have just looked at and the passive of (22): (22) The traitor was executed The passive in (22) is not perfect. Again the non-finite form can occur independently of the periphrasis, as in (23) and (24): (23) They saw the traitor executed (24) She’s having it refurbished And again this periphrasis conforms to the requirements of the prototype as much as the progressive does, even though the non-finite form in (22) to (24) conveys a difference in voice rather than in something temporal. The two properties of perfect and passive are combined in (18), however, where the non-finite form is not part of a periphrasis. This moves the perfect even further from the prototype. With transitive verbs a simple perfect non-finite form appears to occur only in the periphrasis. The common source of perfect and passive means that it is the function verb that signals whether the non-finite is to be interpreted as a perfect or a passive. Compare the progressive vs. passive contrast, where the distinction is signalled by the non-finite. Again, this moves the perfect, and also in this respect, the passive, further from the prototype: their categorization is expressed by the identity of the governing functional verb. As is well-known, these periphrases can occur only in the relative order in (25): (25) They have been being cherished We have seen that progressive cannot precede perfect, as in (20). Similarly neither of them can follow the passive, which in English requires a transitive verb, so neither (26) nor (27) is viable: (26) *She is been cherishing (27) *She is had cherished We might interpret (25) as having the status of what we might call an ‘analytic paradigm’. Each marked term – perfect, progressive, passive – is signalled morphologically, but by an inflection on a separate word form from the others. All of these word forms are non-finite. Again, this seems to be the best a morphologically impoverished language can do in terms of

John Anderson

25

manifesting a multidimensional paradigm – to spread it out over several forms. Denison (1993: Ch.15) documents the slow growth of this ‘analytic paradigm’. It’s again well known that what was particularly belated is the combination of progressive and passive. There was for a time only a neutralization of active and passive progressives. The neutralization involved expressions like (28): (28) X is building The identity of ‘X’ could often disambiguate the construction – if, say, we substitute ‘the house’ rather than ‘Bill’, giving ‘passive progressive’ vs. ‘simple progressive’. But the completion of the ‘analytic paradigm’ in (25) resolves the neutralization by removing a gap in the paradigm. I’m now going to look at a two-word construction in English which perhaps marks the outer limit of what sufficiently resembles the prototype to be called a periphrasis. This is exemplified in (29): (29) Fred used to smoke a cheroot Similar is (30) where interpreted in a similar way to (29), but I’ll concentrate here on (29): (30) Fred would smoke a cheroot For many speakers it meets the prototype requirements of (a) and (b) and is not far short of (d), though the function verb does seem to be inherently past and habitual – not quite minimal. But it does not have the crucial properties in (c), which embody the paradigmaticity crucial to identifying a grammatical periphrasis. The function verb doesn’t seem to be subcategorized for particular terms, merely for a lexical verb: it is the function verb that signals past and habitual. And no existing paradigm is extended in any obvious way by the construction. However, it might be said to extend the finite paradigm in a way, in that it disambiguates the synthetic past. (31) is ambiguous between punctual and habitual: (31) Fred smoked a cheroot Indeed, in isolation a punctual interpretation seems more likely. (29) insists on the one interpretation of (31): the governed verb is interpreted as habitual. But this is the closest the construction gets to involvement in a paradigm of the language. Such a disambiguating periphrasis shows at

26

What are ‘Grammatical Periphrases’?

most the minimum similarity to the prototype that warrants periphrastic status. And there are further doubts introduced as to its status by the observation that used is not a function verb for all speakers of English. If function verb status in English is associated with undergoing the ‘inversion’ in (32), for many speakers used is not one – as the familiarity of (33) shows: (32) Used Fred to smoke cheroots? (33) Did Fred use to smoke cheroots? The same conclusion follows from contemplating the pair with negativeattachment in (34) and (35): (34) Fred usedn’t to smoke cheroots (35) Fred didn’t use to smoke cheroots With used prototypical property (a) is absent in varieties of English. Moreover, used, the lexical verb of (33) and (35), and (for speakers who use the expressions in (33) and (35)) of (29) in that case, is only one of a set of lexical verbs that bring along a habitual interpretation of the verb they govern. The same is true of the examples in (36) and (37): (36) Fred tended to smoke a cheroot (37) Fred liked to smoke a cheroot Used is the most empty semantically, and it is the only one of these verbs that is necessarily past. But that is the limit of its grammaticalization. For a large number of speakers used does not resemble a periphrasis. Habitual would is more consistently functional. I’d like now to move back towards where my worrying about periphrasis started from – so-called “subjunctive periphrases”. But rather than the constructions shown in (1) and (2), I’m going to start with that in (38) and its relation to the synthetic subjunctive in (39): (38) They demanded (that) Fred should resign (39) They demanded (that) Fred resign The should construction also conforms to (a) and (b) in the specification of the prototype, and it shows the semantic bleaching associated with (d).

John Anderson

27

As concerns (d), compare the should of (37) with the deontic should of (40) and the epistemic should of (41): (40) Fred should resign (41) That should be OK The should of (39) and (40) is roughly paraphraseable by ought to in those varieties where both are current, as illustrated by (42) and (43): (42) Fred ought to resign (43) That ought to be OK That is not true of the should in (38): (44) ?They demanded (that) Fred ought to resign However (44) might be interpreted, it is not equivalent to the obvious interpretation of (38). In (38) should has been bleached. However, again the valency of the function verb doesn’t include terms in a morphological paradigm, as specified in prototypical property (c(i)). As a modal, the function verb itself introduces the non-factivity that is also associated with the synthetic subjunctive of (39). The potential periphrasis at least shares an element of sense with an inflection. The subordinate clauses in both of (38) and (39) are not propositional, but non-factive. The entire sentences in (38) and (39) express a proposition containing a directive for a certain situation to come about, to become a fact. That is one reason why I find (44) strange. The two subordinate expressions also share the status of being governed by a certain subclass of verbs. Compare with (38) and (39) the pair in (45) and (46): (45) They said (that) Fred should resign (46) *They said (that) Fred resign The should in (45) is not the same should as in (38). It is the equivalent of the ought to of (47): (47) They said (that) Fred ought to resign Likewise the synthetic subjunctive of (46) is anomalous. This lexicallygoverned status is something shared between the two expressions in (38)

28

What are ‘Grammatical Periphrases’?

and (39). To this extent the putative periphrasis has something in common with an inflected finite – though this is scarcely the relationship of the periphrastic prototype. What then is the relationship between (38) and (39)? For many speakers these are not simply equivalent. Often the two-word construction is described as ‘more tentative’. We can associate this with a property of all the modals based on the historical preterite subjunctive: might, could, would, should, ought – but not must, which has shifted allegiance, so that we get the pairings in (48): (48) may – might, can – could, will – would, must – should/ought The second member of each pair is less assertive than the first. And this is analogous to the perceived difference between (38) and (39). (38) is a more robust and constant alternative signal of non-factivity than the synthetic subjunctive, which is seriously defective in morphological expression. But it also retains something of the meaning of the full modal. The should of (38) is the only signal, if any, of non-factivity for those speakers who substitute for (39) simply (49), with no morphological difference from the indicative, except, in this case, for the (perfectly acceptable) apparent discrepancy in tense between the two verbs in the sentence: (49) They demanded (that) Fred resigns And even the latter is absent in the alternative in (50): (50) They demanded (that) Fred resigned The situation is obviously fluid. But it may be that for many speakers should approaches simple analytic status. However, although non-factivity is marked by the head of the twoword construction in (38) and not on the complement, the non-factivity is at least indicated morphologically on the simple verb of (39), for speakers who retain this form. The periphrasis shares this function with morphology in the same language system. And it shows the paradigmatic property of being subject to rection – and rection exercised by the same set of verbs as govern the synthetic expression. The construction interacts with paradigmaticity more than the habitual construction with used in (29). This potential periphrastic subjunctive is closer to the prototype. Let’s have a look at (2) in relation to (3): (2) I wish that it would rain

John Anderson

29

(3) I wish that it rained occasionally These are both not merely non-factive but counterfactual. The propositions ‘It will rain’ and ‘It rains occasionally’ are pre-supposed not to be true at the time of speaking. English has two subjunctive forms, that exemplified by resign in (39) and that manifested in rained in (3). Whatever their historical sources, the subjunctives now differ in being non-factive vs. counterfactual. (2) also expresses counterfactuality, but contrasts with (3) in being future rather than present. We have a sort of ‘paradigm’ if we add to (2) and (3) the past in (51): (51) I wish (that) it had rained Past in (51) is expressed by the inflection of the complement in the perfect periphrasis had rained, and counterfactual is expressed by the inflection of the finite verb have. Futurity is often contaminated with modality, and thus modal expressions are a common source, via bleaching, of expressions dedicated to future reference – or at least predictive status. In (2), (3) and (51) we have the counterfactual equivalent of the noncounterfactual set of (52) to (54): (52) I think (that) it will rain (53) I think (that) it rains occasionally (54) I think (that) it rained For the same reason of lack of modal contamination in the case of (54) and (53), (51) and (3), because of their tenses, are more strongly counterfactual than (2). Something that is untrue at the time of speaking cannot come true in the past or in the present. Whereas a wish for something that is at present untrue to become true in the future may be realized. We have in (2) a ‘weak’ counterfactual. Again the presence of the subjunctive, synthetic or analytic, is governed, as is common with inflectional systems. But what is governed is counterfactuality, not the individual members of the tense ‘paradigm’ represented by (2), (3) and (51). They are freely selected. Nevertheless, the would of (2) is not far from the prototypical periphrasis – though even (52) to (54) has been controversial as a paradigm, given the asymmetry between past and future. (2) still doesn’t conform to (c) in the specification of prototypicality; no non-finite morphology is involved in this ‘periphrasis’. Nor is it fully bleached. But there is interaction of an

30

What are ‘Grammatical Periphrases’?

appropriate type of syntactic construction with the synthetic paradigm, and some kind of extension of it. That takes us back to (1): (1) If it rained, I would be happy Again the synthetic subjunctive rained and the modal construction with would are counterfactual. But instead of contrasting in tense they are in complementary distribution. And the synthetic expression in the protasis is indifferent to the future vs. present distinction, so it may be either ‘strongly’ or ‘weakly’ counterfactual, as perhaps in (55) vs. (56): (55) If my aunt had wheels, she would be a bicycle (56) If it rained tomorrow, it would be good for the crops And the would in the apodosis may be predictive or deductive more than temporal. If we want to insist on futurity in the protasis, we can have recourse to another ‘periphrasis’ such as in (57): (57) If it were to/should rain, it would be a relief The function verbs in (57) seem to have the ‘disambiguating’ role of used in (29). Again there is a lack of conformity of conditional would to prototypicality conditions (c), but the periphrasis does enable counterfactuality to be expressed in the main clause – a capacity that the synthetic subjunctive has lost, as I observed initially. This development is nicely illustrated by the apodoses of the successive sentences in (58) from Thackeray’s The History of Henry Esmond, which also illustrates in both sentences the ‘inversion’ type of protasis in place of that marked by the initial presence of if: (58) Had she been a Whig, he had been one; had she followed Mr. Fox, [...] no doubt he would have abjured ruffles and a periwig It is not misleading to talk here of ‘grammatical periphrasis’, I think. These modal periphrases, despite their non-prototypicality, illustrate a productive source of analytic expressions of distinctions that are morphologically relevant in the language. Rather than using the non-finite paradigmatic resources of the language, however, they rely solely on the semantic bleaching of modal expressions. And I think they conform to the paradigm-enhancing sense of periphrasis I’m struggling towards, though

John Anderson

31

they lack characteristics of the prototype. Indeed, they are prevalent enough to be a challenge to the dominance of the non-finite paradigm as the prototypical source of periphrasis. In their case it is the presence of the functional verb that through bleaching interacts with the morphological paradigm. I’ve looked at something of the variety of grammatical periphrases in English, perhaps enough to encourage me to return to studying how they develop. As a group they stand apart from the rest of the non-idiomatic modal constructions in English, for instance, as exemplified in (59): (59) They must/may resign The latter conform to prototype properties (a) and (b) and to some extent (d), but they do not enhance the finite paradigm or interact with the morphology, let alone exhibit (c). They are not periphrastic in the sense we’ve been looking at. But I obviously need to fill out the picture of nonprototypical periphrases – no doubt comparatively. And at this point I’m left with a lot more questions. Some quite particular ones like: how periphrastic is ‘periphrastic do’? And more general ones like, rather crucially: why is almost all the literature on grammatical periphrasis concerned with verbs? Is this a “historical accident of scholarship”, as Haspelmath (2000) thinks, or does it tell us something about the relationship between periphrasis and the ‘parts of speech’? As already indicated, there is some discussion of these and other such questions in Anderson (2011). Overall, despite remaining uncertainties, I think I can return to contemplating the development of ‘subjunctive periphrases’, somewhat reassured, but also even more aware of the complexities that still have to be confronted.

Acknowledgments I am grateful to Fran Colman for her very helpful comments on and discussions of several versions of this paper, as well as to a number of other participants in the Pécs conference for their encouraging and thought-provoking comments after the presentation of the paper, and particularly after the session itself.

32

What are ‘Grammatical Periphrases’?

Works Cited Primary Sources The Macquarie Dictionary. 1981. Edited by Arthur Delbridge et al. St. Leonards, New South Wales: Macquarie Library.

Secondary Sources Anderson, John M. 2004. “Syntactic Categories and Syntactic Change: The Development of Subjunctive Periphrases in English”. New Trends in Historical Linguistics: An Atlantic View ed. by Isabel MoskowichSpiegel Fandino & Begoña Crespo García, 31-73. Universidade da Coruña. —. 2011. Morphology, Paradigms, and Periphrases. (The Substance of Language vol. II.) Oxford: Oxford University Press. Brinton, Laurel J. & Leslie K. Arnovick 2006. The English Language: A Linguistic History. Oxford: Oxford University Press. Denison, David. 1993. English Historical Syntax. London & New York: Longman. Dik, Simon C. 1987. “Copula Auxiliarization: How and Why?” Historical Development of Auxiliaries ed. by Martin Harris & Paolo Ramat, 237256. Berlin: Mouton de Gruyter. Harris, Martin & Paolo Ramat eds. 1987. Historical Development of Auxiliaries. Berlin: Mouton de Gruyter. Haspelmath, Martin. 2000. “Periphrasis”. Morphologie/Morphology ed. by G. E. Booij, Christian Lehmann & Joachim Mugdan, vol. I, 654-664. Berlin: de Gruyter. Hockett, Charles F. 1958. A Course in Modern Linguistics. New York: Macmillan. Kiparsky, Paul. 2005. “Blocking and Periphrasis in Inflectional Paradigms”. Yearbook of Morphology 2004 ed. by Geert Booij & Jaap van Marle, 113-135. Dordrecht: Springer. Mustanoja, Tauno. 1960. A Middle English Syntax, I: Parts of Speech (Mémoires de la Société Néophilologique de Helsinki 23.) Helsinki: Société Néophilologique. Triantafyllidis, Manolis et al. 1941/1991. Neoelliniki Grammatiki (tis Dimotikis). Athens: ȅ.Ǽ.Ȉ.Ǻ, 1941. (Reprinted, with corrections, Thessaloniki: Aristotle University, 1991.)

John Anderson

33

Vincent, Nigel. 1987. “The Interaction of Periphrasis and Inflection: Some Romance Examples”. Historical Development of Auxiliaries ed. by Martin Harris & Paolo Ramat, 237-256. Berlin: Mouton de Gruyter. Warner, Anthony. 1993. English Auxiliaries: Structure and History. Cambridge: Cambridge University Press. Yawamoto, Tomonori. 2010. “On the Semantic and Syntactic Development of Periphrastic Modal Verb + Infinitive Constructions in OE: Comparing the Versions of Gregory’s Dialogues, the OE Boethius, and Psalter Glosses”. Aspects of the History of English Language and Literature: Selected Papers Read at SHELL 2009, Hiroshima ed. by Osamu Imahayashi, Yoshiyuki Nakao & Michiko Ogura, 225-239. Frankfurt am Main: Peter Lang.

MORPHOLOGICAL PRODUCTIVITY AS A TRIGGER OF THE ÞURFAN / NEED REPLACEMENT LUCÍA LOUREIRO-PORTO

1. Introduction We know that OE preterite-present þurfan was a prominent pre-modal verb and that it might have developed into a Present-Day English (PDE) modal auxiliary, as was the case with OE *sculan > PDE shall, OE cunnan > PDE can, etc. However, it did not follow this evolution, being replaced in Middle English by need (see Loureiro-Porto 2009, among others). Several possible reasons have been suggested to explain this replacement, which remains a mystery. The most common of these is of a phonological nature (first posited by Visser 1963–1973: 1423, §1343, and supported by a number of writers, including Molencki 2005, and also tentatively by myself in earlier work, Loureiro-Porto 2009: 83). According to this hypothesis, ME thurven (the evolution of OE þurfan) disappeared because it was phonologically confused with durren (> PDE dare, another preterite-present verb). The partial morphophonological overlapping of these two verbs is indeed observed in ME texts, as in (1): (1) Ne dar he seche non oþer leche. not need.3SG he search no other doctor ‘He need not look for any other doctor’. c1390 Castle of Love (Horstmann, Minor P. Vernon MS) 733, as quoted by Visser (1963–1973: 1423, §1343) In (1) we observe that the third person singular form dar exhibits morphophonological features of durren (the initial consonant is a dental plosive /d/, and the final one is not a labio-dental fricative /f/, but simply the liquid /r/). However, it is not a word form of durren, but is instead a word form of the paradigm of thurven (the form we would expect in this case is tharf). Examples such as these support the hypothesis that thurven

Lucía Loureiro-Porto

35

actually disappeared from the language so as to avoid phonological confusion with durren, which did indeed survive to Present-Day English (PDE dare). Convincing as this hypothesis may seem, though, the phonological overlapping between thurven and durren is not expected to have taken place in all verbal forms (all persons in present and past), so avoidance of phonological confusion does not seem to be the only reason for the thurven / need replacement. In addition, the þurfan / need replacement, which is achieved by the end of the 14th century (see Loureiro-Porto 2009: 209), was preceded by the replacement of the corresponding nouns þearf and neod, as found in verbo-nominal constructions beon neod / habban neode and beon þearf / habban þearfe, since the construction beon neod replaced beon þearf more than two centuries before the verbal replacement, at the very beginning of the Middle English period (12th century) (see Loureiro-Porto 2010: 379). I have suggested elsewhere (Loureiro-Porto 2010: 395) that this nominal replacement contributed to the subsequent replacement of the verbs, triggering it or paving its way. However, a question remains: why does the noun neod replace þearf? The noun þearf was not phonologically confused with any other noun, so the phonological confusion alleged for the verbs cannot be considered the trigger for this nominal replacement. The semantic and syntactic possibilities of the constructions involving both nouns were similar (cf. Loureiro-Porto 2010), so neither phonology, nor semantics or syntax can account for this development. The present study, then, explores a new hypothesis based on morphology, the only linguistic level of analysis which has not yet been considered, and which may help to shed some new light on the reasons for such a strange replacement. The focus will be on morphological productivity, a phenomenon closely related to psycholinguistics. This is the case because when speakers select one of two available morphemes that share function and meaning, the selection depends on the degree of psycholinguistic activation of these morphemes: those with a high resting activation will be accessed more readily than morphemes with a low resting activation (cf. for example, Plag 2005). Several factors determine the degree of activation of morphemes in the mental lexicon, the most salient of which is frequency (cf. Baayen 1992, Hay 2001, Hay & Baayen 2002). The aim of the current study, therefore, will be to calculate the degree of productivity of the morphemes neod and þearf in Old English, taking into account their frequency, but also the frequency of words derived from them. This is necessary because, as we know, they feature not only in the isolated homonym nouns (see Loureiro-Porto 2010) and in the verbs neodian and þurfan (see Loureiro-Porto 2009), but also in many different complex

36

Morphological Productivity as a Trigger of the þurfan / need Replacement

words, as we will see. By using the productivity measures available in the literature (Baayen 1992, Hay 2001, Hay & Baayen 2002), I will explore the Dictionary of Old English Corpus (diPaolo et al. 2000), which contains all extant texts from this period, around 3 million words, and will analyse each of the complex words containing neod and þearf. This constitutes a new approach in the study of productivity. Whereas other studies focus mainly on the productivity of affixes (Hay 2001, Hay & Baayen 2002, Plag 2006: 544-546, Palmer 2009), here I will calculate the productivity of the two roots involved, neod and þearf, so as to determine which one of these two morphemes is more productive. It could be claimed that roots should not be given the same treatment as affixes, because affixes may compete if they stand in variation and exhibit the same meaning and function (e.g. the affixes -ity and -ness in words such as productivity and representativeness). However, as has been seen, neod and þearf also exhibit these features, because they stand in variation in verbo-nominal constructions (beon neod, beon þearf) as well as in several OE words (see nearoþearf ‘misery, trouble, affliction’ and nearoned ‘great/dire adversity’). In addition, the semantic and syntactic differences between them (even if significant in the OE verbs neodian and þurfan) are inexistent in the OE verbo-nominal constructions (cf. Loureiro-Porto 2009, 2010), and thus function or semantics cannot account for the survival of neod-words and the disappearance of þearf-words. The hypothesis is that the morpheme neod is more productive than the morpheme þearf in Old English and, therefore, its psycholinguistic activation in the mental lexicon acted as one of the forces pushing out its competitor in verbo-nominal constructions, when both convey similar meanings and have the same function (Loureiro-Porto 2010). With the aim of testing this hypothesis, the paper is structured as follows. Section 2 provides an account of morphological productivity and the factors that play a role in it, describes its relation with psycholinguistics, and provides ways of measuring the productivity of morphemes. Section 3 describes the methodology used to extract the data from the Dictionary of Old English Corpus (diPaolo et al. 2000), including problems encountered and how they were overcome. Section 4 presents the findings, based on different frequency analyses of types, tokens and hapax legomena. Finally, section 5 includes a discussion and draws some conclusions.

Lucía Loureiro-Porto

37

2. Theoretical Foundations 2.1. Productivity Productivity is the property of a morphological process to create new words on a systematic basis (cf. for instance, Plag 2005: 121). For example, the morpheme -ness is said to be very productive in English because it is used to create new words such as us-ness, over-the-top-ness, etc., as opposed to the morpheme -ity which has the same meaning (‘quality of’), but is very seldom used to coin new words (cf. Baayen 1992: 116). This is so because, as introduced above, -ness is more readily available for speakers, who decide to create new words by resorting to the morphemes with a higher resting activation. The factors that contribute to the productivity of a given morpheme are classified by Plag (2006: 547549) as follows: parsability, relative frequency and semantic and phonological transparency. Parsability refers to the possible decomposition of a complex word in the mental lexicon. In principle, complex words can be stored in two ways: a) as single entries (whole-word route, see Plag 2006: 547), or b) decomposed into morphemes (decomposition route, see Plag 2006: 547),1 as seen in Figure 1:

ENTRIES IN THE MENTAL LEXICON

Whole-word route drive driver teach teacher smoke smoker dive diver produce producer drink drinker eat eater etc.

Decomposition route drive teach smoke dive produce drink eat -er

Figure 1. Possible ways of storing complex words in the mental lexicon 1 The terms whole-word route and decomposition route are borrowed by Plag from McQueen & Cutler 1998.

38

Morphological Productivity as a Trigger of the þurfan / need Replacement

The theory postulates that storing morphemes in the mental lexicon is more economical as far as storage is concerned, but is less economical as far as the understanding (or activation, in psycholinguistic terms) of a given word. Thus, more frequent words are more easily activated by speakers, and they are therefore said to have a higher resting activation. Less frequent words have a lower resting activation. In Plag’s (2003) words: How does frequency come in here? As mentioned above, there is a strong tendency that more frequent words are more easily stored and accessed than less frequent words. Psycholinguists have created the metaphor of resting activation to account for this phenomenon (and other phenomena). The idea is that words are sitting in the lexicon, waiting to be called up or ‘activated’ when the speaker wants to use them in speech production or perception. If such a word is retrieved at relatively short intervals, it is thought that its activation never completely drops down to zero in between. The remaining activation is called ‘resting activation,’ and this becomes higher the more often the word is retrieved. Thus, in psycholinguistic experiments it can be observed that more frequent words are more easily activated by speakers, such words are therefore said to have a higher resting activation. Less frequent words have a lower resting activation. (Plag 2003: 49, emphasis original).

That is to say, whenever a low frequency word is heard, speakers decompose it into morphemes and store each of them in their mental lexicon. If another low-frequency word sharing one of the morphemes of the previous one is heard, the representation of the morpheme is strengthened in the lexicon, and this makes it readily available to coin a new word, which is likely to be a low-frequency word, and hence the cycle continues, as in Figure 2: low-frequency complex neologism

decomposition in mental lexicon coinage of new neologisms strengthening of the representation of the affix

the affix is readily available for use with other bases

Figure 2. The cycle of low frequency and productivity

Lucía Loureiro-Porto

39

Given this theoretical framework, the least frequent words of a language contain the most productive morphemes, while the most frequent words will contain unproductive morphemes. In Plag’s (2003: 54) words: unproductive morphological categories will be characterized by a preponderance of words with rather high frequencies and by a small number with low frequencies. With regard to productive processes, we expect the opposite, namely large numbers of low-frequency words and small number of high-frequency words.

Thus, frequency plays a very important role in parsability. In this respect, it must also be noted that the relative frequency of a derived word regarding the frequency of its base has also been considered of vital importance (cf. Hay 2001, Hay & Baayen 2002), and constitutes the second factor in the productivity of a morpheme. According to Hay (2001) and others (Hay & Baayen 2002, Palmer 2009) a complex word is more likely to prefer the decomposition route if it is less frequent than its base. The relative frequency of a derived word is calculated according to the following formula: Frequency of the derivative (2) Relative frequency = Frequency of the base For example, Hay (2001: 1048) reports that the derived word exactly is more frequent than its base exact (2,535 to 532 occurrences in her database), while the derived word arrogantly is less frequent than its base arrogant (17 to 116 occurrences). Exactly, then, has a higher relative frequency than arrogantly and, therefore, it is more likely to be stored along the whole-word route; that is, it will not be parsable. The degree of parsability of arrogantly, on the other hand, is very high since its relative frequency is lower. According to Hay, for a word to be parsable not only must its own frequency be taken into account, but also the frequency of its components: “A low-frequency form is likely to be nontransparent if it is composed of even-lower-frequency parts” (2001: 1066). A final factor in the productivity of the morphemes found in a complex word is the semantic and phonological transparency of the morphemes involved. If a given complex word is semantically transparent, it will most likely be stored through the decomposition route, while the chances of parsing a semantically opaque word are minimal (cf. Hay 2001: 10551059; Plag 2006: 548). For example, the meaning of dishorn is maximally transparent (‘to deprive of horns’), while dislocate does not exhibit the

40

Morphological Productivity as a Trigger of the þurfan / need Replacement

same degree of semantic transparency: it exhibits different meanings (‘to displace’, of a bone: ‘to remove from its normal connections’, ‘to put out of joint’), and its base (locate) is not included in its definition (cf. Hay 2001: 1056). Thus, semantic opacity is in line with high relative frequency: both lead to the whole-word route. As for phonological transparency, Hay & Baayen (2003) study the phonotactics of morpheme boundaries in 80 PDE affixes, and they also find a correlation between transparency and decompositionability. For example, the phonotactics of the morpheme -hood makes it very likely to be decomposed in words such as childhood, womanhood, since the phoneme combination /dh/ or /nh/ (resulting from junctural combination of the two morphemes involved in each word) is extremely rare in monomorphemic English words. By contrast, the morpheme -ism, as in socialism or communism, is phonologically opaque, because the phoneme combination /lܼ/, /nܼ/ is highly frequent in monomorphemic English words (e.g. parsley, funny). This implies that “[w]ords with monomorpheme-like phonotactics are therefore much more likely to overtake their bases in frequency than words with low probability or illegal junctural phonotactics” (Hay & Baayen 2003: 121). That is to say, words featuring illegal junctures are likely to be decomposed, which confers on their morphemes a higher resting activation and, therefore, a probable higher degree of productivity. Summing up, the more semantically and morphologically transparent a given morpheme is, and the lower the relative frequency of the word in which it features, the more likely it is that the word is parsable and, therefore, the more readily available the morpheme is to create a new word; this in turn implies a higher degree of productivity. Figure 3 summarizes this idea.

High semantic transparency

+ High phonological transparency

Æ Parsability Æ Productivity

+ Low relative frequency

Figure 3. Factors influencing the productivity of a given morpheme

Lucía Loureiro-Porto

41

2.2. Measuring Productivity In the previous section we have seen the factors that may contribute to the productivity of a given morpheme. Although it might appear that productivity is an all-or-none property of a given morpheme, it is, on the contrary, a gradual phenomenon. That is, morphemes may be more or less productive. In order to determine how productive a given morpheme is, a reliable way of measuring productivity is needed. In this section we will see how different degrees of productivity may be measured. In morphological theory, different ways of measuring productivity have been proposed (see Haspelmath 2002: 109-110). These include: (i) the number of actual words formed according to a pattern, (ii) the number of possible words that can be formed according to a pattern, (iii) the ratio of actual words to possible words, (iv) the number of neologisms attested over a certain period of time. The number of actual words formed according to a pattern does not really reveal how frequently a given morpheme is used to create new words, but, rather, how frequent the morpheme was used in the past. For example, the morpheme -th is found in a series of words in English (width, strength, length, warmth, etc.) but it is no longer used to form new words. In other words, it is no longer productive even though it was used to form quite a number of English words in the past (cf. Plag 2006: 541). Therefore, studying the number of actual words does not necessarily lead to a full understanding of the productivity of a morpheme. In fact, it is the fourth measure listed above, the number of neologisms attested over a certain period of time (see Carstairs-McCarthy 2002 and Plag 2003), that is used most often these days as a quantitative measure of productivity. To measure the productivity of a given morpheme by looking at the number of neologisms formed according to a pattern, a representative corpus is needed, because neologisms tend to exhibit a relatively low frequency and are not very likely to feature in small corpora. The BNC (British National Corpus), used by Plag, Dalton-Puffer & Baayen (1999), is highly representative because it covers different text types and registers and because it contains 100 million words (tokens) representing around 940,000 different words (types). Along with observations based on tokens and types, studies of productivity in corpus linguistics also tend to consider hapax legomena (or simply hapaxes), words which occur only

42

Morphological Productivity as a Trigger of the þurfan / need Replacement

once in a corpus.2 Hapaxes in a small corpus may actually be frequent words, and in a non-representative corpus they may be words which belong to a text type which is not included in the corpus. Nevertheless, if the corpus is sufficiently large and representative, hapax legomena usually correlate with neologisms, as shown by Baayen & Renouf (1996: 74): “large numbers of hapax legomena are a sure sign that an affix is productive”. Such a correspondence makes the work of the morphologist much easier and faster, because identifying hapaxes is much more straightforward than identifying neologisms. Working from the assumption that neologisms correlate with hapaxes, Baayen (1992: 115) introduces a way of quantitatively calculating the productivity of a given morpheme on the basis of the following formula: N. of hapaxes with suffix x (3) Productivity in the narrow sense = N. of tokens with suffix x For example, by using this formula in his study on quantitative aspects of morphological productivity Baayen (1992: 116) manages to quantify the productivity of the English morpheme -ity, which is 0.0007 (42,252 tokens and 29 hapaxes), while the productivity of the competing morpheme -ness is 0.0044 (17,481 tokens and 77 hapaxes).

3. Methodology: Problems and Solutions What has been said in the previous paragraphs refers to Present-Day English. The aim of the present study, however, is to measure the productivity of two OE morphemes, neod and þearf within the same theoretical framework. This obviously poses some problems concerning the number of texts available for the period and the difficulty in identifying neologisms. The representativeness and size of the corpus is undoubtedly of great importance here, because, as seen above, the identification of neologisms may be skewed. For this reason, the largest OE corpus available must be used, which in this case is the Dictionary of Old English Corpus (DOEC) (compiled by diPaolo et al. 2000), which, at around three million words, contains all extant OE texts. Nevertheless, the representativeness of this corpus may be questioned in at least two ways. Its size is small when compared to a corpus such as the BNC (100 million words), which is used 2

These are sometimes referred to as nonce formations.

Lucía Loureiro-Porto

43

for some PDE studies on productivity. Also, the number of text types included in the DOEC is restricted to those which have survived over the centuries and is obviously limited by the fact that a very small percentage of the population was literate at that time. That is, the language found in these texts might not correspond to the language actually used by OE speakers. Counter-arguments to these two drawbacks can easily be found: 1. Concerning the size of the corpus, present-day studies of the morphological productivity of morphemes found in living languages are sometimes done with corpora even smaller than the DOEC. Baayen (1992), for example, uses a 600,000-word Dutch corpus. In addition, the DOEC contains all extant OE texts, which implies that if any study on morphological productivity can be done on Old English, this is, by default, the most reliable corpus. 2. As for the limited number of registers and text types found in the DOEC and its degree of representativeness regarding the actual speech of the OE speakers, we might argue that if the DOEC is considered representative for any other linguistic analysis (lexical, grammatical, phonological, etc.), why would it not be reliable for a study of productivity? The DOEC may not be the perfect corpus for a complete and faithful description of Old English, but it is the best we have and the best we will ever have. A further number of problems have been encountered in this study, especially with the identification of neologisms. In Present-Day English identifying neologisms is a relatively simple issue, since comprehensive dictionaries such as the OED record the first occurrence attested for each entry. Whenever a word is not found in a dictionary like the OED, the chances are that it is a neologism. In addition, we can count on speakers’ intuition in the identification of new words. For example, the word Monicagate is a neologism first used in 1998 to refer to a scandal involving the President of the US, and can be found in the Corpus of Contemporary American English, compiled by Mark Davies. It is not found in dictionaries yet, and speakers of that time could easily identify it as a new word. Old English, however, is a dead language with no living speakers, and, in addition, dictionaries such as Bosworth & Toller record each single word of the OE vocabulary found in the surviving texts. In this scenario, how can neologisms be identified in Old English? The answer, according to Baayen & Renouf (1996: 74), involves hapax legomena, as mentioned above. Therefore, hapaxes will be the key in the study of the productivity of neod and þearf.

44

Morphological Productivity as a Trigger of the þurfan / need Replacement

Resorting to hapaxes may solve some problems, but it also adds a new methodological difficulty. Although measuring the productivity of a PDE morpheme such as -able requires a very simple search (of the type *able in any concordancer), the orthographic conventions of Old English were far from being standardized, and in other cases identifying tokens and hapaxes, including the morphemes, requires a refined manual search. For example, the OE morpheme neod- has variants ned-, nead-, nid-, nied- and nyd-, and the morpheme þearf- has variants þurf-, þarf-, þorf- (see Loureiro-Porto 2009: 142, 71). Given this high degree of orthographic variation, two possible procedures were considered. The first entailed identifying all tokens, including any of these variants, and analysing only those that do indeed constitute word forms of complex words with the morphemes neod and þearf. This procedure was rejected, not least because it would have been extremely time consuming, due in particular to two factors: 1) orthographic variation also affected the other morphemes found in combination with neod and þearf, which would imply an exponential increase of the word forms to be filtered, and, 2) the sequence ned, neod, nid, etc. is highly frequent in many OE words, regardless of its morphemic nature, so the number of tokens in which it features is uncountable. A second approach, then, was adopted. Since OE dictionaries contain all words recorded in surviving texts, I determined to find all possible lexemes including these two morphemes in one dictionary. Once all possible lexemes had been identified, I would only have to look for orthographic and morphological variants of each of the lexemes in the corpus. The OE Thesaurus (Edmonds et al. 2005) was used because, unlike dictionaries such as Bosworth & Toller, the online version of the Thesaurus contains a wildcard tool which allows searching for any word which includes a given sequence of letters in any position in a word: beginning, middle and end, as shown in Figure 4.

Lucía Loureiro-Porto

45

Figure 4. OE Thesaurus Wildcard tool used for the identification of lexemes [http://libra.englang.arts.gla.ac.uk/oethesaurus/wordsearchmpage.html] Thus, I used this wildcard tool and searched for all spelling variants of my two morphemes in initial, middle and final position in the word. This yielded an extremely long list of words. Some of these were excluded from the study, according to the following criteria: a) The most basic verbs derived from the nouns (þearfan, þurfan, neodian, neden, etc.) were not included. The morphological operation behind these verbs cannot be considered to be proper derivation, but would, in Bauer’s (2004) terms, merely have undergone transposition. Old English being a highly inflectional language, noun to verb conversion of the type bottle > to bottle was not found. Instead, the verbal ending -an had to be added to nominal / adjectival bases so as to create a verb (cf. for example, Hogg 1992: 157). Even if the ending -an is traditionally considered derivational, it is not a real lexicon-expanding morpheme, since it does not create a new concept but only allows for the use of a noun / adjective in the typical syntactic slot of a verb (very much like the PDE conversion process in to bottle). For this reason, the tokens representing word forms of the basic verbs above were not considered. Nevertheless, verbs derived through clear lexiconexpanding processes, by means of the addition of derivational prefixes such as be- or ge- (as in beþurfan or genieden), were taken into consideration, since the function of these prefixes is indeed to form a new, different verb. So as to avoid unnecessary counting,

46

Morphological Productivity as a Trigger of the þurfan / need Replacement

the past participles of ge-verbs which had a non-ge- variant (e.g. neodian vs. geneodian) were not taken into consideration. b) The wildcard searches yielded not only words featuring my morphemes, but also any other word including a homographic sequence. Obviously, words not containing the morphemes neod and þearf were also excluded. Such is the case with efeneadig ‘happiness’ and neodspearuwa ‘sparrow’, for example. c) The OE Thesaurus, as well as other OE dictionaries, includes different entries for words such as the adjectives neodful and niedful, and for þearf(ig)end and þearfend. For the sake of simplicity, I have not made any distinction between them and thus neodful and niedful, as well as all their possible inflectional and orthographic variants, have been considered forms of the same lexeme. The same criterion has been applied to þearf(ig)end and þearfend and other related forms.

4. Results: Analysis of the Corpus 4.1. Factors Conditioning Productivity: Semantic and Phonological Transparency and Relative Frequency Before looking at the results of the quantitative measure of the productivity of each of the morphemes, it is worth considering all the factors which could influence their productivity, as summarized above in Figure 3. Firstly, regarding semantic transparency, it is worth noting that the definition of all the words with the morpheme þearf (34 out of 34 possible lexemes) include a reference to meanings such as ‘necessity’, ‘poverty’ or ‘force’. As for neod, 79.4% of its lexemes (85 out of 107 lexemes) also make a reference to the meanings ‘force’, ‘necessity’, ‘compulsion’ or ‘oppression’. According to these percentages, all words including the morpheme þearf would be very likely to be parsed and decomposed in the mental lexicon. Secondly, the analysis of the phonological transparency of morpheme junctions reveals quite a different finding: only 44.1% of the lexemes including þearf (15 out of 34 lexemes) feature illegal phonological combinations (such as /hș/ in feorhþearf), as opposed to 85% of the lexemes featuring neod (such as /dș/ in niedþing). A third factor influencing the parsability of a given morpheme was its relative frequency regarding its base. In this respect, it must be noted that it is not always easy to identify the base of complex words including neod and þearf. Whereas sometimes the base is clearly neod or þearf, as in geniedelic ‘compulsory, forced’, or þearfleas ‘without need’, on some

Lucía Loureiro-Porto

47

other occasions the base is more difficult to identify, as in niedwædla ‘a poor person’, which appears to be a compound of nied and wædla ‘beggar’. For this reason, and with the aim of avoiding confusion, the relative frequency of all complex words has been calculated on the basis of the frequency of the base nouns neod and þearf. All possible inflected forms of these two nouns were searched for in the DOEC; neod occurs 726 times and þearf 1,433 times. The relative frequency of each of the complex words in which they feature was calculated following the formula in (2) above, and, since none of them exhibits as high a level of frequency as their corresponding bases (neod 726, þearf 1,433), all levels are below unity, which, according to Hay (2001), means that all these words are likely to be parsable and decomposed in the mental lexicon. For reasons of space, I cannot show here the relative frequency of each of the complex words, so I have calculated the average relative frequency of the words featuring each of the morphemes so as to show that the frequencies of words featuring neod are below the levels of those featuring þearf: 0.011 to 0.034. From the analysis of the three factors that affect productivity, it can be concluded that neod-words are likely to be parsed and, hence, decomposed in the mental lexicon because: a) they are phonologically transparent, b) they are lexically transparent, and c) they have a very low relative frequency. Concerning þearf-words, they are also likely to be parsed, though they exhibit higher levels of phonological opacity and a slightly higher relative frequency.

4.2. Measuring the Productivity of neod and þearf Having showed that both neod and þearf are apparently equally likely to be parsed and decomposed in the mental lexicon, let us now quantify their degree of productivity in Old English, by means of identifying the number of lexemes, tokens and hapax legomena found in the Dictionary of Old English Corpus. Table 1. Lexemes, tokens and hapaxes with the morphemes neod- and þearf- as found in the Dictionary of Old English Corpus

neodþearf-

LEXEMES 107 34

TOKENS 888 1692

HAPAXES 55 9

48

Morphological Productivity as a Trigger of the þurfan / need Replacement

Table 1 summarizes the main findings, which appear to show that neod- tends to feature in less frequent words than þearf, since – although it is represented by three times as many lexemes – it only features in half of the tokens. Likewise, the number of hapaxes with neod is a lot greater than those of words with þearf. If we apply the formula under (3) above for the quantification of productivity, we find that neod is far more productive than þearf: the productivity of neod is 0.062 (i.e. 55/888) and that of þearf’s is 0.0053 (9/1692). These numbers illustrate that neod is indeed more productive than þearf, but it is perhaps worth providing further statistical analysis to show the frequency of lexemes including both morphemes. Table 2, then, includes the number of occurrences of each of the lexemes featuring neod and þearf, and Figure 5 graphically illustrates their distribution. Table 2. Frequency of complex words with neod- and þearf(calculated on number of lexemes) MORPHEMES

neodþearf-

1 55

2 13

3 9

LEXEME FREQUENCY 4 [5,10] [11,19] [20,49] [50,99] 8 9 5 4 2

9

5

4

2

3

3

3

1

100+ 2 4

Figure 5. Complex words with neod and þearf (calculated on number of lexemes) Both Table 2 and Figure 5 clearly show that there are many more lowfrequency lexemes with neod than with þearf- (especially as far as

Lucía Loureiro-Porto

49

lexemes with a frequency 1 to 4 are concerned). This has a psycholinguistic consequence: each time one of these very low frequency words is heard, speakers decompose it into morphemes, making the morphemes involved more active (higher resting activation), which in turn makes the morpheme (neod in this case) more available for the creation of new words, following the pattern in Figure 2 above. Likewise, Figure 5 appears to suggest that þearf- features in more highly frequent words (over 100 occurrences) than in low frequency ones. This is seen more clearly if the number of tokens is used in the calculation, as in Table 3 and Figure 6. Table 3. Frequency of complex words with neod- and þearf(calculated on number of tokens) TOKEN FREQUENCY MORPHEME

1

2

3

4

[5,10]

[11,19]

[20,49]

[50,99]

100+

NEOD

55

26

27

32

56

76

115

149

352

ÞEARF

9

10

12

8

24

48

99

57

1425

Figure 6. Complex words with neod and þearf (calculated on number of tokens) Table 3 and Figure 6 show that, were it not for very high frequency tokens (100+ occurrences), þearf would scarcely be represented in the corpus, and also that neod would lose more than a third of its occurrences (352 out of 888). Indeed, if we redistribute these frequencies into four

50

Morphological Productivity as a Trigger of the þurfan / need Replacement

groups (hapaxes, frequencies [2,19], frequencies [20,99] and frequencies 100+) and calculate the percentage of representation of each group based on tokens including these two morphemes, as in Figures 7 and 8 below, we see that high frequency words containing þearf- constitute more than 84% of its occurrences, while those of neod- constitute less than 40%.

Figure 7. Percentage frequency of þearf-words

Figure 8. Percentage frequency of neod-words A proportion of 40% of high frequency words with neod appears to be too high for a morpheme which clearly tends to feature in low frequency words (as clearly seen in Figure 5), and this might therefore raise the question of whether neod indeed has a preference for low-frequency words. A closer look at the data reveals that the most frequent neod- word

Lucía Loureiro-Porto

51

is niedþearf, curiously enough a compound word featuring the morpheme þearf. If this word is removed from the final count, the total number of tokens recorded with neod- is 661, and the percentage of neod- words calculated on the number of tokens is shown in Figure 9.

Figure 9. Percentage frequency of neod-words (excluding niedþearf) Niedþearf is an interesting word, whose parsability yields an apparently psycholinguistic clash: if it follows the decomposition route, both morphemes, neod and þearf, get activated in the mental lexicon and hence are readily available to coin new words. However, as the comparison of Figures 8 and 9 shows, niedþearf is responsible for 50% of highly frequent words with neod. It is indeed the most frequent word featuring the morpheme neod (227 occurrences out of a total of 888, a relative frequency of 0.3). Its relative frequency, as well as its semantic and phonological transparency, make it susceptible to parsability, but the truth is that niedþearf disappears from the language at the same time as þearf does (before 1250, as shown by Loureiro-Porto 2010: 378), which seems to show that þearf’s low degree of productivity washed it away.

5. Discussion and Conclusions The replacement of the preterite-present þurfan by the weak verb neden in Middle English is not successfully accounted for in the literature, which focuses rather on phonological matters as being the single factor triggering the disappearance of þurfan, specifically its confusion with durren (Visser 1963–1973, Molencki 2005). This argument has been seen as insufficient, and it has also been shown that the morphologically related

52

Morphological Productivity as a Trigger of the þurfan / need Replacement

nouns þearf and neod undergo a similar replacement more than two centuries before þurfan disappears (Loureiro-Porto 2010). The hypothesis that the nominal replacement triggered the verbal one is plausible, but a question remains: why would the noun neod substitute for the noun þearf? Loureiro-Porto (2010) shows that there is no semantic or syntactic reason for such a replacement, and the present study has therefore taken a new direction, looking at the phenomenon from the perspective of morphological productivity and its relation to psycholinguistics. According to current models of morphological processing, whenever two semantically and functionally similar morphemes are available for speakers to form a new word, speakers tend to select the morpheme that is psycholinguistically more active, that is, the morpheme with a higher resting activation. Studies of morphological productivity in Present-Day English have shown that the most productive morpheme of a group tends to feature in many low frequency words, while less productive morphemes feature in few high frequency words. Since low frequency correlates with parsability (and decomposition in the mental lexicon), morphemes featuring in less frequent words have a higher resting activation and are, therefore, more readily available for the creation of neologisms. A further property of productive morphemes is that they attract members of other classes. For example, the plural morpheme -s in Present-Day English is not only found in many low frequency words (neologisms and borrowings), but is also attracting members from the -en or -ae classes (notice the coexistence of the plural forms formulae and formulas). Since neologisms are the principal symptom of productivity, the most accurate way of measuring productivity stems from counting them in a given corpus. As repeatedly shown in previous studies, hapax legomena in a representative corpus roughly correlate with neologisms. Within this framework, the morphemes þearf and neod have been studied with the aim of finding out if one of them was more productive and, therefore, more active psycholinguistically than the other. The initial hypothesis was that neod replaces þearf because it is psycholinguistically more active, and therefore not only is it used to create new words more often than þearf, but also replaces þearf words in constructions in which they stand in variation (as the verbo-nominal constructions beon neod, beon þearf), in the same way as the plural ending -s replaces -ae in words such as formulae. In order to measure the degree of productivity of neod and þearf, the Dictionary of Old English Corpus was used and the three factors highlighted by Plag (2006) were measured: semantic transparency, phonological transparency and relative frequency of each of the complex

Lucía Loureiro-Porto

53

words featuring these morphemes. High levels of semantic and phonological transparency and low levels of relative frequency are said to lead to parsability, which in turn implies higher productivity. The results show that both neod and þearf exhibit features of parsable morphemes, though the parsability level of þearf is somewhat lower, owing to its phonological opacity (i.e. its integration with the other morphemes yields legal phonotactic junctures) and to its relatively higher average relative frequency (0.034, as opposed to the 0.011 relative frequency for neod words). In a second stage, the actual productivity of neod and þearf was measured, following the formula introduced by Baayen (1992). The findings clearly show that: a) neod- tends to feature in many more lexemes than þearf (Tables 1 and 2); b) 50% of the lexemes in which neod occurs are hapax legomena (Table 1); c) þearf- is well represented in the corpus because it occurs in many high-frequency words (over 100 occurrences), as seen in Table 3; and finally, d) neod’s productivity is 0.062 (i.e. 55/888), and þearf’s productivity is 0.0053 (9/1692). All these findings clearly show that the morpheme neod is more productive than þearf, but this also leads to two more questions: 1) is there any exclusively linguistic reason why neod would be more productive than þearf? 2) can this difference in the degree of productivity account for the replacement of þearf words by neod words in verbo-nominal constructions? The answer to question 1 may be found by looking at the place usually occupied by neod and by þearf in complex words. While þearf tends to occur at the end of words (it does so in 26 out of 34 lexemes, in the compound niedþearf, for example), neod exhibits the opposite tendency, that is, it most commonly features at the beginning of words (in 83 out of 107 lexemes in which it features, as seen in the same example). This seems to point to the analysis of neod as a modifier in compounds, while þearf would instead be the semantic head (see Plag 2002: 135-137 for an analysis of the notion of head in compounds). If we consider it a semantic modifier, it makes sense to find neod attached to a great many words,

54

Morphological Productivity as a Trigger of the þurfan / need Replacement

which may be nonce formations, with the meaning ‘necessary, urgent’. In this respect, neod’s role as a modifier would make it psycholinguistically more active and, as a consequence, more productive. As for question 2, I think the higher degree of productivity of neod can indeed account for its substitution for þearf in verbo-nominal constructions. As noted above, productivity and morphological processing, the activation of a morpheme in the mental lexicon and its availability to be used to coin a new word, are a matter of psycholinguistics. Since both morphemes, neod and þearf, stood in variation (as seen in some complex words and, most especially, in the verbo-nominal constructions beon neod and beon þearf), it seems to make sense to hypothesize that once one of the morphemes is activated and readily available to coin new words, it is also ready to be used instead of a similar competing monomorphemic word. This paper has shown that the morpheme neod was far more productive than the morpheme þearf in Old English. The psycholinguistic implications of this difference in the level of productivity suggest that neod was more active in the mental lexicon of speakers and, therefore, more readily available for the formation of new words and the replacement of þearf in contexts in which they stood in variation, as was the case with verbonominal constructions (beon neod, beon þearf). The replacement of the noun, which is now justified on psycholinguistic grounds, would then have paved the way for the replacement of the verb þurfan by the verb neden in the 14th century, which, in addition, was phonologically threatened by its confusion with durren. The synergistic effects of psycholinguistic activation based on morphological productivity, the phonological confusion of the verbs þurfan and durren, and the previous replacement of the noun þearf may, then, help us understand the reasons for the mysterious disappearance of þurfan.

Acknowledgements For generous financial support I am grateful to the European Regional Development Fund and the Spanish Ministry for Science and Innovation (grant FFI2011–26693-C02-02). I would also like to thank Dr. Cristina Suárez-Gómez for her comments on an earlier version of this paper.

Lucía Loureiro-Porto

55

Works cited Primary sources COCA = Corpus of Contemporary American English (compiled by Mark Davies), , accessed 01-10-2010. DOEC = The Dictionary of Old English Corpus (TEI-P3 conformant version) compiled by Antonette diPaolo Healey, Joan Holland, Ian McDougall & Peter Mielke. Toronto: DOE project on CD-Rom, 2000. Edmonds, Flora, Christian Kay, Jane Roberts & Irené Wotherspoon. 2005. A Thesaurus of Old English. University of Glasgow. [Electronic version of the 2nd impression of A Thesaurus of Old English (Amsterdam: Rodopi, 2000), first published by King’s College, London (1995)] http://libra.englang.arts.gla.ac.uk/oethesaurus/, accessed 20-08-2009. OED = The Oxford English Dictionary. 1989. Edited by James A. H. Murray, Henry Bradley, William A. Craigie, & Charles T. Onions. (2nd ed.) CD-ROM version. Oxford: Oxford University Press.

Secondary sources Baayen, Harald. 1992. “Quantitative Aspects of Morphological Productivity”. Yearbook of Morphology 1991 ed. by Geert Booij & Jaap van Marle, 109-149. Dordrecht: Kluwer. Baayen, Harald, & Antoinette Renouf. 1996. “Chronicling The Times: Productive Lexical Innovations in an English Newspaper”. Language 72.69-96. Bauer, Laurie. 2004. “The Function of Word-formation and the Inflection– Derivation Distinction”. Words in their Places. A Festschrift for J. Lachlan Mackenzie ed. by Henk Aertsen, Mike Hannay & Rod Lyall, 283-292. Amsterdam: Vrije Universiteit. Carstairs-McCarthy, Andrew. 2002. An Introduction to English Morphology. Words and Their Structure. Edinburgh: Edinburgh University Press. Haspelmath, Martin. 2002. Understanding Morphology. London: Arnold. Hay, Jennifer. 2001. “Lexical Frequency in Morphology: Is Everything Relative?” Linguistics 39:4.1041-1070. Hay, Jennifer & Harald Baayen. 2002. “Parsing and Productivity”. Yearbook of Morphology 2001 ed. by Geert Booij & Jaap van Marle, 203-235. Dordrecht, Boston & London: Kluwer.

56

Morphological Productivity as a Trigger of the þurfan / need Replacement

—. 2003. “Phonotactics, Parsing and Productivity”. Rivista di Linguistica 15:1.99-130. Hogg, Richard M. 1992. “Phonology and Morphology”. The Cambridge History of the English Language, vol. I: The Beginnings to 1066 ed. by Richard M. Hogg, 67-167. Cambridge: Cambridge University Press. Loureiro-Porto, Lucía. 2009. The Semantic Predecessors of need in the History of English (c750–1710). (= Publications of the Philological Society, 43). Oxford: Wiley-Blackwell. —. 2010. “Verbo-nominal Constructions of Necessity with þearf n. and need n.: Competition and Grammaticalization from OE to eModE”. English Language and Linguistics 14:3.373-397. McQueen, James M. & Anne Cutler. 1998. “Morphology in Word Recognition”. The Handbook of Morphology ed. by Andrew Spencer & Arnold Zwicky, 406-427. Oxford: Blackwell. Molencki, Rafaá. 2005. “The Confusion between thurven and durren in Middle English”. Rethinking Middle English: Linguistic and Literary Approaches ed. by Nikolas Ritt & Herbert Schendl, 147-160. Frankfurt am Main: Peter Lang. Palmer, Chris. 2009. Borrowings, Derivational Morphology, and Perceived Productivity in English, 1300–1600. Ph.D. dissertation, University of Michigan. Plag, Ingo. 2003. Word-Formation in English. Cambridge: Cambridge University Press. —. 2005. “Productivity”. Encyclopedia of Language and Linguistics (2nd ed.) ed. by Keith Brown, vol. 10, 121-128. Oxford: Elsevier. —. 2006. “Productivity”. Handbook of English Linguistics ed. by Bas Aarts & April McMahon, 537-556. Oxford: Blackwell. Plag, Ingo, Christiane Dalton-Puffer & Harald Baayen. 1999. “Morphological Productivity across Speech and Writing”. English Language and Linguistics 3.209-228. Visser, Fredericus T. 1963–1973. An Historical Syntax of the English Language (4 vols.). Leiden: Brill.

ASPECTUAL LOSS AND RENEWAL: ONGINNAN, BEGINNAN, START LYNN SIMS

1. Introduction 1.1. An Overview of Aspect Beginning with the distinction between movements and actualities during the Classical period, the concept of aspect has received a great deal of attention in the literature. While aspectual expressions continue to be analyzed, they are now often discussed in terms of two distinct domains: lexical aspect and grammatical aspect.1 Lexical aspect is unmarked and reflects the inherent temporal property of a verb. At the most basic level, languages appear to express in their lexicons the same four classes of lexical aspect: states (e.g., know, believe), activities (e.g., run, walk), accomplishments (e.g., paint [a picture]), write [a letter]), achievements (e.g., find, realize). Grammatical aspect, on the other hand, is more subjective, reflecting the speaker’s perspective on the internal temporal quality of an event (cf. Comrie 1976; Nordlander 1997). Smith (1997: 61) describes the function of grammatical aspect in terms of a camera lens that makes “objects visible to the receiver”. The object on which the grammatical lens is focused is an event, and this aspectual lens can focus on all or part of the event structure. Thus, grammatical aspect does not introduce an event but rather hooks into and shifts the temporal focus of an existing event. In English, grammatical aspect is formally marked by periphrastic constructions and inflectional morphology. This is best illustrated by the progressive, example (1), which indicates temporally continuous or ongoing events, and the perfect, example (2), which indicates complete or temporally specified events: 1

Terminology for these two domains varies, reflecting the respective approach (i.e. semantic, syntactic, functional, formal). Alternate terms for lexical aspect include aktionsart, situation aspect, or inner aspect. Alternate terms for grammatical aspect include viewpoint aspect or outer aspect.

58

Aspectual Loss and Renewal: onginnan, beginnan, start

(1) LaVesta is laughing. (2) LaVesta has completed the assignment. Similar to the structure of grammatical aspect, ingressive aspect in Modern English (ModE) is expressed periphrastically and morphologically, as in (3), using the markers begin or start. (3) LaVesta began/started singing. (4) LaVesta began/started to sing. In addition to occurring with state, activity, and accomplishment complements, ingressives can occur with achievement verbs if the event is repeatable, as in (5), or if the subject is an unspecified plural or mass noun, as in (6) (cf. also Freed 1979). In these examples, the periphrastic construction results in an iterative-ingressive expression. (5) She began to guess/guessing. (Brinton 1988: 85, ex. 8g) (6) Friends/people began to arrive at the party. (Brinton 1988: 85, ex. 8h) As for the function of ingressive aspect, it also mirrors that of grammatical aspect: an ingressive marker hooks into an existing event and what the lens focuses on is, to use Smith’s terminology, the entry into the event expressed by the non-finite complement. English ingressives tend to develop from lexical verbs that express one of two spatial concepts: (1) movement of the agent toward or into an event or (2) movement of the event toward the agent (Brinton 1988: 114-115). Both concepts involve the semantics of motion, but the second also includes the notion of taking or receiving, as is evident in the Old English (OE) ingressives fon, tacan, and niman, which basically mean ‘to take, seize, receive’. As verbs in both groups grammaticalize, Brinton proposes that their semantics undergo a metonymic shift in focus rather than a gradual weakening of semantics. The metonymic process, as Hopper & Traugott (1993: 87) write, “involves specifying one meaning in terms of another that is present, even if only covertly, in the context”. That some ingressives do not experience significant semantic bleaching as they become less lexical and more grammatical is not problematic. Semantic bleaching occurs later in the grammaticalization process as a verb acquires

Lynn Sims

59

a more functional meaning (Hopper & Traugott 2003; see also Cardinaletti & Giusti 2001, who show that certain motion verbs in Southern Italian dialects retain their semantic content when used periphrastically). Excluding onginnan, OE ingressives shift into and remain in an overlapping boundary (in the sense of Heine et al. 1991: 221-229), somewhere between full verb and auxiliary (Sims 2008). Since they do not move any further than this hybrid-like position, the concept of movement which is inherent to the lexical form remains fairly intact. It is also important to add here that phonological reduction is a characteristic not a requirement of the grammaticalization process. As will be discussed below, onginnan is the only ingressive that experiences significant phonetic loss (gan), but evidence from ModE shows that begin is also susceptible to this process. Diachronically, membership in the English ingressive category reflects a degree of instability. Old English, for example, uses a number of Germanic forms to express the initiation of an event, particularly, -ginnan (and its prefixed variants onginnan, aginnan, ingannan, beginnan), fon (and its prefixed variants gefon, onfon, and underfon), weorþan, tacan, and niman (cf. Mustanoja 1960; Visser 1969, 1973; Brinton 1988; Sims 2008). According to the OED (s.v. begin), the most common OE marker is onginnan; beginnan and (on)fon are rare and weorþan, tacan and niman are even rarer. Though some ingressives fall out of use during the Middle English (ME) period, this period also sees the introduction of both native (e.g., setten, breken, bresten) and borrowed (e.g., commencen, becomsen, proceden) forms to the group. This cycle continues into Modern English, where we see the introduction of resume, recommence, get, and start. Although the native forms, begin and start, are used in similar syntactic constructions, major changes in the distribution of begin during the last twenty years suggest a decline in the use of this ingressive marker.

1.2. Data and Methodology I have examined major changes involving ingressives by looking at the secondary literature on English, such as Callaway (1913), Visser (1969, 1973), Mitchell (1985). I also used online sources such as the University of Toronto Dictionary of Old English electronic corpus of Old English, various Middle English electronic texts, the Helsinki Corpus, the Corpus of Dialogues, the Corpus of Contemporary American English, and the Corpus of Historical American English. I have tried to be as descriptive as possible in this paper, focusing mainly on the cycle of loss and renewal. Therefore, I have kept linguistic theory in the background. This paper is informed, however, by a generative

60

Aspectual Loss and Renewal: onginnan, beginnan, start

framework, particularly a Minimalist framework. I assume the presence of an expanded Inflectional Phrase (IP) that includes an Aspect Phrase (AspP), and also the presence of an expanded Verb Phrase (VP) shell that includes a small v. I adopt Butt’s (2003) proposal that elements in small v can be interpreted as lexical, functional, or a mixture of both. I have included theoretical explanations when called for in the analysis.

2. Old English: Differences Emerge between Ingressive Forms The *ginnan stem is considered a Germanic verb with the root meaning of ‘to open, open up’ (OED, s.v. begin). *Ginnan is also thought to be cognate with ON gína and OE gínan, meaning ‘to gape, yawn’ and coming from a stem *gi-, which also appears in Old Slavonic zij-ati and Latin hi-Ɨre and means ‘to gape, open’. Thus, inherent to the root meaning of *ginnan is the concept of motion, and, as the OED states, the shift in meaning from ‘open up’ to ‘begin’ occurs frequently (e.g., ‘open’ a speech, ‘open’ fire, ‘open up’ negotiations). The use of a *ginnan form to mark ingressive aspect in early Germanic languages is well documented, and counterparts to OE on-/be-ginnan include: Gothic du-ginnan, Old Saxon bi-ginnan, Old High German in-/bi-ginnan, Old Frisian bi-ginna, and Middle Dutch ont-/be-ghinnen. Today, prefixed -ginnan forms are still used in English, German, Frisian, Dutch, Danish, and Norwegian. Old English ingressive constructions, similar to the early Germanic counterparts, involve a bare infinitival complement, with the use of the toinfinitive arising during the OE period. Ingressive + forto constructions begin to occur in early ME, as well as ingressive + -ende/-ing constructions. Thus, I begin with the form of the infinitival complement.

2.1. Complement Patterns: Bare Infinitive vs. to-infinitive As is well known, both the bare and to-infinitives are found in OE texts. However, the bare infinitive is often restricted to perception verbs, hatan and lætan, and the ancestors of ModE modals (cann, dearr, mæg, mot, sceal, þearf, uton, wile) (Warner 1993: 136-139). With -ginnan forms in later OE texts, it is not unusual for a bare infinitive and a to-infinitive to co-occur in the same clause:

Lynn Sims

61

(7) Ða ongunnon ealle ða næddran to ceowenne heora flæsc Then begun all the snakes to chew their flesh and heora blod sucan and their blood to suck (ÆHom. 11.488 [Sims 2008: 116]) The choice of complement type is often considered stylistic rather than syntactic. For instance, Brinton (1988: 270, n.20) argues that “no fundamental difference between the [infinitive] forms was felt” by the writer. More recently, Los (2000, 2005), examining ingressive constructions in Ælfric, proposes that the complement type (and the semantic strength of the ingressive marker) is determined by the discourse function of on-/beginnan in þa V (finite verb in second position) and V1 (finite verb in first position) environments. Los (2005: 94) concludes that “ginnan-verbs only express ingression when followed by a to-infinitive, and cannot express ingression when followed by a bare infinitive”. Thus, with bare infinitives, onginnan and beginnan are semantically bleached auxiliaries that express perfective aspect. While I basically agree with Los’ proposal that the toinfinitive strengthens the concept of ingressiveness, I disagree with her conclusion that onginnan and beginnan are semantically bleached auxiliaries when they occur with a bare infinitive. This conclusion does not hold across the board, especially since beginnan rarely functions as a non-aspectual auxiliary. Looking at the earliest examples of English ingressives, we see that the periphrastic use of -ginnan occurs c700, involves the on-prefix, and takes a bare infinitival complement: (8) oð ðæt an ongan fyrene fremman until one began atrocities to perform (Beowulf 101) With onginnan, the bare infinitive appears to be the original form of the complement (Callaway 1913: 67). That the bare infinitive continues to be onginnan’s main complement in OE is evident from the writings of Alfred (late 9th cent.). Of his 248 ingressive constructions, only two occur with a to-infinitive (see Table 1). The following examples, from Alfred’s Pastoral Care (Cotton manuscript), are typical of early English onginnan constructions.

62

Aspectual Loss and Renewal: onginnan, beginnan, start

(9) þa ongan ic ongemang oðrum mislicum & monigfaldum bisgum then began I among other variously & manifold troubles ðisses kynerices þa boc wendan on Englisc this kingdom the book to translate in English ‘then I began, among other various and manifold troubles of this kingdom, to translate the book into English’ (PC Pref. 6.17 [Sims 2008: 98]) (10) ðæt hie eft onginnen gitsian & reafian for hiora wædle that they again begin to desire & to steal for their poverty ‘that they begin again to desire & to steal because of their poverty’ (PC 45.341.3 [Sims 2008: 100]) Moving to the writings of Ælfric (approximately a century later than Alfred), we see that while his choice of complement with onginnan reflects some variation, his predominant pattern, like Alfred, involves the bare infinitive (77%). Looking next at beginnan, we see that this form appears c1000 in the writings of Ælfric (OED, s.v. begin; see also Visser 1978: 1373). (11) þa begunnon hi to cidenne mid micelre ceorunge ongeane then begun they to quarrel with great complaint against þone ælmihtigan god the almighty god (ÆHom 21.304 [Sims 2008: 116]) With this ingressive marker, however, Ælfric’s preferred complement is the to-infinitive (74%), which is almost the exact opposite of his usage with onginnan. From the outset, then, beginnan occurs mainly with the toinfinitive. Thus, I agree with Callaway’s (1913: 66-67) suggestion that the to-infinitive is beginnan’s original complement. The to-infinitive marks the aspectual event as [+]perfective with beginnan shifting focus to the initiation of that event. While the bare infinitive is found with beginnan up to approximately 1440, its use remains minimal compared to the toinfinitive. The difference between complement type with onginnan, aginnan, and beginnan is provided in Table 1 (from Sims 2008: 113, Table 4, compiled from data in Callaway 1913: 279-287).

Lynn Sims

63

Table 1. OE -ginnan forms in periphrastic constructions bare infinitive

%

to-infinitive

246 82 649 977

99% 77% 98% 96%

2 25 10 37

1% 23% 2% 4%

28

85%

5

15%

0 19 9 28

-26% 75% 33%

0 54 3 57

-74% 25% 67%

%

onginnan Alfred Ælfric Other texts Total onginnan

aginnan Total other texts

beginnan Alfred Ælfric Other texts Total beginnan

After Ælfric, the use of beginnan continues to increase while the use of onginnan decreases. However, just as beginnan makes its appearance, so does the phonologically reduced aginnan (< onginnan). Early examples of this phonological reduction are found in Skeat’s (1887/1970) edition of The Gospel According to Saint Matthew (Matthew), which combines the Northern, Mercian, and two West Saxon versions of Matthew into one text. The Northern version (Lindisfarne Gospels) dates from c950 and is an interlinear gloss of a Latin original. The Mercian version (Rushworth Glosses) dates from c980 and is also an interlinear gloss. The two West Saxon versions are based on six different manuscripts but mainly on the Corpus (c1000) and Hatton (c1150) manuscripts, respectively. As we see from the following (from Sims 2008: 105), the West Saxon versions use aginnan. (12) a. 7 agynð beatan hys efen-þeowas. 7 yt 7 drincþ mid druncenum (Corpus Matt 24:49) b. 7 aginð beaten his efen-þeowas. 7 ett 7 drincð mid druncenan (Hatton Matt 24:49) c. 7 ongann slae heafudlinges his he ett uutedlice 7 dranc mið ðruncnum (sic) (Lindisfarne Matt 24:49)

64

Aspectual Loss and Renewal: onginnan, beginnan, start

d. 7 onginnaþ slan efnþeu his manducat (sic) him þonne 7 drinceþ mid druncennum ‘and begins to strike his servants, and to eat and drink excessively’ (Rushworth Matt 24:49) While the Northern and Mercian versions do not use aginnan (Sims 2008), it appears twice in the Northern version of The Gospel of Luke – once as a full verb and once as an ingressive marker with the to-infinitive (Ogura 1977: 415): (13) ðonne agann bituih him getalade to coæðanne then began within himself thought to say (Lk 11:38 [Ogura 1997: 415]) A search of Luke shows that the Mercian counterpart to (13) does not have a reduced form of onginnan, but it does use the to-infinitive: ðonne ongan ... to cweoðanne. The bare infinitive tends to be aginnan’s main complement type up to a1200, but then the to-infinitive becomes more frequent. Finally, around 1200, onginnan disappears from the textual data. Aginnen disappears by the end of the 14th century; the last example is from 1340, in a text that also makes frequent use of beginnen (Mustanoja 1960: 610; cf. MED, s.v. aginnen; Visser 1969: 1372-1373). (14) ne dorre naвt aginne wel to done not dare not begin well to do ‘dare not to begin to do well’ (Ayenbite 32) Regardless of complement type, the -ginnan forms in (7) to (14) are not semantically weak – the concept of ‘begin(ning)’ is clear. But, by hooking into an existing event, on-/a-/be-ginnan shifts to a more general, more functional role – marking ingressive aspect. The reduction of onto a-ginnan also suggests that this form has shifted further into the grammaticalization process than beginnan. Another sign of grammaticalization involves a verb’s argument structure, to which I now turn.

Lynn Sims

65

2.2. Impersonal Constructions In general, impersonal constructions have an objective case subject rather than a nominative case subject. In English, non-nominative subjects occur through approximately the 15th century (van Kemenade 1993: 163; see also Elmer 1981). In OE, onginnan often occurs with a nonnominative subject (Denison 1990, 1993), as in the following example: (15) Þa ongan me langian for minre hæftnyde then began me-ACC to grieve for my imprisonment ‘then I began to find my imprisonment tedious’ (LS 35 (Vit Patr) 330 [Denison 1990: 148]) In (15), the semantic role ‘experiencer’ is assigned to the subject me, suggesting that onginnan’s argument structure is defective and that the verb is beginning to grammaticalize. Here, onginnan has a single argument (the infinitival ‘event’) and the thematic function of the subject is assigned by the lower predicate (langian). Thematic transparency is considered a characteristic of the group of verbs which eventually become auxiliaries in English (cf. Denison 1993: 441; Warner 1993: 123). Thematic transparency is also a feature of onginnan in impersonal verb constructions. While examples with onginnan in this construction are numerous, examples with beginnan are not. In the COE (Healey & Venezky 1980), I find only one, given below: (16) ða geseah he færlice þone strangan wind then saw he suddenly the powerful wind and begann hine to ondrædenne and began him-ACC to fear ‘then suddenly he saw the powerful wind and [he] began to be afraid’ (ÆCHom II. 28 227.195 [Sims 2008: 91]) Denison also makes no mention of OE beginnan in this construction, but he states that it occurs “apparently once” in ME (1993: 235): (17) Him bigon to gremien, & o grome gredee him-OBL began to grow-angry and in rage cried-out ‘He began to grow angry and cried out in rage’ (St Marg [1] 42.10 [Denison 1993: 235])

Aspectual Loss and Renewal: onginnan, beginnan, start

66

Warner (1993) also provides what he considers a possible ME beginnen example: (18) Þo bigan ham alle to agrise then began them all to fear ‘then they all began to be terrified’ (Otuel, EETS, ES 39, 65-116 [Warner 1993: 128]) However, he does not consider the impersonal construction to be a strong criterion for isolating a potential auxiliary group. Warner (1993: 129-130, 256, n.26) also suggests that beginnan, in addition to a few other OE ingressive markers, are “derived historically from senses in which they presumably selected their subjects”. Consequently, he suggests that not all OE ingressives become transparent aspect markers – a suggestion with which I agree, particularly with beginnan. Unlike onginnan, beginnan remains a hybrid-like form. The evidence for beginnan in non-nominative subject + impersonal verb constructions is extremely rare; therefore, I do not consider this to be a common construction for this ingressive marker. However, the examples do suggest that beginnan also has a defective argument structure and has, thus, entered the grammaticalization process. Structurally, in these constructions, I consider both onginnan and beginnan to be generated in v (inside an expanded VP-shell). Locating both aspect markers in v recognizes two facts: (a) they are still interpreted as lexical because they retain their semantic content and show some agreement features (person and number) and (b) they show signs of being interpreted as functional because of their inability to theta-mark the subject and specify nominative case.

2.3. Non-aspectual Function The non-aspectual use of an ingressive marker happens when the form becomes semantically empty. We do begin to see this with onginnan in OE, but it is rare and only a few examples appear in the manuscripts (Brinton 1988): (19) ge

þa

sciran miht deman ongunnon

‘you then judged the bright power [of Christ]’ (*began to judge) (El 310b-11a [Brinton 1988: 160])

Lynn Sims

67

(20) Witodlice ... ongann se hiredes ealdor to agyldenne þone pening ‘Certainly repaid (*began to repay) the elder of the house the penny’ (ÆCHom II, 5 46, 137 [Brinton 1988: 160]) Because non-aspectual onginnan has few, if any, semantic features, it is not relevant to the argument structure and is, thus, auxiliary-like in its function. Structurally, this means that non-aspectual onginnan merges into a functional position (AspectP) inside an expanded IP. To recap the OE data: we begin to see important differences between onginnan and beginnan; onginnan prefers the bare infinitive, it can occur in non-nominative constructions, and can be used as a semantically empty auxiliary. Beginnan behaves just the opposite in all of these cases. Moving into the ME period, onginnan has already experienced more grammaticalization than beginnan.

3. From Middle English to Early Modern English: A Period of Continued Change Although (a)gan and begin continue into the ME period, begin becomes the most frequently used form, as we will see below. Though minimally used, the bare construction is lost as a complement to begin at roughly the same time (a1200) that an -ing complement is introduced. During this period, gan’s infinitival complement becomes more restricted, to which I now turn.

3.1. The Issue of Tense: Preterit vs. Present As the use of on-/a-ginnan decreases, the use of gan increases. Gan appears in the Northern dialect around 1200 (the Northern and West Midland variants are can/con) and in the Southern dialect in the 1300s. Eventually, gan becomes a frozen preterit construction and its syntactic environment (and genre type – poetry rather than prose) narrows. The function of gan has been the focus of many studies, and a full discussion is beyond the scope of this paper (see Sims 2008 for a summary of key stylistic, metrical, and discourse proposals). By 1500, gan is considered archaic and eventually “passes out of use” (Visser 1978: 1572). The predominance of preterit gan has been noted in the literature, and it is suggested that this may reflect the conventions of narrative style (cf. Mustanoja 1960: 610). However, Ogura (1997, 1998) examines the tense of OE onginnan and offers an alternative proposal. Ogura shows that the preterit form is used in significantly higher numbers compared to other

68

Aspectual Loss and Renewal: onginnan, beginnan, start

forms and suggests that a connection exists between the frequency rate of preterit onginnan and its gradual shift to an auxiliary – a suggestion with which I agree. From a grammaticalization perspective, the preterit becomes a frozen construction, and the over-extended use of this form contributes to its continued bleaching and eventual loss of semantic content. That a preterit form grammaticalizes before (or instead of) a present tense form is not unique. Van Gelderen (2004: 180-199), for example, shows that while present tense see, a ModE perception verb, does not grammaticalize, preterit saw (with a bare infinitive complement) does. In this case, saw functions as an (evidential) auxiliary, checks [+]perfective, and expresses the completed nature of the aspectual event. In Table 2, I summarize Ogura’s (1997: 405-409) data. Table 2. Preterit vs. Non-Preterit onginnan (summarized from Ogura 1997: 405-409) Preterit onginnan + bare infinitive Poetry Bede WSGospels (Corpus ms.) ÆCHomI

177 78 39 26

Non-Preterit onginnan + bare infinitive 27 3 4 0

During ME, present tense gan generally allows an ingressive reading (cf. Funke 1922; Häusermann 1930; Koziol 1932; Mustanoja 1960; Visser 1969; Brinton 1988). Likewise, in early ME texts such as Ormulum (c1200), example (21), and Layamon’s Brut (c1205–1250), example (22), preterit gan can also retain its ingressive semantics. For instance, the Brut has 380 occurrences of preterit gan (gunne(n)/gonnen/gan/gon) and 21 occurrences of preterit agon (see Table 3), and the majority of these examples allow an ingressive reading of the aspect marker (Sims 2004). (21) 7 he gann þennkenn off himm sellf, And he began to-think of himself 7 off hiss miccle riche 7 of his great power ‘And he began to think about himself and his great power’ (Ormulum 3274-3275 [Sims 2008: 142])

Lynn Sims

(22) 7 hit and it

agon daiвen began to-dawn

69

and deor gunnen waвeвen. and deer began to-stir (Brut 13447 [Sims 2008: 143])

But, as we move through the ME period, preterit gan, unlike its present tense counterpart, frequently does not allow an ingressive reading (cf. Mustanoja 1960: 611-615): (23) Syr Gawen his leue con nyme Sir Gawain his leave [grammatical] to-take ‘Sir Gawain [did take] [took] his leave’ (SGGK 993 [Funke 1922: 25]) Like gan, preterit beginnen is common in ME, but unlike gan, its semantics do not weaken. In fact, Brinton finds only four non-aspectual examples: (24) he bigan the fish to kippe ‘he seized (*began to seize) the fish’ (Havelock 894 [Brinton 1981: 185]) Thus, as a competitor to gan, the lack of non-aspectual beginnen strengthens its position as the predominant ingressive marker during the ME period.

3.2. Complement Types: Changes Continue In addition to a narrowing tense construction, gan’s infinitival complement becomes more restricted, occurring mainly with the bare infinitive (contra Terasawa 1974; cf. Lichtsinn 1913: 33). Various forms of preterit gan occur frequently in Layamon’s Brut (Caligula) and nicely illustrate the narrowing of complement type. Table 3. Preterit gan and its complement type in Layamon’s Brut (Caligula) gunne(n)/gonnen/ gan/gon agon (agan once)

bare inf. 316 (83%) 21 (95%)

to-inf. 64 (17%) 0

forto-inf. 0

-ing 0

1 (5%)

0

total 380 22

70

Aspectual Loss and Renewal: onginnan, beginnan, start

A similar pattern occurs in King Horn (KH) (c1250–1300), a poem fairly contemporary with the Brut. Approximately seventy preterit gan forms are in this poem, and the majority occurs with a bare infinitive. In the Brut, beginnen’s use is minimal (43 total) compared to the gan forms, but like gan, its main complement is the bare infinitive (63%). However, we begin to see the decline of bare infinitives with beginnen as early as c1200 in Ormulum. Of the approximately 46 beginnen occurrences in this text, none occur with a bare infinitive (cf. also Ogura 1997). By the mid-15th century, beginnen + bare infinitive constructions basically disappear. A perusal of Kato’s (1974, s.v.v. began, begyn, begynnyth) concordance to the works of Malory (1469) supports this. Malory uses beginnen + bare infinitive only once compared to approximately fifty beginnen + to-infinitive examples. After the mid-15th century, the toinfinitive is the predominant complement. As we will see next, beginnen begins to lose its bare infinitival complement at the same time (a1200) that forto and -ing complements are introduced. Returning to the data in Table 3, we see that the Brut has one forto complement with agon; KH also has three forto complements with gan: (25) Horn gan for to ride Horn began for to ride (Horn, Gg. 1631 [Sims 2008: 148]) Beginnen also occurs with this complement. Though Visser (1969: 1373) and the OED (s.v. begin) state that forto is first attested with beginnen a1300, I find an early occurrence in Ormulum (c1200): (26) Biginnenn for to spellenn begin for to preach (Ormulum 10887 [Sims 2008: 156]) Although forto infinitives occur with ginnan forms during the ME period, their use is minimal compared to other complement types. After a1500, forto + -ginnan constructions rarely appear (cf. Visser 1978: 1375); later occurrences (c1813), such as (27), are considered dialectal (Northern): (27) Auld Saunders begoud for to wink Old Saunders began for to wink (A. Wilson, Hogmenae Wks. 295 [Visser 1978: 1375]) In addition to forto complements, -ing complements begin to appear with beginnen during the early 13th century. Unlike a to-infinitival

Lynn Sims

71

complement, the -ing complement marks the aspectual event as [–]perfective. The first examples appear in Layamon’s Brut: (28) he sone bi-com in-to Brutlonde & anan bigon ræuinge he soon came into Britain and immediately began pillaging uppen Basian þene kinge over Basian the king ‘and soon he arrived in Britain and immediately began pillaging Basian’s land’ (Brut, Caligula 5267-5268 [Visser 1973: 1890]) (29) Vmben ane stunde; heo bigunnen striuinge By a little time they began fighting ‘after a little while, they began fighting’ (Brut, Caligula 7765 [Visser 1973: 1890]) In early English, constructions involving a preposition and durative infinitive ending in -ende/-ing are often used to express a continuous event, as we see from the following OE and ME examples, respectively: (30) on feohtende wæron oþ niht on fighting were until night ‘[they] were fighting until night’ (AS Chronicle C, D, E, 871 Thorpe 1861: 138-9 [van Gelderen 2004: 204]) (31) þær he wes an slæting there he was on hunting ‘he was hunting there’ (Brut, Caligula 6139 [van Gelderen 2004: 204]) The OE construction, (30), is considered to involve a copula verb + adjective; whether or not this construction is a precursor to the ModE -ing progressive, with ME -ing, (i.e. (31)) a continuation of (30), is debated (cf. Jespersen 1940: 415). Similar ME constructions (possibly dialectal) also occur with beginnen:

72

Aspectual Loss and Renewal: onginnan, beginnan, start

(32) Quen þai vnderstode þis worde A soruing þai bigan When they understood this word on-sorrowing they began ‘When they understood this word, they began grieving’ (Cursor M. 15518 [Visser 1973: 1890]) In addition to two gan + -ende/-ing examples, Visser (1973) includes ten begin + -ende/-ing examples from c1205 to c1553. In some instances, different versions of the same text (i.e. Brut Caligula vs. Otho) vary between the two affixes. Some of Visser’s (1973: 1890) examples follow: (33) fro þat he bi-gan syttynde in a churche-вerd Amonges oþer men ‘from the time that he began sitting in a church yard with other men’ (Life St. Alexius, Laud MS, 108) (34) Tubal-Cain . . . began first graving in metallis ‘Tubal-Cain first began engraving in metal’ (Capgrave, Chron. [ed. Hingeston] 8) (35) some of them begin already giuing no credence to no man but if it be ‘some of them already stopped giving any credence to any man unless he be’ (St. Th. More, Wks, 687 D 19) As we move from ME to early ModE, there is a “gap in the [-ende/ing] evidence” from 1553 to 1813, according to Visser. This gap is confirmed by examining the Corpus of English Dialogues (Kytö & Walker 2006). In this 1.2 million corpus of dialogues between 1560 and 1769, there is one instance from 1716, given in (36), whereas the corpus contains over 200 instances of begin + a to-infinitival complement. (36) thereupon they immediately began breaking Mr Hurst’s windows (CED-D4WOxford) Two possibilities explain this gap: (a) it is accidental or (b) the -ing complement was dying out. What speaks for (a) is that of the ten begin + -ing/ -ende examples provided by Visser, only example (35) is questionable from a Modern English interpretation (and this may be due to the double negative construction). Thus, the aspectual nuances associated with the ME begin + -ing/-ende examples are no different from the 1716 example,

Lynn Sims

73

the Jane Austen (1813) example below, and the later examples (up to 1962) provided by Visser (1973: 1890). (37) Unable to contain herself, she began scolding one of her daughters (J. Austen, Pride & Prejudice II. 10 [Sims 2008: 162]) To recap the ME and early ModE periods: the context in which gan is used narrows (poetry rather than prose), the form becomes a frozen preterit construction, and its ability to express ingressive aspect declines. The use of begin to mark ingressive aspect increases; however, this form loses its bare complement but gains a for-to complement in addition to an -ende/-ing complement.

4. The Modern English Period: The Rise of start and Decline of begin As discussed above, the to-infinitive remains and an -ing competitor is introduced. This latter complement, however, has a slow start. Below, we will see that ingressive start is introduced and becomes a competitor to begin. An overview of the ingressives and their complements as we leave the early ModE period is illustrated by a 588,820 word sample from the Helsinki Corpus, EMOD 1500–1700 (* indicates present tense endings): (38) begin* to: 39; begin* -ing: 0; began to: 54; began -ing: 0 start* to: 0; start* -ing: 0; started to: 0; started -ing: 0 As the numbers from this corpus show, begin occurs only with a toinfinitive and start as an ingressive aspect marker does not occur. However, the use of start as a lexical verb with the sense of ‘to initiate, begin’ appears as early as 1595 (Sims 2005): (39) Do but start an echo with the clamor of thy drumme. (1595, Shakespeare, John, V.ii.167) (40) He started a discourse of a talk he hears about the town. (1666, Pepy’s Diary, 24 June)

4.1. The Introduction of start The origin of start is not exactly clear and its semantic history is complex. Similar to the *ginnan stem, the semantics inherent to start involve the concept of motion. Start comes from OE styrtan, which, in

74

Aspectual Loss and Renewal: onginnan, beginnan, start

turn, is derived from the Germanic verb *sturtjan, meaning ‘to overthrow, precipitate, overturn’ and (intransitive) ‘to rush, to fall headlong, to gush out’ (OED, s.v. start). Only one occurrence of styrtan is found in OE (Durham Ritual, a1000). Early periphrastic constructions with start do not express ingressive aspect, but they do demonstrate its inherent notion of movement and its grammatical-like function, as we see in (41) and (42): (41) who so is nexte shulde sterte to geete her hoole fro hem ‘whoever is nearest should move to guard her den from them.’ (c1410, Master of Game, xxxiv [Sims 2005]) (42) Atte the dredfulle day he wolle axe acomptes where as there shalle none sterte to yelde ansuere ‘On that dreadful day he will demand account and none shall escape to yield an answer.’ (1450, Knight de la Tour (1868) 113 [Sims 2005]) It appears that during the 1600s the meaning of start is extended and the new meaning expresses a more abstract notion of movement, one of initiating or beginning an action. With the added notion of ‘initiating/ beginning’, start is now available for use in syntactic constructions similar to other ingressives (e.g., begin). In the mid-1800s, we begin to see examples of start + V-ing and start + (in) to-V constructions, as the following examples from Sims (2005) show. (43) I had before this written to Rose how we had best start agitating (1833, J. H. Newman, Letters & Correspondence [OED]) (44) Alexander finally started rehearsing the play in December (1893, Henry James, Guy Domville [Lit. On-Line]) (45) I saw a dog of this kind start to nibble at a flea (1869, Mark Twain, Innocents Abroad, [OTA]) (46) There would be no chance of crossing it [the river] for some days ... even if it started to go down at once. (1891, C. Roberts, Adrift Amer., 181 [OED]) Although start is used as an ingressive marker by a1800, its use is considered ‘colloquial’, ‘vulgar’, and ‘irregular’ by early grammarians (Visser 1978: 1381; 1984: 1895). This attitude toward start may help explain its low frequency rate in texts from the 1800s. This early attitude

Lynn Sims

75

may also be associated with formality levels which seem to be attached to begin vs. start. (My students indicate that they tend to use begin in more formal situations and start in less formal. A continued review of American newspapers suggests that papers tend to mirror this formality pattern: news and economic sections generally use begin, but leisure and sport sections use start.) In addition to formality perceptions, the verb classes to which start/begin belong may play a role in usage patterns: start is a weak verb, marking the preterit and past participle with an -ed ending, while begin is a strong verb, marking the preterit and past participle by an internal vowel change.

4.2. A Phonetically Reduced begin At about the same time start begins to be used as an ingressive marker, a reduced form of begin appears. The gan + to-infinitive construction is said to be rare by the early ModE period, and the last example in Visser (1978: 1576) is from Shakespeare, given in (47). However, it is unclear if this is the reduced form of onginnan or begin: (47) some ... ‘gan to look The way that they did, and to grin like lions (Cymb. V, iii, 35) A search of the Corpus of Historical American English (COHA), a corpus of 400(+) million spoken and written words, shows that gan also occurs in texts from around 1800. Two examples are provided below. I consider these examples to be a reduced form of begin. (48) When round me ‘gan the air to melt (1813, W. Allston, Sylphs Seasons [COHA]) (49) Ho, ho – a feast! – I ‘gan to croak (1848, J. M. Legare, Orta-undis [COHA]) While (48) and (49) may be influenced by poetic meter, this is not the case with similar constructions in the 1893 edition of Voodoo Tales (VT). The stories in VT were collected by the author from original sources in Missouri, and as is evident from the examples below, she attempts to represent the dialect in which the tales were told to her.

76

Aspectual Loss and Renewal: onginnan, beginnan, start

(50) Bimeby dey laigs ‘gin ter trimmle an’ dey eyes ‘gin ter budge (bulge) ‘By-and-by their legs begin to tremble and their eyes begin to bulge’ (VT 1893: 41) (51) Dat am mos’ gin’ly de way, honey, w’en folkses ‘gin ter ‘mire deyse’fs ‘That is most generally the way, honey, when folks begin to admire themselves’ (VT 1893: 81) (52) I ‘gun ter study ‘bout hit in de night, las’ night ‘I began to think about it in the night, last night’ (VT 1893: 120) (53) he feel so gay he ‘gun ter whistle de onles’ chune dat he know ‘he felt so gay he began to whistle the only tune that he knew’ (VT 1893: 124) It might be argued that gin/gun in (50) to (53) represent a phonological feature of certain varieties of American English, specifically the reduction of a weak or unstressed initial syllable (e.g., ‘fraid < afraid). However, examples from COHA, above, and also from the Corpus of Contemporary American English (COCA), (54) below, argue against this conclusion since they do not all reflect non-standard speech. Even though reduced forms are not common in the data, they appear as late as the 1990s: (54) Mitch said with vigor as he gan to move (1993, Talismans Shannara [COCA]) While I consider the ModE gan examples to be a reduced form of begin, its use is not standard and reflects a more informal and/or poetic style. However, this reduced form indicates that begin is susceptible to phonological reduction, further illustrating begin’s grammaticalized status.

4.3. The Decline of begin: Recent Changes in Distribution Patterns As we move towards the end of the 1990s, we begin to see changes in the distribution of begin and also its competitor start. Figures 1 and 2 represent present and preterit tense ingressives found in COHA, a 400(+)

Lynn Sims

77

million word corpus of written texts. Looking at these figures, we see that present tense begin + to infinitive usage (Figure 1) has decreased sharply since the 1990s while preterit began + to infinitive usage (Figure 2) has decreased dramatically since the 1940s. Even though begin/began + toinfinitive shows significant changes in usage, it still appears to be the preferred ingressive marker with this complement. 60 50 40 begin to-V 30 20

begin -ing start to-V start -ing

10

18 1 18 0 4 18 0 60 18 8 19 0 0 19 0 2 19 0 4 19 0 6 19 0 8 19 0 9 20 0 00

0

Figure 1. Changes in present tense ingressive patterns (percent per million, COHA) On the other hand, when the -ing complement is involved, start has replaced begin in the rate of frequency, especially in the present tense. As Figure 1 shows, the use of present and preterit tense start increases with both the to-infinitive and -ing complement, but the -ing complement experiences a dramatic increase, in both tenses, beginning around 1920.

Aspectual Loss and Renewal: onginnan, beginnan, start

78

250 200 began to-V

150

began -ing 100

started to-V started -ing

50

18 1 18 0 40 18 6 18 0 8 19 0 00 19 2 19 0 4 19 0 6 19 0 80 19 9 20 0 00

0

Figure 2. Changes in preterit tense ingressive patterns (percent per million, COHA) Freed (1979: 69ff) demonstrates that begin and start do not share many semantic characteristics, with begin being more semantically restricted than start. Specifically, a person can start to do something (e.g., sneeze) and then not do it, but a person cannot begin to do (e.g., sneeze) something and then not do it. With a to-infinitival complement, it appears, then, that the aspectual focus of start is on the onset of the event but the focus of begin is on the nucleus of the event. With -ing complements, neither ingressive marker allows the subsequent negation of the non-finite event (Freed 1979). Thus, the semantic restrictions associated with begin may influence the introduction of the new ingressive marker start. However, this marker is not a serious competitor to begin until a1920. Structurally, I consider both ModE ingressives to be semi-lexical forms located in v moving to an expanded IP (i.e. M(ood) or A(spect)) to pick up to([+]perfective) or -ing ([–]perfective) and then moving to T(ense). Although both forms have grammaticalized, neither has moved as far along the grammaticalization process as the earlier ingressive onginnan.

5. Concluding Remarks In this paper, I have shown that clear differences emerge between the early ingressives onginnan and beginnan in terms of complement type, impersonal constructions, and non-aspectual functions. These differences

Lynn Sims

79

suggest that onginnan is already more grammaticalized than beginnan during this period. This, in turn, contributes to the decline of onginnan as an ingressive marker and the rise of beginnen as the main ingressive marker during the Middle English period. Although beginnan grammaticalizes, it never reaches the same stage as onginnan – an empty, non-aspectual form. Changes continue during the ME period, mainly in the type of complement, with beginnen losing the bare infinitive but also acquiring the -ing complement. Semantic restrictions associated with begin may have influenced the introduction of the new ingressive marker start during the (early) Modern English period. More recently, distribution patterns suggest that the begin/began + to-infinitive construction is declining while these ingressive markers are holding steady in their frequency rates with the -ing complement. Start/started, on the other hand, continues to gain ground, especially with the -ing complement. How this competition between begin and start will end is, at this point, uncertain.

Works Cited Primary Sources CED = Kytö, Merja & Terry Walker. 2006. Corpus of English Dialogues. Stockholm: Uppsala Universitet. Available online at http://www.engelska.uu.se/corpus.html. COCA = The Corpus of Contemporary English by Mark Davies, 2010. Available online at http://corpus.byu.edu/coca. COE = Healey, Antonette diPaolo & Richard L. Venezky. 1980. A Microfiche Concordance to Old English. Toronto: Pontifical Institute of Medieval Studies. COHA = The Corpus of Historical American English by Mark Davies, 2010. Available online at http://corpus.byu.edu/coha. CSE = Corpus of Spoken Professional American English. Not available online, but see http://www.athel.com. [2 million words; divided in COMM, FACMT, WH] DOE = Dictionary of Old English Corpus. Available online at http://www.doe.utoronto.ca. HC = Helsinki Corpus. Available through the Oxford Text Archive (OTA), and the Penn-Helsinki Parsed Corpus. OED = Oxford English Dictionary. 1933. Oxford: Oxford University Press, and OED Online.

80

Aspectual Loss and Renewal: onginnan, beginnan, start

Secondary Sources Brinton, Laurel. 1981. The Historical Development of Aspectual Periphrases in English. Ph.D. dissertation, University of California, Berkeley. —. 1988. The Development of English Aspectual Systems. Cambridge: Cambridge University Press. Butt, Miriam. 2003. The Light Verb Jungle. Retrieved from http://ling.sprachwiss.uni-konstanz.de/pages/home/butt/main/papers/ harvard-work.pdf. Callaway, Morgan. 1913. The Infinitive in Anglo-Saxon. Washington, D.C.: Carnegie Institution. Cardinaletti, Anna & Giuliana Giusti. 2001. “‘Semi-lexical’ Motion Verbs in Romance and Germanic”. Semi-lexical Categories: On the Content of Function Words and the Function of Content Words, ed. by Norbert Corver & Henk van Riemsdijk, 371-414. Berlin: Mouton de Gruyter. Comrie, Bernard. 1976. Aspect: An Introduction to the Study of Verbal Aspect and Related Problems. Cambridge: Cambridge University Press. De la Cruz, Juan M. 1972. “Transference and Metaphor in Middle English Verbs Accompanied by a Locative Particle”. Orbis 21.114-135. Denison, David. 1990. “Auxiliary + Impersonal in Old English”. Folia Linguistica Historica 9:1.139-166. —. 1993. English Historical Syntax: Verbal Constructions. London & New York: Longman. Elmer, Willy. 1981. Diachronic Grammar: The History of Old and Middle English Subjectless Constructions. Tübingen: Max Niemeyer Verlag. Freed, Alice. 1979. The Semantics of English Aspectual Complementation. Dordrecht: D. Reidel. Funke, Otto. 1922. “Die Fügung ginnen mit dem Infinitive im Mittelenglischen”. Englische Studien 56.1-27. Gelderen, Elly van. 2004. Grammaticalization as Economy. Amsterdam & Philadelphia: John Benjamins. —. 2008. “Where Did Late Merge Go? Grammaticalization as Feature Economy”. Studia Linguistica 62:3.287-300. Gonda, Jan. 1962. The Aspectual Function of the Rgvedic Present and Aorist. The Hague: Mouton. Häusermann, Hans Walter. 1930. Studien zu den Aktionsarten im Frühmittelenglischen. (Wiener Beiträge zur englischen Philologie, 54.) Wien: Braumüller Wilhelm.

Lynn Sims

81

Heine, Bernd, Ulrike Claudi & Friederike Hünnemeyer. 1991. Grammaticalization: A Conceptual Framework. Chicago: University of Chicago Press. Hopper, Paul & Elizabeth Traugott. 1993/2003. Grammaticalization. Cambridge: Cambridge University Press. Hiltunen, Risto. 1983. The Decline of the Prefixes and the Beginnings of the English Phrasal Verb. Turku: Turun Yliopisto. Jespersen, Otto. 1940 [1970]. A Modern English Grammar on Historical Principles, Part V: Syntax. London: Allen & Unwin. Kato, Tomomi. 1974. A Concordance to the Works of Sir Thomas Malory. Tokyo: University of Tokyo Press. Koziol, Herbert. 1932. Grundzüge der Syntax der Mittelenglishen Stabreimdichtungen. (Wiener Beiträge zur englischen Philologie, 58.) Wien: Braumüller Wilhelm. Lichtsinn, Peter. 1913. Der syntaktische Gebrauch des Infinitivs in Layamon’s Brut. Ph.D. dissertation, Königlichen Christian-AlbrechtsUniversität zu Keil. Kiel: H. Fiencke. Los, Bettelou. 2000. “Onginnan/beginnan with Bare and to-Infinitive in Ælfric”. Pathways of Change: Grammaticalization in English ed. by Olga Fischer, 251-274. Amsterdam & Philadelphia: John Benjamins. —. 2005. The Rise of the to-Infinitive. Oxford: Oxford University Press. Mitchell, Bruce. 1985. Old English Grammar I and II. Oxford: Oxford University Press. Mustanoja, Tauno F. 1960. A Middle English Syntax. Helsinki: Société Néophilologique. Nordlander, Johan. 1997. Towards a Semantics of Linguistic Time. Sweden: Umeå University Press. Ogura, Michiko. 1997. “On the Beginning and Development of the begin to Construction”. Studies in Middle English Linguistics ed. by Jacek Fisiak, 403-428. Berlin: Mouton de Gruyter. —. 1998. “The Grammaticalization of beginnen in Medieval English”. Advances in English Historical Linguistics ed. by Jacek Fisiak & Marcin Krygier, 293-314. Berlin: Mouton de Gruyter. Owens, Mary Alicia. 1893. Voodoo Tales. New York: Knickerbocker Press. Sims, Lynn. 2004. “The Function of gan in Layamon’s ‘Brut’”. Unpublished paper. —. 2005. “Arrested Grammaticalization: A Diachronic Analysis of start”. Paper presented at Studies in the History of the English Language (SHEL 4), Arizona, September 2005.

82

Aspectual Loss and Renewal: onginnan, beginnan, start

—. 2008. The Grammaticalization of Ingressive Aspect in Early English. Ph.D. dissertation, Arizona State University. Skeat, Walter W. ed. 1887/1970. The Gospel According to St. Matthew in Anglo-Saxon, Northumbrian, and Old Mercian Version. Darmstadt: Wissenschaftliche Buchgesellschaft. Smith, Carlota. 1997. The Parameter of Aspect. Dordrecht: Kluwer. Terasawa, Yoshio. 1974. “Some Notes on ME gan Periphrasis”. Poetica 1. 89-105. Visser, Fredericus T. 1969. An Historical Syntax of the English Language (Part 3, First Half). Leiden: E. J. Brill. —. 1973. An Historical Syntax of the English Language (Part 3, Second Half). Leiden: E. J. Brill. Warner, Anthony. 1993. English Auxiliaries. Cambridge: Cambridge University Press.

LITTERAL SUBSTITUTION SETS IN MSS OF THE MAN OF LAW’S TALE JUSTYNA ROGOS

1. Modern ‘Letter’ vs. Medieval Littera Impressionistically, spelling systems attested in Middle English manuscripts reflect the scribes’ highly unsystematic attempts at ‘graphemicizing’ the dialectal diversity of the English language of the time. Indeed, from the perspective of a modern reader, accustomed to the conventionalizing force of an external standard, medieval English orthography displays few symptoms of conscious planning. However, this “surface nubbliness”, as Laing & Lass (2003: 257) refer to the apparent chaos of Middle English spelling forms, proves to be quite systematic, once a non-anachronistic approach is adopted, i.e. one which takes into account the differing expectations towards the written text that characterize the modern and Middle English reader. In particular, tools available for synchronic studies of modern orthographies fail to embrace the inherent multivalence of medieval spelling systems. Graphemic theory provides a methodology for approximating but not for communicating the dual nature of the medieval ‘letter’ and it is an artefact of modern taxonomy to present inscribed characters as elements of written language exclusively. In this perspective, the prodigality of Middle English orthographic forms must appear chaotic and confusing to the present-day reader. Notably, contemporary preconceptions about the systematicity (or lack thereof) of the polyvalent Middle English orthographies are a case of anachronistic reductionism, which presents the medieval littera as formally and functionally equivalent to the modern ‘letter’. However, the latter is part of a highly conventionalized and formalized (near-)economical writing system, whose goal is to approximate the ‘one spelling–one sound’ ideal. The post-classical littera, in turn, as laid down in Donatus’ Ars Maior, is a suprasystemic unit, belonging to both the written and the spoken domains. This “smallest unit of articulated sound” (or, “the abstract or superordinate notion of the letter”, cf. Laing & Lass 2003: 258)

84

Litteral Substitution Sets in MSS of The Man of Law’s Tale

is a tripartite entity, characterized by nomen (name), figura (shape) and potestas (sound value). The three structural elements point to the ‘transsystemic’ properties of the littera, which allow “multivocal relationships in the mapping of sound to symbol” (Laing & Lass 2009: 2) in Middle English spelling systems. This multivalence is manifested on two levels: first, a number of litterae (realized by variable figurae) can occur in variation for the same potestas, or, inversely, one littera can map multiple phonic values. The former type of relationship generates Litteral Substitution Sets (LSS) and the latter is responsible for Potestatic Substitution Sets (PSS) (cf. Laing 1999; Laing & Lass 2003, 2009). Laing and Lass (2009) illustrate the two types of scribal prodigality evoking the practice of hand D of MS Cambridge, Trinity College B.14.39 (323). Thus, the scribe’s use of ܌‬for the sound values [h], [x], [j], [w] and [ܵ] is an example of a Potestatic Substitution Set, whilst spellings like , attested for the reflexes of Old English , comprise one Litteral Substitution Set (Laing & Lass 2009: 2). A distinct advantage of this framework is that it allows for ‘unpacking’ and systematizing the network of relations between manuscript symbols and their vocal referents as well as accounting for apparent inconsistencies and unpredictable spelling-to-sound mappings. It also provides context for a systematic comparison of orthographic forms attested in various witnesses of polytextual works, which can contribute to more accurate reconstructions of the transmission history of such works.

2. The Analyzed Corpus An illustrative example of a polyvalent work is constituted by manuscripts of Chaucer’s Canterbury Tales, which are characterized by both complex textual history and complex orthographies. The extant eighty-plus copies of the text are witnesses to diatopically and diachronically differing linguistic contexts in which they were produced and both types of variation are reflected in the orthographic tissue of the manuscripts. The present paper takes a closer look at a snapshot of this inherently intricate network of cross-referencing between the spoken and written systems of Middle English, derived from a group of manuscripts of The Man of Law’s Tale (henceforth MLT). The following discussion presents fragmentary results of a pilot study for a larger project, whose goal is a comparative analysis of orthographic systems in a group of genetically related manuscripts of one of Chaucer’s Canterbury Tales. The basis for the present discussion is constituted by seven early- to late-15th

Justyna Rogos

85

century manuscripts, representative of two textual traditions of the Canterbury Tales (cf. Manly & Rickert 1940). Table 1. The corpus MS Fitzwilliam McClean Harley 1239 Manchester English 113 Petworth Royal 18 C II Egerton 2863 Lichfield

sigil Fi

family a

date 1450–1469

Ha1 Ma

a a

1450–1470 1483–1485

Pw Ry2 En2 Lc

d d d d

1420–1430 1420–1450 1430–1450 1430–1450

3. Research Methodology The following study is based on the analysis of diplomatic transcriptions of manuscript facsimiles, prepared for The Man of Law’s Tale Project.1 The relevant forms from the transcripts were subsequently checked against the facsimiles. The manuscripts were analyzed in terms of the distribution of three orthographic features, later systematized in the form of Litteral Substitution Sets, i.e. “a set of litterae in variation for the same potestas” (Laing & Lass 2009: 2). In the focus of the present study are litteral realizations of the potestates /‫ݕ‬/, /x/ and /ș~ð/ and their implications for the prodigality of the scribes of the MLT manuscripts. Although it is not possible for such a fragmentary study of manuscript contents to yield an exhaustive diagnosis of the spelling habits of the scribe, yet operating within the closed unit of a tale allows for observing tendencies and patterns that surface in the continuous process of copying from – in most cases – a single exemplar. Moreover, thus restricted orthographic repositories are more manageable for a systematic comparison of litteral and potestatic attributions in a larger group of manuscripts, which in turn allows for evaluating the scope of reference, or prodigality, of manuscript characters. 1

The Man of Law’s Tale Project (2007–2010) was a research initiative of the Department of the History of English at the School of English of Adam Mickiewicz University in PoznaĔ (Poland). Its aim was to prepare an electronic edition of all extant manuscripts of Chaucer’s The Man of Law’s Tale.

86

Litteral Substitution Sets in MSS of The Man of Law’s Tale

4. Litteral Substitution Sets for /‫ݕ‬/ Litteral representations of the /‫ݕ‬/-sound in the analyzed manuscripts fall into two groups, depending on the grammatical category of the lexeme in which they were attested. On closer examination of relevant data, there seem to exist two distinct sets of mappings for the potestas in question: one for lexical words and another for function words. Insofar as in the latter lexical set the potestas /‫ݕ‬/ is realized by either of two litteral substitutions (see below), lexemes belonging to the latter group can be said to have their ‘islands of prodigality’ as far as the spelling for /‫ݕ‬/ is concerned. This potential for orthographic variability is evident from the example of eight nonce-occurring items (that is, items that occur only once in each examined manuscript), which in most cases encode the potestas /‫ݕ‬/ with the litterae . Accordingly, the following forms have been attested in the corpus: (1) abasshed Bysshop(es) Englissh Fisshes mannyssh venquysshed wasshe wysshe There are, however, exceptions to this general pattern. Notably, in MS Fi (a later ‘a’-family manuscript) the /‫ݕ‬/-sound is normally rendered with and, on one occasion, with , returning a LSS : (2) abasscheѐ Bysschopթ Englyssch Fyssches wassch wyssche venquyssed The scribe of another ‘a’-manuscript, Ha1, invokes a three-element LSS in the same orthographic contexts, i.e. :

Justyna Rogos

87

(3) abasched Bychopis Fych wasch wych (WASH) vnquissed As for the group of function /‫ݕ‬/-words, their orthographic forms fall into two categories, which cannot be said to coincide with the genetic affiliation of the manuscripts, although group ‘a’-manuscripts are more likely to spell words like SHE, SHAL, SHOLDE, with , whereas in ‘d’-manuscripts (with the notable exception of MS En2) the same context triggers the -spelling. All in all, taking into account both the predominant and incidental spellings, the following LSS for /‫ݕ‬/ have been attested in the manuscripts under consideration: (4) MS Ma, Lc, En2, Ry2: MS Ha1: MS Fi: MS Pw: It is worth pointing out that the above-presented groupings do not coincide with genetic affiliations, as in the first example, a group of ‘d’manuscripts is joined by an ‘a’-manuscript (MS Ma), whereas the remaining ‘a’-MSS as well as the earliest ‘d’-manuscript (MS Pw) constitute categories of their own.

5. Litteral Substitution Sets for /x/ The /x/-sound is rendered by either of the two mutually substitutive litteral representations: or ܌‬, as in doughter vs. douвtere (although in MS Ry2 the latter symbol appears only twice, both times in a spelling variant of nought and right respectively). In manuscripts Ma, En2 and Ha1, however, where the spelling is otherwise predominant, the scribes on a few occasions deviate towards fairly idiosyncratic choices of litterae. Thus, MS Ma has forms like cauth, tauth, thowth and wrowth/wrouth in variation with caught, taught, thought and wrought respectively. This type of litteral substitution is shared by yet another group ‘a’-manuscript, Ha1, which returns forms like fythe (where Ma has fight), kauth (caught in

88

Litteral Substitution Sets in MSS of The Man of Law’s Tale

Ma), myth (for might) and owth (for ought). Ha1 also substitutes with zero, as in myty or ryt for mighty and right respectively. For En2 – a ‘d’-manuscript – litteral substitution for the potestas /x/ is much more incidental and limited to a mere three examples: lowh, sawh and slowh, where other group ‘d’-manuscripts have low(e), saugh (MSS Ry2, Lc) or soghe (Pw) and slough (Lc), sclawe (Pw) or slow (Ry2) respectively. Interestingly enough, only two ‘d’-manuscripts group together in terms of sharing a LSS for /x/. The substitution sets for particular manuscripts can be summarized in the following: (5) MS Ma: MS Ha1: MS Fi: MS Lc, Pw: ܌‬ MS En2:

6. Litteral Substitution Sets for /ș~ð/ The litterae and quite predictably render the /ș~ð/ sounds. Similarly to LSS for /‫ݕ‬/, yet again MS evidence implies the differentiation in mapping the potestates /ș~ð/ in function words, as opposed to lexical words. Thus, for function words like THE, THAT, THIS, THOU, MLT manuscripts alternate between the more frequent - and the rarer forms (except MS Fi where -spellings predominate). With lexical words, however, the opposite scenario is the case: only in MS Pw are nearly all /ș~ð/ contexts rendered by means of the littera . In most relevant cases in the remaining MSS either the -forms are prevailing or some -spelled words appear alongside their -spelled equivalents, e.g.: (6) boþe ~ bothe (MSS Lc, Ma) comeþ ~ cometh (MS Lc) faþer ~ fader (MS Ma) haþ ~ hath (MS Ry2) moþer ~ modere (MS Ma) oþer ~ other (MSS Ry2, Ma) raþer ~ rathir (MS Ma) þanketh ~ thonketh (MS Lc) þennes ~ thennes (MSS Lc, Ha1) þerfore ~ therfore (MS Fi)

Justyna Rogos

89

þider ~ thider (MS Lc) þilk ~ thilke (MSS En2, Ry2) þinges ~ thinges (MS Ry2) þousande ~ thousand (MS Ry2) þre ~ thre (MS Ry2) This summary of the potestas-to-littera projections presents a fairly static model of ‘litteral substitution pairs’, whose distribution is largely a matter of free variation. In effect, it appears that there is little potential for prodigality in mapping the /ș~ð/ sounds in the analyzed corpus. In one manuscript, however, this stasis is interrupted by a littera which is a rather unusual choice for encoding the sound under consideration. On two occasions MS Ha1 extends the mapping of ܌‬beyond what other MSS alternatively encode as or . Thus, in line 43, which in other MSS reads , Ha1 has , and from line 536 is spelled ܌‬. Obviously, two isolated instances of a rare spelling is too weak a piece of evidence to venture any generalizations. However, even such idiosyncratic spellings are potentially indicative of more general tendencies and show the possible extension of the limits of prodigality in MLT manuscripts.

7. Conclusions Despite the limited range of the above-presented pilot study, it appears that at least two tentative conclusions can be formulated on the basis of the preceding discussion. To begin with, some potestates trigger multiple litteral representations, with the /‫ݕ‬/-sound generating the most prodigal substitution set and the /x/-sound following suit (). Importantly, not all members of LSS are statistically significant, i.e. it is usually the case that no more than two litteral substitutions are part of the scribes’ active repertoire in the spelling of particular forms and scribes only sporadically select from a larger pool of orthographic variants. This is perhaps what should be expected in 15thcentury manuscripts, particularly in those from the latter part of that period: although late Middle English spelling systems were still tolerant of a considerable variability (also on the intradialectal level), the standardizing forces of supraregionalization start encroaching upon the written output of later medieval English. In effect, with the choice of available spellings thus restricted, some litterae appear to be ‘more multivalent’ than others, as it were. That is, certain litterae tend to recur as members of substitution sets for different potestates. For instance, ܌‬is a member of LSS for /x, ș/, as well as for /j/ and /g/ (e.g. a вeyn as a variant of ageyn in Lc).

90

Litteral Substitution Sets in MSS of The Man of Law’s Tale

Likewise, , apart from substituting for in the rendition of /ș~ð/, is also applicable as a representation of /x/ (in MS Ma). Overall, however, Litteral Substitution Sets attested in the manuscripts under consideration point to a rather restricted prodigality of the MLT scribes, who clearly operate within the confines of conventionalized spellings, admitting of little variability. With such a limited set of litterae to choose from, it is to be expected that those litterae will constitute multivalent phonic signifiers. Therefore, it seems that a perspective complementary to the framework outlined in the present discussion, i.e. a study of potestatic substitution sets attested in the MLT manuscripts, is necessary for a comprehensive description of the spelling systems in the 15th-century manuscripts of the Canterbury Tales.

Works Cited Primary Sources British Library MS Egerton 2863 (En2) British Library MS Harley 1239 (Ha1) British Library MS Royal 18 C.II (Ry2) Fitzwilliam Museum MS McClean 181 (Fi) John Rylands Library MS English 131 (Ma) Lichfield Cathedral Library MS 29 (Lc) Petworth House MS 7 (Pw)

Secondary Sources Laing, Margaret. 1999. “Confusion wrs Confounded: Litteral Substitution Sets in Early Middle English Writing Systems”. Neuphilologische Mitteilungen 100.251-270. Laing, Margaret & Roger Lass. 2003. “Tales of the 1001 Nists: the Phonological Implications of Litteral Substitution Sets in Some Thirteenth-Century South-West Midland Texts”. English Language and Linguistics 7.257-278. —. 2009. “Shape-Shifting, Sound Change and the Genesis of Prodigal Writing Systems”. English Language and Linguistics 13.1-31. Manly, John M. & Edith Rickert. 1940. The Text of the Canterbury Tales: Studied on the Basis of All Known Manuscripts. Vol. 2: Classification of the Manuscripts. Chicago: University of Chicago Press.

PART II SYNTACTIC AND SEMANTIC VARIATION AND CHANGE

SYNTACTIC AND SEMANTIC VARIATION: MIDDLE ENGLISH CAUSATIVES AND COMPLEMENT TYPES BRIAN LOWREY

1. Introduction My intention in this paper is to examine Middle English (henceforth ME) analytical causative constructions. To begin with, I shall be looking at the individual causative verbs and the principal complement types, showing how they interact to construct causative meaning, and which semantic factors influence the choice of verb and of complement. Previous attempts have been made, of course, to explain the distribution of ME complement types, on essentially semantic grounds. They have typically met with only partial success, however, as I shall demonstrate here. I shall argue that this relative failure can be attributed to the fact that insufficient account has been taken up to now of variation. We know, since Weinreich, Labov & Herzog (1968), that variation is an inherent property of natural language, not only at the level of the speech community at large but also at the level of the individual speaker. Speakers, when confronted with a choice between two or more variables, do not display systematically uniform behaviour, and another of my objectives here will be to stress the need to allow for variation of this type when dealing with phenomena such as complementation patterns. To do so, I shall call upon the notion of ‘variational space’ as defined by Smith (1996: 44-47). Of course, it would be impossible, from our modern vantage point, to measure the degree of variation in the everyday speech of an individual ME speaker. What we can do, however, is measure variation across individual ME texts. For this purpose, I shall use a restricted corpus here, made up of a small number of relatively substantial 14th- or early 15thcentury works: Gower’s Confessio Amantis (henceforth CA), prologue and liber I-V (over 55,000 words), Chaucer’s Canterbury Tales (CT), with the exception of the prose tales (over 140,000 words), a selection of Chaucer’s prose works (CP), comprising the Parson’s Tale, the Tale of Melibee,

Brian Lowrey

93

Boece, and the Treatise on the Astrolabe (over 127,000 words), the ‘B’ text of Langland’s Piers Plowman (PP, 72,000 words), the York Mystery Plays (YP, over 90,000 words), and a selection of Richard Rolle’s works (RR), made up of The Form of Living, Ego Dormio, The Bee and the Stork, The Commandment, Desyre and Delit, The English Psalter, Gastly Gladnesse, and the Meditations (67,000 words). For each individual corpus, every occurrence of each analytical causative has been recorded, along with the complement type, and the thematic role assigned to both causer and causee. These findings are presented in section 3.3. Given that ME displays considerable dialectal diversity, there is, as one might expect, some degree of regional variation which also needs to be taken into account. I have therefore chosen to examine the distribution of causatives across texts from three different dialect areas: the North (RR and YP), the (West) Midlands (PP), and the South (CA, CP, and CT). Variation, of course, is also considered to be one of the principal causes of language change, and any account of ME causatives that failed to mention the changes which affect the causative sub-system between the years 1150 and 1500 would necessarily be incomplete. I shall therefore summarise some of the most important innovations here, from a broadly variationist perspective. In particular, I shall show that the very same factors which govern the choice of causative verb and of complement type also seem to drive many of the major changes that take place in ME.

2. Middle English Causative Structures I shall begin by presenting the linguistic tools at the disposal of ME speakers to express causation.

2.1. ME Causative Verbs For the purposes of the discussion here, I shall restrict the causative group to ‘implicative’ verbs, to borrow a term from Karttunen (1971). I shall consider to be implicative those causatives that fit the definition given by Shibatani (1976: 1-2): “The truth of the caused event holds whenever that of the causing event holds”. This parameter separates verbs like make and cause in Present-Day English (PDE) from other, manipulative verbs, such as order or ask, as the contrast between (i)/(ii) and (iii)/(iv) shows: (i)

*I made him open the door, but he didn’t do it

94

Syntactic and Semantic Variation

(ii) *I caused him to open the door, but he didn’t do it (iii) I asked him to open the door, but he didn’t do it (iv) I ordered him to open the door, but he didn’t do it To say that someone made or caused an event to take place necessarily implies that the event did indeed take place. This is an entailment, and cannot therefore be cancelled, as (i) and (ii) show. Verbs like order or ask can of course be used implicatively, as in: (v) I ordered/asked him to open the door and walked out of the room Here, the context makes it clear that the event described by the complement actually took place. This is a pragmatic inference, however, rather than an entailment, and can as such be cancelled, as in (iii) and (iv). Applying the same parameter to ME, we can identify do, gar, let, and make as the implicative causatives, while cause joins this group towards the end of the ME period. I shall not be concerned here with so-called ‘semi-causatives’ such as haten or bidden/beden which, like their PDE counterparts, fail the implicativity test.

2.2. ME Causative Complements Several verbs, then, were available to ME speakers to express causation, and these verbs could combine with a variety of complement constructions. Most of the verbs listed above appear at some stage, at least, in ME with most of the complement types listed below, although distribution varies considerably, both diatopically and diachronically: VOSI1: (1) For ther ben manye felle frekes my feeris to aspie- / Bothe Priour and Suppriour and oure Pater Abbus; / And if I telle any tales, thei taken hem togideres, / And doon me faste Frydayes to breed and to watre; ‘For there are many cruel men, to spy out my companions, both Prior and Sub-prior and our Father Abbot, and if I tell any tales,

1

VOSI: For ‘Verb + Object or Subject + Infinitive’, a term first coined by Visser (1963-73: §2055) to allow for uncertainty as to the status of the second NP.

Brian Lowrey

95

they take them all together and make me fast on Fridays, on just bread and water;’ (Langland, Piers Plowman)2 VOSI[to]: (2) fro a litil errour out of þis weye . may a man come bi grace aвen / but myche errour wiþ errour of wit: makiþ man to growe in more errour ‘from a little error, astray from this path, a man may with [God’s] grace come back, but great error, with error of understanding, makes one grow still more in error’ (Wyclif, De Officio Pastorali) VOSI[for to]: (3) For it was wynter time tho, / And wynter, as be weie of kinde / Which stormy is, as men it finde, / Ferst makth the wyndes forto blowe, ‘For it was winter then, and winter, which by its very nature is stormy, as men find, first makes the wind blow,’ (Gower, Confessio Amantis) V+ÞAT (4) Of Accidie comth first that a man is anoyed and encombred for to doon any goodnesse, and maketh that God hath abhomynacion of swich Accidie, as seith Seint John. ‘Of sloth comes first that a man is unwilling and slow to do anything good, and makes (that) God abhor(s) such sloth, as saint John says.’ (Chaucer, Parson’s Tale) V+I3 (5) This Emperour leet enclose the chirche of seynt Sepulcre + walle it within the cytee, þat before was withoute the cytee long tyme beforn

2

All the translations given here are those of the author, who obviously accepts full responsibility for any inaccuracies or errors. 3 V+I: for ‘Verb + Infinitive’ (term borrowed from Denison 1993: 165).

96

Syntactic and Semantic Variation

‘This emperor had the church of St Sepulchre enclosed and walled within the city, (a church) that was outside the city wall for a long time beforehand’ (Mandeville’s Travels) VOSI corresponds to the superficial structure found with active make in PDE, with a ‘plain’ or ‘unmarked’ infinitive, and VOSI[to] to that used in Modern English with cause and get. A similar construction in which for to precedes the infinitive, as in (3), becomes relatively frequent during the 14th century. I shall follow Fischer (1995: 2-3) here in treating VOSI[for to] as a variant of VOSI[to]. Example (4) is included to show that phrasal complements introduced by that, common in Old English (OE) with certain causatives (Lowrey 2010: 70-71), still occur in ME, albeit rarely (Lowrey 2002: 177-183). This is the case both in our restricted corpus here, as well as in the much more extensive sample of ME texts used in Lowrey (2002), which contains but a handful of examples, chiefly involving do and make. The structure illustrated by (5), also inherited from OE, in which the causee is left unexpressed, remains frequent throughout the ME period, particularly in verse texts but also in prose works such as Mandeville’s Travels or Thomas Malory’s late 15th-century Arthurian cycle.

3. Factors Governing the Choice of Verb and Complement My aim here will be to determine which factors, if any, govern the selection of individual causative verbs and their complement structures. As far as the choice of verb is concerned, the considerable variation that one observes can be at least partially explained on dialectal or diatopical grounds. As Mustanoja (1960: 601) points out, gar is commonly used as a causative in texts of Northern origin, while do is frequent in the East, and let is the most widely used causative in the West and the South. Stylistic or artistic considerations also have a part to play, especially in verse texts. For example, it is the 14th-century renaissance of the OE alliterative poetic tradition that seems to explain the appearance of gar in (6), rather than do or make, the most frequent causatives in Langland’s verse: (6) Grammer for girles I garte first write, / And bette hem with a baleys but if thei wolde lerne. ‘A grammar for the young I first had written, and beat them with a stick if they would not learn.’ (Langland, Piers Plowman)

Brian Lowrey

97

In similar fashion, the appearance or otherwise of (for) to before the infinitive in a causative complement can sometimes be explained on purely metrical grounds, the gain or the loss of a syllable providing the poet with a readily available device for balancing his verse. But beyond such artistic considerations, to what extent is it possible to explain variation in lexical and grammatical choices? I do not think that the choice of one verb or complement type over another can be seen a case of free variation. Certain semantic parameters determine, at least to an extent, which complement and which verb will be selected in a given situation. It is not possible, however, to explain each individual choice by rigidly applying such semantic criteria. I shall demonstrate that variation, an essential element in the linguistic system and an important source of linguistic evolution, also needs to be taken into account. In the following discussion, I shall use the term ‘causer’ to designate the subject of the causative verb, and ‘causee’ to refer to the subject of the infinitive. I shall assume that the causer receives the thematic role of Agent in the case of volitional causation, and Cause when this is not the case.

3.1. The V+I Construction 3.1.1. Choice of Causative I shall start by looking at which verbs appear in the V+I construction. A first example with let was presented in (5), and another is given in (7): (7) Arður asechen lette; alle þa riche / kinges and eorles; and þa riccheste beornes. / þa i þan fehte weoren i-slaвen; and idon of lif-dæвen / he lette heom burien; mid baldere pruten. ‘Arthur had all the rich kings and earls, and the richest men that were slain in the battle, and deprived of life, sought out. He had them buried with due honours.’ (La‫܌‬amon, Brut) Let V+I occurs frequently in Western and Southern texts. The same structure, however, appears with other verbs in different dialects: gar is the usual Northern form, while do is common in the East. In each case it is the verb most closely associated with agentive causation that is found in V+I. The choice of construction can sometimes determine the meaning to be assigned to the verb: let is used as a causative almost exclusively in V+I. Let VOSI, apart from a limited number of occurrences in early ME,

Syntactic and Semantic Variation

98

nearly always expresses the so-called ‘permissive’ sense, or perhaps more accurately the “cessation of impingement” to borrow a term from Talmy (1985: 301).4 3.1.2. The Semantic Profile of V+I The syntax of subjectless complement clauses has received some attention in the literature, and various explanations have been put forward, whose respective merits I shall not debate here. The reader is referred to the discussion in Fischer (1990: 257), Roberts (1993: 286-292), or Van Gelderen (2000: 148), for a variety of perspectives on this question. Instead, I shall focus on the semantic characteristics of causative V+I. Indeed, V+I structures differ from other types of infinitive complement in that they express, in the great majority of instances, a certain type of causation, whose chief characteristics can be summarised as follows: -

-

the infinitive is generally transitive; the causer is an Agent; the causative verb appears most often in the preterit or the imperative, and denotes a single action; the non-expressed causee has the features +AGENT, +HUMAN; the complement denotes an ‘accomplishment’ according to Dowty’s classification (1979: 52-70): it is usually a ‘telic predicate’ as defined by Hopper & Thompson (1980: 251): “an action viewed from its endpoint”; the focus is on what happens to the object of the infinitive, rather than the role played by the implicit causee.

These constructions can often be glossed by have + past participle in PDE, as the proposed translations of (5), (6) and (7) suggest. Indeed, it is only following the loss of V+I in the first part of the 16th century that have + past participle starts to become a frequent causative construction in English (Lowrey 2002: 465-485). Although some exceptions can be found, mostly with make in early ME and with cause in late 14th-century Southern texts, most occurrences of V+I seem to fit this profile.

4

I disregard the grammaticalised, optative let VOSI construction here.

Brian Lowrey

99

3.2. VOSI and VOSI[to] Constructions 3.2.1. Choice of Causative VOSI and VOSI[to] occur with each of the causative verbs listed in 2. In the texts to which I refer here, beyond the broad dialectal distribution suggested by Mustanoja, gar, do, and let find themselves either in competition or in complementary distribution with make within each dialect area. The latter verb seems to be present in all ME dialects. In a more or less Midland text such as PP, for example, the verb do shares the task of expressing causation with make, whereas in a Northern text such as YP gar is used alongside make. In both cases, the choice of causative is at least partially motivated. Two features, especially, seem to be relevant. Firstly, as a glance at Tables 2 and 6 (see section 3.3. below) will show, an agentive causer makes gar or do the more likely choice in a VOSI-type construction, with or without to, than make. The subject of do is an Agent in 25 cases out of 30 in PP (83.3%), while make has an agentive subject in only 16 cases out of 29 (55.1%) in the same text. A similar pattern emerges in YP, where the causer is an Agent in 33 cases out of 39 (84.6%) when gar is used, but in only 16 out of 35 (45.7%) occurrences of make. Agentive causation, then, is more readily expressed by do or gar VOSI or VOSI[to], according to the dialect, than by make VOSI or VOSI[to]. As for the second factor, a close study of our corpus shows that, in early texts especially, make is often associated with generic contexts, as in (8): (8) For idelnesse is luðer; on ælchere þeode / for idelnesse makeð mon; his mon-scipe leose / ydel-nesse makeð cnihte; forleosen his irihte / idelnesse græiðeð; feole uuele craften. ‘For idleness is hateful to all peoples, for idleness makes a man lose his manhood; idleness makes a knight neglect his duty, idleness prepares the way for many evil deeds.’ (La‫܌‬amon ,Brut) The Brut is a lengthy work, consisting (in its longest version) of more than 16,000 lines (over 128,000 words), the majority of which relate scenes of combat and the exploits of Arthur and his knights. Occasionally, however, the author expresses philosophical or moral sentiments of a more general nature. In these instances, it is not the usual let V+I which appears, but make VOSI.

100

Syntactic and Semantic Variation

A definite tendency, therefore, is beginning to emerge. One cannot claim that do and gar are used exclusively in VOSI/VOSI[to] with agentive causers, and make with non-agentive ones. This would simply be too strong. It is possible, however, to affirm that do or gar are more likely to appear in specific contexts, or with an agentive causer, while make is more common in generic contexts, or in the presence of a non-agentive causer. 3.2.2. Choice of Complement Type: VOSI or VOSI[to]? The alternation between infinitival complements with and without to has already given rise to much debate. Previous work on the subject has often focused on the syntactic or distributional properties of bare and toinfinitival complements, such as Svartvik & Quirk (1970), or Warner (1975), while Warner (1982: 123) suggests that the choice of the form with or without to is often “a result of the lexical preference exercised by the matrix verb”. Another, more semantically-based explanation for the alternation, which often resurfaces under different forms, is suggested by Bock (1931: 220): the idea that there exists a closer link between the higher verb and its complement in the absence of to. Where the causatives are concerned, this often translates into the idea of ‘direct’ or ‘indirect’ causation, an idea taken up in two papers by Fischer (1995, 1996), devoted to infinitive complements in late and early ME respectively. Of course, one of the problems posed by this kind of approach is that the very notion of what constitutes ‘direct’ causation proves notoriously difficult to define (see for example Baron 1974: 333). Fischer rejects the idea that the alternation of infinitive complements with or without to constitutes a case of free variation, or that the choice of complement is lexically determined by the higher verb. Her approach is probably the most complete of its type, and particularly relevant to our discussion here, in that she attempts to define (in)direct causation, and proposes a number of semantic criteria likely, in her view, to favour the use of either VOSI or VOSI[to] complement clauses. I have reproduced here those criteria listed by Fischer (1995: 7-18) which appear relevant as far as the causatives are concerned. Each of the following is considered likely to favour the appearance of to: I. II.

Non-simultaneity of tense domains. Indirect causation: a) subject of matrix verb does not directly cause what is expressed in the infinitive clause. b) the subject/causer is inanimate and as such more of an instrument than a cause.

Brian Lowrey

101

c) what is caused is a process in which the causee himself takes/must III. IV.

take an active part: (he is) animate and represented as ‘highly volitional’. Non-actuality. Passive constructions.

Figure 1. Factors likely to favour the choice of VOSI[to] over VOSI, based on Fischer (1995) I agree with Fischer that the alternation between VOSI and VOSI[to] is at least partially determined by semantic factors. Most of the factors mentioned here do indeed prove to be relevant at some stage, at least, during the ME period. Only IIc seems to be genuinely problematic, as I shall show in 3.3.2. below. Care needs to be taken, however, when applying criteria such as those listed above, and the kind of explanation suggested in Fischer (1995) for the distribution of causative complements with or without to seems incomplete. The claim appears to be that the VOSI/VOSI[to] alternation can be explained on each occasion by means of the criteria mentioned in Figure 1. But such an approach rapidly runs into a number of problems, some of which I shall mention here. The first would have to be the ease with which counter-examples can be found for each criterion. For instance, Fischer (1995: 11) quotes the following example, from CT, to illustrate criterion IIa (my translation): (9) 2507: Sire, wolde he seyn, an emperour moot nede / 2508: Be vertuous and hate tirannye - / 2509: For which he in a bath made hym to blede / 2510: On bothe his armes, til he moste dye. ‘Sire, he would say, an emperor must of necessity be virtuous and hate tyranny, for which he [Nero] made him [Seneca] bleed, in a bath, from both his arms, until death.’ (Chaucer, Monk’s Tale) Fischer calls this “a very clear example”, claiming that to is present “because we know indeed that Nero did not kill Seneca with his bare hands, but pressed him into committing suicide” (1995: 11). Some kind of intermediate factor intervenes between the causer and the result. At first sight, this all seems highly plausible. However, if we accept this explanation, how do we account for the absence of to in example (10)?

102

Syntactic and Semantic Variation

(10) 2511: This nero hadde eek of acustumaunce / 2512: In youthe agayns his maister for to ryse, / 2513: Which afterward hym thoughte a greet grevaunce; / 2514: Therefore he made hym dyen in this wise. ‘This Nero had also in youth the custom of rising up against his master, which afterwards seemed to him to be a source of great grievance, therefore he made him die in this manner.’ (Chaucer, Monk’s Tale) I have left the line numbers (according to Benson’s 1987 edition) in place to show just how close the two examples are in the original text, and that (10) is the direct continuation of (9). The referents of the pronouns he and hym (Nero and Seneca, respectively) are identical in both cases, and both causative structures describe the same ‘real world’ event. Crucially, the relationship between the protagonists, the factor which, according to Fischer (1995) determines the choice of VOSI[to] in (9), is the same in (10). This time however, plain VOSI is chosen. Unfortunately, the other criteria also yield a number of apparently inexplicable counter-examples. It would obviously be impossible to quote them all here: the reader is referred to Lowrey (2002: 327-59) for a more complete sample. I shall give just one further contrast, taken from CA, in which VOSI[for to] alternates with VOSI: (11) The Soldan in gret privete / A goldring of his dowhter tok, / And made hire swere upon a bok (…) / That sche schal thilke man obeie ‘The sultan, secretly, took one of his daughter’s gold rings, and made her swear upon a holy book (...) that she would obey this man’ (Gower, Confessio Amantis) (12) Bot ferst sche made hem forto swere / That thei the children scholden bere / Unto the See, ‘But first she made them swear that they would carry the children to the sea,’ (Gower, Confessio Amantis) Here again, one might want to justify the choice of bare VOSI in (11) using criteria IIa and IIb in Figure 1: the subject of the matrix cause is the

Brian Lowrey

103

direct cause of the event expressed in the infinitive clause, and is also an animate entity, factors which ought to favour the absence of to before the infinitive. But in this case, one needs to explain why the same reasoning does not apply in the case of (12). In both instances, the relationship between the causer and the causee appears to be of a very similar nature: both imply that the causer exerts some degree of coercion over the causee. Obviously, other factors are probably important here, such as rhythm and meter, the kind of artistic considerations alluded to in 3. But even this would indicate that whatever semantic factors may oppose marked and unmarked infinitive complements, they are not strong enough to override such considerations. Clearly, the suggested semantic opposition does not tell the whole story. It fails to explain all cases of alternation, and cannot therefore be as clear-cut as Fischer (1995) seems to suggest. Secondly, the kind of analysis suggested here offers a largely static vision of ME complementation patterns. After all, if the alternation VOSI/VOSI[to] did serve to express such clearly defined semantic parameters, it is unclear why any change would have taken place. And yet changes have obviously occurred. Active make no longer accepts VOSI[to] in PDE (except in exceptional circumstances: see Lowrey 2002: 452-60). One might even expect any change to lead to a more systematic marking of semantic differences, such as the opposition between let VOSI, which expresses the ‘permissive’ sense and let VOSI[to], which almost always expresses the idea of ‘preventing’ or ‘hindering’ in ME, as in (13): (13) "That is sooth," seide Scripture; "may no synne lette / Mercy al to amende" ‘That is true,’ said Scripture, ‘may no sin prevent Mercy from repairing everything’ (Langland, Piers Plowman) A further objection concerns the fact that, ultimately, such criteria inevitably come into conflict. For example, Fischer (1996: 253-54) identifies a further set of criteria, based upon the factors used by Hopper & Thompson (1980: 252) to define transitivity. Fischer finds that some of the elements linked by Hopper & Thompson to low transitivity can also be associated with indirect causation and the appearance of VOSI[to]. She claims that the following elements in particular are likely to trigger the appearance of to before the infinitive: - Generic contexts - Non-volitional causers

104

Syntactic and Semantic Variation

Once again, I agree, broadly, with Fischer. These parameters do indeed seem to explain, at least in part, the distribution of causative VOSI and VOSI[to] complements. The criteria she identifies are particularly interesting, given that the very same elements play an important role, as we have seen, in the choice of causative verb, confirming that they are indeed likely to be relevant to the choice of complement and to the expression of causation in general. We shall see in section 5 that they also influence many of the changes which take place within the ME causative sub-system. However, consistently adding further semantic criteria in this way not only tends to weaken the explanatory potential of the theory, it also increases the risk of the criteria coming into conflict. This becomes clear from the following example (borrowed from Fischer 1996: 262): (14) hwet is win in wunde; Win makeð wunde smerte. Ah þe smertinge clenseð þe wunde; ... Al so hali schrift bið in ure wunde hwan we scule festen ‘What is wine in a wound; wine makes the wound smart. But the smarting cleans the wound, just as holy confession is in our wounds when we shall feast(?)’ (Lambeth Homilies) Given the generic context and the non-volitional causer, Fischer’s criteria clearly predict VOSI[to] here. And yet it is VOSI which appears in the text. Fischer tries to explain the phenomenon by applying, apparently, ‘in reverse’ her IIc, from Fischer (1995): “[t]he object of what is caused is also inanimate, so that it cannot play an active part, which makes the bare infinitive again the logical choice” (Fischer 1996: 262). And yet there seems to be no ‘logical’ reason why the inanimate nature of the causee should determine the choice of complement rather than the equally inanimate nature of the causer, or the generic context. What, therefore, do we do when our semantic criteria directly contradict one another? One could perhaps attempt to establish some kind of hierarchy among the various criteria, to determine which will take precedence, but given the considerable variation we observe in the distribution of VOSI and VOSI[to], this is likely to prove an impossible task. Rather than explaining their distribution, the approach seems to further complicate what is already a complex situation. The kind of alternation observed in (9) and (10), or in (11) and (12) appears rather to be authentic cases of variation, a factor which is missing from the approach based purely on semantic criteria as set out in Figure 1. VOSI and VOSI[to] are not in complementary distribution under causative verbs. Instead, their uses seem to ‘overlap’.

Brian Lowrey

105

What is required is a framework that can take into account both intuitions, such as Fischer’s concerning agentivity and genericity, which do indeed seem basically sound, and also the variation which characterises ME complementation patterns.

3.3. A Dynamic Model of Variation and Change The framework which I propose to adopt here is the ‘dynamic’ model set out by Smith (1996), inspired by Weinreich et al. (1968), which recognises the need to factor variation into any account of overlapping forms such as causative VOSI and VOSI[to]. Variation will occur across the speech community, on a broad scale, but also at the level of the individual speaker, the “synchronic oscillation in the speech of individuals” identified by Weinreich et al. (1968: 166). Indeed, from this perspective, variation is seen as “communicatively necessary” (Smith 1996: 45), in this instance, presumably, to take account of differing individual perceptions of complex events such as causation. To illustrate the phenomenon, we can call upon the concept of ‘variational space’, described by Smith (1996: 44-47). As Smith points out, “variation is possible because individual linguistic items occupy fairly broad slots in the langue within which they are situated, and their realisations in parole can therefore vary quite widely” (1996: 44). The variational space of a linguistic item, then, is made up of all the contexts in which the item appears. I shall represent the distribution of causative VOSI and VOSI[to] in terms of variational space, as in Figure 2:

+DIRECT

m

VOSI

VOSI [to]

o

+INDIRECT

Figure 2. Variational space, causative VOSI and VOSI[to] The circles represent the variational spaces of each construction, the sum of all the contexts in which they occur. The variational space of VOSI will contain more instances of ‘direct’ causation, according to criteria such as those listed in Figure 1, and that of VOSI[to] more cases of indirect causation. It is important, however, not to view VOSI and VOSI[to] as

106

Syntactic and Semantic Variation

being in fully complementary distribution. Their respective spaces overlap in the middle, the zone within which individual speakers will hesitate between the two forms. As Smith (1996: 45) points out, “such overlaps are typically the sites of change in the linguistic system”. The variational space of a given item can shift over time, expanding or shrinking and causing variables to overlap to a greater or lesser extent. We shall see in a moment how the variational spaces of causative verbs and constructions change during the ME period. This kind of approach expressly predicts that some degree of variation is inevitable, and emphasises the need to take a much broader view of variables such as complement patterns. It is not enough, if we wish to make sense of the distribution of causative VOSI and VOSI[to], to simply take isolated examples of each construction and try to explain them on a one-by-one basis, in order to show how individual occurrences might be made to fit criteria such as those listed in 3.2.2. Such an approach is ultimately doomed to fail, however relevant the criteria may be, because it fails to take the overlap zone into account. A more productive strategy would be to define the contours of each construction’s variational space. To this end, I examined the distribution of thematic roles across our sample of ME texts, in an attempt to determine what kind of causation is expressed by each complement type when used with each particular verb. The results are given in Tables 1-6 below, where [Ag] denotes an Agent, [Th] a Theme, [Exp] an Experiencer, and [Cs] a Cause, a non-agentive causer. The most common complement types are described in 2, while the remaining abbreviations are explained in the list given in the Appendix.

Brian Lowrey

107

Tables 1-6. Distribution of Causers/Causees by thematic role: Table 1. RR, Works, Richard Rolle, Northern dialect Causer [Ag]

Causee

[Cs]

[Ag]

5

6

[Ag] o

[Th] [Exp]

Other

[Ag]

gar VOSI VOSI PASS

14

9

5

3

3

2

2

Total

16

11

5

7

3

3

3

2

1

3

1

VOSI[to] VOSI[to] PS

11

5

6

8

2

Total

15

7

VOSI

1

1

1

1

Total

1

1

1

1

2 6

make VOSI

1

1

1 8

1 1

12

3

do

Syntactic and Semantic Variation

108

Table 2. YP, York Mystery Plays, Northern dialect (Yorks.) Causer

Causee

[Ag] [Cs]

[Ag]

[Ag] o

[Th] [Exp]

Other

[Ag]

gar VOSI VOSI PASS

37

31

6

21

14

3

20

2

2

V+I

13

13

Total

52

46

6

21+13

16

VOSI

10

2

8

6

6

VOSI[to]

25

14

11

13

8

5

1

7

Total

35

16

19

19

14

5

1

8

VOSI

1

1

VOSI[to]

2

1

1

1

1

Total

3

2

1

2

1

2 [13]

[13] 3

20+13

make 1

do 1

Brian Lowrey

109

Table 3. CA, Gower, Confessio Amantis (Lb I-V), London/South-East Midlands Causer [Ag]

Causee

[Cs]

[Ag]

[Ag] o

[Th] [Exp]

Other

[Ag]

make VOSI

34

22

12

15

9

10

3

10

VOSI[to] VOSI[for to]

14

7

7

2

8

2

2

1

11

6

5

4

8

4

2

1

V+ÞAT

4

V+I

1

V+PRT

1

2 1 [1]

4

2

2

68

38

30

23

26

13+1

5

13

VOSI

8

6

2

1

4

3

1

1

VOSI[to] VOSI[for to]

6

5

1

3

2

2

13

10

3

2

1

9

1

1

V+I

10

10

2

2

39

33

V+I

20

20

[21]

[21]

V+V+I

13

13

[13]

[13]

Total

33

33

[34]

[34]

Total do

V+ÞAT Total

[10]

[10] 2

6

6+10

9

14

2

2+10

let

cause VOSI[to]

1

1

1

V+[to]+I V+[for to]+I

1

1

1

1

1

[1]

Total

3

3

[1]

1 2

1

Syntactic and Semantic Variation

110

Table 4. CP, Chaucer, prose works: Tale of Melibee, Parson’s Tale, Boece, Treatise on the Astrolabe Causer [Ag]

Causee

[Cs]

[Ag]

[Ag] o

[Th] [Exp]

Other

[Ag]

make VOSI

9

5

4

5

4

1

18

7

11

4

7

6

1

1

V+ÞAT

5

1

V+I

1

1

34

15

19

VOSI

2

1

1

VOSI[to]

1

Total

3

1

V+I

4

4

[5]

[5]

Total

4

4

[5]

[5]

VOSI[to] VOSI[for to]

Total

3 1

2

1 4

2

3

[1] 9+1

[1] 14

10

1

5+1

do 2

1

1

2

2

1

let

Brian Lowrey

111

Table 5. CT, Chaucer, The Canterbury Tales, London/South-East Midlands Causer

Causee

[Ag] [Cs]

[Ag]

[Ag] o

[Th] [Exp]

Other

[Ag]

make VOSI

42

34

8

23

11

9

21

VOSI[to] VOSI[for to] VOSI PASS

27

16

11

11

14

3

7

14

6

8

3

9

1

V+ÞAT

1

1

1

1

1

29

1

8

1

7

5

2

1

92

57

35

42

37

14

12

9

3

4

8

2

VOSI[to]

3

3

2

1

V+PRT

1

1

19

19

1

1

36

33

15

15

[15]

[15]

3

3

[3]

[3]

18

18

[18]

[18]

3

1

Total do VOSI

V+I V+ÞAT Total

4 2

[19]

[19] 1

3

6+19

9

2

1

6+19

let V+I V+V+I Total cause VOSI[to] VOSI[for to]

2

1

1

1

1

V+[to]+I

1

1

[1]

Total

5

4

2+1

1

1

1

1

1

1

1

Syntactic and Semantic Variation

112

Table 6. PP, Langland, Piers Plowman, B-Text, W. Langland, SouthWest Midlands Causer [Ag]

Causee

[Cs]

[Ag]

[Ag] o

[Th] [Exp]

Other

[Ag]

gar VOSI

3

3

1

2

1

VOSI[to]

4

4

1

3

1

V+I

1

1

[1]

Total

8

8

2+1

5

VOSI

11

7

4

8

4

3

VOSI[to]

18

9

9

12

4

4

1

6

1

9

[1] 2+1

make

V+ÞAT Total

2

2

3

2

31

16

15

20

10

7

23

20

3

19

4

3

16

5

4

1

1

1

4

1

2

1

1

2

1

V+I

12

11

1

[12]

[11]

Total

42

36

6

22+12

V+I

7

7

[7]

[7]

Total

7

7

[7]

[7]

do VOSI VOSI[to] VOSI[for to]

5

7

18+11

let

Owing to cases of infinitive coordination, the number of Causees may exceed the number of Causers. Obviously, the factors highlighted by statistics of this kind do not explain on their own the distribution of causative verbs and complement types in ME. Other contextual factors almost certainly play a part, too: genericity, notably, does not show up in the tables presented here. Nonetheless, we can use the information given above to reach a better understanding of the relative distribution of both verbs and complements. In particular, Tables 1-6 illustrate to what extent agentivity, both that of

Brian Lowrey

113

the causer and of the causee, helps to define the variational spaces of causative verbs and complements in ME. 3.3.1. Agentivity of the Causer Firstly, the tables confirm the importance of causer agentivity in determining the choice of causative verb, not just in VOSI-type constructions but with all complement types. Agentive causers are commonest with the ‘central’ causative of each dialect, the verb that appears most frequently in the V+I construction. In each case, this verb can be opposed to make. As Table 2 shows, gar occurs in YP on 52 occasions (all complement types). In 46 of these instances (88.7%), the causer is an Agent. Make on the other hand is less frequent, especially with agentive causers. In only 16 of the 35 instances of causative make in YP (45.7%) is the causer a volitional Agent. A not dissimilar pattern emerges from RR, although the number of occurrences of causative make here is proportionally higher. Gar, as Table 1 shows, appears 16 times, with an agentive causer on 11 occasions (68.7% of the total). The same corpus contains 15 examples of causative make, 7 of which (46.6%) have an agentive causer. The fact that gar is proportionally less frequent here can probably be explained by the more religious, philosophical content of RR, involving a higher proportion of non-agentive causers generally, leading to a drop in the number of occurrences of gar. In PP, a predominately Midlands text, and CA, a text of more southerly origin, a similar tendency emerges (see Tables 6 and 3). In both texts, make is opposed to do with regard to causer agentivity. PP contains 42 instances of causative do, 36 of which (85.7%) have an agentive causer, and 31 instances of make, 16 (51.6%) with an agentive causer. In the more southerly dialect of Gower and Chaucer, the situation is slightly different in that make has already started to become the central causative. Causative do is therefore less frequent here. Nevertheless, we still find 39 occurrences in CA of causative do, 33 of which (84.6% of the total) present an agentive causer, whereas in the case of make only 38 causers out of 68 (55.9%) are Agents. The variational spaces of gar, do, and make, particularly in the North and the Midlands, could therefore be represented, with regard to causer agentivity, as in Figure 3:

114

Causer [Agent] m

Syntactic and Semantic Variation

do/gar

make

o Causer [Cause]

Figure 3. Variational space, do/gar and make with Agentive causers Although causer agentivity is clearly a very important factor in determining which verb will be used, do and gar are not used exclusively to express agentive causation. In other words, there is no perfect match between semantic criteria of this type and the distribution of the verbs concerned. Once again, when faced with a choice between two variables, speakers will not systematically select the same option. Variation of this type is an inherent property of natural language. Interestingly, Kemmer & Verhagen (1994: 143) observe a not dissimilar pattern in modern Dutch, another Germanic language, concerning the alternation between causative doen and laten. The latter verb is associated in 99% of cases in their corpus with an animate causer, whereas only 42% of causers with doen are Agents. 3.3.2. Agentivity of the Causee The overall statistics for CA (see Table 3) also bring another important tendency to light. They show not only that in Gower’s dialect a form of opposition is beginning to appear between make VOSI and make VOSI[to] in terms of the agentivity of the causer, but also that the choice of VOSI is more likely still if the causee is also an Agent. Table 3 shows a relatively high proportion of agentive causees in the VOSI construction without to. Of the 34 occurrences in CA of make VOSI, the causee is an Agent on 15 occasions (44.1%). Only two of the causees are Agents among the 14 instances of make VOSI[to], however. This, of course, contradicts Fischer’s IIc in Figure 1, and confirms the need to look at the overall distribution of overlapping forms if we are to understand how they are related to one another. The same corpus, CA, also contains a relatively high number (10 tokens out of a total of 34) of make VOSI structures expressing [Agent]ĺ[Agent] configurations, cases of what Kemmer & Verhagen (1994: 120) call “inducive causation”, where one agentive subject carries out an action in response to a causative stimulus from another. This tendency is even more marked in CT (21 cases out of 42: see Table 5). A similar phenomenon is to be found in other dialects, too. In PP, for

Brian Lowrey

115

example, do VOSI is used on 16 occasions out of 23 (69.5%) to evoke [Agent]ĺ[Agent] configurations, and gar VOSI appears in YP in the same type of context in 20 cases out of 37. It would appear, then, in all three dialects represented here, that [Agent]ĺ[Agent] configurations start to become particularly compatible with VOSI (without to) as of the latter part of the 14th century. Gar, being generally associated with agentive contexts, is not at all common in the VOSI[to] construction. Most of the exceptions to this occur in texts like PP, in a dialect where gar is not the central agentive causative, and can therefore probably be considered atypical uses.

4. Change within the ME Period Having shown how the agentivity parameter helps shape the variational spaces of both causative verbs and complement types, we shall now turn to the question of the extent to which it combines with the other factors mentioned here, genericity and [Agent]ĺ[Agent] interaction, to explain some of the changes which take place during the ME period.

4.1. Causative Make Perhaps the most obvious development to occur in ME, as far as the causatives are concerned, is the rise of causative make. Make first appears as a causative in late OE, used with ‘small clause’ complements (Lowrey 2010: 68). Causative use of make with an infinitive complement, however, is a ME innovation. The earliest occurrence quoted in the Oxford English Dictionary (OED) is dated c1175. We have already mentioned the fact that initially, at least, it tends to appear in generic contexts, such as in (8), (14) or again in (15): (15) A vorbisne is of olde iwurne, / Þat neode makeþ old wif eorne / & yet ich habbe an oþer onswere ‘A proverb was coined a long time ago, that need makes the old woman run, and yet I have another answer’ (Owl and Nightingale) Even among those (relatively few) instances of V+I that do not fit the ‘profile’ set out in 3.1.2., the matrix verb is most often make:

116

Syntactic and Semantic Variation

(16) Vor hit is soþ, Alured hit seide, / & me hit mai ine boke rede: / "Eurich þing mai losen his godhede / Mid vnmeþe & mid ouerdede." / Mid este þu þe miвt ouerquatie, / & ouerfulle makeþ vvlatie: ‘For it is true, king Alfred said so, and one may read it in a book, everything may lose its goodness, through immoderation and through excess. You might over-glut yourself with pleasure, and over-filling makes [one] feel disgusted:’ (Owl and Nightingale) In CP and CT, make can still be found in generic contexts, with inanimate causers, as in (17) and also in cases of what would presumably be deemed ‘indirect’ causation, such as (18), where the NP hire lecherye seems to function as a kind of intermediary: (17) And forther over, it maketh hym that whilom was sone of ire to be sone of grace; and alle these thynges been preved by hooly writ. ‘And furthermore, it makes him that was formerly the son of anger the son of grace, and all these things are demonstrated by the Holy Scriptures.’ (Chaucer, Parson’s Tale) (18) Fy! spek namoore – it is a grisly thyng – / Of hire horrible lust and hir likyng / Of clitermystra, for hire lecherye, / That falsly made hire housbonde for to dye ‘Enough! Speak no more – it is a grisly thing – of her horrible lust and her desire, of Clytemnestra, who through her lechery treacherously made her husband die (caused her husband to die?) (Chaucer, Wife of Bath’s Prologue) (17) and (18) show that the variational space of make has shifted considerably since Chaucer’s time. Make in the PDE equivalent of (17) would almost certainly be accompanied by a ‘small clause’, the presence of be in an infinitive complement generally producing an agentive reading. And in (18) cause, at least in the author’s native variety of English (Southern British Isles), would probably be preferred to make in PDE.

Brian Lowrey

117

At the same time as examples such as (15)-(18), however, we also find an increasing number of [Agent]ĺ[Agent] configurations expressed by make, usually with a VOSI complement, as in (19): (19) He yaf me al the bridel in myn hond, / To han the governance of hous and lond, / And of his tonge, and of his hond also; / And I made hym brenne his book anon right tho. ‘He gave me free rein, in my own hands, to have the governance of his house and land, and of his tongue and his hand, too; and I made him burn his book right then, straight away.’ (Chaucer, Wife of Bath’s Tale) Make in (19) apparently has much the same coercive value as in PDE, thanks largely to the supposition that there exists an opposition between the intentions of two volitional subjects. The use of make in this type of context represents an innovation compared to its earliest uses as an analytical causative. Once again, as Tables 3 and 5 show, inducive uses of make are more frequent with plain VOSI complements. The situation in CA and CT could therefore be summed up as in Figure 4:

Agentive m Causer/Causee

make VOSI

make VOSI[to]

o Non-agentive Causer/Causee

Figure 4. Variational space, make VOSI and make VOSI[to] in CA/CT This corresponds nicely to what we know of the subsequent evolution of causative make. The two variational spaces will gradually draw further apart along the agentivity axis, with a marked acceleration as of the late 16th and early 17th centuries (see Lowrey 2002: 418-452), as VOSI comes to be increasingly associated with agentive contexts. As for the question why make within the causative group should have evolved in this manner, it seems that once again agentivity, which Song (1996: 144-46) shows to be an important factor in determining the nature of causative constructions cross-linguistically, plays a significant part. A simplified representation of the variational space of the different causative

Syntactic and Semantic Variation

118

constructions in PP and YP, taking just the agentivity parameter into account, appears in Figure 5:

VOSI + Agentive m

o -Agentive

make do/gar

V+I

Figure 5. Variational space, do/gar and make, PP and YP Both do and gar will shortly leave the causative group. Do is already on its way to acquiring auxiliary status. Gar, largely restricted to Northern dialects, seems to be perceived as a regionalism. The advent of the printing press and the spread of a new written standard based to a fairly large extent on the Chancery standard (Samuels 1963: 81-94) will serve to isolate it still further, and it rapidly disappears from written texts in the early 16th century. The consequence of these changes will be for the variational space of make to expand leftwards on the diagram in Figure 5, to fill the gap left by the first two verbs. This process already appears to be under way in the South at the time of CT and CA. Here, as Tables 3 and 5 show, causative do is already becoming rarer, and the variational space of make VOSI has begun to shift leftwards, drawing apart as it does so from that of make VOSI[to]. Again, we could represent the situation in this dialect, at this particular stage, as in Figure 6: do +Agentive m

make VOSI

make VOSI[to]

let V+I

Figure 6. Variational space, CA and CT

cause o -Agentive

Brian Lowrey

119

Once again, the loss of causative do as well as that of causative let once V+I disappears as a syntactic option will cause the variational space of make, and make VOSI in particular, to expand leftwards, towards more agentive contexts. As a result, a gap appears at the right of the diagram for a new causative verb to fill. And indeed, we find that cause first appears as an analytical causative, in Southern dialects, during the 14th century, in precisely the same kind of generic, non-agentive contexts as those in which make first appeared some 150–200 years earlier: (20) For every creature bore, / If thou him yive, is glad therfore, / And every gladschipe, as I finde, / Is confort unto loves kinde / And causeth ofte a man to spede. ‘For every creature born, if you give to him, is glad thereof, and all gladness, or so I find, is a comfort unto the very nature of love, and often causes a man to prosper.’ (Gower, Confessio Amantis) The earliest recorded instance of cause used as a verb in the OED is from Gower, although Lowrey (2002: 359) quotes 13th-century uses of cause as a noun. The chronological data would therefore seem to support the claim that cause was pressed into service as a verb to compensate for the shift in the variational space of make away from non-agentive contexts. We can observe that agentivity, as well as being a crucial factor in the choice of causative verb and complement type, also lies at the heart of the changes which take place during the ME period, agentive structures leaving the causative system to be replaced by new, non-agentive forms at the opposite end of the scale.

5. Conclusion The data presented in this paper show that what appear to be semantic universals with regard to causation, such as agentivity and genericity, clearly determine to a large extent the choice of analytical causative and of causative type in ME. Certain verbs are used with certain complement types to express a particular type of causation. Nevertheless, excessively rigid attempts to apply criteria of this type in order to explain the distribution of individual verbs or complement types invariably encounter problems, as we have seen. Such an approach can only be made to work if variation is fully integrated into the account. Elements such as the choice of one causative among a set of verbs, or one type of complement structure rather than another, are precisely the kinds of variables which will lead to

120

Syntactic and Semantic Variation

hesitation on the part of the speaker. Such hesitations clearly show up quite clearly in the figures given here. We have also seen that agentivity is one of the driving forces behind some of the major changes which take place within the ME causative subsystem. English loses, between the beginning and the end of the ME period, the three principal agentive causatives let, gar, and do. In each case, it is the verb make whose variational space is extended to fill the void left by the element which disappears. Make becomes increasingly associated with agentive causers and causees. Signs are even beginning to appear in CA and CT of a shift in the relative variational spaces of VOSI and VOSI[to]. Make VOSI is even used by Chaucer with inducive or coercive meaning, implying a high degree of [Agent]ĺ[Agent] interaction. The development in the use of make over time is interesting in that the verb enters the causative group at the non-agentive, generic end of the scale (factors associated with ‘indirect’ causation), and moves across the whole range of causative values to end up at the agentive, coercive end, where we find cases of more ‘direct’ causation. As we have seen, this allows a new verb, cause, to enter the group, at the same end of the scale as make. We know (Lowrey 2002: 385-412) that cause will also undergo a very similar shift, albeit briefly, towards agentive uses, before returning to the type of role in which it first appears. There seems to exist a tendency for new causatives to emerge in what might prove to be more ‘marked’ non-agentive contexts, before being extended to agentive contexts, although further research will obviously be required to fully explore this question.

List of Abbreviations CA: Confessio Amantis, corpus (see Introduction) CP: Chaucer, prose works, corpus (see Introduction) CT: Chaucer, Canterbury Tales, corpus (see Introduction) PP: Langland, Piers Plowman, corpus (see Introduction) RR: Richard Rolle, corpus (see Introduction) YP: York Mystery Plays, corpus (see Introduction) V+I: ‘Verb+Infinitive’ (see 2.2.) V+[to]+I: ‘Verb+to+Infinitive’, construction similar to V+I with pre-infinitival to, as in: Wyn, wo, or chaungynge of complexioun / Causeth ful ofte to doon amys or speken. (Chaucer, Canterbury Tales)

Brian Lowrey

121

V+[for to]+I: As above, but with pre-infinitival for to, as in: Ricth-wise men he louede alle, And oueral made hem forto calle (Havelok) V+PRT: ‘Verb+Past Participle’, as in: but god, of his mercy, / And youre benyngne fader tendrely / Hath doon yow kept (Chaucer, Canterbury Tales) V+ÞAT: ‘Verb+finite that-clause’ (see 2.2.) V+V+I: ‘Verb+Verb+Infinitive’: the so-called ‘double causative’ construction, as in: it befell upon a chance, / That he for Jupiteres sake / Unto the goddes let do make / A sacrifice (Gower, Confessio Amantis) VOSI: ‘Verb+Subject / Object+Infinitive’ (see 2.2.) VOSI[to]: construction similar to VOSI with pre-infinitival to (see 2.2.) VOSI[for to]: construction similar to VOSI with pre-infinitival for to (see 2.2.) VOSI PASS: VOSI construction with passive complement, as in: He makys his saute in vs, and bost byfore God, to gere vs be dampned yf he may overcome vs in any temptacioun (Rolle, The English Psalter) VOSI[to] PS: VOSI[to] construction with passive complement, as in: Besyde þat welle kyng Roboas let make .IJ. CALUEREN of gold + made hem to ben worschipt (Mandeville’s Travels)

Works Cited Primary Sources Chaucer, Geoffrey: The Canterbury Tales (c1380–1385). Source: Larry D. Benson (ed.) 1987. The Riverside Chaucer (3rd ed.) Oxford: Oxford University Press. [E-text: University of Virginia] Gower, John: Confessio Amantis (a1393), MS Bodleian Fairfax 3. Source: George C. Macaulay (ed.) 1899–1902. The Complete Works of John Gower. Clarendon Press: Oxford. [E-text: University of Michigan] Havelok (c1280), MS Laud 108, Bodleian Library. Source: Walter W. Skeat (ed.) 1868. The Lay of Havelok the Dane, Composed in the Reign of Edward I, about AD 1280. Early English Text Society ES4. London: Trübner. [E-text: University of Michigan] Lambeth Homilies (a1225), Lambeth MS 487. Source: Richard Morris (ed.) 1867–1968. Old English Homilies and Homiletic Treatises. Early English Text Society OS 29, 34. London: Trübner. [E-text: University of Michigan]

122

Syntactic and Semantic Variation

Langland, William: Piers Plowman (c1400 (c1385)), B-Text MS B.15.17, Trinity College, Cambridge. Source: A. V. C. Schmidt (ed.) William Langland: The Vision of Piers Plowman. A Critical Edition of the BText Based on Trinity College Cambridge MS B.15.17 with selected variant readings, an Introduction, glosses, and a Textual and Literary Commentary. London & New York: J.M. Dent & E. P. Dutton. [Etext: University of Michigan] La‫܌‬amon: Brut (a1225 (?a1200)), MS Cotton Caligula A ix (C), British Library. Source: George L. Brook & Roy F. Leslie (eds.) 1963–1978. Layamon: Brut. Early English Text Society 250, 277. Oxford: Oxford University Press. [E-text: University of Michigan] Mandeville’s Travels (a1425 (c1400)), MS Cotton Titus C XVI, British Museum. Source: Paul Hamelius (ed.) 1919–1923. Mandeville’s Travels: Translated from the French of Jean d’Outremeuse, Early English Text Society OS 153, 154. London: K. Paul, Trench, Trübner & Co. [E-text: University of Michigan] OED = Oxford English Dictionary (2nd ed., 1989) The Owl & the Nightingale (c1250 (c1190)), MS Cotton Caligula A ix (C), British Library. Source: Eric Stanley (ed.) 1960. The Owl and the Nightingale. London: Nelson’s Medieval and Renaissance Library. [Etext: University of Virginia] Writings of Richard Rolle (?c1343–a1349). Source: Hope E. Allen (ed.) 1931. The English Writings of Richard Rolle, Hermit of Hampole. Oxford: Clarendon Press Wyclif, John: De Oficio Pastorali (a1500(?c1378)). Source: Frederic D. Matthew (ed.) 1880. The English Works of Wyclif Hitherto Unprinted. Early English Text Society OS 74. London: Trübner. [E-text: University of Michigan]

Secondary Sources Baron, Naomi S. 1974. “The Structure of English Causatives”. Lingua 33.299-342. Benson, Larry D. (ed.) 1987. The Riverside Chaucer. (3rd ed.) Oxford: Oxford University Press. Bock, Hellmut. 1931. “Studien zum präpositionalen Infinitiv und Akkusativ mit dem to-Infinitiv޵. Anglia 55.114-249. Denison, David. 1993. English Historical Syntax: Verbal Constructions. London & New York: Longman. Dowty, David R. 1979. Word Meaning and Montague Grammar. Dordrecht: Kluwer.

Brian Lowrey

123

Fischer, Olga. 1990. Syntactic Change and Causation – Developments in Infinitival Constructions in English. (Amsterdam Studies in Generative Grammar 2.) Amsterdam: University of Amsterdam Press. —. 1995. “The Distinction between to and Bare Infinitival Complements in Late Middle English”. Diachronica 12.1.1-30. —. 1996. “Verbal Complementation in Early ME: How do the Infinitives Fit In?” English Historical Linguistics 1994: Papers from the 8th International Conference on English Historical Linguistics (8th ICEHL, Edinburgh, 19–23 September 1994) ed. by Derek Britton, 247270. Amsterdam: John Benjamins. Gelderen, Elly van. 2000. A History of English Reflexive Pronouns. Amsterdam: John Benjamins. Hopper, Paul J. & Sandra A. Thomson 1980. “Transitivity in Grammar and Discourse”. Language 56.251-299. Karttunen, Lauri. 1971. “Implicative Verbs”. Language 47:2.340-58. Kemmer, Suzanne & Arie Verhagen. 1994. “The Grammar of Causatives and the Conceptual Structure of Events”. Cognitive Linguistics 5:2.115-156. Lowrey, Brian. 2002. Les verbes causatifs en anglais: une étude diachronique, du moyen-anglais à l’anglais moderne. PhD thesis, Université de Lille III. —. 2010. “Causative Verbs in West Saxon Old English”. Bulletin des Anglicistes Médiévistes 78.57-88. Mustanoja, Tauno F. 1960. A Middle English Syntax. Helsinki: Société Néophilologique. Roberts, Ian G. 1993. Verbs and Diachronic Syntax: A Comparative History of English and French. Dordrecht: Kluwer. Samuels, Michael L. 1963. “Some Applications of Middle English Dialectology”. English Studies 44.81-94. Smith, Jeremy J. 1996. An Historical Study of English: Function, Form and Change. London & New York: Routledge. Shibatani, Masayoshi. 1976. The Grammar of Causative Constructions: a Conspectus. New York: Academic Press. Svartvik, Jan & Randolph Quirk. 1970. “Types and Uses of Non-finite Clauses in Chaucer”. English Studies 51.1.393-411. Talmy, Leonard. 1986. “Force Dynamics in Language and Thought”. Chicago Linguistic Society 21:2.293-337. Song, Jae J. 1996. Causatives and Causation. London & New York: Longman. Visser, Fredericus T. 1973. An Historical Syntax of the English Language. Leiden: Brill.

124

Syntactic and Semantic Variation

Warner, Anthony. 1975. “Infinitive Marking in the Wyclifite Sermons”. English Studies 56.207-214. —. 1982. Complementation in Middle English and the Methodology of Historical Syntax. London: Croom Helm. Weinreich, Uriel, William Labov & Marvin I. Herzog 1968. “Empirical Foundations for a Theory of Language Change”. Directions for Historical Linguistics ed. by Winfred P. Lehmann & Yakov Malkiel, 95-195. Austin: University of Texas Press.

V-1 MAIN DECLARATIVE CLAUSES IN OLD ENGLISH AND OLD HIGH GERMAN BIBLICAL TRANSLATIONS ANNA CICHOSZ

Both in Old English and Old High German, the V-initial order is used predominantly in imperative and interrogative clauses. However, V-initial structures can also be encountered in main declarative clauses, though this pattern is usually regarded as marked. In the case of Old English, the V-1 pattern is often treated as a purely stylistic or emphatic device. Nonetheless, it has been noted that V-1 is used very extensively in Old High German translations, but the exact mechanism of its usage and its dependence on Latin source texts remains a controversy. The present paper aims to analyse and compare the incidence of V-1 main declarative clauses in Old English and Old High German biblical translations, trying to determine whether the structure appears in the texts independently of the Latin original (which would suggest its native character) or if it only surfaces in the target text as a copy of the Latin word order pattern. The overall objective of the analysis is to determine if there are any visible discrepancies between the languages in the application of this pattern and – provided that some differences indeed exist – whether they are related to divergences between the Old English and Old High German syntactic systems or to a different degree of Latin interference in the investigated target texts.

1. The V-1 Pattern in Old English and Old High German The V-initial structure, as opposed to the V-final pattern, is said to have functioned as a marked word order already in Proto-Germanic (Smith 1971). It was inherited by other Old Germanic languages, including Old English (OE) and Old High German (OHG), but the structure appeared mainly in other than declarative clause types:

126

Old English and Old High German Biblical Translations The verb-initial pattern, then, represents a marked order used principally in imperatives, clauses with dramatic force, and conjoined clauses including relatives (Smith 1971: 96).

Even though the V-1 pattern is characteristic of imperative and interrogative clauses, it is also possible to identify V-initial structures in Old Germanic main declarative clauses, e.g. in OE poetry1: (1) Wæs seo tid cumen þæt þær fæge men feallan sceoldon was the time come that the doomed men fall shall (Mald 104) (2) cymeð him seo ar of heofonum comes (to) him forgiveness from heaven (Sea 107) (3) onwendeð wyrda gesceaft weoruld under heofonum changes turn of fate world under heavens (Wan 107) However, this pattern is relatively rare and it is generally agreed that the predominant word order pattern in OE main declarative clauses is V-2: For the classical Alfredian and even Aelfrican periods of OE, we can sum up by generalising and saying that the predominant word order in OE was OV, with V2 in main clauses (Traugott 1992: 277).

Traugott admits that V-1 is also used in main declaratives and underlines the importance of this pattern (though in her analysis she includes both verb-initial V-S clauses and V-S clauses introduced by adverbials such as ða) in the narrative structure of OE texts: ... in narrative contexts there continue to be some striking examples of the use of verb-initial word order to convey pragmatic information ... The word order signals that a new segment of information is beginning, and is a device for reintroducing for renewed attention a topic that has already been mentioned (Traugott 1992: 278-279).

Mitchell (1985: 974-976) states that it is very difficult to find a precise explanation for the use of this pattern. Among possible factors influencing its occurrence he lists: emphasis or “special declarative effect”; individual 1

In all the examples provided in the article, the finite verb is underlined and the subject is shown in bold.

Anna Cichosz

127

preference of certain authors for the pattern (it appears to be exceptionally common in the Aelfredian Bede and some of the poetry); its frequent cooccurrence with verbs of saying and sentences with an impersonal verb + a noun clause subject; the influence of verse tendency to place the verb before the first stress; stylistic considerations such as balance or rhythm; as well as the function of the V-1 pattern in the narration, marking a turning point, a transition or a change of pace in the prose (similarly to Traugott’s observation). Neither of the authors mentions a possibility of Latin influence in this context, treating V-1 main declaratives as rare but native constructions. In research on OHG syntax, the significance of this word order for text types other than poetry has often been disregarded. Otto Behaghel (1932: 30), in his fundamental book Deutsche Syntax, claims that ‘Spitzenstellung’ is a rare and exceptional pattern without any great importance for Old High German. Nonetheless, more recently it has been noted that V-1 is used very extensively in Old High German translations (cf. Axel 2007, Robinson 1996), but the exact mechanism of this phenomenon remains a controversy. Some linguists see it as a Latin influence: Unter dem Einfluȕ des Lateinischen kann das Verb nicht nur in subjektlosen Aussagesatz in Anfangsstellung vorkommen (Penzl 1986: 101). [Under the influence of Latin, the verb could occupy the initial position not only in subjectless declarative clauses.]2

Other scholars claim that V-1 was used for ‘stylistic reasons’ but its precise function remains unknown (Lockwood 1968), while Robinson, who analysed the Isidor translation, claims that: (...) the Isidor translator had fixed on verb-first as a marker for the Biblical portions of the text, imposing it even in those cases where the Vulgate does not give a direct model (Robinson 1996: 25).

Another scholar who analysed this phenomenon was Lippert (1974), also quoted by Robinson, who clearly does not agree with Lippert’s conclusions: Although Lippert (1974: 53) characterises verb-first as a (little-utilized) native pattern for the Isidor translator, a pattern perhaps supported by (but not caused by) Latin syntactic models, I would characterise it as a distinctively foreign pattern, which because of this very distinctiveness led

2

Translation mine, A. C.

128

Old English and Old High German Biblical Translations the Isidor translator to use it as a flag to mark Biblical quotations in the overall text (Robinson 1996: 25).

Robinson claims that this structure was used exactly because of its unfamiliarity: it was supposed to mark passages quoted from the Bible (Isidor is a religious treatise with frequent references to the Scripture), where lack of familiarity is not a flaw but rather a sign of the uniqueness of the sacred book. Lippert, however, claims that V-1 is a native German pattern, used when the Latin structure was impossible to render word by word. If one considers the fact that in Old Germanic poetry the V-1 pattern is used relatively often, the latter hypothesis seems quite plausible. It would hardly be possible to use the V-1 pattern so extensively if it had not existed in the language at all, at least as a very strongly marked pattern. The translator may have used an infrequent but accepted pattern to render the unfamiliar structure, substituting “unfamiliar and hard to understand” with “unfamiliar but possible to understand”. As Lockwood (1968: 256) puts it: “[t]he verb in this position [V-1] doubtless had stylistic value”. It was used in poetry, so from there it could have been transferred to other text types, and treated as an elegant solution to some translation problems. Hinterhölzl & Petrova 2005 and Petrova & Solf 2008, representing a new discourse-based approach applied to the analysis of Old High German syntax, disagree with scholars who see V-1 as a foreign structure and claim that the native Germanic V-1 pattern played a very special role in discourse, encoding a coordinating discourse relation: In this view, the text of the Old High German Tatian allows for a dynamic model of discourse structuring where verb placement serves as a major device in distinguishing the type of discourse relations. (...) it may be concluded that verb-initial structures establish coordinative discourse relations whereas verb-second clauses signal subordinating linkage to the previous discourse part. In this sense, a verb-initial occurrence within the text, even involving an already established discourse referent, may be perceived as a signal that the utterance quits a previous passage of subordination and returns to the main line of the discourse (Hinterhölzl & Petrova 2005: 3).

In the case of OE, a similar view was expressed by Traugott 1992 and Mitchell 1985, quoted earlier in this section. What needs to be mentioned here, however, is that Hinterhölzl & Petrova decided to analyse only ‘deviations’, i.e. clauses which demonstrate the V-1 pattern contrary to the Latin model, which means that they were not interested in investigating the degree of dependence of this pattern on Latin.

Anna Cichosz

129

Another factor mentioned in the context of V-1 declarative clauses is related to the verb itself. It has been noted that in OHG certain verbs cooccur with the pattern more often than others: Furthermore, verb-initial structures with discourse-given material regularly occur with certain groups of main verb predicates. The most common among these are motion verbs, verbs of saying as well as verbs of sensual or cognitive perception, the latter especially encoding an inchoative meaning, that is, implying the initiation of a new state of affairs (Petrova 2006: 159). In OHG, declarative verb-first order is e.g. attested in the context of motion verbs and with sentential negation expressed by proclitic ni which attaches to the finite verb (Axel 2007: 33).

Mitchell (1985: 974) also mentions verbs of saying as often cooccurring with the pattern in OE, so there are parallels in the use of the structure between the two analysed languages. Thus, there is evidence suggesting that V-1 main declarative clauses found in OHG translations may be a native structure, at least partially independent of the Latin original. Since the V-1 pattern is attested in both OE and OHG main declaratives, and its use may have been strengthened by Latin, it should be possible to trace this structure in translations created in the two languages (where the Latin influence should be both visible and equally strong). The incidence of the structure should also be compared in detail, taking into consideration all the factors mentioned in this section, in order to identify any possible discrepancies. According to Davis, no important differences between the languages should emerge as OE and OHG had exactly the same syntactic systems: The patterns of word order exhibited by all the Old Germanic languages may be regarded as identical. This is tantamount to saying that the syntax of these languages is the same in all its important points (Davis 2006: 53).

In their book devoted entirely to the two languages in question Davis & Bernhardt (2002: 1) make the following claims: a) both Old English and Old High German word order is governed by “conventions which approach the status of rules and may be precisely described”; b) these conventions are “effectively identical”, which entails that “it is meaningful to speak of a common syntax of West Germanic, which must have existed in Proto-West-Germanic and which is evidenced in the major languages in the group”;

130

Old English and Old High German Biblical Translations

c) as a consequence, “the study of the syntax of one Old West Germanic language is an indicator of the syntax of others; in particular the syntax of the better-recorded Old English acts as a guide to the syntax of Old High German”. If OE and OHG are supposed to behave in a similar, if not identical, way, then their similarity should be even more pronounced if we compare texts from the same domain, i.e. biblical translations (the Tatian Gospel Translation used in the present study was also the material used by Davis & Bernhardt in their book; the material they analysed for Old English were homilies by Aelfric). However, if some important differences appear, it needs to be determined if they are caused by the source text interference or if they can be treated as actual differences between the Old English and Old High German syntactic systems.

2. Research Questions and Methodology The following research questions have been formulated for the purposes of the study: a) Do V-1 main declaratives appear in the translations independently of the Latin original or are they (mostly) syntactic calques? b) If V-1 pattern is employed in the target text, does the clause need to fulfil a certain (set of) condition(s)? c) If the Latin original represents the V-1 pattern, is it usually copied or modified in the translated text? In order to answer these questions, a sample of OE and OHG biblical translations has been analysed in detail. Table 1. The study corpus Language Text

Date of composition No. of clauses in the source text No. of clauses in the target text

Old English Heptateuch (The Book of Genesis) Late 10th century 522 492

Old High German Tatian Gospel Translation Early 9th century 505 507

Anna Cichosz

131

As shown in Table 1, the sample collected for the purpose of the study comprises ca. 500 clauses for both translations, and the equivalent number of clauses from the Latin originals. Biblical translations were selected in order to eliminate any potential differences related to style or attitude of the translator towards the translated text. When translating the Bible, both the OE and the OHG translators had to be very cautious not to intervene in the contents of the sacred text, so the possibility of Latin interference may be regarded as equal, or at least close. However, there were other problems related to the proper selection of texts which proved to be more difficult to eliminate. First of all, the date of composition of both translations is different, the OHG text being at least 150 years older. Unfortunately, it was impossible to organize the corpus in a different way since there is no OHG biblical translation composed later, and all OE translations of the New or Old Testament were created in the 10th century (except for some earlier translations of psalms). It was equally impossible to investigate the use of V-1 main declaratives on the basis of other translations because all OHG translations are religious in character, so there is no comparative material for e.g. the OE version of Bede’s Historia Ecclesiastica Gentis Anglorum. Thus, in order to conduct a comparative study of translations in OE and OHG, it is necessary to accept the 150-year difference, taking into consideration the fact that this discrepancy may have some influence on the results of the study. On the one hand, some differences which are due to the diachronic development of one or both languages may be identified. When the OHG translation was being created, the language had hardly any literary traditions, i.e. Tatian is one of the oldest texts codifying OHG, whereas in the 10th century OE already had a long tradition of glossing, translating and writing, which may have given the translator much more confidence. In addition, the translator of the OE text is not anonymous; for the purposes of the study, the first 6 chapters of the Book of Genesis were analysed, and it is known that Aelfric himself is responsible for the creation of this part of the translation (Marsden 1991, Stanton 2002). Nonetheless, the problem of comparability of OE and OHG sources has been known for decades (for a detailed discussion see Smith 1971 and Cichosz 2010) and, since there is no perfect solution to this problem, it is necessary to choose texts as carefully as possible and accept the unavoidable weaknesses of any research design. All the methodological problems mentioned here will have an influence on the interpretation of the results, which are provided in the following section.

132

Old English and Old High German Biblical Translations

3. Results and Analysis At the beginning of the analysis, clauses were annotated syntactically. Then, source and target clauses were linked, showing to what extent the translation was done clause by clause (in Tatian this technique was used in 98% of all clauses), how often the translator omitted Latin clauses (this proved to be a relatively frequent situation in the OE text, where the translator often shortened long catalogues of genealogical relations), paraphrased the source clause or added clauses without any direct syntactic equivalent in the source text. All the techniques are summarised in Table 2. Table 2. Relations between source and target texts text

1 to 1

%

omissions

%

paraphrasing

%

additions

total

Tatian Genesis

496 442

98% 85%

1 45

0% 9%

8 35

2% 7%

0 19

505 522

For the purpose of the analysis, only clauses in 1 to 1 equivalence will be considered, since this is the only context which makes it possible to analyse the possible Latin interference (when a single Latin clause is translated by a single OE or OHG clause, the same word order patterns in both clauses may suggest an influence, whereas any divergences prove that the translator felt the need to modify the original word order). Since the study is concerned with V-1 main declarative clauses, the corpus needed to be narrowed down to this specific clause type. As a result, the final size of the analysed sample was 238 clauses for OHG and 260 clauses for OE (plus the same number of equivalent Latin clauses in both cases).

3.1. The V-1 Pattern and the Weight of Subjects As shown in Table 3, the position of the finite verb in main declarative clauses turns out to be visibly different in both translations under investigation. The main discrepancy concerns the relative frequency of the V-1 and V-2 pattern.

Anna Cichosz

133

Table 3. Position of the finite verb in main declarative clauses in the target texts text Tatian Genesis

V-1 109 37

% 46% 14%

V-2 93 162

% 39% 62%

V-3 27 51

% 11% 20%

V-F 9 10

% 4% 4%

total 238 260

In Tatian as many as 109 clauses (46% of the whole sample) demonstrate the V-1 pattern, which is a very high frequency for a structure claimed to be marked. In the OE pattern only 37 clauses can be counted as V-1 (14% of the sample). In the OE translation the dominant pattern is V2, whereas in OHG the proportions are reversed. Nonetheless, before reaching any final conclusions, it is necessary to check if the V-1 clauses in question contain an expressed subject, since a V-1 clause with an unexpressed subject is by no means a violation of OE or OHG syntactic rules; there is an empty subject slot before the verb which could theoretically be occupied by the subject, as e.g. in: (4) & het ðæt leoht dæg and called the light day (OE, cl. 11)3 (5) & bletsode hi and blessed them (OE, cl. 65) (6) quad iru: told her (OHG, cl. 115) (7) inphieng thô antvvurti fon themo heilagen geiste received then answer from the holy spirit (OHG, cl. 408) Out of the 109 V-1 clauses found in the OHG translation, only 35 contain an expressed subject, as in:

3

Numbers assigned to clauses in the database created for the purposes of the project (separate numbering for source texts and target texts).

134

Old English and Old High German Biblical Translations

(8) Inti quad Zacharias and said Zacharias

zi themo engile to the angel (OHG, cl. 55)

(9) Uuard thô gifullit heilages geistes Helisabeth was then filled with holy spirit Elizabeth (OHG, cl. 135) (10) Uuarun thô were then

hirta in thero lantskeffi shepherds in the land (OHG, cl. 345)

In the case of the OE text, only 4 V-1 clauses contain an expressed subject: (11) & wæs ða geworden æfen & mergen oðer dæg and was then made evening and morning the second day (OE, cl. 23) (12) & wæs and was

geworden æfen & mergen se ðridda dæg made evening and morning the third day (OE, cl. 40)

(13) & wæs geworden æfen & mergen se feorða dæg and was made evening and morning the fourth day (OE, cl. 58) (14) Nis na god not-is at all good

ðisum men ana to wunigenne for this man alone to live (OE, cl. 160)

Examples (11)-(14) clearly show that all the clauses fulfil the same condition: they contain a very long, heavy subject, either a complex phrase (examples (11)-(13)) or a non-finite clause (example (14)). Thus, weight must have played a role in the application of this pattern, the heavy phrase being postponed till the end of the clause, as: ... light, i.e. phonologically short, often adverbial or pronominal, forms, were preferred clause-initially, especially in the middle period around 1000, and heavy elements, typically complex phrases or subordinate clauses, were preferred sentence-finally (Traugott 1992: 276).

Anna Cichosz

135

All the other OE V-1 clauses do not contain an expressed subject, and they are mostly conjoined (35 out of 37 clauses – 94%). In the case of OHG, conjoining seems to be much less prominent – only 50 out of 108 clauses are conjoined (46%), and subjects are much shorter (cf. examples (8)-(10)); in 3 cases the subjects are light (pronominal): (15) inti nemnis thû and call you

sinan namon Iohannem his name John (OHG, cl. 44)

(16) giforhtun sie im thô in mihhilero forhtu were afraid they REFL then with great fear (OHG, cl. 351) (17) brahtun brought

sie they

inan him

thô then

in Hierusalem to Jerusalem (OHG, cl. 396)

Therefore, it can be concluded that while in OE a heavy subject could result in a main declarative clause showing a V-1 pattern, in OHG weight seems not to exert any influence on the usage of the pattern in question.

3.2. The V-1 Pattern and the Verb As far as verbs co-occurring with the V-1 pattern are concerned, their usage is demonstrated in Table 4. Table 4. Verbs co-occurring with the V-1 pattern (subject expressed) motion verbs verbs of saying verbs of perception ‘to be/become’ negated verbs other TOTAL

Old English 3 1 4

Old High German 10 5 3 11 6 35

Table 4 indicates that both languages under investigation use the V-1 pattern with the verbs ‘to be/become’, though in OE other categories are not well represented because of the low frequency of the structure in the analysed corpus. Thus, it is impossible to determine if this difference is

136

Old English and Old High German Biblical Translations

due to a real discrepancy between the two languages (which is rather improbable since V-1 structures co-occur quite often with motion verbs in Anglo-Saxon poetry or with verbs of saying, which was mentioned by Mitchell 1985) or if it is a simple result of underrepresentation of the structure in this specific text (which seems more plausible).

3.3. The V-1 Pattern and Latin Interference So far, it has been demonstrated that there are some visible differences between the two analysed samples. The next factor which should be taken into consideration is Latin interference. It needs to be determined whether the use of V-1 main declaratives is a direct result of copying or if the structure appears in the texts independently. In the case of OE clauses, all four V-1 clauses are not word by word translations: (18) & wæs ða geworden æfen & mergen oðer dæg and was then made evening and morning the second day (OE, cl. 23) et factum est vespere et mane dies secundus and made was evening and morning the second day (L, cl. 23) (19) & wæs geworden æfen & mergen se ðridda dæg and was made evening and morning the third day (OE, cl. 40) factumque est vespere et mane dies tertius made-and was evening and morning the third day (L, cl. 40) (20) & wæs geworden æfen & mergen se feorða dæg and was made evening and morning the fourth day (OE, cl. 58) et factum est vespere et mane dies quartus and made was evening and morning the fourth day (L, cl. 60) (21) Nis na god ðisum men ana to wunigenne not-is at all good for this man alone to live (OE, cl. 160) non est bonum not-is good

esse hominem solum to be a man lonely (L, cl. 162)

Anna Cichosz

137

In clauses 18-20, which all contain a complex verb phrase, we can see a regular correspondence between the OE finite – non-finite and the Latin non-finite – finite order (though in both languages the complex verb phrase can be found at the beginning of the clause, so there is a similarity between the target and the source clauses). Example (21) shows that the Latin clause was completely restructured (OE uses an indirect object followed by a non-finite clause acting as a subject, whereas in Latin the whole message is carried by the subject), though the position of the finite verb is the same. However, the differences in the position of other clause elements seem to exclude copying as the sole source of the structure; weight of subjects seems to play a much more important role in this respect, though the influence of Latin cannot be excluded. In the OHG text, 9 out of 35 V-1 clauses are not copied, demonstrating a different position of the finite verb than the original: (22) Uuard thô gifullit heilages geistes Helisabeth was then filled with holy spirit Elizabeth (OHG, cl. 135) Et repleta and filled

est spiritu sancto Elisabeth is with spirit holy Elizabeth (L, cl. 135)

(23) vvuntorotun thaz wondered (about) this

thô then

alle all (OHG, cl. 191)

et mirati and amazed

sunt universi were all (L, cl. 191)

(24) Gioffonota sih thô sliumo sîn mund inti sîn zunga opened REFL then immediately his mouth and his tongue (OHG, cl. 192) Apertum est autem ilico os eius et lingua eius opened is then immediately mouth his and tongue his (L, cl. 192) (25) Uuard thô was then

forhta ubar alle iro nahiston fear over all their neighbours

Et factus est and made was

(OHG, cl. 195) timor super omnes vicinos eorum fear over all neighbours theirs (L. cl. 195)

138

Old English and Old High German Biblical Translations

(26) uuas siu fundan was she found inventa est found was (27) vvurÿun were

sô scaffaniu fon themo heilagen geiste pregnant by the holy spirit (OHG, cl. 297) in utero habens de spiritu sancto in womb having by spirit holy (L, cl. 297)

taga gifulte days fulfilled (OHG, cl. 339)

impleti fulfilled

sunt dies were days (L, cl. 339)

(28) Uuarun thô were then

hirta in thero lantskeffi shepherds in the land (OHG, cl. 345)

Et pastores erant in regione eadem and shepherds were in region same (L, cl. 345) (29) Vvurbun turned

thô then

thie hirta the shepherds

heimuuartes homeward (OHG, cl. 384)

Et reversi sunt and returned are

pastores shepherds (L, cl. 382)

(30) uuard was

imo him

ginemnit namo Heilant given name Saviour (OHG, cl. 392)

vocatum called

est was

nomen eius Ihesus name his Jesus (L, cl. 390)

In some cases (examples (22), (26), (27), (30)) both the source and the target clause contain a complex verb phrase and the order of the finite and non-finite forms is regularly reversed (like in OE); the influence of Latin is not strong enough to violate this pattern. In other cases (examples (23), (24), (25), (29)) Latin contains a complex verb phrase which is translated by a single verb in OHG. Since in all such examples in the Latin original the complex verb phrase takes the first position in the clause (sometimes

Anna Cichosz

139

introduced by a coordinating conjunction), it may be concluded that there is some kind of V-1 model structure in the original text, so even though there is no word by word copying, Latin might have influenced the use of the V-1 pattern in these cases, as opposed to V-2. Only in example (28) is the situation different; in the Latin clause the subject precedes the verb, whereas OHG reverses their order. Here, and only here, the influence of Latin can be safely excluded. The remaining 26 OHG V-1 clauses copy the position of the finite verb from the original, as e.g. in: (31) Quad said

tho then

Maria zi themo engile: Mary to the angel (OHG, cl. 111)

Dixit said

autem Maria ad angelum: then Mary to angel (L, cl. 111)

(32) Uuoneta Maria mit iru nah thri manoda lived Mary with her for three months (OHG, cl. 167) Mansit autem Maria cum illa quasi mensibus tribus lived also Mary with her about three months (L, cl. 167) (33) brahtun sie brought they

inan him

thô then

in Hierusalem to Jerusalem (OHG, cl. 396)

tulerunt illum in Hierusalem brought him to Jerusalem (L, cl. 394) The only intervention in the original structure is the provision of a pronominal subject, as in example (33); in Latin the subject is not expressed (such a situation can be found in 3 OHG clauses). The examples show that the subject does not need to be heavy to appear in a V-1 structure, unlike in OE. Summing up the situation in the OHG translation, it is possible to claim that in most cases the use of the V-1 pattern coincides with the same (or very similar) structure in Latin, and only 1 out of 9 cases of modifying the position of the finite verb may be treated as truly independent. In the OE text the structure is rare, and when it is used, there is always at least a slight modification of the original order, though when the complex verb

Old English and Old High German Biblical Translations

140

phrase takes the initial position in the source clauses, Latin interference cannot be fully excluded.

3.4. Translating V-1 Pattern from Latin V-1 main declarative clauses seem very rare in the OE translation and quite widespread in the OHG text, even though the pattern is equally frequent in the two source texts; V-1 structures are very well represented in biblical Latin, though they are rendered differently in the two translations under investigation, as shown in Table 5. Table 5. Latin V-1 clauses in translation text

Tatian Genesis

V-1 to V-1 82 31

%

89% 23%

V-1 to V-2 10 76

%

11% 58%

V-1 to V-3 0 22

%

0% 17%

V-1 to V-F 0 3

%

total

0% 2%

92 132

Table 5 clearly indicates that in Tatian V-1 clauses are mostly copied (89%), whereas in the OE translation they are usually changed into V-2 (58%). In order to understand the situation fully, however, it is necessary to focus on V-1 clauses with an expressed subject. In the translation from Latin to OE, 58 out of 132 Latin clauses contain an expressed subject, as shown in Table 6. Table 6. Latin V-1 clauses and their OE translations Latin V-1 without an expressed S

Rendered as OE V-1 without an expressed S

Rendered as OE V-1 with an expressed S

Rendered as OE V-2, V-3 or V-F with an expressed S

74 Latin V-1 with an expressed S

30 (40%) Rendered as OE V-1 without an expressed S

0 (0%) Rendered as OE V-1 with an expressed S

42 (57%) Rendered as OE V-2, V-3 or V-F with an expressed S

58

0 (0%)

1 (2%)

57 (98%)

Rendered as OE V-2, V3 or V-F without an expressed S 2 (3%) Rendered as OE V-2, V3 or V-F without an expressed S 0 (0%)

Anna Cichosz

141

In the case of the remaining 74 clauses without an expressed subject, the OE translator in 30 cases copies the order and translates them into V-1 clauses without an expressed subject, as e.g. in: (34) & sealde and gave

hyre were to her husband (OE, cl. 212)

deditque gave-and

viro suo to husband hers (L, cl. 211)

However, in 42 cases the translator provides a pronominal or nominal subject in the translation, changing the original V-1 pattern into V-2 or V3: (35) & ðu and you

ytst eat

ðære eorðan wyrta earth’s herbs (OE, cl. 268)

et comedes and eat

herbas Terre herbs of earth (L, cl. 264)

(36) Witodlice truly

Matusalem Matusalem

gestrynde Lamech begot Lamech (OE, cl. 410)

et and

genuit begot

Lamech Lamech (L, cl. 439)

Interventions such as the one shown in example (36) are quite extensive and change the whole structure of the clause. A pronominal subject is provided in 28 cases (out of 42; 67%), in the remaining instances a full noun is used in the translation regardless of its lack in the source clause. It is only in 2 cases that neither the target clause nor the source clause contains an expressed subject, even though the position of the finite verb is slightly modified: (37) & hine and him

gehalgode sanctified (OE, cl. 111)

et sanctificavit illum and sanctified him (L, cl. 114)

142

Old English and Old High German Biblical Translations

(38) & on ableow on his ansyne lifes orðunge and on breathed in his face life’s air (OE, cl. 127) et inspiravit and breathed

in faciem eius spiraculum vitae in face his air of life (L, cl. 131)

In example (37) the position of the object is changed, whereas in example (38) a particle ‘on’ is added in the translation, though it is so closely connected to the verb that the modification is practically negligible. In the case of the 58 Latin V-1 clauses with an expressed subject, the subject is also always expressed in the OE translation, though its position is changed in 57 cases except for the clause which has already been presented, i.e. (39) Nis na god not-is at all good

ðisum men ana to wunigenne for this man alone to live (OE, cl. 160)

non est not-is

bonum good

esse hominem solum to be a man lonely (L, cl. 162)

A much more common situation is illustrated with examples (40)-(42), where the Latin V-1 order is changed into V-2, V-3 or V-F respectively: (40) God God

geseah ða saw then (OE, cl. 38)

et and

vidit saw

Deus God (L, cl. 38)

(41) & God ða gefylde on ðone seofoðan dag his weorc and God then finished on the seventh day his work (OE, cl. 106) conplevitque finished-and

Deus die septimo opus suum God day seventh work his (L, cl. 109)

Anna Cichosz

(42) & God and God

143

hi bletsode them blessed (OE, cl. 85)

benedixitque blessed-and

illis Deus them God (OE, cl. 87)

In the OHG text the variation is much more limited – V-1 clauses are only rendered as V-1 or V-2 in the translation. As shown in Table 7, out of 92 Latin V-1 clauses, only 26 contain an expressed subject. Table 7. Latin V-1 clauses and their OHG translations Latin V-1 without an expressed S 66 Latin V-1 with an expressed S 26

Rendered as OHG V-1 without an expressed S 56 (85%) Rendered as OHG V-1 without an expressed S 0 (0%)

Rendered as OHG V-1 with an expressed S 3 (4%) Rendered as OHG V-1 with an expressed S 23 (88%)

Rendered as OHG V-2 with an expressed S 7 (11%) Rendered as OHG V-2 with an expressed S 3 (12%)

Rendered as OHG V-2 without an expressed S 0 (0%) Rendered as OHG V-2 without an expressed S 0 (0%)

In the case of the 66 clauses without an expressed subject, only in 10 clauses does the OHG translator provide a subject in the target text, e.g.: (43) giforhtun sie im thô in mihhilero forhtu feared they REFL then with great fear (OHG, cl. 351) et timuerunt and feared

timore magno with fear great (L, cl. 350)

(44) thu fundi huldi mit gote you find kindness with God (OHG, cl. 102) invenisti find

enim gratiam apud deum then kindness with God (L, cl. 102)

144

Old English and Old High German Biblical Translations

Only 3 out of the 10 clauses follow the V-1 pattern in the translation (like in example (43)); in 7 cases the expression of subject triggers the V-2 pattern (like in example (44)). All 10 subjects are pronominal, unlike in OE, where the translator provided a nominal subject against the source text in 33% of cases. When the subject in Latin V-1 clauses is expressed, the OHG translation always repeats it, e.g.: (45) framquam gibot fon ÿemo aluualten keisure came edict from the omnipotent Cesar (OHG, cl. 328) exiit edictum a CĊsare Augusto came edict from Cesar Augustus (L, cl. 328) (46) Thô then

quad Maria said Mary

Dixit said

autem Maria then Mary

(OHG, cl. 124)

(L, cl. 124) In 3 cases (like in example (46)) the V-1 pattern is changed to V-2; in 23 cases the translation follows the original (like in example (45)). Summing up, it is visible that the OHG translation follows the original much more closely, though modifications also appear. The influence of Latin seems much stronger here, but it cannot be completely excluded in the case of the OE text, either.

3.5. Summary – Answers to Research Questions The study aimed to find answers to three basic research questions: a) Do V-1 declaratives appear in the translations independently of the Latin original or are they (mostly) a syntactic calque? In the OE translation there are 37 V-1 main declarative clauses, but only 4 of them (11%) contain an expressed subject. In all these cases the subject is heavy, which must have influenced the use of the structure. In addition, all 4 clauses may have been modelled on the Latin original: one of them has exactly the same position of the finite verb (example (14)), which is clearly related to negation; in the other three cases the Latin

Anna Cichosz

145

original has a different order of the finite and non-finite verb form, but the whole complex verb phrase can be found at the beginning of the clause. In the OHG translation there are 109 V-1 declaratives and 35 of them (32%) contain an expressed subject. Within this group, 26 clauses have exactly the same position of the finite verb, and in 8 clauses Latin uses a complex verb phrase at the beginning of the clause (examples (22)-(27) and (29)-(30)), which means that there is a possible source of interference, even though the position of the finite verb itself is different. There is only one case (example (28)) in which the influence of Latin can clearly be excluded. b) If V-1 pattern is employed in the target text, does the clause need to fulfil a certain (set of) condition(s)? In the case of OE, it seems that a very heavy subject licenses the use of the V-1 structure in main declarative clauses. This tendency in one case (example (14)) is additionally strengthened by negation. Most uses of the V-1 pattern involve lack of an expressed subject (33 out of 37) and conjoining (35 out of 37). In the OHG translation there seems to be no straightforward correlation of the V-1 pattern with subject weight (subjects are mostly single nouns, 4 out of 35 subjects are pronouns). Conjoining is also much less common; only 11 out of 35 V-1 clauses with an expressed subject are conjoined. However, it is possible that other factors are involved here: the V-1 pattern often coincides with specific verbs (cf. Section 3.2) or may be related to discourse structure. According to Hinterhölzl & Petrova (2005) V-1 main declaratives are used when a new storyline is introduced into the narration, which seems to work in most cases, e.g. in clause 22 below (even though in this study not only deviations were analysed, contrary to Hinterhölzl & Petrova’s methodology): (47) Inti thaz lioht in finstarnessin liuhta and the light in darkness shines (OHG, cl. 20) inti finstarnessi thaz ni bigriffun. and darkness it not penetrates (OHG, cl. 21)

146

Old English and Old High German Biblical Translations

Uuas in tagun Herodes thes cuninges Iudeno sumer was in days of Herod the king of Judea a biscof namen Zacharias bishop called Zacharias fon themo uuehsale Abiases inti quena imo from the house of Abia and wife his fon Aarones tohterun of Aaron daughter (OHG, cl. 22) However, there are cases when the subject is pronominal, which implies that the person/object mentioned in the sentence is not a new element of the narration, as e.g. in clause 396 below: (48) Inti after thiu gifulta uuarun taga sinero subarnessi and after fulfilled were days of his purification after Moyseses euuu, according to Moses’ law (OHG, cl. 395) brahtun sie inan thô in Hierusalem, brought they him then to Jerusalem (OHG, cl. 396) This is definitely not a new storyline, “they” are Joseph and Maria, mentioned a few clauses earlier. The discourse function of V-1 main declarative clauses seems to be less pronounced in the case of pronominal subjects, and 4 out of 35 OHG subjects are pronouns. c) If the Latin original represents the V-1 pattern, is it usually copied or modified in the translated text? In the OE translation, V-1 Latin clauses without an expressed subject are often copied (30 out of 74 clauses; 41%); in the remaining instances the subject is expressed in the translation against the Latin original, and the provided subjects are both pronominal (28 out of 42) and nominal (14 out of 42). However, when the subject is expressed, OE consistently changes its position, modifying V-1 into V-2, V-3 or V-F (only in 1 clause

Anna Cichosz

147

out of 58 is the subject inverted, but the inversion is caused by the negative adverb ne). In the OHG text, if the Latin clause does not contain an expressed subject, the translator follows the original order in 51 out of 66 cases (there are only 10 clauses in which the subject is expressed against the source model and all these subjects are pronominal). When the Latin subject is expressed, the OHG translation mostly copies the V-1 pattern (23 out of 26 cases); only 3 clauses change the structure from V-1 to V-2 in this case.

4. Conclusions and Directions for Further Research The analysis presented above indicates that there seem to be some visible correlations between the use of V-1 in the OE and OHG sample and the original word order of the Latin clause. Nonetheless, a very important question which should be asked here is when a structure which seems to be a copy is a syntactic calque and when is it just a coincidental correspondence, representing native syntax of both languages, which seems to be the case e.g. in example (14). The analysis of certain cases seems very straightforward: when the translation modifies the original word order, there is no foreign interference, and when the translation copies the original and the resulting clause seems exceptionally rare when compared to evidence from other text types (e.g. poetry or original prose), we can easily assume that the structure is a calque. However, the solution is much less clear when the order of both clauses is the same, or very similar, and yet the OE or OHG clause seems very ‘Germanic’ in character and if it appeared in an original prose or verse text, no one would suspect it of being modelled on Latin. Discussing the word order of Tatian, Dittmer & Dittmer (1998: 18) notice: In most cases the word order of Old High German is the same as the one in the Latin, and in such cases it is not possible to determine whether the structure is native or not.

The clause which has the same word order as its Latin equivalent is not necessarily ‘wrong’. Nonetheless, in the case of OHG, which is known mostly on the basis of translated sources, a very popular methodology is to analyse ‘deviations’ only, i.e. clauses where the translator modified the order of the original clause (as practised by Dittmer & Dittmer 1998 and Petrova & Solf 2008). This is not done in OE, though, probably because we feel much more secure about the quality of the OE translations (cf. Stanton 2002, Reszkiewicz 1966) and there are many other texts which

148

Old English and Old High German Biblical Translations

may be treated as points of reference. By ignoring the ‘syntactic coincidences’ in OHG, we may be excluding some valuable material from the analysis, because “... we must then note also that ‘literalism’ does not necessarily involve the writing of bad, let alone incomprehensible, OE prose” (Marsden 1991: 322), and it does not need to involve the writing of ‘bad’ OHG prose either. On the whole, it seems that OE and OHG V-1 main declarative clauses cannot be treated simply as a syntactic calque, first and foremost because the pattern (though marked) is present in all Old Germanic languages and is widely attested in Germanic poetry, where the possibility of Latin influences is very small. However, it must be noted that Latin drastically increases the frequency of the pattern in the analysed sample of the OHG translation. Despite the high frequency of the structure in both source texts, OE seems much more immune to the Latin influence and very consistently modifies the structure, most probably because of the later time of composition of the OE translation (the V-1 pattern may have become more marked around 1000 AD so it started to be avoided, in contrast to OHG which uses it freely) and Aelfric being a more experienced translator, well educated and aware of the nature of his task. Thus, the best summary of the nature of the structure seems to be: A (little-utilized) native pattern (…), a pattern perhaps supported by (but not caused by) Latin syntactic models (Lippert 1974: 53).

The V-1 structure in main declarative clauses was a marked native pattern, clearly avoided in the OE translation, and in the OHG translation strongly promoted by Latin syntactic models. The difference between the two samples may be related both to the growing markedness of the structure in OE (Mitchell 1985 mentions its frequent usage in Bede, which is an earlier translation, and poetry, which must have existed in oral tradition long before its codification) and to the lesser independence of the OHG translators, who may have felt less confident because of the virtually nonexistent literary tradition of OHG at the beginning of the 9th century. A more detailed study of a wider selection of different OE and OHG translations, which our project team plans to undertake, should shed more light on the problem and it will make it possible to test the above-mentioned hypothesis on a more varied sample of texts. It seems quite certain, though, that there are some visible differences between the OE and OHG word order found in the analysed samples (e.g. the influence of subject weight), so the claims about a common “West Germanic” or even “Old Germanic” syntax (Davis & Bernhardt 2002, Davis 2006) seem to be an overgeneralization; OE and OHG were independent linguistic systems and

Anna Cichosz

149

they should be treated as such, because transferring observations from one language to the other, tempting as it sounds, may prevent us from noticing some specific characteristics related to their individual development.

Acknowledgements The present study has been completed as part of the project N N104 379140: A corpus-based study of the influence of Latin syntax on the word order of selected OE and OHG translations, financed by the Polish Ministry of Science and Higher Education.

Works Cited Primary Sources Crawford, Samuel J. (ed.) 1922. The Old English Version of the Heptateuch. London: Early English Text Society 160. Electronic version retrieved from the Dictionary of Old English Corpus on CDROM. Sievers, Eduard (ed.). 1966. Tatian: Lateinisch und altdeutsch mit ausführlichem Glossar. Schöningh: Unveränderter Nachdruck Paderborn. (Electronic version retrieved from Titus: Thesaurus Indogermanischer Text- und Sprachmaterialien, http://titus.uni-frankfurt.de/texte/etcs/ germ/ahd/tatian/tatia.htm) Weber, Robert & Roger Gryson (eds.). 1994. Biblia Sacra Vulgata. Editio quinta. Stuttgart: Deutsche Bibelgesellschaft. (Electronic version retrieved from The Scholarly Bible Portal of the German Bible Society, http://www.academic-bible.com/en/online-bibles/vulgate/)

Secondary Sources Axel, Katrin. 2007. Studies on Old High German Syntax: Left Sentence Periphery, Verb Placement and Verb Second. Amsterdam & Philadelphia: John Benjamins. Behagel, Otto. 1932. Deutsche Syntax: eine geschichtliche Darstellung [Band IV: Worstellung. Periodenbau]. Heidelberg: Carl Winter. Cichosz, Anna. 2010. The Influence of Text Type on Word Order of Old Germanic Languages: A Corpus-based Contrastive Study of Old English and Old High German. Frankfurt am Main: Peter Lang.

150

Old English and Old High German Biblical Translations

Davis, Graeme. 2006. Comparative Syntax of Old English and Old Icelandic: Linguistic, Literary and Historical Implications. Oxford: Peter Lang. Davis, Graeme & Karl A. Bernhardt. 2002. Syntax of West Germanic: The Syntax of Old English and Old High German. Göppingen: Kümmerle Verlag. Dittmer, Arne & Ernst Dittmer. 1998. Studien zur Wortstellung – Satzgliedstellung in der althochdeutschen Tatianübersetzung. Göttingen: Vandenhoeck & Ruprecht. Hinterhölzl, Roland & Svetlana Petrova. 2005. “Rhetorical Relations and Verb Placement in Early Germanic Languages. Evidence from the Old High German Tatian Translation (9th century)”. Salience in Discourse: Multidisciplinary Approaches to Discourse 2005 ed. by Manfred Stede, Christian Chiarcos, Michael Grabski & Luuk Lagerwerf, 71-79. Münster: Stichting/Nodus. Lippert, Jörg. 1974. Beiträge zu Technik und Syntax althochdeutscher Übersetzungen unter besonderer Berücksichtigung der Isidorgruppe und des althochdeutschen Tatian. München: Wilhelm Fink. Lockwood, William B. 1968. Historical German Syntax. Oxford: Clarendon Press. Marsden, Richard. 1991. “Ælfric as Translator: The Old English Prose Genesis”. Anglia – Zeitschrift für englische Philologie 109.319-358. Mitchell, Bruce. 1985. Old English Syntax (Volume II). Oxford: Clarendon Press. Penzl, Herbert. 1986. Althochdeutsch. Bern, Frankfurt am Main & New York: Peter Lang. Petrova, Svetlana. 2006. “A Discourse-based Approach to Verb Placement in Early West-Germanic”. Working Papers of the SFB632, Interdisciplinary Studies on Information Structure (ISIS) 5 ed. by Shinichiro Ishihara, Michaela Schmitz & Anne Schwarz, 153-182. Potsdam: Universitätsverlag Potsdam. Petrova, Svetlana & Michael Solf. 2008. “Rhetorical Relations and Verb Placement in Early Germanic: A Cross-linguistic Study”. ‘Subordination’ vs. ‘Coordination’ in Sentence and Text: a Crosslinguistic Perspective ed. by Cathrine Fabricius-Hansen & Wiebke Ramm, 329-352. Amsterdam & Philadelphia: John Benjamins. Reszkiewicz, Alfred. 1966. Ordering of Elements in Late Old English Prose in Terms of their Size and Structural Complexity. Wrocáaw, Warszawa & Kraków: Wydawnictwo PAN. Robinson, Orrin W. 1996. Clause Subordination and Verb Placement in the Old High German Isidor Translation. Heidelberg: Carl Winter.

Anna Cichosz

151

Smith, Jesse. 1971. Word Order in the Older Germanic Dialects. Unpublished Ph.D. Dissertation, University of Illinois. Stanton, Robert. 2002. The Culture of Translation in Anglo-Saxon England. Cambridge: D. S. Brewer. Traugott, Elizabeth C. 1992. “Syntax”. The Cambridge History of the English Language Vol.1: The Beginnings to 1066 ed. by Richard M. Hogg, 168-289. Cambridge: Cambridge University Press.

THE GRAMMATICALIZATION OF OLD ENGLISH WHETHER IN THE LEFT PERIPHERY VÍCTOR PARRA-GUINALDO

1. Introduction The introduction of functional categories during the 1980s (TP = tense phrase, DP = determiner phrase, CP = complementizer phrase) was of particular interest to grammaticalization theory, as these were considered the locus of change. It is not until the turn of the century though that some generativists have tried to provide answers to the grammaticalization process within the view of the Minimalist Program. Literature concerning the grammaticalization of the TP and DP abound, but the CP has been neglected until very recently. The grammaticalization of whether is a case in point. To date only three studies are worth mentioning: Ukaji (1997) provides a merely descriptive account of the evolution of this word in English, Allen (1980) has a very brief article based on the then current Government and Binding framework, and van Gelderen (2009) uses whether as an example of an element grammaticalizing into the left periphery of the clause, thus explaining the process using minimalist concepts. The aim of this paper is to review the theoretical framework under which a proper, more current, study ought to be carried out in order to explain the evolution of whether over the history of English, but with an emphasis on Old English (OE); to clearly show Ukaji’s typological exposition of the evolution of whether; and finally, as a conclusion, to propose briefly that further research be carried out in order to explain the grammaticalization process this element is claimed to have undergone based on economy as a main motivating factor under the Minimalist Program following van Gelderen’s (2004 and later) economy principles: Late Merge and Head Preference.

Víctor Parra-Guinaldo

153

2. Theoretical Framework 2.1. The Minimalist Program The Minimalist Program (MP) is the most recent instantiation of generative grammar, and it evolves from the Principles and Parameters (P&P) model. As its name indicates, it is not a theory, but a program that allows for different minimalist models. MP was first outlined in Chomsky’s (1993) seminal article “A Minimalist Program for Linguistic Theory” and in Chomsky (1995) The Minimalist Program, as well as in later and related work. MP continues to address Plato’s problem (how children can acquire grammatical competence despite the impoverished nature of the primary linguistic data or PLD, which is the input to this process), but it succeeds its earlier P&P models (Government and Binding (GB) in particular) in that it rids itself of superfluous constructs, such as unnecessary levels of representation (the internal levels: deep structure (DS) and surface structure (SS)), relations of government, trace theory, theta structure, construction specific rules for wh-movement, raising, and the like. Thus MP highlights the importance of economy, reminiscent of Chomsky’s earliest work on generalized transformations of 1955. Two main ideas involving economy were developed under MP, namely economy of derivation and economy of representation. The former refers to the fact that movement only occurs in order to match interpretable with uninterpretable features; and the latter requires that grammatical structures exist for a purpose, that is, no larger or more complex than required. Under MP, the derivation includes the selection of a set of lexical items from the numeration; then, the mapping between the numeration and the output representations PF (phonetic form) and LF (logical form) entails a series of applications of two core syntactic processes, merge and move (or internal merge) (cf. Chomsky, 2001), which create the configurations in which the features of the lexical items are checked and eliminated. A probe–goal system would account for this feature checking. Thus, a probe seeks a goal in order to match features that establish agreement: the uninterpretable features of each get valued by the interpretable features of the other in an operation called agree. Once valued, the uninterpretable features get deleted via move. For the features of both probe and goal to engage in these two operations, they have to be active, ensuring they have been valued and deleted. Syntactic structures are built up one phase at a time, and the phase interpretability condition (PIC) ensures that the elements within the domain of a phase head are inert or impenetrable to an outside probe.

154

The Grammaticalization of Old English whether in the Left Periphery

The aspect of MP most relevant to the present paper is its use within historical linguistics. It is worth pointing out that generative grammar in general finds in acquisition the locus of linguistic change, or in other words, the acquisition of a grammar distinct from the output of the preceding generation. In early P&P, language change reflects an abrupt shift, which comes about through the resetting of parameters, such as prodrop, headedness, and wh-movement. For MP, syntactic change occurs in the lexicon (Pintzuk et al. 2000); more precisely, change is a consequence of the reorganization of the featural content of the lexical items of the language. Language change is therefore confined to the properties of the lexicon. The new interpretation of minimalist parameters as choices of feature specifications during the child’s acquisition of a lexicon is encapsulated in the Borer–Chomsky conjecture, as has been coined by Baker (2008). All parameters are encoded in the lexicon and they determine linearization, thus accounting for language variation. Recent work has moved away from a focus on an abrupt parametric change towards an emphasis on the gradual syntactic change as interpreted within the P&P paradigm. If the empirical data for the study of diachronic change could create a methodological problem for MP, as these data are Elanguage, that is, external as in historical texts, it is currently accepted by many that aspects of E-language data can be interpreted as the output of an underlying grammar; thus, the analysis of variation in E-language can reveal information about the nature and organization of the grammar.

2.2. The Complementizer Phrase The X’ theory (X-bar theory)1 is an attempt to identify syntactic features common to all human languages; according to this theory, for any category X, there is a fixed hierarchy of units. It constructs phrase structure as the syntactic projection of the argument structure of a lexical item; a head X projects a maximal constituent XP by being optionally combined with a complement, a number of modifiers or adjuncts, and a specifier. The head of a clausal structure introduced by a complementizer is the complementizer itself, allowing thus every syntactic structure to be a

1

The X’ theory was first proposed by Chomsky in 1970 and further developed by Jackendoff in 1977 in order to eliminate the redundancy of the Aspect model (Chomsky’s 1965 Aspect of the Theory of Syntax), where the information about possible phrase structures (PS) is coded twice, once in the PS rules and a second time in the lexical entries.

Víctor Parra-Guinaldo

155

projection of a head word.2 C merges with the Tense Phrase TP (or Inflection Phrase IP in earlier literature), that is, the projection of T, C becoming the Head of the overall clause. In other words, C takes a projection of T as its complement, and T takes VP (the Verbal Phrase) as its complement. Clauses introduced by complementizers have thus the status of CP (Complementizer Phrase), and this generates two implications: (1) phrases, clauses, and sentences are all projections of H words, and (2) phrases and clauses are derived (formed) in a bottom-up fashion. Following Rizzi (1997), I contend that C splits into several Heads, each projecting a separate phrase. The complementizer system is to be conceived of as a structural zone consisting of different heads and their projections,3 in the same fashion as the TP (Tense Phrase) and the DP (Determiner Phrase). Rizzi postulates a fixed component within the CP, namely heads for Force and Finiteness, and an accessory or optional component, namely heads for Topic and Focus. As for the former component, they can both coalesce into a single Head; and as for the latter, they need to be activated before they are used by means of topicalization or focalization. Rizzi (2001) adds an Int(errogative) component, and Top is optional in different positions between Force and Fin, as in (1); the complementizer system is thus viewed as the interface between a propositional content (TP) and the superordinate structure (a higher clause), where Force would consist of elements looking outside and Fin would have elements looking inside, and Top the constituent for old information and Foc the one for new information. (1) Force (Top*) Int (Top*) Foc (Top*) Fin

2.3. Grammaticalization Grammaticalization is defined by Hopper & Traugott (2003: xv) as “the change [rather than process] whereby lexical items and constructions come in certain contexts to serve grammatical functions and, once grammaticalized, continue to develop new grammatical functions”. Heine

2

We owe this notion to the pioneering work of Stowell (1981) and Chomsky (1986). 3 Cinque (1999) already explored the expanded CP, for which he coined the cartographic approach, in order to accommodate a wide range of adverb types: speech act, such as frankly; evaluative, such as unfortunately; evidential, such as allegedly; and modal, such as affixes.

156

The Grammaticalization of Old English whether in the Left Periphery

& Kuteva (2002: 2) characterize this phenomenon by means of these four interrelated mechanisms: (2) Desemanticization: semantic bleaching or loss in meaning content Extension: context generalization Decategorialization: loss in morphosyntactic properties Erosion: phonetic reduction or attrition Notice that the mechanisms in (2) cover the different aspects of language: semantics, pragmatics, morpho-syntax, and phonetics respectively. Core to grammaticalization is also the concept of unidirectionality, where changes tend to be, as is claimed, from less to more grammatical. Heine & Kuteva (2002) also include layering as an important factor, which refers to the persistence of older forms and meanings alongside the newer ones. Some scholars have proposed the inclusion of additional characteristics (sometimes referred to as principles) in the definition of grammaticalization, but they essentially refer to the principle characteristics listed above in (2). For example, Heine, Claudi & Hünnemeyer (1991) include recategorialization along with decategorialization.

2.4. Economy For Roberts & Roussou (2003), grammaticalization is a case of parameter change, but they consider it epiphenomenal. They postulate though that new functional material is created via structural simplification, a fact which van Gelderen (2004: 18, 28) addresses by means of her two economy principles: Head Preference and Late Merge. (3) Head Preference or Spec to Head Principle (HPP): Be a head, rather than a phrase (4) Late Merge Principle (LMP): Merge as late as possible Head Preference refers to the fact that incorporation, that is, checking between two heads, is more economical than checking between a specifier and a head. Late Merge is also stated as “merge over move,” where it is less economical to merge early and then move than to wait as long as possible before merging (van Gelderen 2004: 29).

Víctor Parra-Guinaldo

157

In (5) below, van Gelderen (2009) indicates the economy principles as moving forces of what has been known as the CP cycle. (a) represents the movement of a phrase from the bottom of the structure into the specifier of the CP; in (b) LMP ensures that the phrase be base generated within the CP rather than the VP, thus skipping the move operation; (c) indicates the reinterpretation of the phrase in the Spec of the CP into the head of the CP. After these three steps, a new lexical item may be incorporated within the VP in order to reinforce the function of the head of the CP, triggering the whole cycle all over again. It is important to note that the CP cycle may not always reach completion, as it may be interrupted at any point and a new lexical item may never get to replace the initial item affected by the grammaticalization process. (5) The CP Cycle (van Gelderen 2009: 152)

The cycle has also been restated in terms of feature reanalysis, where semantic features are in effect in its initial stage, as in (a) above, then there are interpretable features ([i-Q]) as in (b), and finally uninterpretable features ([u-Q]) as in (c). Thus the loss of semantic features is represented formally (van Gelderen 2009). The cycle of whether can be represented as in (6) in terms of semantic features (see van Gelderen 2009: 153): (6) whether > pronoun semantic

whether > CP specifier [i-Q]

whether head [u-Q]

158

The Grammaticalization of Old English whether in the Left Periphery

3. Whether 3.1. Etymology Before delving into the ‘true sense’ of the word, as the Greeks would put it, let us draw attention first to what it is not. Although of apparent similarity, English whether is unrelated to its homophones weather and wether. The former goes back to OE weder, in turn to Old Germanic *wedro-m, and to Proto-Indo-European *wƟ-dhro- (see OED s.v. weather); the OE equivalent of the latter is weðer, and it may be related to Latin vitulus of the same meaning (see OED s.v. wether). The etymology of the word which is relevant to the present study, whether, can likewise be traced back in time, as (7) shows. (7) OE hwæðer < PGmc *Ȥwaþaraz (*Xwe-) < PIE *kwo-teroNotice that the PIE etymon is composed of the element *kwo ‘what’ plus the comparative suffix *-tero-, which originally had the basic function “to set off a member of a pair contrastively from the other member of the pair” (Fortson 2004: 122). The comparative suffix is relevant to understanding the distributive nature of hwæðer ‘whether’ as it was used to introduce direct as well as indirect questions indicating alternatives: affirmative as opposed to negative in the case of yes/no questions, and one of two options in the case of disjunctive questions. The PIE etymon could then be interpreted as ‘what more’. Coincidentally, Sanskrit has a cognate ending in a superlative suffix with a possible interpretation of ‘which of many’ (Maurer 1995: 434). OE hwæðer has the following representative cognates indicated in (8). The reconstructed PGmc form is *Ȥwaþaraz, and related words in other IE languages are ʌóIJİȡȠȢ in Greek, utrum in Latin, and katará- in Sanskrit. (8) OFris hwed(d)er; OSax hweðar; OHG (h)wedar; ON hvaðarr; Goth hwaþar

3.2. Derivatives of hwæðer OE hwæðer had a series of derivative words that may have influenced the behavior of hwæðer during its grammaticalization process. I will mention these derivatives here (9) only in passing, as they ought to be explored in future research.

Víctor Parra-Guinaldo

159

(9) OE nǀhwæðer [ne ‘not’ + Ɨ/ǀ ‘always’ + hwæðer ‘whether’] ‘neither of two persons or things’ > nǀðer > (regional) nother4 > PDE nor OE Ɨhwæðer, Ɨwðer, ǀwðer, Ɨðer ‘either’

3.3. Spelling One important methodological aspect to be considered in the analysis of whether through the history of the English language is its great orthographic variety. Whereas at its first stage, one has to account for the different endings attached to whether to indicate its case morphology, later on, especially during the Middle Ages, when spelling reflected the idiosyncrasy of scribes and regional peculiarities, and orthography was rather unstable and to some extent unpredictable, whether would take one of many possible spellings.5

3.4. Evolution and Classification The outline presented in the following section is based on Ukaji’s (1997) study on whether, where he described the different functions of this word through its various stages in its development. He states that OE hwæðer had two main uses, one as an interrogative disjunctive pronoun, and another one as an interrogative conjunction, and – although he suggests that its interrogative use seems likely to be the original one based on its etymology – he claims that “[i]t is empirically impossible to determine which of these two uses was the original one in OE” (Ukaji 1997: 1236). 3.4.1. Pronoun 3.4.1.1. Interrogative Pronoun: Declinable In this use, OE hwæðer is declined as a strong adjective and it introduces a wh-question. The word order is the same as with all other

4

Not to be confused with nother which came about from the reanalysis of another/an other = a nother. 5 The Oxford English Dictionary (OED Online) lists almost a hundred different spellings altogether for whether under its entry ‘Illustration of Forms’.

160

The Grammaticalization of Old English whether in the Left Periphery

independent wh-questions, namely wh-+V+S6. Both the case ending and the common word order can be seen in (10). (10) Gif Þu nu deman moste, hwæðerne woldes Þu If you now judge must which-ACC.SG7 would you Deman wites wyrðran, judge punishment.GEN worth-CMPR ‘If you are now to judge, which would you judge worthier of punishment?’ (c888 Alfred Boethius 38.122-28-29; Ukaji 1997: 1237) 3.4.1.2. Interrogative Pronoun: Indeclinable The use of hwæðer as an indeclinable interrogative pronoun is considered a new development accompanying a series of innovative changes: it begins to acquire an adverbial character and therefore it is not in argument position anymore, and it refers cataphorically to, for example, a prepositional phrase, as shown in (11). (11) Hwæðer wille ge ðæt ic cume to eow, ðe mid Which wish you that I come.SBJV to you, DISTR with gierde ðe mid monnðwære gæste? rod DISTR with kind spirit? ‘In which way do you wish me to come to you, with a rod or with a kind spirit?’ (c897 Alfred Past. Care 117.7-8; Ukaji 1997: 1238) 3.4.1.3. Interrogative Pronoun: Indefinite The meaning of hwæðer further develops into an indefinite ‘whichever’, but this time it usually refers to an antecedent anaphorically. The environment in which it is found is restricted to the distributive

6

Notice though that the order is reversed when hwæðer occupies the subject position, in which case the verb would follow hwæðer. 7 The following glossing abbreviations have been used in this paper: ACC = accusative; CMPR = comparative; COMP = complementizer; DISTR = distributive; GEN = genitive; Q = question marker; SBJV = subjunctive; and SG = singular.

Víctor Parra-Guinaldo

161

expression swa ... swa ‘so ... so’ and to desiderative verbs, such as willan ‘will, wish’. (12) below is an example of this use. (12) Nim ðonne swa wuds swa wyrt, Take then DISTR tree DISTR plant swa hwæðer swa þu wille, DISTR whichever DISTR you will ‘Take therefore tree or plant, whichever you will,’ (c888 Alfred Boethius 34.91.19-22; Ukaji 1997: 1240) 3.4.2. Conjunction When hwæðer introduces a question rather than a phrase, it functions as an interrogative marker. It may be the case that it introduces an independent question, in which case Ukaji (1997: 1240) warns that “it is not a conjunction in a strict sense of the word, because [hwæðer] does not connect [the independent question] to any preceding or following sentence to form a larger sentence unit”. He then goes on to call hwæðer in this function an interrogative particle instead. 3.4.2.1. Independent Question (Particle) The difference between the two types of independent question marker indicates whether a second alternative is made explicit or not. If it is, Ukaji calls these simple yes/no questions (see 13), if not, they are simply disjunctive (see 14). (13) Hwæðer ge nu secan gold on treowu? Q you now seek-SBJV gold on tree ‘Do you seek now gold on trees?’ (c888 Alfred Boethius 32.73.25-26; Ukaji, 1997: 1241) (14) Hwæðer ðe ðu hi forseo, and þines Q COMP you them despise-SBJV and your agnes þonces hi forlete buton sare, þe þu own thought them give-up without sorrow or you gebide hwonne hi ðe sorgiendne forlæten? wait-SBJV until they you sorrowing leave

162

The Grammaticalization of Old English whether in the Left Periphery

‘Are you to despise them and give them up of your own accord without sorrow, or are you to wait until they leave you sorrowing?’ (c888 Alfred Boethius 8.20.30-32; Ukaji 1997: 1242) Notice that the word order varies when hwæðer functions as a question particle. There is usually (but not always) a hwæðer+S+V order in simple yes/no questions, where the verb is conjugated in the subjunctive, and usually (but not always) a hwæðer+V+S order in disjunctive questions. 3.4.2.2. Dependent Question (Conjunction) The use of hwæðer as a conjunction8 is the only one to have survived until the present time. The word order is hwæðer+S+V, where the verb may appear in either the indicative or the subjunctive. The conjunction is here an alternative to if ‘if’. (15) Simple yes/no question: Þa sende se halga wer swyftne ærendracan to þæs Then send the holy man swift messenger to the bioscopes ceastre, þæt he sceolde geaxian hwæðer Bishop.GEN city that he might learn.by.inquiry COMP he lifes wære. he alive were. ‘Then the holy man sent a swift messenger to the bishop’s city, that he might learn by inquiry whether he were alive.’ (c1000 Ælfric Cath. Hom. 184.34-186.1; Ukaji 1997: 1243) (16) Disjunctive question: Deofol mot ælces mannes afandigan, hwæðer he aht sy, Devil may every man try COMP he aught be

8

Although I use here Ukaji’s (1997) terminology, hwæðer with this function is referred to as a complementizer in current minimalist literature and this seems more appropriate for the view proposed here as well.

Víctor Parra-Guinaldo

163

oððe naht or naught ‘The devil may try every man, whether he be aught or naught.’ (c1000 Ælfric Cath. Hom. I.268.11-14; Ukaji 1997: 1243) Ukaji’s (1997: 1251) typology of OE hwæðer is summarized in (17) below. (17) Pronoun Interrogative pronoun: declinable Interrogative pronoun: indeclinable Interrogative pronoun: indefinite Conjunction Independent Question (particle) simple yes/no question disjunctive question Dependent Question (conjunction) simple yes/no question disjunctive question

4. Concluding Remarks It is important to acknowledge the value of past studies; in particular, Ukaji’s (1997) study provides a variety of examples representative of the different evolutionary stages of whether in the history of English. His work though, I contend, is in need of revision and further analysis for the following reasons: it is essentially a descriptive work, and thus it has no explanatory adequacy, to put it in minimalist words; although it suggests that one of the uses is more likely to have preceded the other, it does not show conclusive evidence that would prove it; finally, the suggested classification is somewhat contrived in that the interrogative particle is included under the conjunction category, when it clearly does not belong there. Van Gelderen’s economy principles (2004) may be the starting point to determine whether the word under investigation goes through a process of grammaticalization following the expected unidirectional cycle, albeit incomplete. The change or variation in word order and the progression from declinable to indeclinable and to indefinite may prove essential in the overall interpretation of the cycle. I suggest that insights about the incipient stages of the grammaticalization cycle of whether can be gained

The Grammaticalization of Old English whether in the Left Periphery

164

once the behavior of its cognates in older Germanic languages is observed and analyzed9. Several are the challenges that such an extended analysis may encounter; for example, given the controversy over the definition of grammaticalization, it is imperative that a definition be provided which takes into consideration both the already assumed characteristics as well as any criticisms that may question any of these. Also, the selection of texts ought to be as variegated and wide as possible so as to be able to ascertain whether word order variation or frequency of use, for example, is or is not related to texts of different regional linguistic varieties.

Works Cited Primary Sources [c888 Alfred Boethius] King Alfred’s Old English Version of Boethius: De Consolatione Philosophiae. Ed. from the Mss., with Introduction, Critical Notes and Glossary by Walter John Sedgefield. Oxford: The Clarendon Press, 1899. Reprinted Darmstadt: Wissenschaftliche Buchgesellschaft, 1968. [c897 Alfred Past. Care] King Alfred’s West-Saxon Version of Gregory’s Pastoral Care. Ed. by Henry Sweet, Part I and II. EETS OS 45 and 50, 1871. Kraus Reprint, 1978. [c1000 Ælfric Cath. Hom.] The Sermones Catholici, or Homilies of Ælfric. In the Original Anglo-Saxon, with an English Version. Ed. by Benjamin Thorpe, 2 Vols. London, 1844 and 1846. Reprinted Georg Olms, 1983.

9

The grammaticalization of question markers into complementizers is common to many other languages. Indonesian, for example, employs one and the same word to introduce yes/no questions as well as embedded questions, and this word can also mean ‘or’ as the following examples show in (i) and (ii) (examples provided by Dinny R. Aletheiani, personal communication). (i)

(Apa) Kamu mau teh? Q you like tea ‘Would you like some tea?’

(ii)

Saya bertanya apa(kah) kamu mau teh (apa tidak). I wonder COMP you want tea or not ‘I wonder whether you want tea (or not).

Víctor Parra-Guinaldo

165

Oxford English Dictionary. 2000. OED Online. Oxford: Oxford University Press. http://dictionary.oed.com/

Secondary Sources Allen, Cynthia. 1980. “Whether in Old English”. Linguistic Inquiry 11:4.789-793. Baker, Mark. 2008. The Syntax of Agreement and Concord. Cambridge: Cambridge University Press. Chomsky, Noam. [1955] 1975. The Logical Structure of Linguistic Theory. Chicago: University of Chicago Press. —. 1965. Aspects of the Theory of Syntax. Cambridge, Mass.: MIT Press. —. 1970. Current Issues in Linguistic Theory. Berlin & New York: Mouton de Gruyter. —. 1986. Barriers. Cambridge, Mass.: MIT Press. —. 1993. “A Minimalist Program for Linguistic Theory”. The View from Building 20. Essays in Linguistics in honor of Sylvain Bromberger ed. by Kenneth L. Hale & S. Jay Keyser, 1-52. Cambridge, Mass.: MIT Press. —. 1995. The Minimalist Program. Cambridge, Mass.: MIT Press. —. 2001. “Beyond Explanatory Adequacy”. MIT Occasional Papers in Linguistics 20. Cambridge, Mass.: MITWPL. Cinque, Guglielmo. 1999. Adverbs and Functional Heads. Oxford: Oxford University Press. Fortson IV., Benjamin W. 2004. Indo-European Language and Culture: An Introduction. (Blackwell Textbooks in Linguistics, 19.). Malden, Mass.: Blackwell. Gelderen, Elly van. 2004. Grammaticalization as Economy. Amsterdam: John Benjamins. —. 2009. “Renewal in the Left Periphery: Economy and the Complementizer Layer”. Transactions of the Philological Society 107.131-195. Heine, Bernd & Tania Kuteva. 2002. World Lexicon of Grammaticalization. Cambridge: Cambridge University Press. Heine, Bernd, Ulrike Claudi & Friederike Hünnemeyer. 1991. Grammaticalization: A Conceptual Framework. Chicago, Ill.: University of Chicago Press. Hopper, Paul J. & Elizabeth Closs Traugott. 2003. Grammaticalization. (2nd ed.) Cambridge: Cambridge University Press. Jackendoff, Ray. 1977. X-Bar Syntax. A Study of Phrase Structure. Cambridge, Mass.: MIT Press.

166

The Grammaticalization of Old English whether in the Left Periphery

Maurer, Walter Harding. 1995. The Sanskrit Language: An Introductory Grammar and Reader. Richmond, Surrey: Curzon Press. Pintzuk, Susan, George Tsoulas & Anthony Warner. 2000. “Syntactic Change: Theory and Method”. Diachronic Syntax. Models and Mechanisms ed. by Susan Pintzuk, George Tsoulas & Anthony Warner, 1-22. Oxford: Oxford University Press. Rizzi, Luigi. 1997. “The Fine Structure of the Left Periphery”. Elements of Grammar ed. by Liliane Haegeman, 281-337. Dordrecht: Kluwer. —. 2001. “On the Position ‘Int(errogative)’ in the Left Periphery of the Clause”. Current Studies in Italian Syntax: Essays Offered to Lorenzo Renzi ed. by Guglielmo Cinque & Giampaolo Salvi, 287-296. Amsterdam: Elsevier. Roberts, Ian & Anna Roussou. 2003. Syntactic Change. Cambridge: Cambridge University Press. Stowell, Tim. 1981. Origins of Phrase Structure. Ph.D. diss., MIT, Institute of Linguistics and Philosophy. Ukaji, Masatomo. 1997. “A History of Whether”. Studies in English Linguistics: A Festschrift for Akira Ota on the Occasion of His Eightieth Birthday ed. by Masatomo Ukaji, Toshio Nakao, Masaru Kajita & Shuji Chiba, 1236-1261. Tokyo: Taishukan.

THE LOSS OF LEXICAL CASE IN THE HISTORY OF ENGLISH: WHAT IS BEHIND TRANSITIVATION? FUYO OSAWA

1. Introduction In this paper, I claim that the productivity of transitive constructions in Present-Day English is due to the loss of the lexical case. More specifically, it is because cases of arguments are separated from their thematic roles and are assigned syntactically by case assigning categories such as T and small v, former AgrO, which have emerged in English (cf. Osawa 2003). In doing so, I also examine the validity of the current case system, or feature valuation and deletion procedure, as a means of accounting for these diachronic changes. It is not that I deny the entire attempt of the Minimalist Program for the Present-Day English case system. Rather, I suggest that a different view is necessary to account for the difference between Old English and Present-Day English. I claim that case is not uninterpretable in earlier English unlike in Present-Day English. That is, case is interpretable in earlier English. Then, I propose that a change in the case system took place and that this change triggered many syntactic changes in the history of English, including the proliferation of transitive constructions and the loss of impersonal constructions. This paper is organized as follows: in the next section, I introduce the preliminaries of transitivation research drawing on Visser (1970). In Section 3, the nature of Present-Day English transitive constructions will be semantically examined in detail. I show that some present-day transitive constructions are deviant from prototypical transitive sentences. Some of them have no transitive properties at all. In Section 4, I will show that this deviance which is observed in Present-Day English was not found in Old English. Then, the thematically-motivated nature of the Old English case system will be clarified. In Section 5, the notion of lexical-thematicity is explained concerning this thematically-motivated

168

The Loss of Lexical Case in the History of English

case system. In a lexical-thematic language, every constituent must be thematically related. This explains why the Old English case system is thematically motivated. In a lexical-thematic language, impersonal constructions, which are apparently deviant from the norm, are allowed. In Section 6, the demise of the thematically-motivated case system and the appearance of a new case system are discussed. In Section 7, I argue the issue of the uninterpretable case. It will be shown that cases in earlier English were not uninterpretable. I propose a new case hierarchy. Finally in Section 8, I conclude my discussion.

2. Transitive Constructions Transitive constructions are definitely the most productive constructions in Present-Day English. It is often said that this is because transitivisation or transitivation has taken place on a large scale in the history of English: Historically, the most remarkable fact is that, whereas in Old English the number of double-faced or amphibious verbs was far inferior to that of intransitive verbs, in Pres[ent] D[ay] English amphibious verbs far outnumber the intransitive verbs. In other words, transitivation has taken place on a large scale (Visser 1970: 99).

Traditional accounts have been given to explain this transitivation: transitivation is assumed to have been triggered by a phonological decay of Old English prefixes. In Old English, transitive verbs were derived by prefixation from intransitive verbs such as birhtan ‘to shine’ vs. gebirhtan ‘to make bright’, restan ‘to repose’ vs. gerestan ‘to give rest to’, etc. When the prefix ge- finally disappeared, verbs with ge- and without ge- appeared in an identical form in which the transitive and intransitive functions were combined (cf. Visser 1970: 127). A second cause is presumed to be phonological coalescence, which happened in the Middle English period. In Old English, the internal vowel alternation, known as ablaut, differentiates intransitive from transitive verbs: sincan (intransitive) ‘to sink’ vs. sencan (transitive) ‘cause to sink’, brinnan (intransitive) ‘to burn’ vs. bærnan (transitive) ‘to burn, set fire to’ (cf. sit vs. set, rise vs. raise, etc. in Present-Day English). They became phonologically identical due to the loss of vowel alternation. Factors such as phonological reduction and morphological simplification surely have played important roles in transitivation (Visser 1970: 99, 127). However, I focus on the third cause, i.e. the demise of

Fuyo Osawa

169

morphological case distinctions, which has also played a crucial role in the overall picture. This will be discussed in more detail later in this paper.

3. Transitivity 3.1 What is Transitivity? I start this section with a fundamental question of what transitivity is. Is it primarily a syntactic or semantic notion? It is not so easy to give a satisfactory answer to this simple question. Crystal (1980: 364-365) defines ‘transitivity’ as follows: A category used in the grammatical analysis of clause/sentence constructions, with particular reference to the verb’s relationship to dependent elements of structure. The main members of this category are ‘transitive’, referring to a verb which can take a direct object, and ‘intransitive’, where it cannot.

If a verb can take a direct object, it is called ‘transitive’, or, is it possible to say that if a verb has transitive properties, it can take a direct object? Even if it is possible, we still have to define the exact nature of the transitive properties which a verb has. Formally, transitivity means that the situation denoted in a predication involves at least two participants, or two arguments, i.e. two nominals in two semantic roles, such as Agent and Patient. Then, the criterion of the transitive construction is often decided based on whether passivisation of a given sentence is possible or not. However, in a generative framework, it is assumed that there is no derived relation between active and passive sentences (Chomsky 1995, 2005). They are generated independently of each other by the operation Merge. It is true that transitive and passive constructions are still thought to be cognitively related to each other, but even in the historical domain, the traditional criterion of passivisation does not seem to have much meaning. In the Old English period, the verbs that took dative-marked nouns as their objects did not have ‘passive’ counterparts. Some researchers classify those verbs as ‘intransitive’, although they took object nouns (cf. Nakao 1979). The notion of transitivity on which the transitive constructions are supposed to be formed is not as easy to define as it seems. In the next section, I clarify the nature of Present-Day English transitive sentences, which will give a very important insight into the diachronic change of the English case system.

170

The Loss of Lexical Case in the History of English

3.2 Present-day Transitive Constructions Although “the term TRANSITIVE is often applied to all verbs which require an object” (Quirk et al. 1985: 54) and this definition is perfect from a syntactic point of view, it does not capture the complicated nature of present-day transitive constructions, since very different types of verbs are involved in transitive constructions. As discussed by Hopper & Thompson (1980), present-day transitive sentences can encode a wide range of states of affairs which semantically differ from the prototypical example of transitivity (cf. Hopper & Thompson 1980; Taylor 2003). In Hopper & Thompson (1980), transitivity is described in terms of eleven semantic properties, which concern the relations encoded by two arguments, i.e. subject and object NPs in transitive constructions. Some of them are shown below (cf. Taylor 2003): (1) a. The two participants, i.e. arguments, are highly individuated, i.e. they are discrete, specific entities, distinct both from each other, and from the background environment. b. The Agent acts consciously and volitionally, and thus controls the event. Thus, the Agent is typically a human being. c. As a consequence of the Agent’s action, something happens to the Patient, i.e. the referent of the object NP. d. After the occurrence of the event, the Patient is in a different state from before the event. Usually, the difference is highly perceptible to an onlooker. e. The Agent’s action on the Patient usually involves direct physical contact, and the effect on the Patient is immediate. f. Typically, Agent and Patient are not only clearly differentiated entities; often they also stand in an adversative relationship. The more of these properties a given construction has, the more prototypical its transitive construction is. Consider the following examples: (2) Bill killed Tom. (3) Brian approached the bridge. (4) John helped Mary. Example (2) may be a prototypical example of transitivity in that this sentence has most of the semantic properties mentioned above. In

Fuyo Osawa

171

example (2) Bill (subject-Agent) and Tom (direct object-Patient) are clearly differentiated entities, and they stand in an adversative relationship: Bill is a killer and Tom is a victim. The Agent acts consciously, and after the killing, the Patient Tom is in a different state from before the killing. Meanwhile, the other two sentences are deviant from the prototypical transitivity: in (3) the patient bridge does not undergo any change as a result of the Agent’s (Brian) action. In (4) the subject and the object are not in an adversative relationship but in a co-operative one. Otherwise, however, they are still highly transitive in that an action is carried out during the event. There are more marginal cases in Present-Day English: (5) He resembles his father. (6) The auditorium holds 5000 people. These two sentences (5) and (6) are distant from the typical transitive sentence in that no action on the side of the subject is involved, and nothing happens to the object. In example (5) He does not have an Agentive role and His father is not a Patient. The auditorium in example (6) is not an Agent either. Neither sentence has a corresponding passive counterpart, as shown below: (7) *His father is resembled by John. (8) *5000 people are held by the auditorium. Nevertheless, they are still syntactically classified as transitive constructions. Like this, the striking feature of Present-Day English is that transitive constructions encode states of affairs which semantically deviate from transitivity.

4. The Old English Case System In the previous section, we have observed that some present-day transitive sentences have no transitive/transitivity properties. However, in Old English, this deviance from transitivity did not often occur. I claim that there are correlations between transitivity of the predicate and morphological cases in Old English. These correlations also concern thematic roles such as Agent or Patient. That is, there was a motivated relation between morphological cases, thematic roles, and transitivity in Old English.

The Loss of Lexical Case in the History of English

172

I argue that in Old English, morphological case was closely related to the thematic roles of argument nouns, that is, morphological cases were assigned on the basis of thematic information only, and there was no structural requirement. All NP arguments of a verb must have one thematic role and carry a case related to its thematic role. No syntactic case assignment (case feature checking) was operative or needed in earlier English. I propose that all the cases in Old English were sensitive to thematic roles, that is, there was no distinction between inherent case and structural case. Later in the Middle English period around the 13th century, structural case assignment became activated. Thematic roles of the argument are based on the meaning of the predicate verb. Therefore, transitivity of the predicates is also related to the selection of morphological cases. What I want to argue is that Old English morphological cases were not assigned to a thematically unrelated NP. For example, the thematic role of the subject of the verb undergo, when it means ‘to bear, to suffer, to subject or to be subjected’ is not Agent, but Patient in Present-Day English, yet it is still assigned nominative case. In the earlier period, this was impossible. This meaning of the verb undergo dates from the 14th century and before that, it meant ‘to work under, to undermine’. At that time, the nominative subject of undergo had an Agentive theta role. Other evidence for this proposal comes from impersonal constructions, which will be discussed in Section 5. Let us examine other cases of these correlations. The prototypical transitive verb in Old English took an accusative-marked noun as its object under the semantic case system. For example, verbs like ofslean ‘kill’ and brecan ‘break’ took accusative objects. The semantic constructions exemplified by sentences using those verbs meet most of the transitivity properties mentioned above: (9) he he

ofslog killed

þone that.ACC.

aldormon ealdorman.ACC.

‘he killed the ealdorman’ (AS.Chronicle Parker MS 46. 21.755) Meanwhile, verbs such as andswarian ‘answer’, œtywan ‘show’, deman ‘judge’, genealœcan ‘approach’, togeþeodan ‘be faithful’, and þancian ‘thank’, took dative nouns as their object arguments. These verbs are equivalent to the Present-Day English ones in examples (3) and (4) in Section 3.2:

Fuyo Osawa

(10) Ða Then

andsuarode answered

he he

173

him him.DAT.

‘Then he answered him.’ (CP 304, 12) Verbs such as (be)þurfan ‘need’, bidan ‘wait for’, blissian ‘rejoice’, helpan ‘help’, þyrstan ‘thirst’ took genitive nouns as their object arguments. Genitive marking tends to encode circumstantial roles rather than full participants. Compared with the direct involvement of accusative- or dative-marked NPs, genitive-marked NPs are involved indirectly in the action/event of the predicate: (11) he fِgnode he rejoiced

ðِs that.GEN.

miclan weorces & great work.GEN. and

fِgernesse beauty.GEN

ðِrre that.GEN.

ceastre city.GEN.

‘he rejoiced at the greatness and beauty of the city’ (CP 39.14-15) (12) Bidaþ Await.PL.

Dryhtnes lord.GEN.

domes doom.GEN.

‘(they) await the Lord’s judgment’ (Bosworth & Toller 1964) As pointed out by Plank (1983: 255), Visser (1970: 355f.) rightly underlines the danger of misrepresentation in Modern English translation in examples such as (12). A closer translation of (12) would be ‘they are in expectation with regard to the coming of the Lord’s judgment’. In brief, a co-operative Patient role or a Recipient role was ideally realised as a dative object, while an adversative Patient role was ideally realised as an accusative object, and an Agent as a nominative subject. Therefore, it was not so easy for a nominal with a Recipient semantic role to carry nominative case. This is the reason why subjectless constructions such as impersonal sentences were allowed in earlier English. In fact, the situation in Old English was much more complicated than this, since most of the texts available now are from the late stage of Old English in which case distinctions had already decayed to some extent. Accordingly, my hypothesis should be relativized to pre-Old English

174

The Loss of Lexical Case in the History of English

stages of the language. My claim goes beyond what the data available to us can fully establish, and my claim involves some idealization of an Old English situation. However, the attempt to give an explanation to changes spanning a longer period necessarily involves referring to unattested periods. Although I notice this problem, there is a necessity of this kind of idealization for historical studies. There is much counter-evidence to my claim and the problems with this should be more closely addressed subsequently. Although arguably, we still say that overt morphological cases are closely related to thematic roles in earlier English. My claim that there was a motivated correlation between morphological cases and semantic roles is well supported by Plank (1981; 1983). Plank (1981; 1983) proposes that there are correlations between the choice of case and thematic role. For example, dative marking of object NPs tends to signal a relatively low degree of opposedness between the referents of object and subject NPs (i.e. the subject-object relationship is not adversative), accusative case marking signals relatively high opposedness (i.e. the subject-object relationship is not cooperative) correlated with the Patient function, while genitive marking tends to encode circumstantial roles rather than full participants (cf. Plank 1983: 255; Denison 1993: 1819).

5. Old English as a Lexical-thematic Language In the previous section we have observed that the Old English case system is thematically motivated. Therefore, it is relevant to discuss the lexical-thematic nature of Old English. Lexical-thematicity is a key notion in understanding the nature of the Old English case system and what happened in the history of English. Lexical-thematicity means that the syntactic structure is completely lexically and thematically constructed. This means that lexical-thematic languages have no functional categories such as DP, or TP, but only lexical categories (N, V, A, P or only content words) and their phrasal projections (NP, VP, AP, PP) (see Abney 1987 and Radford 1990). Note that in lexical-thematic languages, every constituent must be thematically related. This means that only arguments which are required by the meaning of a predicate must be syntactically realised (cf. Osawa 2003). All constitutes in the clause structure are thematically related. There are no thematically unrelated constituents such as expletives. I argue that Old English is one instantiation of lexical-thematicity. It deploys lexical categories and their phrasal projections (NP, VP, AP) and no functional categories, which are non-thematic, as Abney (1987)

Fuyo Osawa

175

suggests. Old English lexical-thematicity already began to deteriorate at the end of the period, but some vestige remained. As is often discussed (cf. van Gelderen 1993, 2004; Osawa 2003, 2009), functional categories developed very limitedly in Old English. I suggest that the clause in Old English is the projection of V, i.e. VP with no functional projections. (13)

In (13) the two NPs are unordered with respect to each other. The parentheses show that NPs are optional. This structure can deal with impersonal constructions in which there is no NP argument such as (14). There is no TP, and therefore, no EPP-feature: hence, there is no subject requirement. (14) norþan from the north

sniwde snowed

‘snow came from the north’ (Seafarer 31) Then, if the meaning of the predicate does not require an Agent, which should be realised as a nominative NP, the clause structure of a given verb lacks the nominative argument NP. This is called a subjectless construction. A subject is not necessarily a priori in a lexical-thematic language. The requirement that all clauses should have a subject is a later development, due to the emergent TP or to the EPP-feature (cf. Chomsky 2005, 2008). Under this semantic-based case system, it is not easy for an NP with a Recipient semantic role to carry nominative case, as mentioned above. This is why, in Old English, indirect passives like (15) were not allowed and impersonal passives like (16) were used instead. (15) He was given presents.

176

The Loss of Lexical Case in the History of English

(16) him næs getiðod ðære him.DAT. not was granted that.GEN.

lytlan small

lisse favour.GEN.

‘he was not granted that small favour’ (ÆCHom. I.23.330.29) In (16) the two internal arguments carry cases related to their theta roles. Him carries the dative case, which is assigned based on its thematic role of ‘Recipient’, as I explained above. There is no nominative-marked NP, since the ‘passive’ sentence is not concerned with the Agent of the action/event. It is not easy for an NP with a Recipient theta role to carry the nominative case. Dative NPs should remain dative. This is the nature of impersonal passives in Old English. In summary, the development of functional categories was very limited in Old English (cf. van Gelderen 1993, 2004; Osawa 2003, 2009). Then, over time, a new functional category was introduced. The emergence of a new functional category brings about many syntactic changes and hence new constructions. The history of English is an instantiation of this shift from a lexical-thematic to a functional stage. Present-Day English has a full-fledged set of functional categories. Their presence determines how the clause structure of Present-Day English is constructed.

6. New Case System In Present-Day English, we have observed that there are many ‘transitive’ constructions which have no transitive properties. How is this possible? What is behind this? I claim that this was made possible due to a change in the case system. The case system changed from a thematically motivated to a syntactic one against the background of the shift from a lexical-thematic to a functional stage, and the case assignment was separated from thematic-role assignment. This separation was made possible due to the emergent functional category. This separation made ‘deviant’ transitive constructions possible. The Middle English period saw the appearance of new constructions such as indirect passives, and the Exceptional Case Marking constructions and the loss of impersonal passives, and productive transitive constructions, as we have observed above. What made these constructions possible were newly emergent functional categories or case assigning elements in English. Due to the demise of case morphology, the thematically motivated case system deteriorated and functional categories such as T

Fuyo Osawa

177

and small v emerged. A new case system, i.e. a structure-based case system appeared instead, and brought about the separation of case from theta role. T can assign (check) nominative case to a thematically unrelated NP and small v can assign accusative case to a thematically unrelated NP. There is no longer semantic constraint. Case is exclusively assigned under structural conditions by T and small v. In other words, case is liberated from semantic roles. This means that the correlations between thematic roles, and transitivity also deteriorated. We can easily understand why transitive sentences are so productive in Present-Day English. In Present-Day English, there is no motivated relation between thematic roles of arguments, their (argument) case, and the transitivity of the predicate verb. Cases, i.e. structural cases, need not be associated with thematic roles. Case can be assigned structurally to a thematically unrelated NP. For example, nominative can be assigned to the subject NP, whatever its thematic role is; Agent, Patient, Experiencer or Theme: (17) Mary killed John. (18) The ball rolled under the desk. (19) Helen suffered a stroke. (20) I believed him to have stolen my money. In (17) the nominative subject Mary is an Agent. In (18) the argument with the theta role Theme functions as the subject, in (19) the nominative case is assigned to the Patient of a stroke, and in (20), which is the Exceptional Case Marking construction, the semantic Agent of the action is assigned the accusative case. This separation, or inconsistency, between case and theta roles is characteristic of Present-Day English. The Exceptional Case Marking construction was not observed in Old English. This absence follows easily under a semantic-based case system. The Present-Day English clause structure of John saw them is something like (21):

The Loss of Lexical Case in the History of English

178

(21)

Thanks to this separation, or liberation, a non-thematic subject position is made available. That is, Present-Day English may have nonthematic constituents due to the presence of functional categories such as T. For example, the emergent TP requires the presence of a subject, irrespective of a thematic role. In Present-Day English, the subject position, i.e. the Specifier of TP, must always be filled due to an EPPfeature, or due to the functional category T. Then, nominal elements with person/number properties are inserted by Merge. The nominal which occupies the Spec, TP positions is a subject, i.e. external argument. The insertion of expletive it is required if there is nothing to occupy the position. Impersonals without nominative NPs, including impersonal passives, are not allowed any more. The transitivation is a process in which an NP argument with any semantic role can occur as the subject or object.

7. Uninterpretable Case 7.1. Present-Day English Case Finally, I discuss the issue of uninterpretability of a case feature. In Present-Day English, within the Minimalist approach, a case feature of a pronoun such as he is said to be uninterpretable, since it does not have much contribution in the semantic interpretation (cf. Radford 2004: 284; Chomsky 2005, 2008). Let me explain more about the current feature valuation and deletion procedure in the Minimalist Program. In Present-Day English, grammatical features of lexical items in a clause are either valued or unvalued when they are in a lexical array before they enter the derivation. Nominal expressions including pronouns

Fuyo Osawa

179

are supposed to be with their ø-features (i.e. person/number) already valued when they enter the derivation, but their case feature unvalued. Meanwhile, finite T elements (e.g. auxiliaries) enter the derivation with their tense features valued (e.g., [+past]), but their ø-features unvalued. These features are later valued through agreement with a nominal goal (cf. Radford 2004: 284; Chomsky 2005, 2008): (22)

This unvalued case feature (u) carried by the goal he is valued by the finite tense features of the probe. We can say that the finite tense features of the probe can assign nominative case to the goal. This valued/unvalued feature is related to the interpretability of features. The case feature of a pronoun like he is said to be uninterpretable because it does not have much contribution. A subject pronoun is realised as nominative, accusative, or genitive depending on the type of clause it is in, without having any effect on meaning, as shown in (23): (23) a. It is said [they were arrested] b. He expected [them to be arrested] c. He was shocked at [their being arrested] (Radford 2004: 288) In other words, a change in case forms does not affect the meaning. Hence, case is said to be uninterpretable. In contrast, ø-features such as person and number of pronouns are interpretable, because he is different

The Loss of Lexical Case in the History of English

180

in meaning from the 1st Pers. Sing. pronoun I. There is a correlation between valued/unvalued and interpretability: features which enter the derivation unvalued are uninterpretable (cf. Radford 2004: 288). Interpretable features play an important role in semantic interpretation, and since uninterpretable features play no role in semantic interpretation, those features should not be sent to the LF.1 The uninterpretable features have to be deleted by a feature-deletion operation so that the uninterpretable features become invisible to the semantic component.

7.2. Earlier English Case In the previous section, we have observed that case is uninterpretable in Present-Day English. However, we see a different situation in earlier English. As we have discussed in the previous sections, morphological case distinctions, although they deteriorated a little by the end of the Old English period, played an important role. Case was interpretable in earlier English. A few more supporting examples are given below. In Old English, a difference in case forms leads to a difference in interpretation (cf. Plank 1981, 1983; Denison 1993: 18, 19): (24) Him folgiað fuglas scyne him.DAT. follow birds brilliant ‘brilliant birds follow him’ (Phoenix 591) (25) ond and

ða then

folgode pursued.ACC.?

feorhgeniðlan deadly-foes

‘and then he pursued his deadly foes’ (Beowulf 2933) Whether folgian means ‘follow’ or ‘pursue’ is expressed by the case of its object NP. In (24), the object NP is dative, and the verb means ‘follow,’ while in (25) the verb with an accusative-marked object means ‘pursue’. Likewise, the verb geefenlæcan with an accusative-marked object is supposed to mean ‘imitate’, but the verb with a dative-marked object means ‘resemble’ (cf. Plank 1981, 1983; Denison 1993: 18, 19): 1

In Chomsky (2008: 142) LF is eliminated as the final internal level. The semantic component may be a more appropriate term.

Fuyo Osawa

(26) and and

þa the

unandgytfullan unintelligent

181

hine geefenlæcen him-Acc. imitates

‘… and the unintelligent … may imitate him’ (27) Gif he geeuenlæcð gode if he resembles god.DAT. ‘if he resembles God’ As these examples cited from Plank (1981: 20, 1983: 247) show, a case form plays a crucial role in semantic interpretation, and hence it has to remain as it is. If this feature is deleted and is not sent to the semantic component, the correct interpretation will not follow. While I do not deny the uninterpretability of case features in Present-Day English, the historical facts suggest the necessity of a different view on case interpretability when we deal with the earlier data. Plank (1983: 247) translates the example (27) using the adjective like as “he is like/resembles God”. This implies that the meaning equivalent to (5), which is mentioned in Section 3.2., might be expressed in adjectival sentences. Indeed, adjectival sentences using (ge)lic ‘alike’ are amply attested as in (28) and (29): (28) Slæp biþ deaþe gelicost ‘Sleep is most like death.’ (Bosworth & Toller 1964) (29) Is byþ gimmum gelicust ‘Ice is most like gems.’ (Bosworth & Toller 1964) This suggests that a state of affairs in which no transitive properties are observed is also compatible with adjectival constructions. Adjectival constructions are supposed to have very low transitivity. We should note that internal arguments which adjectives took carried either dative case as shown above, or genitive case, although arguably. I propose that there is a case hierarchy, which puts morphological cases in the following order: (30) case hierarchy Accusative > Dative > Genitive Transitivity of a predicate: high transitivity

low transitivity

182

The Loss of Lexical Case in the History of English

What this hierarchy means is that high transitive predicates tend to take accusative-marked objects, while genitive-marked objects occur in low transitive predicates, including adjectival constructions. Dativemarked nouns are placed between the two. The nominative case is excluded from this hierarchy. Nominative nouns should be considered differently since nominative-marked nouns are excluded from internal argument positions. This case hierarchy is different from a widespread hierarchy (cf. Primus 1999, Hawkins 2004), which is shown below, except that the genitive case is placed low on both hierarchies: (31) Nominative > Accusative > Dative > Other This widespread hierarchy implicates that once the dative/accusative distinction was lost, the lower case than that, i.e. the genitive case could not survive. My hierarchy is defined in terms of the transitivity of the predicate a case-marked NP is in. One question remains: how to deal with a fifth case, i.e. the instrumental case. The instrumental case is only marked in certain strong adjectival and pronominal declensions in Old English. Although tentatively, I put the instrumental case lower than the genitive case on the above hierarchy. (32) case hierarchy – revised Accusative > Dative > Genitive > Instrumental Although the instrumental case could function as a predicate argument in Old English, the verbs which could take an instrumentalmarked NP as an object were very few. The instrumental case was absorbed by the dative case early in Old English. Thus, I argue that the instrumental case should be placed lower than the genitive case. There may be a counter-argument to this ordering. There is disagreement on the ordering of dative and genitive in the hierarchy among the researchers. There have been analyses suggesting that the genitive is put higher on the case hierarchy than the dative (cf. Calabrese 1998, Blake 2001). In my case hierarchy, the status of the genitive case is made explicit, based on transitivity. As far as the English data are concerned, the genitive case should be placed on a lower rank than the dative case.

Fuyo Osawa

183

8. Concluding Remarks In this paper I have tried to answer the question of why transitive constructions in Present-Day English are so productive. I have shown that what contributed most to this large-scale transitivation is the shift of the English case system from a thematically motivated to a syntactic system. Case assignment was separated from thematic role assignment, and functional categories like T and small v have come to take care of case assignment/checking. This has made possible the transitive constructions which have no transitive properties. Transitivation is a process in which an NP argument with any semantic role can occur as a subject or object. I have also shown that earlier English cases were not uninterpretable and the uninterpretability of cases came about due to the separation of case from thematic roles and this separation made new syntactic constructions possible. It is very difficult to answer “the fundamental question of why languages should have uninterpretable features such as case, which play no role in semantic interpretation” (Chomsky 2005: 18). I hope that my proposal has shed some light on this important issue of case.

Acknowledgements This study was supported by the Japan Society for Promotion of Science (Grant-in-Aid for Scientific Research No. 21520516).

Abbreviations Agr AgrO CP D DP EPP LF small v spec T TP

a head of an agreement phrase AgrP a head of a direct object agreement phrase AgrOP Complementiser phrase Determiner Determiner phrase Extended Projection Principle Logical form a head of light verb phrase vP Specifier Tense Tense phrase

184

The Loss of Lexical Case in the History of English

Works Cited Primary Sources Bosworth, Joseph & Northcote T. Toller, eds. 1964. Anglo-Saxon Dictionary. London: Oxford University Press. Earle, John & Charles Plummer, eds. 1892–1899. Two of the Saxon Chronicles Parallel: With Supplementary Extracts from the Others. Oxford: Clarendon. Gordon, Ida, ed. 1979. The Seafarer. Manchester: Manchester University Press. Klaeber, Frederick, ed. 1950. Beowulf and the Fight at Finnsburg. Heath: Lexington. Krapp, George & Elliot Dobbie, eds. 1931–1953. The Anglo-Saxon Poetic Records: A Collective Edition III. (Phoenix). New York & London: Columbia University Press. Sweet, Henry, ed. 1871. King Alfred’s West Saxon Version of Gregory’s Pastoral Care. London: Trübner for the Early English Text Society. Thorpe, Benjamin, ed. 1844. The Sermones Catholici or Homilies of Ælfric I. London: Ælfric Society.

Secondary Sources Abney, Steven, P. 1987. The English Noun Phrase in its Sentential Aspect. Ph.D. dissertation, MIT. Blake, Barry, J. 2001. Case. (2nd ed.) Cambridge: Cambridge University Press. Calabrese, Andrea. 1998. “Some Remarks on the Latin Case System and its Development in Romance”. Theoretical Analyses on Romance Languages ed. by José Lema & Esthela Treviño, 71-126. Amsterdam & Philadelphia: John Benjamins. Chomsky, Noam. 1995. The Minimalist Program. Cambridge, Mass.: MIT Press. —. 2005. “Three Factors in Language Design”. Linguistic Inquiry 36.1-22. —. 2008. “On Phases”. Foundational Issues in Linguistic Theory: Essays in Honor of Jean-Roger Vergnaud ed. by Robert Freidin, Carlos P. Otero & Maria Luisa Zubizarreta, 133-166. Cambridge, Mass.: MIT Press. Cinque, Guglielmo. 1999. Adverbs and Functional Heads: A Crosslinguistic Perspective. Oxford: Oxford University Press.

Fuyo Osawa

185

Crystal, David. 1980. A First Dictionary of Linguistics and Phonetics. London: Andre Deutsch. Denison, David.1993. English Historical Syntax. London & New York: Longman. Gelderen, Elly van. 1993. The Rise of Functional Categories. Amsterdam & Philadelphia: John Benjamins. —. 2004. Grammaticalization as Economy. Amsterdam & Philadelphia: John Benjamins. Hawkins, John A. 2004. Efficiency and Complexity in Grammar. Oxford: Oxford University Press. Hopper, Paul J. & Sandra A. Thompson. 1980. “Transitivity in Grammar and Discourse”. Language 56.251-299. Nakao, Toshio. 1979. A Linguistic History of the English Language. Tokyo: Shinozaki Shorin. Osawa, Fuyo. 2003. “Syntactic Parallels between Ontogeny and Phylogeny”. Lingua 113:1.3-47. —. 2007. “The Emergence of DP from a Perspective of Ontogeny and Phylogeny: Correlation between DP, TP and Aspect in Old English and First Language Acquisition”. Nominal Determination: Typology, Context Constraints, and Historical Emergence ed. by Elisabeth Stark, Elisabeth Leiss & Werner Abraham, 311-337. Amsterdam: John Benjamins. —. 2009. “The Emergence of DP in the History of English: The Role of the Mysterious Genitive”. Historical Linguistics 2007 ed. by Monique Dufresne, Fernande Dupuis & Etleva Vocaj, 135-148. Amsterdam & Philadelphia: John Benjamins. —. 2010. “Transitivisation in the History of English”. Aspects of the History of English Language and Literature: Selected Papers Read at SHELL 2009, Hiroshima ed. by Imayahashi, Osamu, Yoshiyuki Nakao & Michiko Ogura. Frankfurt am Main: Peter Lang. Plank, Frans. 1981. “Object Cases in Old English: What Do they Encode? A Contribution to a General Theory of Case and Grammatical Relations”. (Ms., 67pp) Englisches Seminar, Universität Hannover. —. 1983. “Coming into Being among the Anglo-Saxons”. Current Topics in English Historical Linguistics ed. by Michael Davenport, Erik Hansen & Hans Frede Nielsen, 239-278. Odense, Odense University Press. Primus, Beatrice. 1999. Cases and Thematic Roles: Ergative, Accusative and Active. Tübingen: Niemeyer.

186

The Loss of Lexical Case in the History of English

Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech & Jan Svartvik. 1985. A Comprehensive Grammar of the English Language. London: Longman. Radford, Andrew. 1990. Syntactic Theory and the Acquisition of English Syntax. Oxford: Blackwell. —. 2004. Minimalist Syntax: Exploring the Structure of English. Cambridge: Cambridge University Press. Rizzi, Luigi. 1997. “The Fine Structure of the Left Periphery”. Elements of Grammar ed. by Liliane Haegeman, 281-337. Dordrecht: Kluwer. Taylor, John. 2003. Linguistic Categorization: Prototype in Linguistic Theory. Oxford: Oxford University Press. Visser, Fredericus Th. 1970. An Historical Syntax of the English Language I, Leiden: Brill.

MODAL ADJECTIVES IN THE HISTORY OF ENGLISH OCTAV EUGEN DELAZERO

1. Introduction A look at the history of modal adjectives in English and at their ancestors and equivalents in other languages shows that they are rare and exceptional cross-linguistically, and are subject to various restrictions. In English, originally they could only modify nouns describing situations, as opposed to physical entities, and the latter usage began to spread no earlier than the 18th–19th centuries. This selectional restriction was inherited from the languages from which the modal adjectives were borrowed. In Present-Day English too, the nouns these adjectives can modify always involve situations: they either denote sets of situations (events: war; or states: peace) or are relational, denoting sets of entities in context (king, bomb, etc.). In the first four sections of this article I present several diachronic, cross-linguistic, and synchronic observations about the contexts of use of modal adjectives and the evolution of their meaning in English and the languages from which they were borrowed into English. The last section offers a formal analysis of the semantics of the modal adjectives and the nouns they modify.1

2. Preamble: The Modal Adjectives as a Class Modal adjectives are intensional and non-subsective. Kamp & Partee (1995: 137-138) define subsectivity vs. non-subsectivity as properties of adjectives based on their meaning postulates: (1) subsectivity: [[Adjective + Noun]] Ž [[Noun]] 1

Thanks are due to Wayles Browne, Molly Diesing, Carol Rosen, and Michael Weiss, for comments on the material and analyses in the present article.

188

Modal Adjectives in the History of English

(2) non-subsectivity: [[Adjective Noun]] Œ [[Noun]] The combination of a subsective adjective with a noun yields a subset of the set of individuals denoted by the noun: a tall tree is a tree, and a skilled surgeon is a surgeon. This is not the case with non-subsective adjectives: a former president is not a president at the reference time i0, and a possible war is not a war in the base world w0: (3) temporal adjective: [[former president]] i0 Œ [[president]] i0 (4) modal adjective: [[possible war]] w0 Œ [[war]] w0 The common denominator of modal adjectives is the notion of possibility. In addition, some of them involve probability (stronger possibility), either always (likely, probable, and verisimilar) or in some of their uses (plausible). The combination of a noun with a modal adjective denotes a set of individuals in a world accessible from the base world. It is useful to distinguish the notion of possibility from that of nonevidentiality: when the lack of evidence is supplemented by doubt, as usually with alleged, putative, or purported, the prevailing nuance is privative, as in fake, false, or pretended. Such adjectives are subsective, as argued by Partee (2007) – for instance, in fake fur, the meaning of fur is expanded to include both genuine and fake fur. Adjectives like alleged and purported make explicit the way in which the noun’s meaning is expanded – namely, by somebody’s statement, qualified as an allegation. Other adjectives which involve the notion of possibility, like drinkable or liable (plus infinitival complement) are subsective: drinkable water is a kind of water in the base world, and a man liable to get hurt is likewise a man in the base world. I will be discussing only the modal adjectives proper, i.e. those adjectives which are non-subsective (4). As far as I could determine, the modal adjectives in Present-Day English are: likely, possible, probable, potential, plausible, prospective, verisimilar, and virtual. In addition, there are the antonyms of these adjectives in this list, which are non-subsective as well: unlikely, impossible, improbable, and implausible.

3. Diachronic Observations The entries of modal adjectives in the Oxford English Dictionary show that all of them, with the significant exception of potential, started being used to modify nouns which describe corporeal entities (henceforth ‘entity

Octav Eugen DeLazero

189

nouns’) no earlier than the 18th–19th centuries; until then, modal adjectives only modified nouns describing situations (henceforth ‘situation nouns’), which can be events (probable victory) or states (possible relationship).2 In (5) below is a list of modal adjectives together with their first attestations with entity-describing nouns, with the thick line indicating the time gap: (5)

Adjective

Relevant sense (OED)

likely

“having an appearance of truth or fact; that looks as it would happen, be realized, or prove to be what is alleged or suggested; probable” “that can be or become (what is denoted by the following noun); that is potentially a…; that could turn into a…” “that is likely to be or become (something specified)” “expected or expecting to be (the object or thing specified) in the future; that is in prospect; proposed, likely, potential” “that is so in essence or effect, although not formally or actually; admitting of being called by the name so far as the effect or result is concerned” “possible as opposed to actual; having or showing the capacity to develop into something in the future; latent; prospective”

possible

probable

prospective

virtual

potential

2

First attestation with an entity-describing noun 1856: likely camping-ground

1736: possible object of Knowledge 1890: probable companion 1853: prospective bridegroom

1844: virtual sovereign

1398: potencial light

I consulted the electronic edition of the Oxford English Dictionary, which is based on several corpora. The entries specify the semantic fields of the nouns modified by these adjectives, and either distinguish entities from situations (in various formulations), or list the earliest examples. In either case, it is quite easy to detect the first use of modal adjectives with an entity noun.

190

Modal Adjectives in the History of English

The adjectives plausible and verisimilar are not attested with entity nouns, although many speakers find collocations like plausible murderer acceptable. In the example with likely, the noun describes a location, which, strictly speaking, is neither an entity nor a situation (relationship), but is arguably closer in meaning to the former than to the latter. Otherwise, the use of likely with entities would be even later. An example with the same kind of noun illustrates the necessity of considering the context in order to determine whether the adjective is indeed used with a modal value:3 (6) lest the devil that guides him should aid him, I will search impossible places (Shakespeare: The Merry Wives of Windsor III.5) The adjective here means in fact ‘inaccessible’, and the noun describes spatial regions of the world in their material aspect; in this instance, the places are not associated with any dimension – political, administrative, cultural, etc. – other than the spatial one. (Perhaps the spatial dimension, to the exclusion of the temporal one, is the only one that is relevant for entities; if so, locations would be by default entities, unless considered under certain other aspect.) As regards the meaning of the adjective, there seems to be a conceptual continuum between inaccessible places and inaccessible worlds: a place can be so inaccessible that it becomes transcendent (and a time can be so remote that it becomes mythical). With situation nouns, these adjectives are attested several centuries earlier:

3

I owe this example to Molly Diesing.

Octav Eugen DeLazero (7)

Adjective

Relevant sense (OED)

likely

“having an appearance of truth or fact; that looks as it would happen, be realized, or prove to be what is alleged or suggested; probable” “that is capable of being; that may or can exist, be done, or happen” “having an appearance of truth; that may in view of present evidence be reasonably expected to happen or be the case” “possible as opposed to actual; having or showing the capacity to develop into something in the future; latent; prospective” of an argument, idea, statement: “seeming reasonable, probable, or truthful; convincing, believable” “expected or expecting to be (the object or thing specified) in the future; that is in prospect; proposed, likely, potential” “having the appearance or semblance of truth or reality; appearing true or real; probable” “that is so in essence or effect, although not formally or actually; admitting of being called by the name so far as the effect or result is concerned”

possible

probable

potential

plausible

prospective

verisimilar

virtual

4

191 First attestation with a situation-describing noun 1426: maner lykly4

1384: thingis … possible 1425: probable [S þat…]

1398: potencial hete

1565: plausabill argument

1809: prospective saving

1681: verisimilar Semblance

1697: vertual renunciation

There is one earlier example, with the adjective in predicative position, and qualifying a clause: sum men þenken licly þat… .

192

Modal Adjectives in the History of English

The examples above refer only to the modal meanings of these adjectives; the original meaning of most of them was non-modal – e.g. plausible was borrowed into English also with its original sense in Latin of “deserving of applause or approval; praiseworthy, laudable, commendable” (OED); and prospective acquired its modal sense only at the beginning of the 19th century. All these adjectives are borrowings: likely from Old Norse, and the rest from Latin or French. No other modal adjective appears to be attested in Old English; prior to the borrowing from Old Norse, English was able to function without expressing modality at DP (determiner phrase) level. Prospective and virtual were borrowed directly from Latin, and became modal in English only several centuries after the borrowing: the former often functions as a temporal adjective close in meaning to future, while the latter acquired its modal meaning almost 250 years after its first attestation with its original meaning of “possessed of certain physical virtues or capacities; effective in respect of inherent natural qualities or powers; capable of exerting influence by means of such qualities” (OED).5 Virtual, when opposed to actual, is not modal, but privative, like fake, false, or pretended. On the other hand, probable and plausible were used with their nonmodal meanings alongside the modal one: in its first attestation (1387), probable meant “that commends itself to the mind; worthy of acceptance or belief, acceptable, believable” (OED), and is first attested as modal in 1425, although the modal meaning was already available in French (the source of the loan). The only modal adjective that has apparently never been used with entities, even in Modern English, is verisimilar. This fact might be due either to its rarity in use, and consequently to the rarity of occasions to combine with entities, but also to its original meaning of ‘similar to the truth’, which can apply to descriptions of situations, but not of entities. I mentioned above that Old English did not have modal adjectives. This appears to be the case for many other languages, and probably for most of them: for instance, Navajo or Hieroglyphic Egyptian never had such adjectives, and in most European languages they were created on foreign models (e.g. Hungarian valószínĦ cf. German wahrscheinlich; Polish prawdopodobny cf. Latin verisimilis and French vraisemblable). Where they appear, modal adjectives seem to constitute areal phenomena. In West European languages, including English, the semantic and 5

The semantic evolution of these two adjectives is particular to English, and was not paralleled in French: prospectif has only the temporal meaning ‘relative to the future’, and virtuel is a recent anglicism in collocations like réalité virtuelle.

Octav Eugen DeLazero

193

syntactic properties of these adjectives are mostly inherited from their Latin and Greek ancestors, which only modified situation nouns, as their descendants continued to do in English until the end of the 18th century; as suggested by the OED evidence, from that time on the modal adjectives started being used with entity nouns as well, this new usage spreading apparently from possible to likely, probable, prospective, plausible, and virtual. The explanation for the chronological gap between the first attestations of modal adjectives with situation vs. entity nouns in English becomes obvious when considering the fact that, at the time when they were borrowed into English, these adjectives were not used with entity nouns in the source language. The combinations of modal adjectives with entity nouns appeared late in the history of English because they did not exist at the time when the modal adjectives were borrowed, and such use had yet to be accommodated. In the next section I will briefly discuss the situation in the languages from which the modal adjectives were borrowed into English.

4. Cross-linguistic Observations Older Indo-European languages did not have adjectives equivalent in all respects to the modal adjectives of the modern languages. When they existed, these adjectives appear to have been used only as underlying predicates, even when they surfaced adnominally; also in English and French, the examples with predicative uses are overwhelmingly more frequent not only in their early modern periods, but until, and including, the 20th century. Most importantly, I have not found in older Indo-European languages any example with modal adjectives that can be shown to qualify entities underlyingly. This was in fact to be expected, given that intensional (modal and temporal) adjectives cannot be used predicatively with entities even in the modern European languages, including English: *the president is probable/future. When adjectives that allow a modal interpretation in some contexts were combined with nouns describing entities, their meaning was never modal. Old Norse líkligr, although often glossed with the modal meaning ‘likely, probable’, does not behave like its Modern English descendant, and all the examples I have found allow a non-modal reading as well. This adjective was apparently never employed attributively, but only predicatively, as it still is predominantly in English. Being derived from the verb líka ‘please, satisfy’, the original meaning of líkligr was ‘satisfying,

194

Modal Adjectives in the History of English

agreeable, favorable’, whence ‘eligible, preferable, suitable’, and eventually ‘expected’, which could become ‘likely’ if the expectation was depersonalized and interpreted as a probability: (8) Þorsteini var þar vel fagnat, sem líkligt var ‘Thorstein was welcomed there, as was expected/likely to happen.’ (Zoëga 1910: s.v. líkligr) In the instances where the neuter Nom.-Acc. sg. líkligt is translatable as ‘expected’ or ‘likely’, it is never entirely clear that this ‘depersonalization’ actually happened, because, as in (8), there is always an experiencer detectable in the background, and the experiencer’s expectation – in the sense of ‘waiting for’ some pleasing event, rather than ‘expecting’ it – is not epistemically generalized to an impersonal probability. The general positive, non-modal, sense in Old Norse passed into English, too: among the earliest attested meanings are “apparently suitable or qualified”, “apparently able or fitted”, “of seemly or comely appearance; goodlooking, handsome”, and “having the appearance, or giving evidence, of vigour or capacity; strong or capable looking” (OED). All the other modal adjectives in English are ultimately of Latin origin, but in classical Latin only possibilis had modal value. This adjective was created in order to translate Greek dunatós, which was the only modal adjective in Greek. Until then, there were no modal adjectives in Latin. In both classical and medieval Latin, possibilis qualified situations, which could be expressed either verbally (9) or nominally (10), but there are no examples with entities (11): (9) non est possibile [talem resolutionem fieri] ‘It is not possible there to be such a solution.’ (Aquinas, Questiones disputatae de ueritate, q.15) (10) possibilis est conformitas inter nos et Deum ‘Conformity is possible between us and God.’ (Aquinas, Distinctio no.48) (11) *Caesar possibilis uictor est. intended meaning: ‘Caesar is a/the possible victor.’ The adjectives possibilis and potentialis both contain the nominal root pot- denoting power due to the preeminence or authority of the *potis ‘master’, plus the copula root (cf. posse ‘can, be able to’, from the same

Octav Eugen DeLazero

195

root *pot-). Their modal value follows from the notion of power as ability to bring about an event; possibilis can perhaps be best translated as ‘which can be effected, feasible’, and potentialis as ‘having the effective power (potentia) to come about’. Only possibilis is attested in classical Latin in the 1st century AD by Quintilian, glossing the Greek dunatón (12), but potentialis could have existed in late Latin (4th–5th centuries), when the related adverb potentialiter is attested in St. Augustine and Sidonius Apollinaris. (12) melius igitur qui tertiam partem duxerunt dunatón, quod nostri possibile nominant ‘Therefore those are better who made out the third part as dunatón, which we call possibile.’ (Quintilian, De institutione oratoria 3.8.25) The adjective probabilis was employed with entities only in the sense of ‘worthy of approval, commendable, acceptable’, or ‘praiseworthy, admirable’, e.g. ingenium probabile ‘commendable talent’ (Cicero, Brutus 180), while its primary meaning ‘which can be proved’ (verbal stem probƗ-) appears only with situations such as described in reports or accusations, e.g. attributive in probabilis ratio ‘plausible reason’ (Cicero, De officiis 1.8) and predicative in (13). This adjective can be said to express modality only to the extent that the suffix -bilis itself can. (13) id conjectura probabile est ‘This can be proved by conjecture.’ (Cicero, Tusculanae disputationes 4.2) I mentioned above that prospective and virtual became modal only in English, so they are irrelevant for Latin. As for plausibilis and verisimilis, originally meaning ‘to be applauded’ and ‘corresponding to the truth’ respectively, they too are scholarly creations. All these Latin adjectives, old and new, were presumably created in order to respond to specific cultural needs. The assimilation of Greek philosophy starting with the 2nd century BC and the scholastic speculation in the late Middle Ages involved a valuation of the truth and of the methods for its assessment that required expression in new terminology. In Greek, the only modal adjective is dunatós, which was the model on which the first modal adjectives were created in Latin, Germanic (Gothic), and Slavic (Old Church Slavonic), and which was used with entities only in its original etymological meaning of ‘strong, capable, able (to)’ (cf. dúnasthai ‘be able to’ and dúnamis ‘power’) – e.g. híppos dunatós ‘good

196

Modal Adjectives in the History of English

horse’ (Xenophon, Hipparchus 3.9), or dunat۸tatós te kaì soph۸tatos katà taNJta ‘very able and very skilled in these matters’ (Plato, Hippias minor 366.d). Otherwise, this adjective is employed only to describe situations – most frequently in predicative position (14), but also attributively, in infinitival raising constructions (15) like English likely in Johni is likely [ti to get hurt], where it does not have a modal value (or has a modal value to the extent that the suffix -tó-, similar to Latin -bil-, has one). Incidentally, (15) shows the probable mechanism whereby impossible could acquire the sense ‘inaccessible’ in the example from Shakespeare in (6). On the other hand, the use with entity nouns in (16) does not allow the reading ‘possible commander’, but can only mean ‘able/competent commander’. I have not found any examples with this adjective qualifying relationships. (14) oukhì es۸samén se, hoƭón te òn kaì dunatón ‘We did not save you when it was possible and feasible.’ (Plato, Criton 46.a) (15) hodòsi dunat‫[ ں‬ti kaì hupozugíois poreúesthai] ‘a road accessible even to the beasts of burden’ (lit. ‘a roadi able [to be walked on ti even by the beasts of burden]’) (Xenophon, Anabasis 4.1) (16) #dunatòs stratƝgós intended meaning: ‘possible general’ The semantic properties of some of these adjectives, before they were borrowed into English, follow from the syntactic properties of the verbs from which they were derived: Old Norse líkligr, Greek dunatós, as well as Latin possibilis, probabilis, are passive deverbal adjectives, and the corresponding verbs (Old Norse líka, Greek dúnasthai, Latin posse and probare) take sentential complements, which identify situations via event identification (Kratzer 1996). The modal adjectives inherited the selectional properties of the verbs, selecting situations, but not entities, and as such could combine either: - as predicates with clausal subjects in Old Norse (8), Latin (9), and Greek (15); or - as attributes with nouns describing situations, as in Latin (10) and (13), and Greek (14). Since the base verbs did not select entity nouns as complements, the derived adjectives could not combine with such nouns, and (11) and (16)

Octav Eugen DeLazero

197

were infelicitous. This was very probably the general situation in the older Indo-European languages that created or borrowed the respective adjectives. The Bible translations from Greek into Gothic and Old Church Slavonic render dunatós by two adjectives derived from cognate roots: Gothic mahts and Old Church Slavonic vǎzmožƱnǎ, both being past participles of verbs derived from the same Indo-European root meaning ‘strength’ (Old Church Slavonic mog-, Gothic mag-). Gothic has at least one adjective translatable by English modal adjectives: mahts ‘able, possible’ (translating Greek dunatós) and perhaps also skulds ‘owing, owed; lawful’, hence ‘necessary’, cf. the discussion in Sturtevant (1925) and Joseph (1981). All extant examples have these adjectives in predicative position. Skulds is modal only in that it is semantically related to a modal verb (skal ‘be obliged / be about to / shall, will’) that can denote necessity; syntactically it parallels mahts in the raising construction (17)-(18). Combinations of the adjective with modal meaning plus an entity noun (19) are not attested: (17) jahi ni mahta was [fram ainomehum ti galeikinon] ‘and shei could not [be healed ti by any]’ (Luke 8.43) (18) skulds ist [ti ushauhjan] [sa sunus mans]i ‘[the Son of Man]i must [be lifted ti up]’ (John 12.34) (19) #mahts þiudans intended meaning: ‘possible king’ Old Church Slavonic vǎzmožƱnǎ, when not used with its original meaning ‘strong’, as in vǎzmožƱna sǎtvoriti jego ‘to make him strong’ (Codex Suprasliensis 302.24-25), is employed with the meaning ‘possible’ (20) in the same way as its Greek counterpart in (14), whereas usages like (21) are not attested. (20) vƱsČ vǎzmožƱna vČrujІštjumu ‘everything is possible to the believer’ (Codex Zographensis, Mark 9.23) (21) #vǎzmožƱnǎ vojevoda intended meaning: ‘possible army commander’

Modal Adjectives in the History of English

198

For the sake of comparison, I will mention the Sanskrit adjective Ğakya- ‘capable, feasible, possible’, derived from the root Ğak- ‘power, might / be powerful, able, capable’, which behaves much like Greek dunatós: (22) like (15), and (23) like (14). As in all other ‘gerundives’ like rama۬Ưya- ‘enjoyable’, drasփ tփ avya- ‘visible / to be seen’, etc., modality is contributed by the suffix (like the Latin adjectives derived with -bilis), but also by the root (‘dynamic’ modality, in the etymological sense of this term: ‘force, power, might’), where the potentiality follows from the physical capabilities and effective power inherent in an agent. (22) Ğakyas khalu e‫܈‬ai… [prajñayƗ ti nivƗrayitum] ‘he can of course be checked by guile’ (lit. ‘hei is of course able [to be checked ti by guile]’) (Coulson 1992: 203) (23) adhunƗ Ğakyam [anena mara۬am api anubhavitum] ‘it is now possible [for him to suffer even death]’ (Coulson 1992: 205) Comparing the adjectives that started out as modal (rather than acquiring the modal meaning secondarily and later), i.e. Latin possibilis, Greek dunatós, Sanskrit Ğakyas, Old Church Slavonic vǎzmožƱnǎ, and Gothic mahts, it can be seen that they share three properties: -

-

-

they are deverbal: vǎzmožƱnǎ and mahts are participles synchronically, and possibilis, dunatós, and Ğakyas are derived from verbal roots with suffixes semantically equivalent to English ‘-(a)ble’ (drinkable, likeable, etc.); the verbal roots/stems denote ‘strength’ (Greek duna-, Sanskrit Ğak-, Gothic mag-, Old Church Slavonic mog-) or, derivatively, ‘effective ability’ (Latin pot-(e)s-), and the derived verbs mean ‘can, be able to’: the infinitives are Latin posse, Greek dúnasthai, Sanskrit Ğaktum, Old Church Slavonic mošti, Gothic magan; and they are passive in meaning, because the base verbs take clausal direct objects (as do also Old Norse líka and Latin probare, mentioned above): the strength or ability denoted by the verb applies to a situation, expressed in a clause, that can be brought about by using that strength; underlyingly, the matrix clauses are subjectless, because the subject is impersonal, the seat of the strength as effective power being the context; the object of the verb in the embedded clause can raise to become the subject of the matrix clause, and the adjective will agree with it:

Octav Eugen DeLazero

199

(24) posse/dúnasthai/Ğaktum/mošti/magan [S effect: ‘bring about…’] ‘to be able to bring about…’ >> objecti dunatós/Ğakyas/mahts [S ti…] The fact that the object of the base verbs is clausal, describing a (possible, potential, etc.) event, explains why: -

-

originally, these adjectives were not used adnominally, because underlyingly they were predicated of clauses, out of which the nouns could raise to become subjects, as in (10), (14), and (16), unless they remained in situ, as in (9) and (17); and when predicated of nouns underlyingly, as in (5), the meaning of those nouns was of the same type as the meaning of clauses, i.e. they only described situations, and never entities; also when substantivized, as in (20) translating Greek tà dunatá ‘those things which are possible’, the resulting nouns could only describe a situation, not an entity (the ‘things’ in the gloss refer not to entities, but to situations which can occur).

This origin of the modal adjectives explains why they never qualified entity-describing nouns underlyingly; when they modified such nouns superficially, as in (10), (14), and (16), the underlying clause out of which the noun moved was the complement of the adjective. The exceptions are English potential, together with its equivalents in modern European languages derived from scholastic Latin potentialis, and virtual, which started out as antonyms of actual, real (or equivalents): what exists only in potentia is not real/actual, so the original meaning of potential and virtual was privative, similar to false or fake, but without the connotation of deceit. When used with modal meaning, dissociated from the negation of reality/actuality, virtual and potential describe not the entity qualified as such, but a situation – namely, the state consisting of the existence of the respective entity. A closer look at the uses of the modal adjectives in Modern English (as in other modern European languages) shows that whenever a modal adjective modifies a noun describing an entity, that entity is always understood in a context, and in this sense the noun is always relational, in the sense that it describes an entity in context.

5. Synchronic Observations All nouns which can combine with modal adjectives are relational in a certain sense. This selectional property of modal adjectives follows from

Modal Adjectives in the History of English

200

the fact that the modality denoted by such adjectives applies to propositions, rather than entities. The propositions being themselves relations (between a subject and a predicate), modal adjectives can be used without restrictions with nouns describing situations, like victory in possible victory: the referent of a definite description like England’s victory is a situation (event).6 It is therefore convenient to distinguish between two selectional properties of modal adjectives based on their use with: -

nouns denoting entities in context, playing a ‘role’ in a relation: adjective[+role]; or nouns denoting situations (events or states, including relationships) as ‘schemas’ or relations: adjective[+schema].

This terminology belongs to Gentner (2005), who refers to relational nouns (describing entities in context) as ‘role categories’, and to situation nouns as ‘schema categories’. In Modern English, almost all modal adjectives can select both roles and schemas, but other languages which are able to express modality by means of adjectives can have different lexical items corresponding to the two selectional variants of possible, potential, and also probable and likely – for instance French possible[+schema] and éventuel[+schema/role] and German möglich[+schema] and eventuell[+schema/role]: möglicher[+schema] Sieg ‘possible victory’ is fine, but # mögliche[+role] Bombe ‘possible bomb’ is not; on the other hand, eventueller[+schema] Sieg ‘possible victory’ and eventuelle[+role] Bombe ‘possible bomb’ are both acceptable. Also in Polish, moĪliwa bomba ‘possible bomb’, can only be interpreted, if at all, with the noun as a calque on French bombe médiatique ‘breaking news’, therefore denoting a schema (event reported by the media), while ewentualna[role] bomba is quite acceptable. The nouns which normally describe entities cannot be used with modal adjectives, unless that entity is typically involved in a relationship, i.e. it performs a role in a schema. For instance, the collocation ?possible cat by itself is quite odd. Nevertheless, a scenario where the sentence This is a possible cat would be felicitous is not unconceivable, as when examining some tissue or DNA sample taken from an animal which died in an 6 The statement that definite DPs like England’s victory refer to situations must be qualified with Zucchi’s (1993: 113) observation that such DPs sometimes seem to refer to propositions, rather than events: in Mary’s arrival occurred on June 1, 1988, but it might have occurred on June 2, 1988, the DP Mary’s arrival in the second clause does not identify any event in the base world.

Octav Eugen DeLazero

201

accident. However, in this scenario, the noun cat does not refer to the tissue or DNA sample itself, but rather to the situation (schema) where the sample (role) will turn out to belong to a cat (role). Here, the noun cat describes a member (role) of the relationship (schema) between the animal and the samples, and it is this relationship that is described as being possible. There are other entity nouns which, as roles, can combine with modal adjectives, but only when what is referred to is some property that is liable to induce a schema, rather than merely the physical existence of those entities. The noun bomb can be felicitously used in the sentence This is a possible bomb only when the bomb is considered in its capacity as an agent (or rather patient; ‘agent’ and ‘patient’ are roles) in an explosion (schema). Otherwise, the only conceivable scenario would be one similar to the examination of the tissue or DNA samples: when disjoined parts of an object suggest that they belong to a bomb, the phrase possible bomb describes the relationship (schema) between the parts and the whole (role) as being possible. Here, the noun bomb is relational, describing the physical object bomb in the context (situation) of its intended use in an explosion. More examples:7 in I have found a possible house, what is described as possible is not the (physical existence of the) house – unlike in I have found a red flower, where what is described as red is the flower –, but the capacity of the building under discussion to serve as a dwelling, i.e. its role, like in We found a possible candidate. Likewise, in This is a possible hat for my mother, what is described as being possible is the suitability of that particular object to be used as a hat; the hat’s existence is beyond doubt, but whether it can serve as a hat – i.e. play a hat’s role in a schema (covering the head, etc.) – must be assessed in a modal expression. Also when used predicatively, as in This hat is possible, said when trying on hats, the meaning of the adjective is ‘suitable (to be used as a hat)’, not ‘possible’ like in possible victory, where the adjective cannot be interpreted as ‘suitable’ because it does not refer to suitability, but to the occurrence of the victory. In possible world, a world is probably not an entity – at least one cannot point out to a world as one can to an object; if nouns are disjunctively and complementarily classified into entities and relationships, world would fall into the latter category (as a set of relations), and it would be a situation: as space-time, a world is a time interval open at both ends, and any situation can be regarded as a time interval described in a true proposition (cf. Barwise & Perry 1983). 7

The examples in this paragraph have been suggested by Molly Diesing.

Modal Adjectives in the History of English

202

To sum up the observations made so far: Modal adjectives can only modify nouns which describe either situations or entities playing a role in a situation (context). These adjectives can be used predicatively only with the former class of nouns; a predicative use of a modal adjective with a role category, such as in #The murderer is possible, is infelicitous – either because the adjective is interpreted by default as selecting a schema (situation), rather than a role (entity), or for reasons to be explained later in this paper. The fact that modal adjectives primarily select for situationdenoting nouns in synchrony correlates with the diachronic observations that: -

-

in English, the first attestations of modal adjectives combined with entities (which are always roles) are almost invariably of later date than their use with schemas; and in the earlier history of Indo-European languages, the adjectives which are semantically closest to the modal adjectives in modern languages could only combine with schemas, their use with roles lacking the modal value of the modal adjectives in modern languages.

Larson (1998) proposes an analysis of non-intersective adjectives, of which the modal adjectives are a subclass, based on the properties of the modified nouns. In the next section I proceed to discuss the semantics of modal adjectives, starting with the denotations of the nouns they can modify, in order to explain the discrepancy in the attestation of combinations of modal adjectives with schemas vs. roles.

6. The Analysis: A Type-driven Approach 6.1. The Denotations of the Relevant Noun Classes A situation noun (schema), like war (event) or fatherhood (state), denotes a set of situations, and its denotation can be formulated as a function from the set of all situations (in the base world) to the set of truthvalues. Such a noun will have the semantic type st of propositions: (25) [[NOUNschema]] = Ȝes.[NOUN(e)] = Ȝes.[e is a NOUN] (26) [[crime]] = Ȝes.[crime(e)] = Ȝes.[e is a crime] (27) [[This act is a crime]] = [[ crime]] ([[ this act]]) = 1 iff this act is a crime

Octav Eugen DeLazero

203

The distinction between nouns describing situations and nouns describing entities is corroborated by the fact that nouns of different semantic types tend not to be conjoinable: (28) is fine, because both nouns describe entities, and have type et, whereas (29) is odd, because dog has type et, but relationship has type st. At the limit, the type mismatch between two conjoined nouns can be exploited to various effects (30). (28) I saw a dog and a cat. (29)

??

I saw a dog and a relationship.

(30) John was hurt by Bill’s axe and his allegations. If semantic types are not inherent properties of the referents, but are assigned in semantics, a noun which in most circumstances would be a schema can be treated as describing an entity and have type et, provided that the corporeality of the members of the set which it denotes can be imagined without too much effort. In (31), the fact that a battle is usually understood as being a situation, rather than an entity, is irrelevant for the battle’s visibility, and battle can be used together with the entity-denoting noun horse as objects of the same verb, without the sentence sounding odd: in this case, seeing the battle is like seeing a painting. (31) I saw a horse and a battle. On the other hand, a relational noun describes an entity which plays a role in a situation (context): a victim is a victim of an attack, an accident, etc. (events), and a brother is a member in a kinship group or a monastic order (states); a person cannot be a victim or a brother in the absence of a contextual situation. The common feature of role categories is that they describe entities in context. As with situation nouns, if the situation (context) is irrelevant, a relational noun is treated as describing an entity. In (32), the murderer’s role in a murder is irrelevant for his visibility, since he is visible as a person (entity), not as a murderer (entity in context). As such, the noun murderer can be conjoined with an entity-describing noun, yet the sentence is slightly odd when uttered out of context, because murderer still implies the occurrence of a murder, where a cat is difficult to imagine in the absence of additional background knowledge. (32) John saw a cat and a murderer.

204

Modal Adjectives in the History of English

A relational noun is essentially an entity-describing noun that is sensitive to context. Given that the context where an entity plays a certain role is by definition relevant for relational nouns, the denotation of such nouns (see (33)) will be a function of type est. In (35), the PP of England has the type s of situations, because its referent is the situation of reign. (33) [[NOUNrole]] = ȜxeȜes.[NOUN(x)(e)] = ȜxeȜes.[x is NOUN in/of… e] (34) [[king]] = ȜxeȜes.[king(x)(e)] = ȜxeȜes.[x is king in/of/… e] (35) [[John is king of England]] = [[ king]]([[England]])([[John]]) = = [ȜxeȜes.[x is king in/of/… e]](England)(John) = 1 iff John is king of England A relational noun being relative to a certain context, a non-relational noun can be regarded as a relational noun for which context is irrelevant: a person is a king (role category) only in a certain political context, but is a person (sortal noun) in most other contexts. In semantic composition, the context can therefore be irrelevant, in which case a relational noun will behave like a non-relational noun describing an entity with no context, and will be assigned the corresponding semantic type et.

6.2. The Denotation of the Modal Adjectives Larson (1998) proposes four “technical moves” for the analysis of adjectival modification: 1) relativize the semantics of common nouns to events / situations; 2) analyze the adjectives as predicates; 3) allow APs to be predicated either of the entity or of the event in modified noun; 4) introduce an event quantifier. This strategy allows him, in the first move, to treat relational nouns as doublets , where ‘x’ is an entity and ‘e’ is the context (situation); for instance, dancer will be a doublet describing a person ‘x’ who performs in a dancing event ‘e’, and father will describe a man ‘x’ in a state of paternity ‘e’. The second move “takes adjectives like beautiful to be predicates of things: […] beautiful is true of an individual x just in case x is beautiful relative to some comparison class C, which I’ll assume here to be given by context, but which may also be given by an explicit for-PP”

Octav Eugen DeLazero

205

(Larson 1998: 8). The third move means that if, for instance, beautiful dancer refers to the physical beauty of the dancer, the adjective will modify the individual ‘x’, and if it refers to the beauty of the dance, it will modify the dancing event ‘e’. The event quantifier helps to account for ambiguities like in my former restaurants, which can be interpreted either as ‘formerly were restaurants, currently are mine’, or as ‘formerly were mine, and are/were restaurants’: -

-

in the first reading, there is a context of function (of the premises as restaurants); the adjective qualifies the function, and has narrow scope over the noun at individual level (the restaurant as such); in the second reading, there is a context of ownership; the adjective qualifies the possession, and has wide scope over the noun at stagelevel (the restaurant for the duration of the ownership).

All modal adjectives are ambiguous in this sense – for instance, my possible restaurants can describe premises that are either ‘possibly restaurants, actually mine’ or ‘possibly mine, actually/possibly restaurants’. I undertook Larson’s first move in subsection 5.1. Analyzing the modal adjectives as predicates involves predicating them of either the entity or the context (situation) implicit in the denotation of a relational noun. Possibility and probability are propositional operators (see (36)-(37)), and as such they can only apply to constituents of type st, i.e. to clauses or to situation-describing nouns. In (36), A represents the accessibility relation. Aw0(w) means that the world ‘w’ is accessible from the base world w0.8 (36) [[ ¡ ]] w0 = Ȝpst.[ws[Aw0(w) ĺ [[ p]] w =1]] (37) [[ ¡ p ]] w0 = 1 iff for at least one possible world w accessible from w0, [[ p]] w = 1 When a sentence like We could win the war or a phrase like possible victory are uttered, what the speaker actually has in mind is not entire worlds accessible from the base world, but rather situations in these 8 For ease of notation, in the denotations of modality I am ignoring the model variable M (Portner 2009: 18), which is irrelevant for this discussion. A model is defined as a pair which associates a frame F with a valuation function V. A frame is a pair of W (the set of possible worlds) and a relation on W. A valuation function is a function which associates each pair of a member of W and an atomic sentence in the modal logic language with a truth value (Portner 2009: 14-17). The denotation of a modal operator should be understood as being relative to a model.

206

Modal Adjectives in the History of English

accessible worlds: one imagines the situation of winning the war, rather than the state of the entire universe in the event of a victory. As such, it is legitimate to adapt the denotations of the modal operators by replacing the reference to possible worlds by a reference to possible situations, understood as situations in possible worlds. In (38), Aw0(e) means that the situation ‘e’ occurs in a world accessible from w0: (38) [[ ¡ ]] w0 = Ȝpst.[es[Aw0(e) ĺ [[ p]] e =1]] (39) [[ ¡ p ]] w0 = 1 iff for at least one possible situation e accessible from w0, [[ p]] e = 1 The denotation of a modal adjective, when combining with a noun of type st, will be as in (40), with the event argument (the situation described by the noun) bound in a Ȝ-expression. If the adjectives combine with the nouns by functional application, the outputs of the derivations in (41)-(42) are consistent with the intuition. (40) [[possible[+schema]]] w0 = ȜfstȜes.[Aw0(e) & [[ p]] e =1] (41) [[possible war]] w0 = Ȝes.[Aw0(e) & [[war]] e =1] (42) [[That is a possible war]] w0 = 1 iff the referent of that is an event accessible from the base world which is identified as a war (40) is the denotation of modal adjectives in the ancient languages, as well as in English before the combinations of modal adjectives with relational nouns (roles) became allowed. When this happened, the new selectional variant of the existing modal adjectives had to be assigned a secondary denotation. Strictly speaking, an adjective of type (40) is intensional, but not modal. Indeed, although such an adjective combines with an event, that event is a datum, even if it does not belong to the base world: in both War is possible (predicative adjective) and We talked about a possible war (attributive adjective), the noun war describes a given event, in an accessible world, which has already been identified as a war. With relational nouns, the intensional operators have scope over the event argument only: a possible killer is a killer in a context accessible from w0, and a future king is a king in a context subsequent to i0. The denotations of a modal adjective combining with relational nouns will be: (43) [[possible[+role]]] w0 = ȜfestȜxe.[e[Aw0(e) & f(x)(e)]]

Octav Eugen DeLazero

207

(44) [[possible killer]] w0 = Ȝxe.[e[Aw0(e) & killer(x)(e)]] (45) [[John is a possible killer]] w0 = = 1 iff e[Aw0(e) & killer(John)(e)] = = 1 iff there is an event accessible from w0 in which John is a killer (43) is the secondary denotation of modal adjectives, once their combination with relational nouns became possible in the 18th–19th centuries. Given that only (43), but not (40), expresses modality, it can be said that English became able to express modality at DP level only after adjectives of type (43) were created, and that the languages, or stages in the history of individual languages, where (43) did not exist, do/did not have modal adjectives proper, i.e. they were not able to express modality at DP level. The semantics proposed above for the intensional adjectives accounts for the restrictions on their predicative use. As Zucchi (1993: 113) points out, a definite description containing a situation-describing noun is not always referential: in an intensional context like (46), it (referring to Mary’s arrival) has no referent in the base world in the second clause. (46) Mary’s arrival occurred on June 1, 1988, but it might have occurred on June 2, 1988. In That war is possible, the subject that war does not have a referent in the base world; if the predicate adjective is assumed to be generated inside a DP with a deleted noun [DP possible war], the adjective will combine with the underlying noun and yield a denotation consistent with the intuition: the sentence That war is possible is true if and only if the referent of that war is an event accessible from w0 which is identified as a war. Temporal adjectives like former, previous, current, present, and future, which, being likewise propositional operators, combine only with situation nouns (schemas) and relational nouns (roles), are never allowed in predicative position: #That war is future/current/former and #That president is future/current/former. The reason for this seems to be pragmatic, rather than semantic. Indeed, such a sentence would be uninformative, being made redundant by the knowledge of the base world at the utterance time; in these sentences, the subject must be referential, and the test in (46) does not apply: (47) #Mary’s arrival occurred on June 1, 1988, but it occurred on June 2, 1988.

208

Modal Adjectives in the History of English

Since the definite description that president presupposes that the subject in #That president is future/former and #That president is possible is referential (entity existing at the reference time / in the base world), the sentences are contradictory. With current, the sentence #That president is current would be redundant and uninformative, because it is already presupposed by the definite description that president that the subject has a referent at the utterance time.

7. Conclusions The observation that modal adjectives became able to modify nouns describing physical entities only very late in the history of English led to an inquiry into the properties of these adjectives in the languages from which the English ones were borrowed. This inquiry showed that those languages only allowed modal adjectives to modify nouns describing situations, as opposed to nouns describing entities. In Modern English, the entity nouns which can be modified by modal adjectives are always relational, in that they contain an event argument, i.e., they presuppose a situation where the entity described by the respective noun plays a certain role. Modal adjectives that only combine with situation nouns are improperly termed ‘modal’, because the nouns with which they combine describe given situations in accessible worlds. The genuinely modal adjectives are those which can combine also with relational nouns. Those languages which do not allow such combinations, including English until the end of the 18th century, are in fact unable to express modality at DP level. Considering the history of modal adjectives in English, as well as the situation in the languages from which English borrowed these adjectives, reveals significant cross-linguistic generalizations about the syntax-semantics interface in the expression of modality.

List of Symbols i0 w0 Œ Ȝ e s t

reference time base world not included in lambda-calculus the semantic type of nouns the semantic type of time intervals and worlds the semantic type of truth-values

Octav Eugen DeLazero

et st iff >> *X # X [[ X ]] [[ X ]] i0 [[ X ]] w0 [[ X ]] e

209

the semantic type of common nouns the semantic type of propositions if and only if diachronic reanalysis expression X is syntactically wrong expression X is semantically infelicitous denotation of expression X denotation of expression X at the reference time denotation of expression X in the base world denotation of expression X applied to entity ‘e’

Works Cited Primary Sources Aquinas, Thomas. Distinctiones. Aquinas, Thomas. Quaestiones disputatae de ueritate. Cicero. Brutus. Cicero. De officiis. Codex Suprasliensis. Codex Zographensis OED = Oxford English Dictionary Online. Accessed at http://www.oed.com. Plato. Hippias minor. Quintilian. De institutione oratoria. Shakespeare, William. The Merry Wives of Windsor. Xenophon. Hipparchus.

Secondary Sources Barwise, Jon & John Perry. 1983. Situations and Attitudes. Cambridge, Mass.: The MIT Press. Coulson, Michael. 1992. Sanskrit: An Introduction to the Classical Language. London: Hodder & Stoughton. Gentner, Dedre. 2005. “The Development of Relational Category Knowledge”. Building Object Categories in Developmental Time ed. by Lisa Gershkoff-Stow & David H. Rakison, 245-275. Hillsdale, NJ: Lawrence Earlbaum. Joseph, Brian. 1981. “On the So-called ‘Passive’ Use of the Gothic Active Infinitive”. Journal of English and Germanic Philology 80:3.369-379. Kamp, Hans & Barbara Partee. 1995. “Prototype Theory and Compositionality”. Cognition 57:2.129-191.

210

Modal Adjectives in the History of English

Kratzer, Angelika. 1996. “Severing the External Argument from its Verb”. Phrase Structure and the Lexicon ed. by Johan Rooryk & Laurie Zaring, 109-137. Dordrecht: Kluwer. Larson, Richard. 1998. “Events and Modification in Nominals”. Proceedings from Semantics and Linguistic Theory (SALT) VIII ed. by Devon Strolovitch & Aaron Lawson, 1-27. Ithaca, NY: Cornell University Press. Partee, Barbara H. 2007. “Privative Adjectives: Subsective plus Coercion”. Prepublication version accessed at http://people.umass.edu/ partee/docs/ParteeInPressKampFest.pdf. Portner, Paul. 2009. Modality. Oxford: Oxford University Press. Sturtevant, Albert M. 1925. “The So-called Passive Use of the Active Infinitive after the Verbal Adjectives mahts and skulds”. Journal of English and Germanic Philology 24:4.504-511. Zoëga, Geir T. 1910. A Concise Dictionary of Old Icelandic. Oxford: Clarendon Press. Zucchi, Alessandro. 1993. “Incomplete Events, Intensionality, and Imperfective Aspect”. Natural Language Semantics 7.179-215.

PART III ADVERBIALS AND PARTICLES

ADVERBIALS IN THE HISTORY OF ENGLISH: THE CASE OF BEFORE ARTUR BARTNIK

1. Introduction This paper deals with the development and the distribution of before, which was usually rendered by two adverbials in Old English: ær and beforan. By the adverbial I mean an objectless element. Before could also function as a preposition, when it was followed by a noun phrase, and as a connective, when it was followed by a clause. The choice of examining only objectless constituents is not uncontroversial: Jackendoff (1973) already suggested that such adverbs are in fact intransitive, i.e. objectless prepositions. However, traditional grammars such as Quirk et al. (1985) distinguish the adverbial and prepositional use of before. Moreover, the corpora I use in this study also make a distinction between both functions. Therefore, only a thorough analysis of both uses will show whether this division is justified. Examples of the two adverbial forms in Old English are given in (1) and (2) below:1 (1) Ĉa eode se man in beforan to ðam cynge Then went the man in before to the king ‘Then the man went in before to the king’ (coapollo, ApT:14.15.265) (2) & he onfeng þa ilcan gecynde for urum lufon þe he and he received the same nature for our sakes which he ær gesceop before created 1

All the examples are quoted by the title and page number of the electronic text. This is followed by information containing the original coding conventions used in the corpora.

Artur Bartnik

213

‘and he received that same nature for our sakes which he previously had created’ (coblick, HomS_8_[BlHom_2]:23.161.308) In what follows I will show that the two forms have a number of differences: quantitative, as the number of forms found in the corpora varies considerably; semantic, as the two adverbials carry different meanings – temporal, locative or both; grammatical, as ær and beforan do not always allow the same type of modification (see also Molencki 2007, who traces the history of before in Old and Middle English). The paper is organized as follows. In Section 2, I will present the distribution of ær and beforan in the Old English period. In Section 3, I will examine the changes that the two forms underwent in Middle English. Section 4 concludes the article. The present study will be based on The York-Toronto-Helsinki Parsed Corpus of Old English Prose, with 1.5 million words, and the Penn-Helsinki Parsed Corpus of Middle English (second edition), with 1.2 million words.

2. Before in Old English OE beforan had a number of variants that shared the same stem, that is OE foran. In the corpus I found the following forms: foran, ætforan, toforan, wiðforan and foran to. Examples of these forms are given below: (3) þæt sylfe gemet þæt we her beforan cwædon þæs dustes that same measure that we here before spoke of the dust ðysse ylcan wyrte iris Illyrice foran mid ecede gemencged & of the same plant iris Illyrice before with vinegar mixed and gedruncen, hyt fremað þam þe his gecyndelice sæd him drunk, it helps to the one who his natural seed himself sylfwylles fram gewiteþ, þone leahtor Grecas gonorhoeam voluntarily from departs, the malady the Greeks gonorhoeam nemneþ. name ‘that same measure that we before spoke about of the same plant, Iris, mixed to dust with vinegar beforehand and drunk, it helps him whose natural seed voluntarily discharges from him by itself, the Greeks call this malady gonorhoeam.’ (coherbar, Lch_I_[Herb]:158.2.2344)

214

Adverbials in the History of English: The Case of before

(4) ða wæron þa wælisce men ætforan mid þam cynge. then were the Welsh men before with the king ‘Then were the Welshmen with the king beforehand.’ (cochronE,ChronE_[Plummer]:1048.51.2285) (5) Ac þa Wylisce a toforan into muntan and moran ferdan But the Welsh always before into mountains and moors went þæt heom man to cuman ne mihte. that to them man to come not could ‘But the Welsh always went before into the mountains and the moors so that no man could come to them.’ (cochronE, ChronE_[Plummer]:1095.45.3234) (6) Ne tæce we na mid swa geradum bysenum and gebeorhlicum Not mean we not with so well arranged examples and safe larum, þæt he leahtras fyrðrige and weaxan læte, ac þæt he teaching that he vices promote and grow let, but that he snotorlice and mid soþre lufe hy simle wanige, swa swa wisely and with true love them always reduce as he ongyte, þæt hit framian mage his underþeoddum, swa swa he thinks that it do good could his subject people just as we geo ær wiðforan cwædon.2 we already before said ‘[By this] we do not teach that he should allow vices to grow and flourish with such good examples and great knowledge; on the contrary, as we have already said, he should eradicate them prudently and with true love, in the way he thinks best to his subject people.’ (cobenrul, BenR:64.121.6.1174) (7) Þis is swa we ær sædon, þæt seo witegung is of him, and þa This is as we before said that the prophecy is from him and the 2

The only temporal form of wiðforan I have found is in conjunction with ær.

Artur Bartnik

215

þing þe becomon he cydde him foran to, and heora mod things which became he revealed them before and their mind onlihte mid his micclan gife enlightened with his great gift ‘this is so as we before said that the prophecy comes from him and the things which happened he revealed to them before, and enlightened their mind with his great gift’ (coaelhom, ÆHom_7:216.1160) All the forms of beforan represent two main meanings: temporal and locative. The temporal meaning is illustrated in (3)-(7) and in (1) above. The locative meaning is illustrated in (8)-(12) below: (8) Ĉa het Thasia beforan gelædan Theophilum, Then ordered Thasia before to lead Theophilus, Dionisiades gerefan,3 Dionysias’ steward ‘Then Thasia commanded Theophilus, the steward of Dionysias, to be led before’ (coapollo, ApT:50.21.545) (9) Hit scan & berhte, foran swa ymb me uton mid þrymme It shone & glittered before so around me outside with glory ‘It shone and glittered before me and around me in glory’ (coalex, Alex:11.6.83) (10) ða geseah heo licgan ðone hring on ðam wege ætforan mid then saw she lying the ring on the way before with snode mid ealle. headdress with everything ‘then she saw the ring lying on the way before (her) covered with head-dress and everything.’ (cocathom2, ÆCHom_II,_2:14.87.330) 3

Thorpe (1834: 81) also uses the name Tharsia.

216

Adverbials in the History of English: The Case of before

(11) and cwæð þam arleasum þenum: belucað þa wælhreowan and and spoke to the wicked servants: lock the cruel and þa ærenan gatu and toforan onsceotað þa ysenan scyttelsas, the bronze gates and before shoot the iron bolts ‘and said to the wicked servants: lock up the cruel bronze gates and shoot the iron bolts before them’ (conicodA, Nic_[A]:21.1.9.484) (12) And þonne se sacerd smyrað mid þam halgan crisman his And when the priest anoints with the holy chrism his breost & his sceoldru, þonne befehð he þone mann mid breast and his shoulders then surrounds he the man with Godes scylde, þæt se deofol ne mæg ænig his ættrenra God’s shield that the devil not can any his poisonouns wæpna on him arfæstnian, naðer ne wiðforan ne wiðæftan, gif weapons on him fix neither before nor after if he þanonforð þurhwunað on rihtum geleafan. he from that time continues in right belief ‘And when the priest anoints his breast and his shoulders with the holy chrism he then surrounds the man with God’s shield so that the devil cannot fix any of his poisonous weapons on him neither before nor behind if he continues (to live) from that time in the right belief.’ (cowulf, WHom_8b:32.559) The frequency of the variants used in the corpus is not evenly distributed. The temporal meaning is more common than the locative meaning. Out of the temporal forms, beforan is the most common, the other temporal forms being only marginal (toforan, ætforan and wiðforan). As far as the locative meaning is concerned, beforan and foran are equally common. In fact, foran carries almost exclusively the locative meaning. The only exception I found is (3), perhaps because of the analogy with beforan used in the same sentence, a form that is normally used in the temporal sense. However, when foran is followed by to it normally acquires the temporal meaning. Needless to say, the remaining forms are used less frequently in the locative sense.

Artur Bartnik

217

The next point to note is that beforan does not seem to have been modified by any adverbials in Old English. Thus structures such as long before or a little before are never found, at least not with any of the before-type adverbials. The other form that means ‘before’ is ær. The first noticeable difference is that ær, which appears less frequently as ear or er as well, heavily outnumbered beforan in Old English. In my corpus I have found only 123 tokens of beforan (29 locative, 1 directional and 93 temporal cases) as against almost 3000 examples of ær. Further, ær had only a temporal meaning. Thus no locative examples are found. Finally, ær could be modified by other adverbials such as swiðe/swyðe ‘very’, hwene/little/lytle ‘little, somewhat’, (n)æfre ‘never’, wel ‘well’, or other elements such as the quantificational micle/mycle ‘much’. Some examples are given below: (13) þa heo þæt þa geornlice dydon þa lafe þære neahte tide, when they that earnestly did, the remainder of the night-time, þa cwomon swiðe ær in dagunge sume broðor, þa ðe hyre then came soon before in dawn some brothers who their forðfore sægdon, from þære stowe þe heo forðfered wæs. death said from the place which she dead was ‘Now when they had earnestly done so for the rest of the night, there came very early at dawn some brethren, to announce her decease, from the place where she died.’ (cobede, Bede_4:24.340.24.3420) (14) Eac is ðeos bisen to geðencenne, þæt is þæt ælcum men þincð Also is this example to consider that is that to every man seems huniges biobread þy weorodra gif he hwene ær biteres onbirigð. honey’s comb the sweeter if he a little before bitter tastes ‘Also this example is to be considered, that is, that to every man honey seems the sweeter, if he a little before tastes anything bitter.’ (coboeth, Bo:23.52.2.941)

218

Adverbials in the History of English: The Case of before

(15) ðes geares forbarn Lundenburh anre nihte ær Assumptio this year burned London one night before Assumption sancte Mariæ swa swyðe swa heo næfre ær næs syðþan heo of St. Mary as great as it never before was not since it gestabeled wæs. built was ‘This year was London burned, one night before the Assumption of St. Mary, so terribly as it never was before, since it was founded.’ (cochronE, ChronE_[Plummer]:1077.3.2743) (16) ða andswærode him Moyses, Eala Dauid, hwæt is þin neod then answered him Moses, Lo, David, what is your desire swa mucel þet ðu þus ær cumen eart? so great that you thus before come are? ‘Then answered him Moses, Lo, David, what is your so great desire that you have come thus early?’ (corood, LS_5_[InventCrossNap]:70.75) The modifier lange (longe, lenge) ‘long’ is an interesting case. It could premodify and postmodify ær. Consider the examples below: (17) ond he wæs longe ær swiðe earfaðcierre to Godes geleafan. and he was long before very difficult to convert to God’s faith ‘and he was very difficult to convert to God’s faith for a long time before.’ (comart3, Mart_5_[Kotzor]:Jy6,B.2.1071) (18) Þa sendon hie to Læcedemonium & bædon þæt hie gefriend Then sent they to Macedonians and asked that they friends wurden, þeh hie ær longe gefiend wæren, became, though they long before enemies were ‘then they sent to Macedonians and asked to become friends, though they were enemies long before’ (coorosiu, Or_3:7.65.3.1271)

Artur Bartnik

219

It appears that longe ær/ær longe in the examples above is ambiguous because the italicized phrases could indicate either a point in time, in which case the meaning is ‘a long time before’, or duration, in which case the meaning is ‘for a long time before’. The choice of the interpretation is important because longe does not seem to modify ær in the durative reading. Finally, a combination of the two OE adverbials was also possible. Consider (19) below: (19) Se æresta cyning wæs Ninus haten, swa we ær beforan sægden The first king was Ninus called, as we before said ‘The first king was called Ninus, as we previously said’ (coorosiu, Or_2:1.36.25.713) In the corpus, I found 38 ær-beforan (wiðforan) combinations. The reverse order is marginal, with only 3 cases. As might be expected, such combinations are always temporal because ær is present. To sum up, the two forms meaning ‘before’ show clear differences. Ær was more common and always represented the temporal meaning. It could be modified as well. The multi-variant form beforan was relatively rare but could be used in the locative sense beside the usual temporal meaning. Beforan was not normally modified by other adverbials or quantifiers. The two forms could occur together, but only in the temporal sense and normally in the sequence ær-beforan.

3. Before in Middle English The ME period brought a dramatic change: within a very short time, ME before(n) became more common, ousting ME er almost completely. Therefore let me first concentrate on before(n). Since ME spelling was so variable, and because so many word or spelling forms were possible, all the ME variants found in the corpus are given in Table 1, arranged in rows according to the possible changes in their final part of the word. Some slots are left blank, as no corresponding form has been found in the data.

220

Adverbials in the History of English: The Case of before

Table 1. Possible ME variants of before(n) (The numbers in brackets indicate the locative use.) before (5) be-fore be fore byfore by-fore by fore bifore (3) bi-fore (3) tofore to-fore to fore

befor (1) be-for

beforen

byfor by for bifor

beforn be-forn byforn (1)

biforen (3)

by forn biforn (1) /biforne/biforenn (1)

bi-foren (1) toforn/toforne to-foran teforen te-foren

afore (1) beuore (1) be-uore byuore

aforn

biuoren (1) bi uoren touore to-uore for (1), uore

The temporal before with all its spelling variants is the most common form while other forms such as tofore, afore, for, uore and their spelling variants are significantly rarer (with usually only a couple of examples of a given variant). As in Old English, both meanings, temporal and locative, were represented by before(n). Interestingly, the number of locative cases did not noticeably change, although the number of temporal before(n) examples soared dramatically. In Old English I found 29 locative cases against 23 in Middle English (numbers for locative cases are given in brackets in Table 1). Therefore, it is only the number of temporal before(n) tokens that increased in Middle English. There is no clear preference for the locative meaning of any forms, although the bi- forms take this meaning most often. In Middle English, before(n) started to be modified by adverbials and quantifiers, which include longe ‘long’, neuer ‘never’, litel/sumwhat ‘little, somewhat’ or muchel ‘much’. Compared to OE ær, the choice of the modifiers seems to be limited and there are fewer instances of modification. For example, I found only 6 examples of before(n) forms premodified by longe. The reverse order was also possible (1 example).

Artur Bartnik

221

Both word orders, when longe premodifies and postmodifies before(n), are shown below in (20) and (21): (20) but for-asmiche as þe contre was вeuen longe bifore to Horn, but for as much as the country was given long before to Horn, þrouв Vortyger, þo he hade spousede his cosyn; þere he hade through Vortigern, when he had married his cousin there he had made a faire castel þat me callede Horncastel, after his owen name made a fair castle which one called Horncastle after his own name ‘but for as much as the country was given long before to Horn, through Vortigern when he had married his cousin; there he had made a nice castle which was called Horncastle, after his own name.’ (CMBRUT3, 53.1563) (21) Schrift ach to beon bi þocht biforen longe. Confession ought to be considered before long ‘confession should be considered a long (time) before’ (CMANCRIW-2, II.253.117) Semantically, there seems to be no obvious difference between the two sequences: both mean ‘a long time before’, as distinct from Modern English semantics (cf. long before ‘a long time ago’ and before long ‘soon’). However, as indicated above (cf. examples (17) and (18)), ambiguity might appear in the interpretation of before(n) modified by longe. It seems that the point-in-time reading is better in (20), while in (21) the other, durative interpretation (‘for a long time before’) appears to be possible as well. Then longe would not modify before(n). Other modifiers are even less frequent: (22) þe creatur al hol & sownd thankyd hym of hys cher the creature all hale and sound thanked him for his kindness & hys charyte, mech merueylyng & gretly a-wonderyd þat and his charity, much marvelling and greatly astonished that sche felt no peyn & had felt so mech a lytyl be-for. she felt no pain and had felt so much a little before

222

Adverbials in the History of English: The Case of before

‘the creature completely well and uninjured thanked him for his kindness and benevolence greatly amazed and astonished that she felt no pain which she felt so much a little before’ (CMKEMPE, 22.457) (23) Imong alle þere pine þe he for us þolede; ne undude he nefre among all the pain that he for us suffered not opened he never ene his muð mid uuele to-вeines nan of heom al swa þet one his mouth with evil against none of them as the writ seide bi him muchel to-foran writing said by him much before ‘Among all the sufferings that he suffered for us, he opened never once his mouth wickedly against any of them, as the Scripture said of him long before’ (CMLAMBX1, 121.1191) The reverse order is possible, though extremely infrequent: (24) Of þisse stilnesse he spekeð biforen lutel. Of this stillness he speaks before little ‘concerning this stillness, he speaks a little before (that)’ (CMANCRIW-1, II.123.1575) Generally, these modifying structures appeared relatively late in Middle English (Brut, Capgrave’s Chronicle, Aelred of Rievaulx’s De Institutione Inclusarum). The Middle English Dictionary (MED) gives an example from the beginning of the 14th century (example (25) below). In my corpus I also found earlier examples (in the Ancrene Riwle and the Lambeth Homilies). This suggests that before(n) acquired full adverbial properties relatively early, since it could be modified in early texts, though it has to be admitted that it is after the 14th century that one can find more examples of this. (25) King alfred was wisost king þat longe was biuore King Alfred was wisest king that long was before ‘King Alfred was the wisest king that lived a long time ago’ (c1325(c1300) Glo.Chron.A (Clg A.11) 5388 (MED))

Artur Bartnik

223

The adverbial beforen could also postmodify noun phrases. These are late examples. The MED gives an example from around the 14th century. My data generally support this dating. (26) This Pandare, that of al the day biforn Ne myghte han comen This Pandare, that of all the day before not could have come Troilus to se, Troilus to see ‘this Pandare, who all the day before could not come to see Troilus,’ (a1425(c1385) 5.281 Chaucer TC (Benson-Robinson) (MED)) (27) and þe Kyng Edelf come aвeyne þat hade fouвten wiþþe and the king Edelf came again that had fought with Danois þe day toforne to þat bataile. Danish the day before to that battle ‘and the king Edelf, who had fought with the Danes the day before, came again to that battle.’ (CMBRUT3, 107.3252) As I mentioned earlier, er(e) with its variant forms ar(e), ear(e), ær(e) is on the decrease. However, in some ME texts er is the dominant form. For example, I found only 4 before(n) forms (with only one clearly locative) against 34 er forms in the Lambeth Homilies (a West Midlands text). Similarly, another West Midlands text, the Ancrene Riwle, contains only 11 before(n) forms (including 2 clear locative cases) against 41 ear tokens. The Trinity Homilies and the Ormulum, both early East Midlands texts, show the following pattern: the former text contains 38 er forms beside 2 before(n) forms (including 1 locative form), while the latter has 18 er beside 3 before(n) forms (including 1 locative token). It is worth mentioning at this point that there was no rejection of the before(n) form as such at this stage. As a preposition, before(n) appears 105 times in the Ormulum, 57 times in the Ancrene Riwle, 35 times in the Trinity Homilies and 15 times in the Lambeth Homilies. Clearly, it is the function that is a crucial factor here. In other texts in the corpus er is either in the minority or there are too few examples to make definite judgements. For example, there are 8 and 5 instances of ear in Hali Meidhad and St. Katherine respectively, with no before(n) forms in either text. Finally, the OE ær form is quite understandably present in the early texts such as the Peterborough

224

Adverbials in the History of English: The Case of before

Chronicle, the Kentish Sermons or Vices and Virtues, though in relatively small numbers. All in all, er is the major temporal adverbial in some Early Middle English Midland texts although it loses ground very fast. It seems that time is a crucial factor in this respect (the list of texts mentioned in this section involves Early Middle English texts only). Er retains the ability to be modified. The modifiers include the negative neuer ‘never’, longe ‘long’, or quantifiers such as muchel ‘much’ and litel ‘little’. Some examples are given below: (28) but Kyng Arthure and his men quellede so meny Saxones, but King Arthur and his men killed so many Saxons þat neuer er was seyne sochs a slauвter; that never before was seen such a slaughter ‘but King Arthur and his men killed so many Saxons that such slaughter was never before seen’ (CMBRUT3, 70.2115) (29) бe iherden a lutel er on þisse redunge. þet ðe halie gast you heard a little before in this reading that the holy ghost com ofer þa apostlas mid furene tungen; came over the apostles with fiery tongues ‘you heard a little before in this reading that the Holy Ghost descended upon the apostles in the form of fiery tongues’ (CMLAMBX1, 93.813) (30) Efter ham ich iseh towart te patriarches & te prophetes After them I looked towards the patriarchs and the prophets þe makied swuch murhðe þt ha aren nuðe i þt ilke lont of that make such mirth that they are now in that same land of blisse þt ha hefden of feor igret ear on eorðe & seoð nu bliss that they had from far hailed before on earth and see now al þt isoðet. þt ha hefden longe ear icwiddet of ure all that verified that they had long before prophesied of our lauerd as he hefde iscawed ham i gastelich sihðe. lord as he had showed them in spiritual sight

Artur Bartnik

225

‘After them I looked towards the patriarchs and the prophets that make such mirth because they are now in that same land of bliss that they had once hailed from afar on earth and they see now all that is verified which they had long before prophesied of our lord, as he had showed them in spiritual vision.’ (CMSAWLES, 180.220) Postmodification was also possible, as shown below: (31) & on ealle westme swa micel untime wearð swa hit ne wæs and on all fruit so great bad time came as it not was feola gear ær. many years before ‘and in all fruit such unseasonableness came as was not known for many years before.’ (CMPETERB, 47.173) (32) swa swa Symeon hire hwilen ær gewitegode, his slege just as Symeon her a while before prophesied his murder þurheode hire sawle (…) pierced through her soul (…) ‘just as Symeon prophesied to her a while ago, his murder pierced through her soul (…)’ (CMKENTHO, 137.93) It has to be stressed, however, that the pre- and postmodification of er is extremely rare. Suffice to say that each modifier is normally represented by just one example in the corpus (the only exception is neuer ‘never’). It thus seems that such structures, though possible, are on the way out along with the form er itself. Finally, as for the combination of er with before(n), I found only the one example given below which, surprisingly, comes from a relatively late text: (33) Then, yn anoþir nyght, Seynt Nycolas kest yn anoþir somme Then in another night Saint Nicholas cast in another sum of gold ynto þe mannys chambur as he dede er befor; and soo of gold into the man’s chamber as he did before, and so

226

Adverbials in the History of English: The Case of before

þe thryd tyme, tyll þay were all yweded the third time till they were all wedded ‘then another night Saint Nicholas dropped another sum of gold into the man’s chamber, as he did before, and soon the third time until they were all married’ (CMMIRK, 13.347) Generally, since the number of er forms is drastically reduced in Middle English, so is the number of the modifiers. The data show that the structures persist in the lexicon although they are marginal in the ME texts.

4. Conclusion This corpus-based study shows that the two forms meaning ‘before’ in Old and Middle English have a number of differences. Firstly, there are quantitative differences: ær, the dominant form in Old English, became radically reduced in Middle English, though in some Early Middle English texts er remained the basic form. Secondly, there are semantic differences: before adverbials represented two meanings, temporal and locative in both periods, though the locative was clearly of secondary importance. Ær carried only the temporal meaning. Thirdly, there are grammatical differences: before in Old English could not be modified, as distinct from ær, whereas in Middle English it was possible to modify both forms. Finally, only in Old English could both forms be used simultaneously, with only one such example in Middle English. The next step in the research should be a careful analysis of the prepositional function of before, i.e. the one in which it is followed by an object.4 If it turns out that the two functions should collapse into one it might have consequences, among other things, for corpus tagging, which makes use of the distinction between prepositions and adverbs.

Acknowledgements I would like to thank the anonymous reviewers for the numerous comments and questions that helped me improve my paper. I am also grateful to the participants of the 16th International Conference on English Historical Linguistics (Pécs, Hungary, August 2010) for their constructive 4

Beforan as a time preposition is very rare in Old English. B&T gives examples in vol. II only (cf. Pasicki 1987: 137). Therefore beforan is used almost exclusively as an objectless constituent, i.e. as an adverbial in Old English.

Artur Bartnik

227

criticism. Finally, I would like to thank the editors of this volume, especially Dóra PĘdör, for many invaluable suggestions that enabled me to avoid many mistakes. Of course, all the remaining errors are mine.

Works Cited Primary Sources B&T = Bosworth, Joseph & T. Northcote Toller. 1898. An Anglo-Saxon Dictionary based on the manuscript collections of the late Joseph Bosworth. London: Oxford University Press. Kroch, Antony & Ann Taylor, eds. 2000. Penn-Helsinki Parsed Corpus of Middle English (2nd ed.), available at http://www.ling.upenn.edu/histcorpora/PPCME2-RELEASE-3/index.html. MED = Middle English Dictionary, http://ets.umdl.umich.edu/m/med. Taylor, Ann, Anthony Warner, Susan Pintzuk & Frank Beths, eds. 2003. The York-Toronto-Helsinki Parsed Corpus of Old English Prose. Oxford: Oxford Text Archive.

Secondary Sources Jackendoff, Ray. 1973. “The Base Rules for Prepositional Phrases”. A Festschrift for Morris Halle ed. by Stephen R. Anderson & Paul Kiparsky, 345-356. New York: Holt, Rinehart & Wilson. Molencki, Rafaá. 2007. “On the Rise of the Temporal Preposition/ Conjunction Before”. To Make his Englissh Sweete upon his Tonge. Medieval English Mirror (3rd ed.) by Marcin Krygier & Liliana Sikorska, 37-54. Frankfurt am Main: Peter Lang. Pasicki, Adam. 1987. Temporal Adverbials in Old and Middle English. Lublin: Redakcja Wydawnictw Katolickiego Uniwersytetu Lubelskiego. Quirk, Randolph, Sydney Greenbaum, Geoffrey Leech & Jan Svartvik. 1985. A Comprehensive Grammar of the English Language. London & New York: Longman. Thorpe, Benjamin. 1894. The Anglo-Saxon Version of the Story of Apollonius of Tyre. London: John & Arthur Arch.

THE HISTORICAL DEVELOPMENT OF NO SOONER … THAN AND ITS SEMANTIC CHANGES DAISUKE SUZUKI

1. Introduction In Present-Day English, the construction no sooner … than is used with subject-auxiliary inversion and with a combination of the past perfect and past tense in most cases, as shown in example (1): (1) No sooner had we got home than the police arrived. (Huddleston & Pullum 2002: 1133) In data from the British National Corpus (BNC) as Table 1 shows, the use of then or but as a correlative of no sooner offers a fair possibility that the form of such a construction will vary in an actual text. However, a look at the earlier history of this construction can explain this variation. Table 1. Variations and raw frequencies of the correlatives of no sooner (BNC data, based on Suzuki 2010: 44) no sooner (238) hardly (134) scarcely (38)

than (211), when (10), then (2), but (2), no correlative (13) when (93), before (32), than (9) when (24), before (13), than (1)

Example (2a) below illustrates the use of then, (2b) the use of but, and (2c) the non-use of a correlative: (2) a. No sooner had Syd apparently recovered from the knee injury he sustained in New Zealand, then he suffered another major setback. (BNC: K25)

Daisuke Suzuki

229

b. No sooner justice had … But the Norweyan Lord, surveying vantage … Began a fresh assault. (BNC: HPG) c. ... no sooner you got your dinner and had a drink, you had to be back again ... (BNC: HDH) The purpose of this study is to outline the stages of this development and discuss how the construction no sooner … than has changed in meaning over time. We will demonstrate how the evolution of no sooner up to the present day can be characterized in terms of grammaticalization.

2. Previous Studies In this section, I will focus on previous accounts of no sooner ... than. With regard to no less/longer/more/other/sooner/etc. (...) than/but/or, Araki & Ukaji (1984: 489) point out that than and but in these clauses are structural alternatives in Modern English, as is shown in examples (3a, b): (3) a. He could no more but her great misery bemone. (Spenser, F. Q. IV. xii. 12) b. He was no sooner alighted, but he was in haste to be gone; (1766 Goldsmith, Vicar xviii) This is because just like but, than gives a negative property to what follows it, as exemplified by examples (4a, b): (4) a. She is richer than you think. b. You do not think that she is so rich as she really is. (=(4a)) (Jespersen 1917: 37) As for no sooner (…) but, Poutsma (1929: 670) mentions that in Early Modern English, we frequently find but instead of than after no sooner. Mustanoja (1960: 284) also maintains that the construction no sooner but was first found in fifteenth-century texts, and that it was common in Early Modern English. Subsequently, considering no sooner ... than, Fischer (1992: 355) indicates that no sooner ... than was not found in Middle English, and that other expressions were used instead, such as not so soone … that/but, as in example (5):

230

The Historical Development of no sooner … than

(5) And nat so sone [he] departed nas / Tho fro him, that he ne mette … (The House of Fame, III 2068-70) In addition, Rissanen (1999: 314) points out that the earliest instances of no sooner (…) than found in the Helsinki Corpus date from around 1600. According to Terasawa (1997: 1310), no sooner … but was first recorded in 1560 and was obsolete by the early nineteenth century, while no sooner … than could be traced back to 1594. While noting the replacement of but by than, previous studies offered no detailed description of this shift. This paper, therefore, provides some new insights into the development of no sooner … than.

3. Data This corpus-based investigation offers the total numbers of occurrences of expressions such as no sooner … than in my corpus, and these numbers enable us to use quantitative analysis. In this study, I investigated the entire dataset of the second edition of The Oxford English Dictionary (OED) on CD-ROM and used the OED quotations database as a main corpus. As noted in previous studies such as Brewer (2000), Mair (2001, 2004), Hoffmann (2004), and Iyeiri (2010), a collection of dictionary citations is not a balanced or representative corpus, although the dataset covers more than a thousand years of English usage and contains a large amount of material for analysis. No other historical corpora provide sufficient data concerning no sooner … than. Therefore, the use of the OED enables us to examine the development of this construction from different angles. The search in the OED yielded 314 instances of data concerning no sooner … than. In order to analyze the target expression during a particular century in detail, I used different sets of databases as additional means of investigation. First, I downloaded various texts from Project Gutenberg. I examined 22 works that 13 authors published within a restricted timespan, and collected 461 examples of data concerning no sooner … than. Second, since the examples of no sooner in the OED are rather restricted in genre, I investigated the dataset of the London Gazette, which is the official newspaper of record for the United Kingdom, as well as A Corpus of English Dialogues 1560–1760 (CED), which contains various modes of

Daisuke Suzuki

231

speech presentation in Early Modern English. From the London Gazette and CED, 109 and 20 examples were identified, respectively.1

4. Results and Discussion 4.1. Correlatives of no sooner Figure 1 provides a survey of the development of the correlatives of no sooner, focusing on but and than.2, 3 The data for a diachronic overview in Figure 1 spans the period from the late fourteenth century to the twentieth century.4

Figure 1. Proportions of the correlatives of no sooner (OED) 1

Duplicated examples in the London Gazette are counted only once. Following are the earliest instances of the construction no sooner with but and than: (i) He had no sooner made his wysshe but they were set on the table. (1533 Ld. Berners Huon lxxxiv. 264, OED) (ii) They are no sooner inshored, than they yeeld up their ghosts. (1577 Stanyhurst Descr. Irel. in Holinshed’s Chron. (1807–8) VI. 38, OED) 3 In Figure 1, “others” in earlier times include several examples of the construction with that as a correlative. 4 The data in Figure 1 contain few instances from the 14th and 15th centuries. 2

The Historical Development of no sooner … than

232

Data for Figure 1

but

32

1601– 1650 49

than

2

2

6

11

19

33

31

30

others

13

9

7

3

2

2

2

2

Total

47

60

44

32

29

36

34

32

㻃

–1600

1651– 1700 31

1701– 1750 18

1751– 1800 8

1801– 1850 1

1851– 1900 1

1901– 0

At first but was the dominant form, and the use of than was firmly established only in the course of the nineteenth century. The data in Figure 1 also suggest that there was competition between but and than in the first half of the eighteenth century. Examples (6a–d) below illustrate the use of but and than: (6) a. I no sooner rais’d my Head in Sight, but slam came three or four Stones at me. (1726 G. Roberts Four Yrs. Voy. 320, OED) b. But he had no sooner labour’d himself into a tolerable knowledge of the Affairs of it, than he rode triumphant. (1723 Dk. Wharton True Briton No. 48, OED) c. She had no sooner made an end of her speech, but she withdrew. (a1774 Goldsm. tr. Scarron’s Com. Romance (1775) I. 57, OED) d. Dinner was no sooner over, than the loo-table was introduced into the drawing-room. (1789 C. Smith Ethelinde (1814) II. 130, OED) In order to further explore what transformation took place during the first half of the eighteenth century, I would like to use the data taken from the texts from Project Gutenberg, as mentioned in Section 3. Table 2 exhibits the raw frequencies of but and than with no sooner:5 5

Out of the 22 works examined, the following texts contain no example of the construction no sooner: Gulliver’s Travels by Jonathan Swift, The Beggar’s Opera by John Gay, An Essay on Man by Alexander Pope, and Letters to His Son by Philip Dormer Stanhope Chesterfield.

Daisuke Suzuki

233

Table 2. Raw frequencies of but and than with no sooner (Project Gutenberg) 

but

than

1704 Jonathan Swift The Battle of the Books (1)

0

1

1704 Jonathan Swift A Tale of a Tub (1)

0

1

1719 Daniel Defoe Robinson Crusoe (16) 1720 Daniel Defoe The Life, Adventures and Piracies of the Famous Captain Singleton (4) 1720 Daniel Defoe Memoirs of a Cavalier (4) 1722 Daniel Defoe The Fortunes and Misfortunes of the Famous Moll Flanders &c.(7) 1722 Daniel Defoe A Journal of the Plague Year (1)

8

8

3

1

4

0

7

0

1

0

1724 Daniel Defoe The Fortunate Mistress (4) 1736–50 Horace Walpole Letters of Horace Walpole – Volume I (1) 1739–40 David Hume A Treatise of Human Nature (4)

2

2

1

0

0

4

1740–41 Samuel Johnson Parliamentary Debates I. (8)

1

7

1740 Samuel Richardson Pamela (3)

1

2

1742 Henry Fielding Joseph Andrews (66) 1744 Eliza Fowler Haywood The Fortunate Foundlings (37) 1747–49 Samuel Richardson Clarissa (12) 1748 Tobias Smollett The Adventures of Roderick Random (121) 1749 Henry Fielding The History of Tom Jones, a Foundling (165) 1750 John Cleland Memoirs of Fanny Hill (6)

1

65

1

36

5

7

0

121

3

162

5

1

In Table 2, numerous works of Daniel Defoe and John Cleland show the clear predominance of but, while than is the dominant form in the works of Samuel Johnson, Henry Fielding, Eliza Fowler Haywood, and Tobias Smollett (see the underlined numbers). Interestingly enough, a closer look at the data in Table 2 reveals that there are competitive frequencies between but and than within the same work, as seen in Robinson Crusoe, The Fortunate Mistress, and Clarissa, exemplified by examples (7a, b):

The Historical Development of no sooner … than

234

(7) a. ; but no sooner were my eyes open, but I saw my Poll sitting on the top of the hedge; (1719 Daniel Defoe Robinson Crusoe) b. I had no sooner stepped do (sic!) ground, than I plainly saw it was a terrible earthquake, … (1719 Daniel Defoe Robinson Crusoe) Second, I also analyzed the data from the London Gazette and CED. Table 3 is based on the data from the London Gazette, with the use of but and than with no sooner linguistically illustrated in examples (8a, b): Table 3. Raw frequencies of but and than with no sooner (London Gazette) 㻃

but

1665–1700

54

1

1701–1750

9

18

1751–1800

3

24

Total

66

43

than

(8) a. They were no sooner got on the other Side of it, but a great Number of Prussians were falling on them, from which they escaped, by a small Career, to the Rutowski Regiment of Light Horse; (11 June 1745, London Gazette) b. The King was no sooner informed of the Princess’s safe Delivery, than he sent M. Insenstiers, one of his Chamberlains, to compliment their Royal Highnesses in his Name; (28 July 1750, London Gazette) In Table 3, a significant decline in the use of but with no sooner is evident by 1750, while the construction with than is dominant after 1750. With this in mind, I am going to discuss the results of the investigation of CED. The result is presented in Table 4, which shows the raw frequencies in the CED data, as linguistically illustrated in examples (9a, b):

Daisuke Suzuki

235

Table 4. Raw frequencies of but and than with no sooner (CED) but

than

1651–1700

6

0

1701–1750

8

5

1751–

0

1

Total

14

6

(9) a. I WAS no sooner got into the Street, but I saw another Woman come to me. (CED) b. …, and he was no sooner gone, than (^Arentia^) ask’d how long he had been in the Family. (CED) Here again, the dominance of the use of but with no sooner is clear for the period 1651–1700. The important point is that in dialogues as well, but and than were in competition during the period 1701–1750. (Although note that the number of examples is fairly low.) As clearly shown in Tables 2–4, the first half of the eighteenth century is a key period in the development of this construction.

4.2. Other Factors Connected to the Development of this Construction The factors to be considered regarding the construction no sooner are (i) inversion in the first clause, (ii) tense in the first clause, and (iii) the cooccurrence with synonymous expressions such as immediately, instantly, at once, and presently. In this subsection, I make use of the entire dataset of the OED quotations again. First, Table 5 traces the relative frequency of inversion in the construction: no sooner with inversion was less frequently found in the first half of the eighteenth century than in the two earlier periods, though inversion regained and even increased its currency in the subsequent period, 1751–1800. Examples (10a, b) below illustrate the use of inversion in the construction:

The Historical Development of no sooner … than

236

Table 5. Frequencies and proportions of inversion in no sooner clauses (OED) –1600

1601– 1650

1651– 1700

1701– 1750

1751– 1800

1801– 1850

1851– 1900

1901–

Total

47

60

44

32

29

36

34

32

Frequency

2

18

11

5

11

15

21

19

Percentage

4.3

30

25

15.6

37.9

41.7

61.8

59.4

㻃

(10) a. No sooner cleered was the Coast, but that the bidden Guest Steales to her Chamber doore. (1602 Warner Alb. Eng. xii. lxxiv., OED) b. No sooner could a hint appear, But up he started to pickere. (1678 Butler Hud. iii. ii. 448, OED) Second, I focus on tense in these constructions. As shown in Figure 2, a steady decline in the past tense starting from the nineteenth century can be noted, while a significant rise in the frequency of the past perfect tense can be observed in the twentieth century. Thus, it follows that the use of these tenses from 1701 to 1750 is not in accordance with the general historical development. This development of tenses seems to have been involved in the competition between no sooner ... than and as soon as.

Figure 2. Proportions of tenses in the no sooner clause (OED)

Daisuke Suzuki

237

Data for Figure 2

present

14

1601– 1650 28

past present perfect past perfect Total

23

23

23

16

18

19

15

10

6

3

1

0

1

5

3

1

4

6

11

6

5

4

5

14

47

60

44

32

29

36

34

32

㻃

–1600

1651– 1700 9

1701– 1750 10

1751– 1800 5

1801– 1850 8

1851– 1900 11

1901– 7

Third, focus is placed on the possibility of the co-occurrence of no sooner with synonymous expressions, as seen in examples (11a–c): (11) a. A gracious soul no sooner hears where Christ is, but instantly makes out after him. (1703 Burkitt On N.T., Luke ii. 20, OED) b. …, where they were no sooner entered but they were immediately attacked by 7 or 8 other large Gallies, … (1720 London Gazette) c. … and no sooner one arises than the rest immediately follow. (1739–40 David Hume A Treatise of Human Nature) Table 6 demonstrates how frequently the construction no sooner … than co-occurs with synonymous expressions such as immediately, instantly, at once, and presently in the OED quotations. Table 6. Frequencies and proportions of the co-occurrence of no sooner with synonymous expressions (OED)

Total

47

1601– 1650 60

Frequency

3

7

3

6

0

2

0

1

Percentage

6.4

11.7

6.8

18.8

0

5.6

0

3.1

㻃

–1600

1651– 1700 44

1701– 1750 32

1751– 1800 29

1801– 1850 36

1851– 1900 34

1901– 32

238

The Historical Development of no sooner … than

Table 6 shows that the construction was used most frequently with synonymous expressions in the first half of the eighteenth century, a significant period in the development of this construction, as already mentioned. This period witnessed the vacillation between the two variants – but and than –, which also seems to have made the construction no sooner unstable during that period, in response to which the co-occurrence with a synonym helped to improve the meaning of the construction (‘immediately’).

4.3. Semantic Change In this subsection, I suggest that no sooner can be analyzed in terms of the process of grammaticalization. Figure 3 gives an overview of the occurrence of no sooner in clause-initial position on the basis of the OED dataset, with such positioning linguistically illustrated in examples (12a, b). As discussed in 4.1, the data in Figure 3 suggest that the decline observed in the period 1701–1750 also seems to have been connected to the competition between but and than in that period.

Figure 3. Proportions of no sooner occurring in clause-initial position (OED)

Daisuke Suzuki

239

Data for Figure 3

Total

47

1601– 1650 60

Freq.

5

22

㻌

–1600

1651– 1700 44

1701– 1750 32

1751– 1800 29

1801– 1850 36

1851– 1900 34

14

6

11

21

24

1901– 32 19

(12) a. No sooner he began To round and redden, than away he ran. (1807 Crabbe Par. Reg. iii. 553, OED) b. No sooner were the Tarquins dead than the nobles began to void their venom on the people. (1883 L. Villari Machiavelli ii. ii. III. 274, OED) At the same time, this period can be viewed as a crucial stage in the semantic change of no sooner. Brinton & Akimoto (1999: 229) made the following argument: There are some well-known cases in which take replaced nimen and give replaced sell [gislas sellan ~ gislas giefan ‘to give hostages’; tacan to wiue ~ niman to wife ‘to take as wife’]. In the process of replacement, there is usually a stage where two forms or several analogous forms co-exist. Although this co-existent period seems transient, the period is important and can be considered as a pre-stage to idiomatization.

In fact, Figure 3 shows that after the period of competition, no sooner, when positioned initially, continues to spread at a steady rate.6 In this position, no sooner serves as a marker of the topic expressing time such as when and before. Thus, the expression no sooner can be reanalyzed as a single chunk and has fulfilled the function of expressing time explicitly, which accounts for the status of no sooner in Present-Day English. In fact, no sooner can behave just like a conjunction in Present-Day English. Consider examples (13a, b) taken from the BNC: (13) a. ... no sooner you got your dinner and had a drink, you had to be back again ... (BNC: HDH) (=(2c))

6

The sudden decline in the twentieth century may be connected to the fact that the OED quotations in the twentieth century are much more restricted than those in the nineteenth century.

240

The Historical Development of no sooner … than

b. … as soon as they come, we had to open a trench and heel them … (BNC: HER) No sooner in example (13a) lost its correlation with than, and can be replaced with as soon as, as shown in example (13b), so the two variants can be structural alternatives. In other words, no sooner fulfils the same function as a conjunction expressing time. In summary – unlike no less/longer/more/other (…) than – no sooner has undergone a process of grammaticalization, gradually losing its status as an adverb to become a conjunction.

5. Conclusion This paper has investigated varying patterns of the correlatives of the construction no sooner … than from a historical viewpoint. It is well known that Present-Day English shows the use of than as a correlative of no sooner. It is, however, not the only form attested in the history of English. There is an interesting shift from but to than. After the competition between but and than, the use of than became firmly entrenched with no sooner. I have also shown other factors related to this construction such as inversion, tense, and the co-occurrences with synonymous expressions. With respect to the semantic change of the construction, this study has focused on the development of no sooner in clause-initial position. I have demonstrated that the semantic change was triggered by the co-existence of but and than in the first half of the 18th, and that this process of grammaticalization is also valid in analyzing the development of no sooner ... than in this study. Finally, it has to be stated that the use of the OED as a corpus enables us to describe the long-standing competition between no sooner and its correlatives. Therefore, it has been very helpful in treating the OED quotations database as a tool for linguistic research.

Acknowledgements This article is based on the paper presented at the 16th International Conference on English Historical Linguistics (ICEHL-16), held at the University of Pécs on 23–27 August 2010. I would like to express my sincere gratitude to Yoko Iyeiri, Kensei Sugayama, Shuichi Aita and Takuto Watanabe for their helpful suggestions on an earlier version. I am

Daisuke Suzuki

241

deeply indebted to Gábor GyĘri and Ruth Carroll for their invaluable comments on semantic changes. Thanks also go to the audience at the meeting and the anonymous referees for their useful comments and the necessary corrections of the first draft of this paper. This research was supported by the Japan Society for the Promotion of Science Grants-inAid for Scientific Research (No. 22·3983). All remaining errors and inadequacies are, of course, my own.

Works Cited Primary Sources BNC = The British National Corpus CED = A Corpus of English Dialogues 1560–1760 OED = The Oxford English Dictionary on CD-ROM (Version 4.0) The London Gazette

Secondary Sources Araki, Kazuo & Masatomo Ukaji. 1984. Eigoshi IIIA (History of the English Language IIIA). Tokyo: Taishukan. Brewer, Charlotte. 2000. “OED Sources”. Lexicography and the OED: Pioneers in the Untrodden Forest ed. by Lynda Mugglestone, 40-58. Oxford: Oxford University Press. Brinton, Laurel J. & Minoji Akimoto. 1999. Collocational and Idiomatic Aspects of Composite Predicates in the History of English. Amsterdam: John Benjamins. Fischer, Olga. 1992. “Syntax”. The Cambridge History of the English Language, II: 1066–1476 ed. by Norman F. Blake, 207-408. Cambridge: Cambridge University Press. Hoffmann, Sebastian. 2004. “Using the OED Quotations Database as a Corpus – A Linguistic Appraisal”. ICAME Journal 28.17-30. Huddleston, Rodney & Geoffrey K. Pullum. 2002. The Cambridge Grammar of the English Language. Cambridge: Cambridge University Press. Iyeiri, Yoko. 2010. Verbs of Implicit Negation and their Complements in the History of English. Amsterdam: Benjamins. Jespersen, Otto. 1917. Negation in English and Other Languages. Copenhagen: A. F. Høst & Søn.

242

The Historical Development of no sooner … than

Mair, Christian. 2001. “Early or Late Origin for begin + V-ing? Using the OED on CD-ROM to Settle a Dispute between Visser and Jespersen”. Anglia 119.606-610. —. 2004. “Corpus Linguistics and Grammaticalisation Theory – Statistics, Frequencies, and Beyond”. Corpus Approaches to Grammaticalization in English ed. by Hans Lindquit & Christian Mair, 121-150. Amsterdam: Benjamins. Mustanoja, Tauno F. 1960. A Middle English Syntax, I. Helsinki: Société Néophilologique. Poutsma, Hendrik. 1929. A Grammar of Late Modern English. (2nd ed.) Groningen: P. Noordhoff. Rissanen, Matti. 1999. “Syntax”. The Cambridge History of the English Language, vol. III: 1476–1776 ed. by Roger Lass, 187-331. Cambridge: Cambridge University Press. Suzuki, Daisuke. 2010. “No sooner ... than Kobun-no Keishiki-to Kino-ni tsuite (A Formal and Functional Analysis of the Construction No sooner ... than)”. Kansai Linguistic Society 30.37-48. Terasawa, Yoshio. 1997. The Kenkyusha Dictionary of English Etymology. Tokyo: Kenkyusha.

A TURN OF THE SCREW IN THE SEMANTICS OF PHRASAL VERBS: PHRASAL VERBS WITH UP AS A TEST CASE PAULA RODRÍGUEZ-PUENTE

1. Introduction The present study focuses on the description and analysis of the semantic component of phrasal verbs: lexicalized combinations of a verb and a post-verbal particle which function semantically and lexically as a single unit, but to varying degrees (e.g., look out, give up, fall down). Traditionally, phrasal verbs have been classified into one of three semantic groups depending on their degree of semantic transparency (cf. e.g., Hiltunen 1983: 147-148, Quirk et al. 1985: 1162-1163), namely literal, (cf. (1)), figurative (cf. (2)) and non-compositional (cf. turn off ‘make (sb.) bored’ in (3)). (1) I went out and told her what she wanted to know. (BNC CH2 W_newsp_tabloid) (2) And that was the reason that the government brought in a new initiative after the Warner Report. (BNC J3P S_meeting) (3) All that intense passion, the long hours in bed when you made love to me over and over, but never said anything – it just turned me off. (BNC H8F W_fict_prose) Recently, two new semantic categories have been introduced: aktionsart or aspectual phrasal verbs, such as write up in (4) and fade away in (5) (Dehé 2002: 6; Thim 2006: 221-225) and the so-called pleonastic (Thim 2006), redundant (Hampe 2002) or reiterative phrasal verbs, such as rise up in (6).

244

A Turn of the Screw in the Semantics of Phrasal Verbs

(4) We had to write up the parrot sketch in English. (BNC KPT S_conv) (5) The humming and buzzing sounds of summery woods faded away into a dreadful silence. (BNC ADA W_fict_prose) (6) He rose up on tiptoe, looking round for James, Allan, and the rest. (BNC A0N_W_fict_prose) However, a number of phrasal verbs do not seem to fit in any of these five semantic groups, some of them containing the particle up. A closer examination of these combinations with up will be the aim of this paper, in order to try to establish a more fine-grained categorization which can be inclusive of a wider range of structures. For this purpose, I have looked for examples in various historical and contemporary corpora, including ARCHER 3.1, BNC, BROWN, CLMETEV, COCA, FLOB, FROWN, HC and LOB, as well as occasional recourse to the internet. Corpus evidence is supported by data from the OED, as well as from a number of contemporary dictionaries such as CALD, COD, LDOCE and MWO. The article is structured as follows. In Section 2, I provide a description of the main semantic groups of phrasal verbs as found in the literature. In Section 3, the distinction between aspect and aktionsart is introduced, in order to refer to those phrasal-verb combinations which cannot be categorized within any of the five existing groups. Section 4 aims to establish a new semantic category of phrasal verbs. Finally, in Section 5, I deal with the origin for these structures.

2. Semantic Categories of Phrasal Verbs The five main semantic groups of phrasal verbs which are usually described in the literature are all attested in the corpora used in the present study, including literal, aspectual/aktionsart, reiterative, figurative and non-compositional. In literal phrasal verbs, “the meaning of a simple verb combines with the meaning of a simple local adverb homonymous with the particle” (Thim 2006: 222). These combinations are not accepted as ‘true’ phrasal verbs by all scholars, who prefer the term ‘free combination’ to designate them (Quirk et al. 1985: 1152-1154). However, it is undeniable that, although the members of these compounds possess lower cohesion than other combinations, they still show a certain degree of lexicalization. All in all, the verb and the particle do not appear together by chance, but rather

Paula Rodríguez-Puente

245

in a quite fixed manner, very often being the base from which other combinations more restricted syntactically and with more opaque meanings are derived. Literal verb-adverb combinations are attested in the HC as early as the period between the years 850 and 950. Some examples of the category are provided in (7) and (8). (7) Nu now

ic I

þuss gehroren eam & aweg gewiten. thus fallen am and away gone

‘Now I amWKXVIDOOHQDQGJRQHDZD\̓ HC O2: Alfred’s Orosius, 74) (8) [Æ]fter þam hie foron eft ut after that they went again out ‘After that they went out again’ (HC O2: Chronicle MS A Early, 101) In turn, the group of reiterative phrasal verbs includes combinations of a verb and a particle expressing and somehow repeating a semantic element already present in the verb. These combinations have also been referred to as ‘pleonastic’ (Thim 2006) and ‘redundant’ (Hampe 2002), but both labels imply the idea of an extra, unnecessary element, whereas, in fact, the two members of the compound fulfil important functions. For this reason, the term reiterative seems more appropriate, because it emphasizes the idea that there is some kind of repetition, but not unnecessary or redundant repetition. The first attested examples of reiterative phrasal verbs in the HC consist of the particle up combined with a verb indicating upward movement (cf. (9)), and date back to the period between the years 850 and 950. (9) Ða wæron up astigende. they were up rising ‘They were rising up.’ (HC O2: Bede’s Ecclesiastical History, 426)

1

Unless otherwise stated, OE and ME translations are my own.

246

A Turn of the Screw in the Semantics of Phrasal Verbs

The third semantic category of phrasal verbs includes combinations of a verb which maintains its basic or literal meaning plus a particle expressing some kind of aspect or aktionsart. This type of structure seems to have appeared later than literal and reiterative phrasal verbs. The first recorded examples in the HC are found in the period between 950 and 1050 (cf. (10) and (11) below), though according to Denison (1985: 42-44) and Brinton (1988: 224), Old English (OE) particles very often present intermingled aktionsart and directional meanings. (10) & siððan to twelf monðum ne cymð þær nan oðer scur, and afterwards for twelve months not comes there no other shower oð þæt seo ea eft up abrece. until that the river again up breaks ‘And afterwards for twelve months there comes no other shower until the river breaks out again.’ (HC O3: Ælfric’s De Temporibus Anni, 40) (11) [S]wa þæt hy asettan him upp on ænne sið. so that they set him up on a journey ‘So that they set him up on a journey.’ (HC O3: Chronicle MS A Late, 132) Some other phrasal verbs are often used figuratively to express meanings which are quite transparent but have lost their literal connotation. Figurative combinations with up appear only sporadically in the OE period. The first two examples attested in the HC date back to the period between the years 850 and 950 and are shown in (12) and (13) below. (12) [S]ume he bereafað hiora welan swiðe hraðe, þæs þe hi some he deprives their wealth very quickly that which they ærest gesælige weorðað, þy læs hi for longum gesælðum first happily obtain, lest they through extended blessings hi to up ahæbben. themselves too up lift

Paula Rodríguez-Puente

247

‘He [God] deprives some of their wealth very quickly, that which they first happily obtain, so that they do not exalt themselves too much.’ (HC O2: Alfred’s Cura Pastoralis, 133) (13) [Þ]a ongunnon Romane þa mæstan sace him betweonum then began Romans the greatest dispute them between up aræran. up raise ‘Then began the Romans to create the greatest dispute between them.’ (HC O2: Alfred’s Orosius, 232) Finally, there are a number of phrasal verbs that can be classified as non-compositional in the sense that their meaning cannot be deduced from that of their individual components. The first clear examples of noncompositional phrasal verbs with up in the HC are found in the Middle English (ME) period, more precisely between 1150 and 1250. Some of the earliest examples are shown in (14) to (16). (14) [...] for he besæt heom til hi aiauen up here castles. ‘because he besieged them till they gave up their castles.’ (HC M1: The Peterborough Chronicle, 59) (15) And then Seynt Thomas appierid to hym and bade hym and then Saint Thomas appeared to him and bade him holde vppe the right of Holy Churche with alle hys myght. hold up the right of Holy Church with all his might ‘And then Saint Thomas appeared to him and bade him take upon himself the right of the Holy Church with all his might.’ (HC M4: The Life of St. Edmund, 172) (16) Lord has hard my beed; lord my prayere has taken vp. lord has heard my prayer lord my prayer has taken up ‘The Lord has heard my prayer; the Lord has listened to my prayer.’ (HC M4: The Psalter or Psalms of David, 23)

248

A Turn of the Screw in the Semantics of Phrasal Verbs

The five semantic groups into which phrasal verbs have been traditionally classified do not constitute discrete categories and sometimes the same verb-particle combination can be classified into more than one category. Thus, for example, rise up, can be said to be literal in the sense that the meaning of the combination can be deduced from the meaning of the members of the compound; it can also be defined as reiterative because the particle is repeating part of the semantics of the verb, namely ‘upward movement’; and, finally, the combination can also be described as aspectual, because phrasal-verb particles very often combine aspectual and directional meanings (cf. Brinton 1988: 224).

3. The Aspect/Aktionsart Distinction: Classification Problems Notwithstanding the five semantic categories traditionally proposed for phrasal verbs, a high proportion of Present-Day English (PDE) combinations do not seem to belong to any of these. This is the case, for example, with ring up and the two synonymous combinations phone up and call up, as well as many other phrasal combinations in which there seems to be no difference between the simple verb and the phrasal verb (cf. e.g., boil up, hurry up, mix up, splash up, split up, wrap up) and also noun- and adjective-derived phrasal verbs, such as, for example, cake up, clean up, and warm up. Very often such structures have been classified within the aspect/aktionsart group because they consist of a verb which keeps its basic signification plus a particle adding a new, subtle nuance of meaning. However, the fact that the particle up expresses an aspectual distinction in these combinations seems rather questionable.

3.1 Some Facts about the Categories of Aspect and Aktionsart An important part of the problem of the semantic classification of phrasal verbs has to do with the definition of aspect and aktionsart. Aspectual meanings are often attributed to particles, but probably in a very broad manner. It is not my intention to provide an extensive discussion of the topic here, but a word seems necessary relating to the concepts of aspect and aktionsart. As noted by Brinton (1988: 2-4), the main problem in the distinction between aspect and aktionsart stems from the fact that the term ‘aspect’ has traditionally been used to designate both what is meant by aspect and by aktionsart, when in fact they are distinct features. Aspect, also called grammatical aspect or viewpoint aspect (Smith 1983, 1997), is related to

Paula Rodríguez-Puente

249

the way in which a speaker perceives a situation to be unfolding over time (Cappelle 2005: 343). Thus, as noted by Brinton (1988: 3), aspect is “a matter of the speaker’s viewpoint or perspective on a situation” which the speaker may choose to portray “as completed (perfective aspect), or as ongoing (imperfective aspect), or as beginning (ingressive aspect), continuing (continuative aspect), ending (egressive aspect), or repeating (iterative or habitual aspect)”. On the other hand, the German term aktionsart, also called lexical aspect, situational aspect or inherent aspect, refers to “lexically expressed aspectual distinctions” (Guerrero-Medina 2001: 1), that is, “to the characteristics of what is inherent in the lexical items which describe the situation” (Li & Shirai 2000: 3). Situations can be divided into three different dichotomies of aktionsart, namely stative or non-stative, punctual or durative and telic or atelic. Phrasal-verb particles have been variously labelled as markers of aspect (causative, perfective, intensive, terminative, resultative, effective, durative effective, conclusive or completive),2 whereas probably some of these labels should not be used to refer to aspect. Thus, if aspect refers to distinctions in the temporal structure of an event, how can ‘intensive’ be a kind of aspect? Other scholars, however, point out that phrasal-verb particles express telic aktionsart (cf. Brinton 1988: 163).3 According to this view, phrasalverb particles, especially up, express telic aktionsart, that is, they have the function of making situations telic or of converting activities into accomplishments. This can be seen by applying the so-called ‘in/for adverbial test’ (cf. Cappelle 2005: 99) to pairs such as drink – drink up and eat – eat up, combinations often used to exemplify the telicizing effect of up (cf. (17a-b) and (18a-b) below). According to this test, telic verbs can be followed by an adverbial introduced by the preposition in, but not by one introduced by for.

2

For the various terms employed in the literature to refer to the aspectual meanings of particles, cf. Brinton (1988: 167) and Brinton & Akimoto (1999: 6). 3 Brinton also points out that this is not the only function of particles. Others, such as on, along and away, may also have the ability to express iterative or continuative aspect. Nevertheless, the distinction between a particle expressing iterative or continuative aspect “depends on the aktionsart qualities of the verb: with punctual or telic situations, which cannot be continued, or with inherently iterative situations, the particles express iterative [aspect], whereas with durative situations, which can be continued, they mark the continuative” (Brinton 1988: 175).

250

A Turn of the Screw in the Semantics of Phrasal Verbs

(17) a. He drank the coffee [in / for] five minutes. b. He drank up the coffee [in / *for] five minutes. (18) a. He ate his hamburger [in / for] a while. b. He ate up his hamburger [in / *for] a while. However, as noted by Cappelle (2005: 420-421), in many cases the situation described by the verb to which up is added may already be telic without the particle, as shown in his examples, repeated below (cf. (19ab)). (19) a. I wrapped the present [in / *for] twenty minutes.4 b. I wrapped up the present [in / *for] twenty minutes. Moreover, Cappelle (2005: 422-423) also notices that many phrasal verbs typically classified as telic are fully compatible with for prepositional phrases, just like their simple counterparts, as shown in (20ab) below. (20) a. I ground the nuts [in / for] five seconds. b. I ground the nuts up [in / for] five seconds. In view of this, Cappelle comes to the conclusion that up does not really express telic aktionsart and, according to him, it is better described as a particle expressing resultative aspect “in the sense that it directs the reader’s attention, not to the event itself, but to whatever salient result the event produces” (2005: 423). Thus, for Cappelle (2005: 426), phrasal verbs with up can be paraphrased as ‘(cause to) get in a ‘V-ed’ state’, so that “if you heat up the soup, you cause the soup to get in a (literally) ‘heated’ state” (2005: 423). However, just as up cannot be described as a telic aktionsart marker because most simple verbs are already telic, the particle cannot be said to add resultative or causative aspect because most simple verbs are also resultative, as demonstrated in examples (21) and (22).

4

Professor David Denison (personal communication) points out that a prepositional phrase introduced by for would be admissible with the simple verb if describing a tedious process of wrapping a present.

Paula Rodríguez-Puente

251

(21) He heated the soup = ‘He caused the soup to be in a heated state’. (22) He bottled the wine = ‘He caused the wine to be in a bottled state’. As can be observed, there is no agreement as to whether particles express aspect (cf. e.g., Capelle 2005) or aktionsart (cf. Brinton 1988). But even when it comes to aspect, scholars do not seem to agree on exactly which aspectual distinction it is that they add to the verb: perfective, intensive, terminative, resultative, causative, effective, durative effective, conclusive or completive.

3.2 The Case of ring up and Similar Structures As noted above, at first sight ring up might seem to belong to the aspectual/aktionsart group. However, ring up does not fulfil the majority of telicity tests proposed by Brinton (1985: 162-163; 1988: 171-173) and Cappelle (2005: 420-421). In fact, obvious differences can be observed in the results of these tests as applied to ring up in comparison to eat up. Thus, an expression of the type to the end, completely, until it is finished, or all of it/them can substitute for the particle in eat up (Brinton 1988: 169), whereas that of ring up does not allow such a substitution (cf. (23a-b)). (23) a. The children ate up the candy = The children ate all of the candy. b. He rang me up  He rang [*all of] me [*completely/to the end] Moreover, whereas the combination eat up is compatible with the expression take an hour to V (Brinton 1985: 162; 1988: 169), an occurrence of this type with ring up is quite doubtful (cf. (24a-b)). (24) a. It took the children an hour to eat up the candy. b. ?It took him an hour to ring me up. Also, eat up can combine with an adverbial prepositional phrase introduced by in, though not with one introduced by for (Cappelle 2005: 99), whereas the ability of ring up to combine with both types of adverbial seems rather questionable (cf. (25a-b)). (25) a. The children ate up the candy in a minute. vs. *The children ate up the candy for a minute.

252

A Turn of the Screw in the Semantics of Phrasal Verbs

b. ?He rang me up in a minute / *He rang me up for a minute. With respect to the occurrence of the combinations with terminative periphrases containing finish (Brinton 1985:162; 1988: 171), eat up and ring up also show a different behaviour (cf. (26a-b)). (26) a. The children finished eating up the candy. b. *The woman finished ringing up [her husband]. Finally, both phrasal verbs are compatible with a temporal expression introduced by during, but not by all during (cf. (27a-b)). (27) a. The children ate up the candy [*all] during an hour. b. My mate rang him up [*all] during the day. (http://drownedinsound.com/community/boards/social/4263189)

As has been shown, ring up satisfactorily fulfils only one of the five main telicity tests, so that its status, at least in principle, seems rather different from that of eat up.

3.3. The Case of Noun- and Adjective-derived Phrasal Verbs Another group of combinations that also seem to belong to the aspectual/aktionsart group are noun- and adjective-derived phrasal verbs, whose particle Cappelle defines as expressing a causative/resultative type of aspect, a statement that he seems to take from Live (1965: 436) and Declerck (1977). One example of this category is the combination beer up (‘to drink beer; to get drunk, esp. on beer’; OED s.v. beer v.), whose first recorded occurrence in the OED is shown in (28). (28) They made an attack upon the bar, and after ‘beering up’, started for the diamond-field. (OED, 1884 N.Y. Times 26 May 8) This combination is probably derived from the simple verb beer, recorded in the OED as early as in the 18th century (cf. (29) below), which is derived through a process of conversion from the noun beer, itself present in the English language since OE times (cf. (30)).

Paula Rodríguez-Puente

253

(29) He surely had been brandying it, or beering. (OED, 1780-6 Wolcott (P. Pindar) Odes R. Acad. Wks. 1794 I. 105) (30) He ne drincð win ne béor. he not drank wine nor beer ‘He did not drink wine or beer.’ (OED, OE Ags. Gosp. Luke i. 15) Other combinations of phrasal verbs following a parallel path of evolution include, among others, sex (n.) > sex (v.) ‘have sexual intercourse’ > sex up (v.) ‘arouse sexually, have sexual intercourse’ (cf. OED s.vv. sex n. 4.b.; sex v. 2.b.) and fog (n.) > fog (v.) ‘cause to be covered or rendered opaque by condensed water vapour’ > fog up ‘cause to be covered or rendered opaque by condensed water vapour’ (cf. OED s.vv. fog n.2 II.2.a.; fog v.2 add.: [1.b]). A similar state of affairs can be described for certain adjective-derived combinations. Thus, even up ‘level’ (cf. (31)) is derived from the verb even (cf. (32)), which derives from the adjective even (cf. (33); cf. OED s.vv. even a.1.a; even v. I.1.a. and 4.b.). (31) They take what he [Anselm] says of justice as if He [Christ] were engaged to even up the score of penalty. (OED, 1865 Bushnell Vicar. Sacr. Introd. 16) (32) All þatt ohht iss wrang & crumb all that anything is twisted and crooked Shall effnedd beon & rihhtedd. shall levelled be and righted. ‘All the things that are twisted and crooked shall be levelled and righted.’ (OED, c1200 Ormin 9207) (33) Seo burg wæs Ҋetimbred an fildum lande & on swiþe emnum. the city was built on flat land and on very evenǤ ‘The city was built on very flat and even land.’ (OED, OE K. Ælfred Oros. I. ii. §4)

254

A Turn of the Screw in the Semantics of Phrasal Verbs

Other adjective-derived phrasal verbs following a similar line of development include warm (adj.) > warm (v.) ‘make warm’ (OED s.v. warm v. 3.a.) > warm up ‘make warm again’ (cf. OED s.v. warm v. 3.c.) and clean (adj.) > clean (v.) ‘free from dirt’ (OED s.v. clean v. 1.a.) > clean up ‘clean by taking up dust or dirt’ (OED s.v. clean v. 3). Some scholars (cf. e.g., Hampe 2002: 38) mention that in combinations of this type the particle functions in a way similar to suffixes in verbal derivation. This may be true if the phrasal verb were directly derived from the noun or the adjective, where the addition of a particle would function as the trigger for a change in word-class; but the creation of a two-word verb from a simple verb with a similar meaning (which seems to be the case with most noun- and adjective-derived phrasal verbs) without a consequent word-class change seems, as a phenomenon, to be different from derivation. Moreover, the assertion that up adds causative or resultative aspect to the simple verb (cf. 3.1. above) does not account for all phrasal verbs of this type, because they usually originate from a simple verb whose meaning is already causative or resultative. The list below (cf. (34)), drawn from Cappelle’s data (2005: 423n), has been modified with evidence from the OED and other contemporary dictionaries to show that the phrasal verbs already had simple equivalents when they first appeared.5 (34) cake (1607) > *cake up; clean (1681) > clean up (1831); clear (c1374) > clear up (1588); crank (1793) > crank up (1900); dirty (1591) > dirty up (1953); dry (OE) > dry up (c1385); foul (OE) > foul up (1922); jam (1719) > jam up (1794); ready (a1225) > ready up (1864); scent (1697) > *scent up; slop (1557) > *slop up; sober (1709) > sober up (1901); warm (OE) > warm up (1848) The only exceptions to Cappelle’s list6 are pretty up and tidy up. Pretty up has a simple equivalent, namely pretty, with exactly the same meaning, but the two-word combination is recorded approximately thirty years earlier in the OED database (cf. (35) and (36) below). 5 Combinations without a dictionary entry have been considered as recent creations and have been marked with an asterisk. 6 I have excluded the pair jack – jack up. The phrasal verb jack up does not mean ‘to cause to be in the shape/form of a jack’ or in Cappelle’s words ‘to cause to be in a jacked state’ and neither is there a simple verb jack with such a meaning. The verb-particle combination is defined in the OED as ‘to hoist with a jack’ (OED s.v. jack v.1), and thus I consider it to be different from the rest of the verbs in the list.

Paula Rodríguez-Puente

255

(35) It is not necessary to tell how she ‘prettied herself up’ in a worked mull muslin with a rich blue sash, and flowers in her hair. (OED, 1868 Godey’s Lady’s Bk. May 421/1) (36) The Yankees, too, have been at it – splashing, dashing, and in quaintly audacious ways prettying stuffs until you think Paris has had a hand at it. (OED, 1894 Denton (Maryland) Jrnl. 27 Oct. 2/3) Similarly the two-word verb tidy up is recorded earlier (cf. (37)) than the single-word equivalent (cf. (38)) in the OED. (37) I mean to have it whitened and tidied up this summer. (OED, 1821 Miss Mitford in L’Estrange Life (1870) II. 127) (38) Bessie employed me as a sort of under nursery maid, to tidy the room, dust the chairs, &c. (OED, 1847 C. Brontë J. Eyre iv) These data need cross-checking with evidence from historical corpora in order to establish a more precise dating for the appearance of these pairs of verbs, and also to ascertain whether it is true that the phrasal verbs appeared earlier than the simple verbs, which would be quite surprising given that the opposite normally occurs. However, such a task does not concern us here, so that for the time being I will consider pretty up and tidy up as the only exceptions to the rule.7

4. A New Semantic Category of Phrasal Verbs As has been shown, in certain phrasal verbs the particle up cannot be said to express aktionsart or aspect, so that the question of what its actual function is remains unanswered. My aim in this section, therefore, is to try to account for the semantic difference between pairs like beat – beat up, call – call up, phone – phone up, ring – ring up, sex – sex up, warm – warm up or wrap – wrap up, in order to establish a more fine-grained categorization.

7

Examples like tidy up and pretty up, which seem to be derived from the corresponding adjectives rather than from the simple verbs, support the view mentioned above that particles of this kind function as suffixes in derivation (cf. Hampe 2002: 38).

256

A Turn of the Screw in the Semantics of Phrasal Verbs

Some scholars have pointed out that up functions as an “intensive adverb” (cf. Potter 1965: 287-288) or that it expresses “intensity or totality” (Live 1965: 436), or even that its presence denotes a process that is “more complete” or “taken further” (Boers 1996: 136). However, I do not think it is appropriate to describe ‘intensive’ as a sub-type of aspect (cf. Section 3.1. above). I do believe, though, that the particle up has developed the ability to act as a reinforcement of the verb, that is, as a means of adding emphasis or intensity, and even certain connotations of informality, familiarity, colloquiality or empathy to the meaning of the simple verb. To illustrate the emphatic value of up, consider the pair beat – beat up. The definitions given in dictionaries for the two verbs differ very little, as can be seen in Table 1 below. Table 1: A comparison of the meaning of beat and beat up as defined by some English dictionaries beat OED MWO COD

‘strike with repeated blows’

‘knock about savagely, thrash’

‘strike repeatedly’ ‘strike with a series of violent blows’

MPVP CALD

beat up

‘to hit repeatedly’ ‘to hit someone or something many times with your hand, a LDOCE stick etc.’

‘inflict severe physical damage on someone by striking or kicking repeatedly’ ‘hurt someone by hitting or kicking them a lot of times’ ‘to hurt someone badly by hitting or kicking them again and again’ ‘to hurt someone badly by hitting them’

Both beat and beat up are used to indicate a continuous or repeated way of striking someone/something, the only difference being that in the OED, the CALD and the LDOCE the two-word verb is said to denote a more violent type of striking, given the use of the adverbs savagely and badly to describe the type of action implied by the verb. Despite the subtle difference in meaning between beat and beat up recorded in some dictionaries, evidence from the BNC reveals that both verbs are used in a very similar way, as can be seen in examples (39) to (42) below, in which they are employed indistinctively to refer to domestic

Paula Rodríguez-Puente

257

violence, both in written discourse (cf. (39) and (41)) and in the spoken register (cf. (40) and (42)). (39) I’ll beat you, just as my father beat me. And you’ll beat your own children. And they’ll beat theirs. (BNC AC4 W_fict_prose) (40) She’s rude right! Yeah. She’s spoilt, put it that way. They don’t beat her that’s why? I can imagine. They don’t beat her. (BNC KPE S_conv) (41) “He doesn’t beat you up or rape you, does he?” asked Tim. (BNC G1D W_fict_prose) (42) The people who were lived behind Ann there’s a yo young couple in there I think he beat her up. (BNC KB7 S_conv) Similarly, both verbs can be used without any type of distinction to refer to an act of bullying, as can be seen in (43) and (44) below. (43) But their caring didn’t extend to their maid who they regularly beat and bullied at their home in Lincoln. (BNC K6H S_brdcast_news) (44) Then they get their friends on you, they beat you up. (BNC A6V W_non_ac_soc_science) The in/for adverbial test (cf. Section 3.1) also works for both the simple and the two-word verb, so that the particle cannot be described as adding telicity to the verb. Similarly, both verbs can be paraphrased as ‘to cause to get in a beaten state’, so that both qualify as resultative according to Cappelle’s test. Nevertheless, when a higher degree of emphasis needs to be transmitted, the phrasal combination seems to be preferred over the simple verb. Thus, a reprimand sounds much stronger with the two-word verb than with its simple counterpart, as can be seen in the following (made-up) examples in (45a-b). (45) a. Watch your mouth or I’ll beat you up. b. Watch your mouth or I’ll beat you.

258

A Turn of the Screw in the Semantics of Phrasal Verbs

On the other hand, the phrasal-verb particle up is also capable of adding certain connotations to the meaning of the simple verb. Thus, combinations with emphatic up seem to have a more colloquial or even friendly, relaxed or familiar tone, as native speakers confirm when asked about the differences in meaning between phrasal verbs and their oneword counterparts (cf. Hampe 2002: 43-44). For this reason, expressions of the type I’ll ring you up are more frequently heard in family contexts or between friends than, for example, at the end of a job interview. The characteristic of being more typical of the colloquial language, however, is inherent to all phrasal verbs in general (cf. Biber et al. 1999: 408, 409), not only those containing emphatic up. What distinguishes constructions with emphatic up is that their particles seem to have developed the ability to function as a kind of ‘colloquializer’, as an element added to verbs which already exist as simple verbs, not only to provide them with some kind of emphasis, but also a familiar, colloquial tone. Phrasal verbs with emphatic up seem to be used by speakers as a way to empathize with the hearer. In fact, for most of these combinations there is a simple counterpart with practically the same meaning; however, by adding the particle spontaneously to the verb, it is as if the speaker is trying to get closer to the hearer, or to sound more relaxed or comfortable. This would explain why many of these emphatic combinations are not listed in dictionaries, because they are often spontaneous creations, perfectly understandable from the context, in which the only function of the particle is to act as both emphasizer and colloquializer. Consider, for example, the phrasal verbs cake up and scent up. Neither combination appears in the OED or in contemporary dictionaries and no evidence of their existence before 1991 is found in historical corpora or those contemporary corpora consulted. Both the BNC and the COCA, however, show that cake up and scent up are indeed used in PDE with meanings similar to their simple equivalents, namely ‘form itself into a cake’ (OED s.v. cake v. 2) and ‘impregnate with an odour’ (OED s.v. scent v. 4), as illustrated in (46) and (47) from COCA.8 (46) He wasn’t going to let mud cake up on the side of a fiftythousand-dollar car. (COCA FIC Bk: Rattled)

8 A total of eight examples of this kind were found: one in the BNC and seven in COCA.

Paula Rodríguez-Puente

259

(47) I work in an allergy department and the number of patients and co-workers who “scent up” before coming in is amazing. (COCA MAG OutdoorLife) The verbal element of the phrasal verb is generally a Germanic, monosyllabic or disyllabic verb with the accent on the first syllable (cf. Claridge 2000: 54; Thim 2006: 219). Therefore, the fact that compounds with polysyllabic Latin-based verbs, such as assemble up (48) and deliver up (49), occur sporadically in PDE seems to reinforce the idea that up can occasionally be added to simple verbs to provide them with a certain degree of familiarity and, arguably in such cases, to lend verbs a more native-like appearance. (48) For the next seventy-five minutes he will be the most important man on the floor––maintaining discipline and relaying all the instructions given him by the team assembled up in the gallery. (BNC F9Y W_misc) (49) He was ordered to deliver up the certificate to the liquidator. (BNC G3J W_fict_prose) The above arguments all seem to support the idea that a new semantic category of phrasal verbs is necessary. Such a category would include those combinations with a transparent verbal base to which the particle adds some kind of emphasis and colloquial tone. Given the nature of the particle of these combinations, I believe that emphatic phrasal verb might be an appropriate term for this category. As has been shown, then, it is possible to distinguish not five but six different semantic categories of phrasal verbs in contemporary English, namely literal, redundant, aspectual/aktionsart, figurative, noncompositional and emphatic.

5. Possible Origins of Emphatic up The first clear examples of emphatic up can be found as early as the ME period, most notably in combinations with Latin-based verbs (cf. (50) to (53) below). (50) [S]o þat þe kyng of Engelond shulde leve and resigne up to þe kyng of Fraunce (HC M3: Polychronicon Ranulphi Higden, Monachi Cestrensis, Vols. VI, VIII, PVIII, 347)

260

A Turn of the Screw in the Semantics of Phrasal Verbs

(51) [T]emper it vpe with an egge schelle ful of white wyne. ‘mix it up with an egg shell full of white wine.’ (HC M4: The ‘Liber de Diversis Medicinis’, 9) (52) [O]r sende ony persone þat is not sufficiant ne able to watche or withowte sufficiant wepon, Ҋe schall certyfye vp his name (HC M4: The Commonplace Book of Robert Reynes of Acle, 155) (53) And also, alle maner notabyl defautis that Ҋe fynde in Ҋoure craft, wel and trewly Ҋe schal presente them vp onto þe mair. (HC M4: The Commonplace Book of Robert Reynes of Acle, 319) The question, then, is why or how these two-word constructions with an apparently superfluous particle emerge, since one-word equivalents are already available in the language to convey the same meanings, as illustrated in (54) to (57) below. (54) He resignede þe castelles of Berwik and of Rokesburgh to þe kyng of Scotlond for ten þowsand pound. (HC M3: Polychronicon Ranulphi Higden, Monachi Cestrensis, Vols. VI, VIII, 347) (55) Take sulphur vyue. & whit tartre & blak. & grynde him al togeder & tempere it with strong eisel. ‘Take natural sulphur and white and black tartar and grind them all together and mix it with strong vinegar.’ (HC M3: A Late Middle English Treatise on Horses, 105) (56) [C]ertyfyyng Ҋow þat I haue spok wyth John Rwsse. (HC M4: Paston Letters and Papers of the Fifteenth Century, 199) (57) [A]nd presente this sayd book to his good & noble lordship. (HC M4: The Prologues and Epilogues of William Caxton, P70) Given that there were no phrasal verbs in Latin, combinations of the type in (50) to (53) above cannot be considered direct borrowings. It

Paula Rodríguez-Puente

261

could, therefore, be hypothesized that the particle up began to be added to certain simple verbs by analogy with related structures already available in the English language. A likely candidate as the source for analogy is found in reiterative combinations (cf. Section 2). As noted above, in such structures the particle literally repeats part of the meaning of the verb, which, in the case of combinations with up, is ‘upward movement’. It could therefore be argued that at some point up starts to be seen as an element that can be added to verbs conveying upward movement to emphasize and reinforce their meaning, so that its use extends to verbs which do not indicate vertical movement, as with ME tempren and resignen. Just as with reiterative constructions, where the particle up overtly and literally emphasizes the expression of vertical meaning conveyed by the verb, in emphatic combinations, in which the verb does not express vertical movement, up intensifies the verbal meaning and provides it with a more familiar and colloquial tone. However, the emergence of emphatic constructions is probably not down to one single factor but to the interaction of several factors working together. In addition to the important role of analogy, we might also note as crucial contributing factors the tendency of English to become a more and more analytic language (cf. e.g., Finegan 2009: 76) and the general tendency for OE prefixed verbs to be replaced by phrasal verbs (cf. Hiltunen 1983). This would explain the emergence of phrasal verbs where they exist in parallel with simple equivalents, as well as the disappearance of some of these simplexes in favour of two-word combinations. Unfortunately, the limited amount of evidence provided by the HC does not allow for definite conclusions in this respect.

6. Summary and Conclusions In this paper, I have argued in favour of assuming a sixth semantic group of phrasal verbs, namely emphatic phrasal verbs, consisting of combinations of a verb which keeps its basic meaning and the particle up. The particle of these combinations has two main functions, to add emphasis and intensity to the meaning of the simple verb, and to provide the verb with connotations of informality, familiarity, and a more colloquial tone; in some cases even a more native-like appearance is achieved. The addition of emphatic up to certain simple verbs has become a common feature of PDE, to the point that many phrasal verbs containing up are not even listed in contemporary dictionaries, probably because most of them are spontaneous creations whose meaning is perfectly understandable from the context in which they are uttered.

262

A Turn of the Screw in the Semantics of Phrasal Verbs

The first clear examples of emphatic up I have traced seem to date back to the ME period, when the particle up starts to be added to certain verbs, most notably Latin-based verbs. These emphatic combinations might have originated by analogy with reiterative constructions with up, which were already available in the English language from OE times, a process which may have been driven by the tendency of English to become a more and more analytic language. Nevertheless, given the scarcity of the data retrieved from the HC, my arguments here are more tentative than conclusive. Dealing only with those combinations formed with the particle up, this paper covers a tiny fraction of the huge number of existing phrasal verbs, and this study might usefully be extended to include other particles which probably fulfil a similar function in other combinations (cf. e.g., broaden out, level out, veg out). Further research, then, should aim at analyzing a wider range of combinations formed with other phrasal-verb particles which can be used to emphasize the meaning of certain verbs.

Acknowledgements I would like to thank Professors María José López-Couso and David Denison for helpful discussion and valuable feedback on earlier versions of this paper. For generous financial support I am grateful to the European Regional Development Fund and the following institutions: the Spanish Ministry for Science and Innovation (grants HUM2007-60706 and FFI2011-26693-C02-01) and the Autonomous Government of Galicia (grant CN2012/012).

Works Cited Primary Sources Corpora ARCHER = A Representative Corpus of Historical English Registers 3.1. 1990–1993/2002/2007/2010, compiled under the supervision of Douglas Biber & Edward Finegan at Northern Arizona University, University of Southern California, University of Freiburg, University of Heidelberg, University of Helsinki, Uppsala University, University of Michigan, University of Manchester, Lancaster University, University of Bamberg, University of Zurich, University of Trier, University of Salford, and University of Santiago de Compostela. BNC = British National Corpus; accessed online at http://corpus.byu.edu/bnc/

Paula Rodríguez-Puente

263

BROWN = A Standard Corpus of Present-Day Edited American English, compiled by W. Nelson Francis & Henry Kucera. Department of Linguistics, Brown University, 1964; revised 1971; revised and amplified 1979. CLMETEV = A Corpus of Late Modern English Texts Extended Version: 1710–1920, compiled by Hendrik De Smet (Katholieke Universiteit Leuven). COCA = A Corpus of Contemporary American English 1990–2010, compiled by Mark Davies, Brigham Young University. Available at: http://davies-linguistics.byu.edu/personal/ FLOB = Freiburg-LOB Corpus of British English, compiled by Marianne Hundt, Andrea Sand & Rainer Siemund, Englisches Seminar, AlbertLudwigs-Universität Freiburg. FROWN = Freiburg-Brown Corpus of American English, compiled by Marianne Hundt, Andrea Sand & Paul Skandera, Englisches Seminar, Albert-Ludwigs-Universität Freiburg, 1999. HC = Helsinki Corpus of English Texts 850–1710. 1991. Compiled by Matti Rissanen et al. Department of English, University of Helsinki. LOB = Lancaster-Oslo/Bergen Corpus, compiled by Geoffrey Leech, Roger Garside (University of Lancaster), Stig Johansson (University of Oslo) & Knut Hofland (University of Bergen). Dictionaries BT = Bosworth, Joseph & T. Northcote Toller. 1898. An Anglo-Saxon Dictionary. Oxford: Clarendon Press. http://beowulf.engl.uky.edu/~kiernan/BT/Bosworth-Toller.htm CALD = Cambridge Advanced Learner’s Dictionary. http://dictionary.cambridge.org/ COD = Collins Online English Dictionary. http://www.collinslanguage.com/ LDOCE = Longman English Dictionary Online. http://www.ldoceonline.com/ MPVP = Macmillan Phrasal Verbs Plus Dictionary. London: Macmillan, 2005. MWO = Merriam-Webster Online. http://www.merriam-webster.com/ OED = Oxford English Dictionary Online. http://www.oed.com/

264

A Turn of the Screw in the Semantics of Phrasal Verbs

Secondary Sources Biber, Douglas, Stig Johansson, Geoffrey Leech, Susan Conrad & Edward Finegan. 1999. Longman Grammar of Spoken and Written English. London: Longman. Boers, Frank. 1996. Spatial Prepositions and Metaphor. A Cognitive Semantic Journey along the UP-DOWN and the FRONT-BACK Dimensions. Tübingen: Gunter Narr Verlag. Brinton, Laurel J. 1985. “Verb Particles in English: Aspect or Aktionsart?” Studia Linguistica 39:2.157-168. —. 1988. The Development of English Aspectual Systems: Aspectualizers and Post-verbal Particles (Cambridge Studies in Linguistics, 49.). Cambridge: Cambridge University Press. Brinton, Laurel J. & Minoji Akimoto. 1999. “Introduction”. Collocational and Idiomatic Aspects of Composite Predicates in the History of English ed. by Laurel J. Brinton & Minoji Akimoto, 1-20. Amsterdam & Philadelphia: John Benjamins. Cappelle, Bert. 2005. Particle Patterns in English: A Comprehensive Coverage. PhD. Diss. Katholieke Universiteit Leuven. Faculteit Letteren, Departement Linguistik. Claridge, Claudia. 2000. Multi-word Verbs in Early Modern English: A Corpus-based Study. Amsterdam: Rodopi. Declerck, Renaat. 1977. “Some Arguments in Favour of a Generative Semantics Analysis of Sentences with an Adverbial Particle or a Prepositional Phrase of Goal”. Orbis 26.297-340. Dehé, Nicole. 2002. Particle Verbs in English. Syntax, Information Structure and Intonation (Linguistik Aktuell / Linguistics Today, 59.). Amsterdam & Philadelphia: John Benjamins. Denison, David. 1985. “The Origins of Completive up in English”. Neuphilologische Mitteilungen 86.37-61. Finegan, Edward. 2009. “English”. The World’s Major Languages ed. by Bernard Comrie, 59-85. London: Routledge. Guerrero-Medina, Pilar. 2001. “Reconsidering Aspectuality: Interrelations between Grammatical and Lexical Aspect”. Working Papers in Functional Grammar 75.1-11. Hampe, Beate. 2002. Superlative Verbs: A Corpus-based Study of Semantic Redundancy in English Verb-Particle Constructions. Tübingen: Narr. Hiltunen, Risto. 1983. The Decline of the Prefixes and the Beginnings of the English Phrasal Verb (Annales Universitatis Turkuensis, 160.). Turku: Turun Yliopisto.

Paula Rodríguez-Puente

265

Li, Ping & Yasuro Shirai. 2000. The Acquisition of Lexical and Grammatical Aspect. Berlin & New York: Mouton de Gruyter. Live, Anna H. 1965. “The Discontinuous Verb in English”. Word 21:3.428-451. McIntyre, Andrew. 2002. “Idiosyncrasy in Particles”. Verb-Particle Explorations ed. by Nikole Dehé, Ray Jakendoff, Andrew McIntyre & Silke Urban, 95-118. Berlin & New York: Mouton de Guyter. Potter, Simeon. 1965. “English Phrasal Verbs”. Philologica Pragensia 8.285-289. Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech & Jan Svartvik. 1985. A Comprehensive Grammar of the English Language. London & New York: Longman. Smith, Carlota S. 1983. “A Theory of Aspectual Choice”. Language 59.479-501. —. 1997. The Parameter of Aspect. (2nd ed.) Dordrecht: Kluwer. Thim, Stefan. 2006. “Phrasal Verbs in Late Middle and Early Modern English: Combinations with back, down, forth, out and up”. Syntax, Style and Grammatical Norms: English from 1500–2000 ed. by Christiane Dalton-Puffer, Dieter Kastovsky, Nikolaus Ritt & Herbert Schendl, 213-228. Bern: Peter Lang.

PART IV FUNCTIONAL AND REGIONAL VARIATION IN DISCOURSE AND VOCABULARY

DISCOURSE STRATEGIES IN MARGERY KEMPE’S DESCRIPTIONS OF HER PILGRIMAGES FUMIKO YOSHIKAWA

1. Introduction This paper analyzes The Book of Margery Kempe1 (hereafter Margery Kempe, and BMK in quotations) and two travel accounts written in Middle English from a diachronic pragmatic viewpoint and discusses three discourse strategies that Virtanen (1995) investigates in several Early Modern English travelogues. These are temporal, locative, and participant/topic-oriented discourse strategies. They are examined by observing sentence-initial or clause-initial adverbials 2 of time and place and references to the author or to participants in the discourse in the texts that we survey. The term ‘discourse strategy’ is employed in the sense used by Virtanen (1995: 499) used for her examination of Early Modern English travelogues, and is defined as “a goal-oriented weighting of decision criteria” based on Enkvist (1987: 24). Temporal discourse strategy is strategy showing “coherent temporal sequentiality” (Virtanen 1995: 500, 501), locative discourse strategy is strategy showing a temporal or spatial chain in discourse created by a chain of locative elements, and participant/topic-oriented discourse strategy is strategy showing a coherent sequentiality implemented by referring to “characters participating in the chain of actions and events presented in the discourse” (Virtanen 1995: 505) or to the topic often introduced with an existential construction in non-narrative discourse (cf. Virtanen 1995: 500-507). According to Virtanen (2004: 80-82), the sentence-initial position is special in some ways. The slot is usually filled with given or inferable information and the elements in this slot let us know what the sentence is about, but occasionally, they bear “crucial information” (Enkvist 1989: 1

All quotations from Margery Kempe in this paper are taken from Windeatt (2000). According to previous works related to this study (Virtanen 1992, Quirk et al. 1985), even if an adverbial follows a conjunction or another adverbial, it is still considered as clause-initial in this paper.

2

Fumiko Yoshikawa

269

176-178; Virtanen 2004: 81). That is, the text type consequently depends on what is chosen for this slot. It is possible for some types of adverbials to occupy the sentence-initial or clause-initial position, 3 and adverbials of time and place are examples of this kind. The ease with which an adverbial can be moved to this initial position depends on to what extent the adverbial is obligatory with the verb to which it is tied and on how syntactically deeply it occurs (see Virtanen 1992: 19ff). Obligatory adverbials are those that are difficult to move to the initial position of the clause and the syntactically deep adverbials need stronger textual justification to do so. Adverbials of time and place can be both obligatory and optional. When adverbials of time and place are optional, they are more easily topicalized (see Virtanen 1992: 21) and show the setting of the contents of the clause or sentence that they start, or the setting of the contents of the textual unit (see Virtanen 1992: 270). Adverbials topicalized to their sentence-initial position tend to serve as markers of text strategy and of text segmentation (see Virtanen 1992: 16). Based on their function as the markers of text strategy, the examination of sentenceinitial adverbials of time and place is to be utilized for text-type or genre categorization.4 The purpose of this study is to discover how similar Margery Kempe is to travel accounts of that time. Margery Kempe is often classified as mystical writing, but the contents resemble an autobiography and her stories contain many descriptions of her travels to holy places and her pilgrimages. Yoshikawa (2008) examined the three above-mentioned discourse strategies in Margery Kempe and in Julian of Norwich’s Revelations of Divine Love, and showed certain resemblances and differences between the two Middle English women’s mystical works. In general, both texts show the narrative characteristics of coherence being mainly achieved by temporal discourse strategy and participant-oriented strategy, but a difference was recognized in the use of locative discourse strategy. On rare occasions, locative strategy can be recognized in Margery Kempe, and this seems to derive from the travelogic contents. They are not found in Revelations. Lacking sufficient comparison with other travel accounts, Yoshikawa (2008) was unable to capture a total picture of the discourse strategies employed in Margery Kempe. For the purpose of gaining a holistic view of the locative strategies in Margery 3

With regard to the taxonomy of adverbials, refer to Quirk et al. (1985: 479ff on the semantic roles with which they are associated; 489ff on where they are realized in linguistic structures; 501ff on the grammatical functions they fulfill). 4 For example, Virtanen (2004: 94) states that sentence-initial adverbials of time work as text-type or genre markers.

270

Margery Kempe’s Descriptions of Her Pilgrimages

Kempe, this paper examines the sentence-initial adverbial phrases of a well-known Late Middle English travel account, Mandeville’s Travels (hereafter Mandeville), and, in addition, those of The Stacions of Rome. Using Mandeville and The Stacions of Rome for this purpose is somewhat problematic. First, with regard to Mandeville, there are several versions in over forty manuscripts, including a metrical version. It is conceivable that the choice of text could make a difference to any results. We explore an early fifteenth-century prose version of the text, using the most easily accessible edition by Hamelius from MS. Cotton Titus C. XVI in the British Library. It is a translation from the French, and simply comparing it with vernacular English prose might therefore be problematic. It also seems to be a patchwork of some encyclopedic works, and it is doubtful that the narrator of the story, John Mandeville, was a real-life figure or that the stories were written based on actual travels. Therefore, problems may arise when comparing it with Margery Kempe, which is based on her real travel experiences. It is also doubtful whether the other text, The Stacions of Rome, can be called a travel account because it is a pilgrims’ guide. However, it is worth examining it because Moseley (2005: 22) points out in the introduction to his translation of Mandeville that, along with Information for Pilgrims Unto the Holy Land, it is strikingly similar to Mandeville in methodology and style. There are also two versions of The Stacions of Rome, one in verse and one in prose. As the metrical version (from the Vernon MS., ab. 1370 A.D.) is written in rhyming verse, it is feasible that the placement of adverbial phrases would have been affected on account of the writer’s need to satisfy metrical requirements. However, as the prose version (from the Porkington MS. No. 10, ab. 1460–1470 A.D.) is much shorter than the verse version, this paper addresses the text in verse as well. 5 As described above, we understand that there are problems associated with comparing these texts, but still, an analysis of their sentence-initial adverbials is attempted. According to Virtanen (2004: 82), sentence-initial adverbials serve both to create coherence and to mark text segmentation. These functions are required regardless of whether the text is verse or prose. Virtanen (2004: 83) also shows in a modern travel guide that the locative strategy connotes the temporal strategy; that is, the reader would follow the route according to the chain of sentence-initial adverbials of place. As the effectiveness of examining a modern travel guide is clear, such examination seems also to be applicable to the Middle English travel guide The Stacions of Rome. 5

Quotations from the two versions of The Stacions of Rome in this paper are from Furnivall (1867).

Fumiko Yoshikawa

271

Using the markers of discourse strategy as a signal of text segmentation would be also effective in drawing the reader’s attention to the story both in fictional and non-fictional texts. Therefore, it is presumed that such sentence-initial adverbials will perform the same functions in fictional texts such as Mandeville as well.

2. Discourse Strategies in Mandeville and The Stacions of Rome In this section, first, some passages describing Jerusalem and its neighboring cities from the first part of Mandeville will be examined.6 The first few chapters make scant reference to the journey to the Holy Land; rather, they are concerned with the significance of writing travel stories, where sacred relics such as the cross and the crown can be found, and the faith in Greece. Consequently, the chain of the sentence-initial adverbials in these chapters does not show the same tendency as the other chapters which deal with the route to the Holy Land. The sixteenth chapter, which is the last chapter of the first part, is also quite different from the preceding chapters. Therefore, some typical passages describing the journey to the Holy Land and the holy sites in the Holy Land and its neighboring cities are examined here. (1) And .viij. myle from Tyre toward the Est vpon the see is the cytee of Sarphen in sarept of sydonyens And þere was wont for to dwelle helye the prophete & þere reysed he Jonas the wydwes sone from deth to lyf. And .v. myle fro sarphen is the cytee of Sydon of the whiche citee Dydo was lady þat was Eneas wif after the destruction of Troye & þat founded the cytee of Cartage in Affrik & now is cleped Sydon Sayete. And in the cytee of Tyre regned Agenore the fader of Dydo. And .xvj. myles from Sydon is Beruth And fro Beruth to Sardenare is .iij. iorneyes And fro Sardenar is .v. myle to Damask. (Mandeville, Ch. V, p. 18, ll. 21-32)7

Quotation (1) is a passage from Chapter 5, which describes the route from Cyprus to Jerusalem. In this passage, cohesion is achieved by using locative adverbials in sentence-initial positions. The first locative adverbial phrase in this quotation is .viij. myle from Tyre toward the Est 6

All the quotations from Mandeville have been taken from Hamelius’s edition of MS. Cotton Titus C. XVI in the British Library. 7 In quotations, sentence-initial adverbial phrases are italicized by the author.

272

Margery Kempe’s Descriptions of Her Pilgrimages

vpon the see, which marks the distance from Tyre to Sarphen. The City of Sarphen is the place to which the writer is now leading the reader. In the following clauses, by using the locative adverb þere, which succeeds the locative topic introduced in the previous sentence, the writer connects the place as the topic to the history of the Bible. That is, the topic-oriented strategy is also functioning in this passage. The next sentence-initial adverbial phrase (v. myle fro sarphen) is an adoption of the same strategy by using an adverbial phrase expressing distance from the previously mentioned place to the next stop, as we have seen in the first sentence of this quotation. Information on distances between major cities was important for travellers, and they seem often to be mentioned in pilgrims’ travel guides written in the Middle English period. For example, a guide from the fifteenth-century Information for Pilgrims Unto the Holy Land has a list of distances between major cities added to the main text (cf. Duff 1893) and The Stacions of Rome, which we will examine below, sometimes refers to distances in miles as well. Introductory essays to Mandeville (see e.g., Moseley 2005: 19) cite Odoric of Pordenone’s travels as the main source of stories about journeying to the Holy Land, as well as some pilgrims’ guides including Liber de Terra Sancta. It follows that the text of Mandeville contains many similarities to pilgrims’ guides in structure and expression. Quotation (2) is a passage from Chapter 10, which describes the church in Bethlehem, and quotation (3) is from Chapter 11, which is about Jerusalem. (2) Also besyde the queer of the chirche at the right syde as men comen dounward .xvj. greces is the place where oure lord was born, þat is full wel dyght of marble & full richely peynted with gold syluer azure & oþer coloures. And .iij. paas besyde is the crybbe of the ox & the asse. And besyde þat is the place where the sterre fell þat ladde the .iij. kynges Iaspar Melchior & Balthazar, But men of Grece clepen hem þus: Galgalath Malgalath & Saraphie, And the Iewes clepen [hem] in this manere in Ebrew: APPELIUS AMERRIUS & DAMASUS. [...] Also vnder the cloystre of the chirche be .xviij. degrees at the right syde is the charnell of the Innocentes where here bones ly‫܌‬n And before the place where oure lord was born is the tombe of seynt Ierome þat was a preest & a Cardynall þat translatede the bible & the psaultere from Ebrew in to latyn And withoute the mynstre is the chayere þat he satt in whan he translated it. And faste besyde þat chirche a .lx. fedme is a chirche of seynt Nicholas where oure lady rested hire after sche was lyghted of oure lord And for as meche as sche had to meche mylk in hire pappes þat greued hire,

Fumiko Yoshikawa

273

sche mylked hem on the rede stones of marble so þat the traces may ‫܌‬it ben sene in the stones all w[h]yte. (Mandeville, Ch. X, p. 46, l. 8-p. 47, l. 1)

(3) And ‫܌‬ee schull vndirstonde þat whan men comen to Ierusalem here first pilgrymage is to the chirche of the holy Sepulcre where oure lord was buryed þat is withoute the cytee on the north syde But it is now enclosed in with the toun wall. And þere is a full fair chirche all Rownd & open aboue & couered with leed And on the west syde is a fair tour & an high for belles strongly made And in the myddes of the chirche is a tabernacle as it were a lytyll hows made with a low lityll dore And þat tabernacle is made in manere of half a compas right curiousely & richely made of gold & azure & oþere riche coloures full nobelyche made And in the right syde of þat tabernacle is the sepulcre of oure lord And the tabernacle is .viij. fote long & .v. fote wyde & .xj. fote in heghte. And it is not longe sithe the sepulcre was all open þat men myghten kisse it & touche it. But for pilgrymes þat comen thider peyned hem to breke the ston in peces or in poudre þerfore the Soudan hath do make a wall aboute the sepulcre þat noman may towche it. But in the left syde of the wall of the tabernacle is wel the heighte of a man is a gret ston to the quantytee of a mannes hed þat was of the holy sepulcre & þat ston kissen the pilgrymes þat comen þider. In þat tabernacle ben no wyndowes but it is all made light with lampes þat hangen before the sepulcre And þere is a lampe þat hongeth before the Sepulcre þat benneth light & on the gode Fryday it goth out be himself [and on þe Pasch day it lightez agayne by it self] at þat hour þat oure lord roos fro deth to lyue. Also with in the chirche at the right syde besyde the queer of the chirche is the mount of Caluarye where oure lord was don on the cros And it is a roche of white colour and lytill medled with red And the cross was set in a morteys in the same roche & on þat roche dropped the woundes of oure lord whan he was pyned on the cross & þat is cleped Galgatha, And men gon vp to þat Golgatha be degrees. And in the place of þat morteys was Adames hed founden after Noes flode in tokene þat the synnes of Adam scholde ben bought in þat same place And vpon þat roche made Abraham sacrifise to oure lord. And þere is an awtere And before þat awtier ly‫܌‬n Godefray de Boleyne & Bawdewyn & oþere cristene kynges of Ierusalem. And þere nygh where oure lord was crucyfied is this writen in Greew: OTHEOS BASILION YSMON PROSIONAS ERGASA SOTHIAS EMESOTIS GYS. (Mandeville, Ch. XI, p. 49, l. 16-p. 50, l. 22)

In these passages, sentence-initial locative adverbials describe the exact location of the next point in direction and distance from the previous stop. The previous stop stays in the reader’s mind as the topic until the

274

Margery Kempe’s Descriptions of Her Pilgrimages

new location is introduced in an existential sentence. Locative discourse strategy is adopted in the same way as in quotation (1), but the descriptions of Bethlehem and of the Holy Land are in greater detail because the Holy Land is the pilgrims’ destination and Bethlehem is also a major city near Palestine. In quotation (2), the chain of locative sentence-initial adverbials starts with the clause besyde the queer of the chirche at the right syde as men comen dounward .xvj. greces. This clause introduces Christ’s birth place, the previous location links to his birth place, and the next location is then linked to the place by the subsequent adverbial phrase .iij. paas besyde. The crib of the ox and the ass is described in this sentence. Furthermore, the next adverbial phrase, besyde þat, leads the reader’s eyes to the place of the falling of the star that led the three kings to the Christ Child. The reader mentally follows the route, guided by these locative adverbials. After the clause besyde þat is the place where the sterre fell þat ladde the .iij. kynges Iaspar Melchior & Balthazar, there is a gap in the chain of locative adverbials. In this gap, details of the story of the three kings are given. The chain restarts with the phrase vnder the cloystre of the chirche be .xviij. degrees at the right side, and through this phrase, we are led to the charnel of the Innocents. Quotation (3) is a description of the Church of the Holy Sepulchre in Jerusalem. After suggesting that pilgrims come to the sepulchre first, a fair circular church is introduced to the reader using the existential thereconstruction. Virtanen (1995: 504) states in her research into Early Modern English travelogues that existential constructions tend to appear once a stop is introduced. The same tendency is recognized in Mandeville. Taking the circular church as the starting point, we, the readers, mentally turn in the direction guided by the sentence-initial adverbials on the west syde (of the church) and in the myddes of the chirche, with which a tabernacle is introduced as the next point. Then, following the locative adverbial phrase in the right syde of þat tabernacle, the Sepulchre of our Lord is introduced and a description of the tabernacle follows along with an account of the Sepulchre of Christ. The chain of locative adverbials is further connected to the adverbial phrases in the left syde of the wall of the tabernacle, with which a great stone of the Holy Sepulchre is introduced, and in þat tabernacle, following which the lamps hanging in front of the sepulchre are described; one of them, which is said to be miraculous, is introduced into the text by the use of an existential construction. After the description of the miraculous lamp, the writer provides us with a view of the church from a wider angle once again using the phrase with in the chirche at the right syde besyde the queer of the chirche. Immediately

Fumiko Yoshikawa

275

afterwards, Mount Calvary is introduced as a new topic, and subsequently, the rock on which the Cross was placed is described. The altar at the rock is also introduced into the text by the use of an existential thereconstruction. As is pointed out in Virtanen (2004: 82), we can see in these quotations that the sentence-initial adverbials of time and place create coherence and, at the same time, mark text segmentation. In the main structure of the text, we are instructed to follow the route shown by the locative adverbials, and, once a location becomes active in our mind, things to be found in this location, or Bible stories associated with this location, are described. The same writing style involving locative discourse strategy can be seen in present-day travel guides. Virtanen (2004: 83) quotes a passage from a present-day guidebook to the archeological site of Ostia Antica in her quotation (2). It is duplicated in the following as quotation (4). (4) Beyond the Taverns of the Fishmongers along the Decumanus Maximus, two insulae are set side by side; the Insula of Dionysos and Insula of the Eagle. Both of them built in 125–130 A.D., Hadrian’s epoch, they were restored and partly modified around the middle of the III century and this late remodelling considerably changed the face of the old insulae. Inside, several mosaics and frescoes show the considerable refinement of their decoration. In the room at the rear of the Insula of the Eagle is preserved the Mithraeum of the Seven Doors which boasts a fresco representing a lovely garden. [...] (Ostia: Guide to the Excavations, Storti 1985, p. 39 [Virtanen 2004: 83])

This passage from the Ostia guidebook displays a similar text structure marked by locative adverbials which introduce the excavation sites one by one. Once a site is activated in the reader’s mind, the materials or buildings at this site and their historical data or significance are described. Similar to the way in which the distances between cities are denoted in quotation (1), in quotations (2) and (3) spatial distances between objects in the churches are described in fine detail. In quotation (2), the distances between objects in the church are denoted by steps as may be seen in as men comen doun ward .xvj. greces, .iij. paas besyde, and vnder the cloystre of the chirche be .xviij. degrees at the right syde. We also see the scale of feet in quotation (3). Introduced by these locative adverbials, the materials with which the holy places are decorated or the colors of the relics are precisely illustrated in these passages. For example, in quotation (2), the cave of the nativity is described as being made of marble and very richly painted with gold, silver, azure, and other colors. Also in quotation

276

Margery Kempe’s Descriptions of Her Pilgrimages

(3), the colors of the tabernacle and those of the rock on which the Cross was placed are described. The same tendency is seen in The Stacions of Rome, which is a short guide for visitors to the shrines in Rome. As already mentioned, there are two versions of this work in verse and prose respectively. According to Furnivall’s edition (1867), the metrical version consists of 734 lines, and the prose version is much shorter fitting into a few pages. They are very different from a narrative travel account because the main purpose of this text is to show the number of years of pardon a pilgrim gains by visiting each sacred site in Rome, on a particular day or in a certain period, and in a particular way. However, the similarities to Mandeville in the descriptions of relics and places in Rome are notable. Locative adverbials in sentence-initial position function in the same way as they do in Mandeville. (5) Frow sent paullis to sente austens is ij myle of feyre waye : þer is

euery day viij Ml ‫܌‬ere of pardoun, & þe þred part of paynance vndo, granttyd by pope vrban ; & sylvester grant for-geyfnis of wratheþinge of fadore & modore, so he layde no vyolent honde on hem ¶Be-fore þe dore ys þe ston þat sent paullis hede lay on ; & þer be iij wellis of gret vertu ¶And þer ys A chappelle þat men calle schalla cely, þat ys of oure lady, & fele holly bonnys byn vndur þe autur, x Ml merturis in þe tyme of tybure-rya þe emparoure. he þat saythe a mase þer witt good devossyon may brynge a soule out of pulcatorry to heyvyn, & gretly helpe his frende þat is alyue . & iij Ml ‫܌‬ere of pardon ys granttyde by popys xlvij þat liue at sent sebestyande. Conformyde be vrbane, seluester, bennet, leon, & clement (Stacions in prose, p. 31, l. 23-p. 32, l. 2)8

(6) On our lady day þe somsyon ys a Ml [‫]܌‬ere of pardon ¶ On sent pettur

and paullis day ys ij Ml ‫܌‬ere of pardon ¶ On sent marttayn þe vij day was þat place hallowyd. Then ys xxviij Ml ‫܌‬ere of pardon, & so mony lenttis, & þe þrede part & of pennance vndo ¶ When they schowe þe warnakoll, ys iiij Ml ‫܌‬ere of pardon ; to pepule of oþer placys ix Ml; & ‫܌‬efe he pase þe see xiiij Ml , & þe þrede part of synnys fore-geyve (Stacions in prose, p. 31, ll. 3-11)

Quotations (5) and (6) are taken from the prose version. In quotation (5), text segmentation and coherence are mainly implemented by sentenceinitial locative adverbials and the distance between two stations is also sometimes denoted in miles. As has been observed in Mandeville, an 8

All the quotations from Stacions have been taken from Furnivall (1867).

Fumiko Yoshikawa

277

existential there-construction is used when a new stop is introduced in the reader’s mind as the topic. To explain the extra pardon a visitor can gain on a special day or in a special period, temporal discourse markers are sometimes used. These show, in sentence-initial position, the saints’ dates on the calendar, as can be seen in quotation (6). This function of temporal discourse markers is less represented in Mandeville, though these texts show many similarities in the use of locative discourse markers. Before conducting this examination, I was sceptical about the effectiveness of analyzing sentence-initial adverbials in the metrical version, but, in fact, the metrical version also shows the same tendency as Mandeville in the use of locative discourse adverbials. (7) 96

100

112

116

136

148

HEr may we . not longe be To seint Anastace . moste we. Two myle . is holde be-twene. Of feir wey . and of grene. Vche day . ‫܌‬if þou wolt craue. Seuen þousent ‫܌‬er . þer may þou haue Pope Vrban . þat holy syre. So rewardede . men heore huyre Men þat ben schriuen . and verrey contrit. Of alle heore synnes . god makeþ heom quit. [...] Bi-fore þe dore . stont a ston. Seynt poules hed . was leyd þeron. A traitur . smot of his heued. Wiþ a swerd . þer hit was leued. Þer aftur spronge welles þre. Hose is þere . wel may he se. Of water . boþe feir and gode Men . and Wimmen . han had heore bote IN þat place . a Chapel is. Scala celi . cleped hit is. [...] Nou passe we forþ . in vre gate To seinte Marie . þe Nunciate Two Mile is bitwene .I. vnder-stonde. [...] To Fabian and Bastian . passe we Þider we haue . Myles þre (Stacions in verse, ll. 93-148)

Quotation (7) is the equivalent part in the metrical version to quotation (5). The sentence-initial or clause-initial locative adverbials operate

278

Margery Kempe’s Descriptions of Her Pilgrimages

similarly to those in the prose version and to those in Mandeville. The verse version is longer than the prose version, with more detailed accounts of places often added.

3. Discourse Strategies in The Book of Margery Kempe In this section, taking into account the characteristics of the use of locative discourse markers in Mandeville and The Stacions of Rome, we will mainly examine the sentence-initial adverbial phrases in Margery Kempe. Yoshikawa (2008), examining temporal, locative, and participant/topic-oriented strategies in Part One of Margery Kempe, reached the following conclusion: It is a narrative mainly textualized by using temporal discourse strategies, and the most frequently used temporal discourse marker is than (then in Present-Day English). However, locative discourse strategy is also used at certain points. As Virtanen (2004: 83) states, a chain of adverbials of place in sentence-initial position is “inherently temporal in the sense that the reader is instructed to follow the route”. Quotation (8) is an example in which locative discourse markers in sentence-initial (and clause-initial) positions are incorporated into a temporal sequence. (8) Than sche cam down into the halle, and ther stood the Frere Prechowr that had cawsyd hir al that wo. And so sche passyd forth wyth a man of the Erchebischop, beryng the lettyr whech the Erchebischop had grawntyd hir for a recorde, and he browt hir to the Watyr of Humbyr, and ther he toke hys leve of hir, returnyng to hys lord and beryng the sayd lettyr wyth hym ayen – so was sche left alone wythowtyn knowlache of the pepyl. (BMK, Ch. 54, ll. 4529-4535)

This kind of usage of locative adverbials is not seen in Julian of Norwich’s Revelations of Divine Love because it does not include the description of a pilgrimage. The use of locative discourse strategy is also rare in Margery Kempe, but it is notable that locative strategy is conspicuously used in the descriptions of Bible-related sites in Jerusalem, where Margery cries and weeps bitterly. (9) And so ovyr al wher-that-evyr the frerys led hem in that holy place, sche alwey wept and sobbyd wondyrfully, and specialy whan sche cam ther owyr Lord was nayled on the cros. Ther cryed sche and wept wythowtyn mesur, that sche myth not restreyn hirself. Also thei comys to a ston of marbyl that owyr Lord was leyd on whan he was

Fumiko Yoshikawa

279

takyn down of the cros, and ther sche wept wyth gret compassyon, havyng mend of owyr Lordys Passyon. (BMK, Ch. 29, ll. 2317-2323)9

(10) And therfor this creatur had gret desyr to be howselyd in that holy place wher owyr mercyful Lord Crist Jhesu fyrst sacryd hys precyows body in the forme of bred and yaf it to hys discipulys. And so sche was wyth gret devocyon, wyth plentevows teerys, and wyth boystows sobbyngys, for in this place is plenyr remyssyon. And so is in other iiii placys in the Tempyl. On is in the Mownt of Calvarye; another at the grave wher owyr Lord was beriid; the thridde is at the marbyl ston that hys preciows body was leyd on whan it was takyn of the cross; the ferd is ther the holy cros beriid, and in many other placys of Jerusalem. (BMK, Ch. 29, ll. 2341-2350)

In Chapter 29, describing how Margery is guided around Jerusalem by friars, locative discourse strategy is abundantly used. The locations described in this chapter are almost all sacred sites in Jerusalem. In quotation (9), the place where Christ was nailed to the Cross and the marble stone where Christ was laid when he was taken down from the Cross are depicted. In quotation (10), these locative adverbial phrases are not used in sentence-initial position, but the sentences are still clearly connected with locative cohesive devices. These sites include the one where Christ made sacred his precious body in the form of bread and gave it to his disciples, as well as four other important places in the temple. At these points also, Margery feels great devotion and weeps abundantly. (11) Another day, erly in the morwenyng, thei went ageyn[es the] gret hyllys. And her gydes teld wher owyr Lord bare the cros on hys bakke, and wher his modyr met wyt hym, and how sche swownyd, and how sche fel down and he fel down also. And so thei went forth al the fornoone tyl thei cam to the Mownt Syon. And evyr this creatur wept abundawntly al the wey that sche went for compassyon of owyr Lordys Passyon. In the Mownt Syon is a place wher owyr Lord wesche hys disciplys fete, and a lityl therfro he mad hys Mawnde wyth hys disciplys. (BMK, Ch. 29, ll. 2332-2340)

Quotation (11) is another good example of how locative strategies in travel texts are inherently temporal strategies following the route taken by

9

In quotations, emphasis in italics is added by the author.

280

Margery Kempe’s Descriptions of Her Pilgrimages

the participant of the sentences, that is, Margery and her company. Throughout the morning, they are on their way to Mt. Zion, and all along the way, Margery sheds copious tears. Yoshikawa (2008: 238) also points out that cohesion between sentences is achieved in Margery Kempe by using participant-oriented strategy concomitantly with temporal strategies. These strategies are typically seen in narrative, and Virtanen (1995: 508) states that participantoriented discourse strategy and temporal discourse strategy can co-occur. Virtanen (1995: 505) proposes a scale between narrative and descriptive, on which travelogues fall somewhere between these two extremes: if the author writes about what he/she has seen on a particular journey, it becomes more narrative, and if the author writes about what everyone would see along the way, it becomes more descriptive. In this sense, Margery Kempe is typically narrative because it is her biography. Even in Chapter 29, where locative discourse strategy is used more conspicuously than in the other chapters, it is not descriptive (that is, not giving detailed information on the places). Presumably, the shift from temporal discourse strategy to locative strategy in this chapter reflects her main motivation in writing this text: to show that her tears are an authentic gift from God. Showing the authenticity of her tears must have been one of the most important purposes of her writing because she was treated badly by the people around her for weeping abundantly in public. Compared to Margery Kempe, the author of Mandeville seldom appears in the text, and when he occasionally does, it is as if he were trying to convince us that these are indeed true stories. (12) In þat castell duelle all wey to kepe it & for to serue the sowdan mo

þan .vjml. persones þat taken all here necessaries of the sowdanes court. I oughte right wel to known it for I duelled with him as soudyour in his werres a gret while, a‫܌‬en the Bedoynes And he wolde haue maryed me full highly to a gret Princes doughter ‫܌‬if I wolde han forsaken my lawe & my beleue. But I thanke god I had no will to don it for nothing þat he behighte me. (Mandeville, Ch. VI, p. 21, ll. 16-24)

(13) Also whan the prelate of the abbeye is ded I haue vndirstonden be informacioun þat his lampe quencheth And whan þei chesen anoþer prelate ‫܌‬if he be a gode man & worthi to be prelate his lampe schal lighte with the grace of god withouten touchinge of ony man. For euerych of hem hath a lampe be himself And be here lampes þei knowen wel whan ony of hem schall dye. For whan ony schall dye the lyght begynneth to chaunge & to wexe dym And ‫܌‬if he be chosen to ben prelate & is not worthi his lampe quencheth anon. And other

Fumiko Yoshikawa

281

man han told me þat he þat syngeth the masse for the prelate þat is ded he schall fynde vpon the awtier the name written of him þat schall ben prelate chosen. And so vpon a day I asked of the monkes bothe on & oþer how this befell, But þei wolde not tell me no thing Into the tyme þat I seyde þat þei scholde not hyde the grace þat god did hem but þat þei scholde publissche it to make the peple to haue the more deuocioun And þat þei diden synne to hide goddes myracle as me semed, For the myracles þat god hath don and ‫܌‬it doth euery day ben the wytnesse of his myght & of his merueylles as dauid seyth in the psaultere: MIRABILIA TESTIMONIA TUA DOMINE, þat is to seyne: Lord thi merueyles ben þi wytness. And þanne þei tolde me bothe on and oþer how it befell full many a tyme, But more I myghte not haue of hem. (Mandeville, Ch. IX, p. 39, l. 33-p.40, l. 21)

(14) AND fro the chirche of the Sepulcre toward the Est at .viijxx. paas is TEMPLUM DOMINI. It is right a feir hows & it is all round & high & c[u]uered with leed & it is wel paued with white marble, But the sarazines wole not suffre no cristene man ne Iewes to come þerein, For þei seyn þat none so foule synfull men scholde not come in so holy place. But I cam in þere & in othere places þere I wolde for I hadde lettres of the Soudan with his grete seel & comounly oþer men han but his signett. In the whiche lettres he commanded of his specyall grace to all his subgettes to lete me seen all the places & to enforme me pleynly all the mysteries of euery place & to condyte me fro cytee to cytee ‫܌‬if it were nede & buxomly to resceyue me & my companye & for to obeye to all my requestes resonable ‫܌‬if þei weren not gretly a‫܌‬en the Ryall power & dignytee of the Soudan or of his lawe. (Mandeville, Ch. XII, p. 53, l. 22-p. 54, l. 8)

Quotations (12) to (14) are examples from Mandeville where the author makes an appearance in the text. However, most of the stories do not involve participant-oriented strategy and are descriptive, that is, the text of Mandeville is rather close to being a travel guide. Seymour (1968), in his introduction to Mandeville, states that most medieval texts are instructive. The fact that Mandeville is written in descriptive style does not seem to be unconnected with the instructive characteristics of this text. If the writing style of Mandeville is typical as a late medieval travel account, then medieval English travel accounts are rather close to descriptive on the scale set up by Virtanen (1995: 505). Taking all these factors into consideration, the clause in quotation (10) from Margery Kempe “for in this place is plenyr remyssyon” sounds similar to the language use in pilgrims’ guides such as The Stacions of

Margery Kempe’s Descriptions of Her Pilgrimages

282

Rome. The writing style of such a travel guide might have affected her descriptions of her experiences in the Holy Land, though Margery is regarded as illiterate. The places where she was deeply affected are all major sacred sites in Jerusalem that Mandeville also mentions. Guides to the Holy Land in those days must have contained descriptions of these sites.

4. Conclusion Comparing Mandeville with the chapters describing her pilgrimage to the Holy Land in Margery Kempe by examining their sentence-initial or clause-initial adverbials, we see that Mandeville can be positioned as descriptive and Margery Kempe as narrative on the scale of travelogues proposed by Virtanen (1995: 505). There are two main differences between these two texts: in Mandeville, locative discourse strategy is the main strategy in achieving the coherence of the text, but the temporal discourse strategy is overwhelmingly dominant in Margery Kempe; (ii) participant-oriented strategy is consistently used in Margery Kempe, but it is seldom seen in Mandeville. These different choices of discourse strategy create differences in their writing style regarding whether they are descriptive or narrative.

(i)

As has been shown in Virtanen (1995, 2004), locative discourse strategy, which is the main cohesive strategy in Mandeville, inherently functions as temporal discourse strategy in travelogues. Therefore, whether the writer adopts locative discourse strategy to achieve textual coherence or not does not make an essential difference to the style. As has been seen in the previous section, sentence-initial locative adverbials are rare but do occur in Margery Kempe. However, they are always used along with participant-oriented strategy. On the other hand, the same strategy is seldom used in Mandeville (that is, locative discourse strategy is used, but participant-oriented strategy is not used). The combination of these two strategies creates the differences in the writing styles of the two texts. Observing the uses of locative discourse strategy, we recognize that the usage in Margery Kempe is very different from that in Mandeville. The purpose of using locative discourse strategy in Mandeville is rather similar to that in the travel guide investigated by Virtanen (2005: 83). In Mandeville, sentence-initial locative adverbials, which show a sequence of places, evoke in the reader a route map and guide him/her to follow the route mentally. At each location, things that would come into the reader’s

Fumiko Yoshikawa

283

sight are described. In other words, it is descriptive. On the other hand, the sentence-initial locative adverbials which are used in Chapter 29 of Margery Kempe are not for describing the places, but for strengthening the authenticity of her tears. Proving that her tears are a divine gift is her main motivation in writing the text. Therefore, it is natural that a marked discourse strategy is used in this chapter. The co-occurrence of locative and participant-oriented strategies would be a criterion for distinguishing the Margery Kempe type of mystical writing from guidebook types of travelogues. Even though Margery Kempe contains some locative strategy, it is still narrative. A drift of narratives towards autobiography in the late Middle English period has been recognized by Davenport (2004: 281), who gives Richard Rolle, Julian of Norwich and Margery Kempe as some examples. Davenport’s observation is clearly relevant to this study. Adopting a narrative style to describe travel accounts is considered a consequence of the fact that Margery Kempe included her travel experiences in her autobiography. This paper has revealed some keys for clarifying similarities and differences between Late Middle English women’s mystical writing and other contemporary travel texts in respect of discourse strategies with sentence-initial elements. However, to gain a full picture of the relationship between the discourse strategies and literary genres of Middle English mystical writing, we need to apply the same examination to further texts in this genre and also to other related texts for a comparison with mystical writing. Other analytical methods than examination of discourse strategies may also be needed for a more precise mapping of the literary genres related to Middle English mystical writing.

Acknowledgements This study was supported by a Grant-in-Aid for Scientific Research (C) from the Japan Society for the Promotion of Science (KAKENHI 22520281).

284

Margery Kempe’s Descriptions of Her Pilgrimages

Works Cited Primary Sources Furnivall, Frederick J., ed. 1867. The Stacions of Rome, in Verse from the Vernon MS., ab. 1370 A.D., and in Prose from the Porkington MS. No. 10. ab. 1460-70 A.D., and the Pilgrims Sea-Voyage, from the Trin. Coll., Cambridge, MS. R, 3, 19, t. Hen. VI. (= Early English Text Society, Original Series, No. 25.) New York: Greenwood Press. Hamelius, P., ed. 1919. Mandeville’s Travels, Translated from the French of Jean d’Outremeuse. Edited from MS. Cotton Titus C. XVI, in the British Library. Vol. I: Text. (= Early English Text Society, Original Series, No. 153.) London: Kegan Paul, Trench Trübner & Co., Ltd. Windeatt, Barry, ed. 2000. The Book of Margery Kempe. Harlow: Longman.

Secondary Sources Davenport, Tony. 2004. Medieval Narrative: An Introduction. Oxford: Oxford University Press. Duff, Gordon, ed. 1893. Information for Pilgrims unto the Holy Land. London: Lawrence & Bullen. Enkvist, Nils Erik. 1987. “A Note Towards the Definition of Text Strategy”. Zeitschrift für Phonetik, Sprachwissenschaft und Kommunikationsforschung 40:1.19-27. —. 1989. “Connexity, Interpretability, Universes of Discourse, and Text Worlds”. Possible Worlds in Humanities, Arts and Sciences: Proceedings of Nobel Symposium 65 (Research in Text Theory, 14.) ed. by Sture Allén, 162-186. Berlin & New York: de Gruyter. Moseley, Charles W.R.D. 2005. The Travels of Sir John Mandeville. Translated with an Introduction. London: Penguin Books. Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech & Jan Svartvik. 1985. A Comprehensive Grammar of the English Language. London & New York: Longman. Seymour, Michael C. ed. 1968. Mandeville’s Travels. (Edited with modernized spelling and an introduction.) (The World’s Classics, 617.) London: Oxford University Press. Virtanen, Tuija. 1992. Discourse Functions of Adverbial Placement in English: Clause-Initial Adverbials of Time and Place in Narratives and Procedural Place Descriptions. Åbo: Åbo Akademis Förlag. —. 1995. “‘Then I Saw to Antique Heddes’: Discourse Strategies in Early Modern English Travelogues”. Historical Pragmatics: Pragmatic

Fumiko Yoshikawa

285

Developments in the History of English ed. by Andreas H. Jucker, 499513. Amsterdam & Philadelphia: John Benjamins. —. 2004. “Point of Departure: Cognitive Aspects of Sentence-initial Adverbials”. Approaches to Cognition through Text and Discourse ed. by Tuija Virtanen, 79-97. Berlin: Mouton de Gruyter. Yoshikawa, Fumiko. 2008. “Discourse Strategies in Late Middle English Women’s Mystical Writing”. Historical Englishes in Varieties of Texts and Contexts. (Studies in English Medieval Language and Literature, 22.) ed. by Masachiyo Amano, Michiko Ogura & Masayuki Ohkado, 233-244. Frankfurt am Main: Peter Lang.

DIVERGING REGISTERS OF THE DISCIPLINES MEDICINE AND SCIENCE LILO MOESSNER

1. Introduction Language varieties and their manifestations in the form of individual discourses can be analyzed and classified in several ways. In their seminal work on discourse analysis Halliday et al. (1964: 87-94) suggested an upper-level distinction of language varieties according to user and varieties according to use, and a further distinction of the latter, which they called registers, according to the parameters mode of discourse, style of discourse, and field of discourse. The mode of discourse distinguishes written and spoken registers, the style of discourse colloquial and formal registers, and the field of discourse distinguishes registers which are mainly identified by their subject matter. They are divided into technical and non-technical registers, and examples of the former are the registers of the disciplines biology and mathematics. The registers which form the object of the present study are medicine and science.1 Although texts from both disciplines are extant from the Old English period onwards, very little is known about the development of their registers. The compilation of the corpora Middle English Medical Texts and Early Modern English Medical Texts initiated a large number of studies on register properties of Middle English and Early Modern English medical texts (e.g. Méndez-Naya & Pahta 2010, Pahta 2006, Tyrkkö & Hiltunen 2009). Publications on register properties of Early Modern 1

The terms ‘medicine’ and ‘science’ are used with the following meanings: “The science or practice of the diagnosis, treatment, and prevention of disease (in technical use often taken to exclude surgery).” (OED, s.v. medicine, 4a.) “In modern use, often treated as synonymous with ‘Natural and Physical Science’, and thus restricted to those branches of study that relate to the phenomena of the material universe and their laws, sometimes with implied exclusion of pure mathematics. This is now the dominant sense in ordinary use.” (OED, s.v. science, 5b.)

Lilo Moessner

287

English science texts are based on various sets of data (Gotti 1996, Moessner 2006, 2008, 2009b). Long-term diachronic studies were carried out separately on the medical register (Atkinson 1992) and later on the science register (Atkinson 1996, 1999). The data Atkinson used are extracts from the Edinburgh Medical Journal and from the Philosophical Transactions. The time span investigated is 1735–1985 in the case of the medical, 1675–1975 in the case of the science register. In both projects the data were sampled at 50-year intervals with two subcorpora for each century. For his register analyses Atkinson used Biber’s (1988) method of multidimensional analysis. The results were similar for his medical and for his science texts. The late 20th-century texts of both disciplines proved less involved, less narrative, more explicit, less persuasive, and more abstract than the texts of the first period. Diachronic comparisons of the medical and the science register are almost non-existent. Taavitsainen (1994) investigated the dimension involvement in Middle English and Early Modern English texts, charting the frequencies of first and second person pronouns as markers of involvement/affection in medical and in science texts. Her figures (Table 1) document that the frequency difference of these pronouns increased between the two periods, but more so in science than in medicine. With all due caution, this can be interpreted as an indication of a beginning divergence of the registers of science and medicine. The only comprehensive long-term diachronic register comparison which includes medicine and science is by Biber & Finegan (1997). It is based on ARCHER, a historical corpus of English texts ranging from the 17th to the 20th century. It covers ten text categories across seven periods at 50-year intervals, among them medicine and science. The results of the multidimensional analysis of this corpus are displayed in the form of diagrams. In addition to supporting Atkinson’s findings about the separate development of the registers, some of the diagrams (Figs. 5, 7, 9, and 10) allow a direct comparison of the changes of the medical and science registers. They reveal a general tendency towards a closer similarity of the registers of science and medicine in the 20th century. The research situation leaves us in a dilemma. Modest signs of a beginning divergence of the registers in the second part of the 17th century are followed by signs of a growing convergence at the end of the 20th century. This contrast makes it desirable to check what happened in between. The period chosen for the present study comprises the years 1650– 1750. Several indicators suggest that in the middle of the 17th century medicine and science were not yet strictly kept apart: famous members of

288

Diverging Registers of the Disciplines Medicine and Science

the Royal Society (e.g. Sir Kenelm Digby, Nehemiah Grew) wrote on topics of both disciplines, a substantial number of scholars were at the same time members of the Royal College of Physicians, and the first issues of the Philosophical Transactions contained articles of both disciplines. The situation changed with the beginning of the 18th century. Under the heading ‘The Changing Character of Medicine’, Barbara Kaplan (1993: 160) describes the 18th century as “a period of codification of new medical ideas, proliferation of medical ‘systems’, a shift to a secularized, rational perspective, and the adoption of a more managerial and interventionist approach to the treatment of disease”. With the foundation of the Edinburgh Medical Journal in 1733, the discipline got a publication channel parallel to the Philosophical Transactions. It can be assumed that the resulting divergence of medicine and science correlated with a divergence of their registers. This is the hypothesis to be tested in the main part of my paper. It will turn out that my results are not in line with those achieved in previous long-term diachronic register studies and that existing discourse models are inadequate for the explanation of the discrepancies. One important aspect which is lacking in Halliday et al.’s (1964) onedimensional discourse model and which will prove helpful in the discussion of the results of my analysis is the aspect genre. Together with register and discipline it figures as an additional component in Bhatia’s (2004: 30-33) two-dimensional discourse model. Disciplines are identified by the content matter of their discourses, genres by their communicative purpose, and registers by characteristic grammatical and lexical features. All three components are interrelated in complex ways. The most straightforward relation holds between disciplines and registers; it is one of content and form. The position of genre with respect to the other two components is described as “cut[ting] across disciplines in an interesting manner” (Bhatia 2004: 30). Bhatia conceives his model as open-ended, i.e. other components may be added. An even more complex discourse model was proposed by Biber & Conrad (2009). It contains the components register, genre, and style. They are defined as “different perspectives on text varieties” (Biber & Conrad 2009: 2). Register analysis correlates typical lexicogrammatical features of text samples with their situational contexts and communicative purposes. Genre analysis focuses on complete texts and tries to establish their rhetorical structure. Style analysis seeks to discover the linguistic features with an aesthetical value, which characterize text samples of a particular author or period.

Lilo Moessner

289

In the discussion of the results of my analysis I will show that none of these discourse models can satisfactorily explain the discrepancies between the results of earlier register analyses of the disciplines medicine and science and mine, and I will propose a new and more powerful model. In Section 2 I will introduce my corpus and describe the research method which I used. Section 3 is the main part of my paper; it contains the analysis of my data. In Section 4 I will discuss my results in view of the initial hypothesis and relate them to those achieved in earlier studies. Finally I will propose a new discourse model, which accommodates the discrepancies between my results and those presented in earlier research.

2. Corpus and Research Method The choice of texts presented some problems. The names of authors could not be taken as a yardstick, because especially in the 17th century many scholars published in both disciplines (cf. Section 1), and their professional training and academic affiliations and titles could also point in either direction. Therefore close reading of potential texts as well as comments of the authors on their texts were taken as criteria for their assignment to one or the other discipline. The following examples illustrate my approach. Henry Power first took a Bachelor of Arts and a few years later a Master of Arts degree in Cambridge, before he received the degree of Doctor of Medicine nine years later in 1655. He practised his profession in Halifax, and it was only through his friendship with the Townley family that he got interested in pneumatical experiments. This is how his contacts with the Royal Society started, and in 1663 he was elected Fellow of the Royal Society. The only book he ever published has the title Experimental philosophy, in three books: Containing new experiments microscopical, mercurial, magnetical. In its second part he describes his experiments relating the height of quicksilver in a tube to the pressure of the air above it. The first seven experiments were included in my 17th-century science sub-corpus. Robert Boyle, on the other hand, is mainly known for his achievements in physics and chemistry. From his experiments with the newly invented air-pump he derived the law about the inverse relationship between the pressure and the volume of a gas. Boyle’s interest in medicine stemmed from two sources: from his early contact with leading physicians in Paris, and from his own weak constitution. In the preface to the second volume of his work Medicinal Experiments he pointed out that the latter was his main motive for collecting the remedies which constitute this volume.

290

Diverging Registers of the Disciplines Medicine and Science

They are arranged in alphabetical order, and those under the letters N – T form part of my 17th-century medical sub-corpus. The titles of the books do not always clearly indicate to which discipline they are to be assigned. In the middle of the 18th century the adjective physical could either refer to “matter or nature” or to “medicine” (OED, s.v. physical, senses I and III). The title Physical Disquisitions of Tennent’s book is therefore ambiguous. Its qualification as a medical treatise can be derived from its lengthy subtitle and from the summaries of the six ‘disquisitions’. The beginning of disquisition II whose contents is specified as “Concerning the Causes, Nature and Cure of pleuritic and peripneumonic Fevers” was included in the 18th-century part of my medical sub-corpus. After the advances made in the design of microscopes in the 17th century, this instrument was used to study microcosms in nature and man. Therefore the title The Microscope Made Easy of Henry Baker’s book does not immediately allow placing it into one or the other discipline. Its sub-title specifies that the book is about different kinds of microscopes and the discoveries which were made with this new instrument. Baker dedicated his book to the Council and Fellows of the Royal Society, and in his dedication he specified that the purpose of his book was to explain the appropriate use of the new types of microscopes. Chapters VIII–XIV were selected for the 18th-century part of my science sub-corpus. The corpus thus compiled contains 18 texts of 18 different authors. Each text is an extract of about 5,000 words. Care was taken to choose passages from different parts of the longer works to avoid a bias towards introductions or conclusions. The overall size of the corpus is 94,721 words. It was produced from facsimile versions of the original texts in the form of a printed edition (Power) or of pdf-files accessed at Early English Books Online and Eighteenth Century Collections Online (all other texts). It covers the period 1650–1750, and both disciplines are represented by roughly the same number of words. The corresponding sub-corpora are organized into three periods, representing the beginning, the middle, and the end of the period investigated, i.e. the years 1650–1670, 1690–1710, and 1730–1750. Table 1 shows the structure and size of the overall corpus; the texts are identified by their authors, full bibliographical details are given in the reference section.

Lilo Moessner

291

Table 1. Structure and size of the corpus Period 1650– 1670 1690– 1710 1730– 1750 Total

Medicine Elkes (1651) Turner (1654) Maynwaringe (1668) Boyle (1693) Colbatch (1704) Pitt (1704) Cheshire (1735) Tennent (1745) Andree (1747)

Words 5,163 5,197 5,533 5,439 5,396 5,447 5,420 5,075 5,492 48,162

Science French (1651) Pecquet (1653) Power (1663) Gregg (1691) Ray (1692) Wilson (1700) Cotes (1738) Baker (1742) Shaw (1747)

Words 5,146 5,109 5,137 5,337 5,013 5,250 5,185 5,204 5,178 46,559

The research method employed in this paper follows the principles of multidimensional analysis (MD analysis) established by Biber (1988) for 20th-century English. It assumes that -

texts are characterized by a combination of several communicative functions; these functions can be described as dimensions of variation; the dimensions of variation can be derived from the co-occurrence patterns of linguistic features; the co-occurrence patterns can be identified by a statistical factor analysis.

The input for the factor analysis is the frequencies of those linguistic features which are assumed to influence the linguistic and thus the functional profiles of the texts under consideration. The absolute frequencies are normalized (here on the basis of 1,000 words) to avoid skewing of the results by different text sizes, then they are standardized to avoid that those features which stand a bigger chance than others of occurring very frequently carry more weight than those with a smaller chance. Finally the standardized frequencies of the linguistic features on each dimension of variation are summed up. These sums yield the text dimension scores. If several texts are to be considered together, their mean text dimension scores are calculated by summing up the individual text dimension scores and dividing this sum by the number of texts.2 This

2

For a detailed description of these procedural steps cf. Biber (1988: 75-78 and 93-97).

292

Diverging Registers of the Disciplines Medicine and Science

method provides a quantitative measure for the comparison of texts or groups of texts on each dimension of variation. In the original version of MD analysis, the input of the factor analysis consisted of the frequencies of 67 linguistic features, and the resulting factors representing co-occurrence patterns of linguistic features were interpreted in terms of five dimensions of variation. Biber used the following labels for them: Dimension 1: Involved vs. Informational Production Dimension 2: Narrative vs. Non-Narrative Concerns Dimension 3: Explicit vs. Situation-Dependent Reference Dimension 4: Overt Expression of Persuasion Dimension 5: Abstract vs. Non-Abstract Information. They describe situational and functional characteristics of texts. Taking over this descriptive framework in a diachronic study implies that the cooccurrence patterns of earlier texts allow the same interpretation as those of 20th-century texts. This assumption, which was made in the diachronic studies by Atkinson (1992, 1996, 1999), Biber & Finegan (1997) and González-Álvarez & Pérez-Guerra (1998), was at least partly supported by an MD analysis of 18th-century texts (Biber 2001), which was based on a new factor analysis and in which two co-occurrence patterns reflect the same production circumstances and communicative purposes as the first two co-occurrence patterns of the 1988 factor analysis of 20th-century texts. This result justifies the practice adopted in the diachronic studies; I will follow their example in this paper. Some adjustments had to be made nevertheless. They concern the number of linguistic features and their functional interpretation. For some linguistic features of 20th-century texts there are ready-made lists, e.g. time adverbials, place adverbials, conjuncts, downtoners, discourse particles, etc.; corresponding lists of these features in earlier centuries do not exist, and establishing them would have been far too time-consuming. Since the data are not tagged, the identification of features like attributive adjectives, that-deletion, or non-phrasal coordination was also beyond the scope of this paper. Therefore the number of linguistic features was reduced to 17 in the present study. Since the functional difference between relative clauses introduced by wh-pronouns and those introduced by that was not given in earlier periods, it was decided to interpret both types of relative clauses as features “used to specify the identity of referents within a text in an explicit and elaborated manner” (Biber 1988: 110). This function is ascribed by Biber to relative clauses introduced by wh-

Lilo Moessner

293

pronouns only; relative clauses introduced by that figure on his factor 6 (Biber 1988: 90). As in all multivariate statistical methods, the co-occurrence patterns and consequently the dimensions of variation in MD analysis contain a variable number of linguistic features. Dimension 1 is always the dimension with the biggest number of features. The linguistic features which were counted for this study are: -

-

-

on dimension 1: present tense verb, first person pronoun, second person pronoun, be as a main verb, possibility modal (can, could, may, might); on dimension 2: past tense verb, perfective aspect form of verb, third person pronoun; on dimension 3: attributive relative clause with relative marker in subject position, attibutive relative clause with relative marker in object position, other attributive relative clause; on dimension 4: finite conditional clause, necessity modal (should, must, ought), prediction modal (will, shall, would); on dimension 5: passive, past participle construction as nominal postmodifier, finite adverbial clause other than conditional.

3. Analysis 3.1. Dimension 1: Involved vs. Informational Production On this dimension the following features were counted: present tense verbs, second person pronouns (= thou/thee/thy, you/your), first person pronouns, be as main verb, and possibility modals (= can, could, may, might). They belong to the positive features on Biber’s factor 1, and this is how he motivated the label of the corresponding dimension: “All of them can be associated in one way or another with an involved, noninformational focus, due to a primarily interactive or affective purpose and/or to highly constrained production circumstances” (Biber 1988: 105). The most important negative features on this factor were nouns, long words, and attributive adjectives. For the reasons stated in Section 2 they were neglected, but from examples (1) and (2) below the inverse relation between the frequencies of present tense verbs and nouns can be derived. The computation of the mean text dimension scores of the medical and science sub-corpora across the three periods yielded the results presented in Table 2 and Figure 1.

294

Diverging Registers of the Disciplines Medicine and Science

Table 2. Mean text dimension scores across three periods (Dimension 1) Period 1 1.23 0.39

Medicine Science

Period 2 -3.00 1.36

Period 3 -1.10 1.52

2 1 0 Medicine

-1

Science

-2 -3 -4 Period 1

Period 2

Period 3

Figure 1. Development on Dimension 1 across three periods On this dimension the registers of both disciplines are fairly similar in period 1; both show a moderate degree of involvement. The close resemblance of both registers is illustrated by examples (1) and (2) (present tense verbs in italics, second person pronouns underlined). (1) In the time of Pestilence in no wise suffer too great thirst, and if you do thirst drink but measurably, and use cooling drinkes, such as Tysan water mingled with Vinegar, or water of Burrage, Lettice, Rose-water, Scabious, Turmentill, or Dittany; also when you go to bed, shut your doores and windowes close, and have a pan of coales, whereon cast powder of Laurell leaves dryed, Rosemary and Frankinsense. (Turner, 69 words) (2) The artificiall processe is this: Take of what water you please, whether well-water, fountain, river, or rain-water, as much as you please, let it settle three or four houres, untill the slime thereof separates it selfe: then digest it the space of a moneth; after which time evaporate the fourth part, by a very gentle heat, and cast it away being but the flegme, then distill off the remainder of the water. (French, 73 words)

Lilo Moessner

295

The communicative purpose of both texts is not to provide information, but rather to serve as guidelines in a given situation. The situation is described by present tense verbs (if you do thirst, when you go to bed, what water you please, as much as you please). The addressee of the texts is expressed by two or three second person pronouns, the addressor remains in the background. Science texts undergo a slow, but steady development towards a higher degree of involvement. The sharp decline of the mean text dimension scores of medical texts in Period 2 indicates a dramatic change towards less involvement. Yet this development is reversed between Periods 2 and 3. A comparison of the registers of both disciplines between Periods 1 and 3 reveals that they become more different between the middle of the 17th and the middle of the 18th century. This result supports our initial hypothesis.

3.2. Dimension 2: Narrative vs Non-narrative Concerns This dimension captures the degree of narrativity of texts. The following linguistic features were counted: past tense verb forms, perfect aspect forms of verbs, and third person pronouns (the forms his and her were only counted when they referred to animate antecedents). These are the positive features which carried most weight on Biber’s factor 2. Past tense and perfect aspect verbs describe past events, and third person pronouns denote specific animate referents apart from the writer and the addressee. Negative features on this factor, which were not counted for this study for the reasons mentioned above, were present tense verbs and attributive adjectives. Their absence supports the interpretation of this cooccurrence pattern as the narrativity dimension. The mean text dimension scores of the medical and science subcorpora across the three periods are given in Table 3 and Figure 2: Table 3. Mean text dimension scores across three periods (Dimension 2)

Medicine Science

Period 1 -0.11 -4.00

Period 2 1.31 -1.77

Period 3 1.22 -4.23

Diverging Registers of the Disciplines Medicine and Science

296

2 1 0 -1

Medicine

-2

Science

-3 -4 -5 Period 1

Period 2

Period 3

Figure 2. Development on Dimension 2 across three periods On Dimension 2 the mean text dimension scores of the sub-corpora are wide apart in all three periods, but the changes which they undergo are parallel. A development towards a higher degree of narrativity is followed by one in the opposite direction. Comparing the values of Periods 1 and 3 we notice an overall trend towards a higher degree of narrativity in the medical, and towards a lower degree of narrativity in the science register. The registers of the disciplines are more different in Period 3 than in Period 1, and this result again supports the initial hypothesis. The big difference between the registers in Period 3 is illustrated in examples (3) and (4) (third person pronouns in italics, past tense verbs underlined, present tense forms in italics and underlined). (3) This Method revived her Spirits, removed the Palsy of the Face, and restored her Appetite in about ten Days Time. However as she still continued pale and weak, I advised her to be sent into the Country, to take the Pills twice a Week, and the Drops the intermediate Days; by which Means she recovered very fast. (Andree, 59 words) (4) The nitrous Acid, changed to a middle Salt, manifests itself distinctly, upon the Evaporation and Crystallisation of the Lixivium; as then also, on the same Account, the unedulcorated Crocus gives not in the Air. But the little Hepar, mixed in at first with the Mass, appears partly from the Lixivium's staining Silver, partly from its depositing an impure Sulphur. (Shaw, 59 words) The narrative style of example (3) is obvious from the exclusive use of past tense verb forms. It is a report about the recovery of a little girl from

Lilo Moessner

297

palsy. Example (4), by contrast, describes the result of a chemical process. All verb forms are in present tense, and third person pronouns do not occur at all.

3.3. Dimension 3: Explicit vs. Situation-dependent Reference The most important positive features on Biber’s factor 3 were different kinds of relative constructions. They serve to avoid potential ambiguities and to provide explicit text-internal reference. The negative features on factor 3 were place and time adverbials. They situate a text with reference to the text-external spatial and temporal situation in which it was produced. The different types of reference motivated the label for this dimension. For obvious reasons the negative features could not be counted for this study. The linguistic features which were counted as indicators of explicit reference are relative clauses with relative pronouns in subject function, relative clauses with relative pronouns in object function, and other attributive relative clauses. Table 4 contains the mean text dimension scores of the sub-corpora, and Figure 3 illustrates the developments on this dimension: Table 4. Mean text dimension scores across three periods (Dimension 3) Period 1 -1.55 2.66

Medicine Science

Period 2 -2.53 0.05

Period 3 0.29 0.07

3 2 1 Medicine

0

Science

-1 -2 -3 Period 1

Period 2

Period 3

Figure 3. Development on Dimension 3 across three periods

298

Diverging Registers of the Disciplines Medicine and Science

The difference between the mean text dimension scores is very big in Period 1 with science texts scoring high for explicitness, whereas medical texts rely heavily on situation-dependent reference. This difference decreases from period to period, until the registers nearly meet slightly above the dividing-line on the side of explicit reference. This development refutes the hypothesis of the rise of separate registers of the disciplines. The difference between the registers in Period 1 is clearly documented by examples (5) and (6) (relative clauses in italics). (5) If strength doe permit, you may purge down the vicious humors with two drams of Rubarb, infused in halfe a pint of white wine, with Currans and sweet Fennell seeds, and drink it fasting to purge the vicious humors, after purging, take a dram of Diascordium, thus may you purge every other day; for in a week after this, make a drink with running water of two quarts, put seeds of Sorrell, Pumgranat rindes, Knot-grasse, Cumfery roots, Bryer roots, Plantain leaves, Cinamon, Bolarmonicke, Dragons blood, and Sugar. (Elkes, 87 words) (6) The processe, which in part shall be set down allegorically is thus: Take the best Wheat, and the best Wine of each a like quantity, put them into a glasse, which you must Hermetically close: then let them putrefie in horse-dung three dayes, or untill the Wheat begin to germinate, or to sprout forth, which then must be taken forth and bruised in a mortar, and be pressed through a linnen cloth & there will come forth a white juice like Milk. (French, 82 words) In both examples directions are given on how to proceed in a particular situation. But whereas the objects used to achieve the desired effect are only modified by a few adjectives in example (5), the writer of the science text (6) took great pains to describe the process very explicitly, using three relative clauses in addition to attributive adjectives.

3.4. Dimension 4: Overt Expression of Persuasion On Dimension 4 the degree of open persuasion is measured. Persuasion is expressed in terms of the writer’s own point of view or by arguments intended to persuade the addressee. There are no negative features on the corresponding factor. The features which were counted as indicators of persuasion are prediction modals, conditional clauses, and necessity modals. The mean text dimension scores computed on the basis

Lilo Moessner

299

of their frequencies are given in Table 5, and Figure 4 illustrates the development of the two registers on this dimension. Table 5: Mean text dimension scores across three periods (Dimension 4) Period 1 1.31 0.90

Medicine Science

1.5 1 0.5 0 -0.5 -1 -1.5 -2 -2.5

Period 2 -2.18 -1.49

Period 3 -0.77 1.22

Medicine Science

Period 1

Period 2

Period 3

Figure 4. Development on Dimension 4 across three periods On this dimension the registers show parallel developments, starting with a moderate degree of open persuasion in Period 1, becoming much less persuasive in Period 2, when the direction of the development changes again towards a higher degree of open persuasion. In Period 1 the registers of the disciplines are even more similar to each other than on Dimension 1, but from period to period they become more and more different. The hypothesis of a growing divergence of the registers is therefore supported on this dimension. The different degree of open persuasion in the registers of medicine and science in Period 3 is illustrated by examples (7) and (8) (conditional clause in italics, necessity modals underlined). (7) But while the Nights are warm as well as Days, the grosser Parts of the perspirable Matter are thrown out along with the finer, in such a Proportion, that tho’ the Blood is still in some Degree viscid from the Rarity of the Atmosphere, yet its Quantity is but little increased, except in Patients of a plethoric Habit. (Tennent, 58 words)

300

Diverging Registers of the Disciplines Medicine and Science

(8) The Nature of the Object, whether it be alive or dead, a Solid or a Fluid, an Animal, a Vegetable, or a mineral Substance, must likewise be considered, and all the Circumstances of it attended to, that we may apply it in the most convenient Manner. If it be a living Animal, care must be taken to squeeze, hurt, or discompose it as little as possible. (Baker, 66 words) Example (7) is a neutral account of the influence of the seasons of the year on human health. It is not marked for persuasion at all. Example (8) contains directions concerning the appropriate handling of microscopes. They are expressed by two tokens of the necessity modal must, and the accompanying circumstances are specified by a conditional clause.

3.5 Dimension 5: Abstract vs. Non-abstract Information The features on Biber’s factor 5 which were interpreted as markers of an abstract or impersonal style comprise conjuncts, two types of passive constructions, two types of past participle constructions, and adverbial clauses other than conditional clauses. Passive constructions and past participle constructions have the effect of backgrounding or eliminating the agent; conjuncts and adverbial clauses express complex logical relations. The co-occurrence of these features characterizes texts of a highly abstract and technical nature. For the present study passives with and without a by-agent were counted as one feature, and only those past participle constructions were counted which function as nominal postmodifiers. Conjuncts were not considered for the reasons stated in Section 2. The mean text dimension scores based on these features are given in Table 6, and the corresponding developments are illustrated in Figure 5. Table 6. Mean text dimension scores across three periods (Dimension 5) Medicine Science

Period 1 -0.65 -0.94

Period 2 0.11 0.70

Period 3 -0.11 0.82

Lilo Moessner

301

1 0.5 0

Medicine Science

-0.5 -1 -1.5 Period 1

Period 2

Period 3

Figure 5. Development on Dimension 5 across three periods This is the dimension where the registers of the disciplines are most similar in Period 1. Their starting-points are even below the dividing-line between abstract and non-abstract information. Both registers become more abstract in Period 2, but their subsequent developments go into different directions. The trend towards a higher degree of abstractness is continued in the science register, whereas the medical register moves into the direction of less abstract information. As on Dimension 4, the registers become steadily more different, and this is in line with the initial hypothesis. The different degree of abstractness in Period 3 is illustrated by examples (9) and (10) (passive constructions in italics, past participle constructions as nominal postmodifiers underlined). (9) There is in every Constitution, an insuperable Propensity, or natural Tendency to one Disease more than another; and this seems to proceed from the weak Structure of the Fabrick, not equally fortify'd in every Part, from the Inconveniencies, which Man, from the Nature of his Existence, is continually expos'd to. (Cheshire, 50 words) (10) The double reflecting Microscope, described p. 16. is the only Instrument at present made amongst us for this Purpose, wherein such a Combination of Glasses is employed. Here, indeed, the magnifying Power of the Object-Lens is greatly increased by the Addition of two Eye-glasses. (Baker, 44 words) Both texts have an informational focus and are of a highly technical nature. In example (9) reasons are presented why some people are more liable to fall victim to a particular disease than others, whereas in example

302

Diverging Registers of the Disciplines Medicine and Science

(10) the advantages of a special type of microscope are advocated. Its more abstract style is due to a bigger number of passive constructions and participial postmodifiers.

4. Discussion of Results and Conclusion On four out of five dimensions of variation the results of the analysis support the hypothesis that the registers of medicine and science started diverging after the 17th century. The growing difference between the registers manifests itself differently on the individual dimensions. The common feature of Dimensions 1, 2, 4 and 5 is that in Period 3 (1730– 1750) the difference between the registers was bigger than in Period 1 (1650–1670). But only on Dimensions 4 and 5 did this development proceed steadily, whereas on Dimension 1 the biggest change occurred between Periods 1 and 2 and concerned medicine, and on Dimension 2 between Periods 2 and 3, where it concerned science. Dimension 3 is the only one where the registers moved towards greater similarity, and they did so as steadily as they moved apart on Dimensions 4 and 5. In the middle of the 18th century there was hardly any difference between the degree of explicit reference in medical and science texts. Examples (11) and (12) illustrate this close similarity (relative clauses in italics). (11) Blood-letting, is evidently serviceable, in the Beginning of this Disease: for the increased Quantity of Blood, and its Viscidity, are both removed thereby in some Degree; the Viscidity being diminished by the Restitution of the Equilibrium between the Contraction of the Solids and Resistance of the Fluids, which before was lost by the plethoric Habit of the Body: And by the Manner it is performed in, viz. Drawing it from the Arm of the well Side, and if both Sides, or the Breast, are affected, from the Foot, a Revulsion is made from the inflamed Part. (Tennent, 96 words) (12) Tho’ by the preceding Directions the magnifying Powers of Microscope-Glasses may easily be understood, it will still remain uncertain, (if the Objects we examine are exceedingly minute) what the real Bigness of them is. For though we may know they are magnified so many thousand Times, we can make by that Knowledge but a very imperfect Computation of their natural and true Size; nor can we, indeed, come to any certain Conclusion as to that, but by the Mediation of some

Lilo Moessner

303

larger Object, whose Dimensions we really know. (Baker, 88 words) Both examples are of a general nature, outlining the advantages of bloodletting (11) and the mode of operation of microscopes (12). Reference to the extralinguistic spatial and temporal reality is not made, and one relative clause in each example does not provide much textinternal reference. Satisfactory as the results of my analysis may be in view of the initial hypothesis, they ask for an explanation when compared to those derived from the diagrams in Biber & Finegan’s (1997) study for the period investigated in the present study (cf. Section 1). My results agree with theirs only on Dimension 2, where the registers of medicine and science became more different. On Dimension 3, where their results point towards a growing divergence, mine show a growing similarity, and on Dimension 5 the inverse relation holds. Biber & Finegan’s diagram representing the developments on Dimension 1 shows that the difference between the registers of medicine and science remained constant, while my mean text dimension scores clearly reveal a growing divergence of the registers. Dimension 4 was left out by Biber & Finegan; my results show a growing divergence of the registers. Atkinson’s (1992, 1999) results can be compared to mine only separately for medicine and for science, and for the former only in the time span which covers my Periods 2 and 3. Here the same results were obtained only on Dimensions 3 and 5. The results for science texts in the same time span agree on Dimensions 2–5, but for the time span covering my Periods 1 and 2 only on one dimension, namely on Dimension 4. Overall, there are more differences than similarities. Several factors may be considered as possible explanations of these discrepancies. One of them is corpus size. In Atkinsons’s science corpus, each 50year period is represented by about 35,000 words, in his medicine corpus by about 26,000 words. ARCHER, the corpus on which Biber & Finegan’s study is based, contains about 1.7 million words. It comprises ten text categories sampled in seven 50-year periods. The categories medicine and science are represented by ten texts of about 2,000 words per period. My corpus with its 15,000 words per period and discipline occupies a comfortable middle position, and it is not very likely that its expansion would yield different results, since the results obtained at an earlier stage of my research with 10,000 words per discipline and period did not differ greatly from those presented here. The individual texts in the corpora of the other linguists are shorter, but they represent the writing-styles of more authors. It could be argued that in a corpus with fewer authors the results

304

Diverging Registers of the Disciplines Medicine and Science

of the analysis might be skewed by the linguistic idiosyncrasies of an individual author. This possibility can never be excluded, but in an earlier study (Moessner 2009a) I could show that two parallel corpora with 10 texts by 10 different authors did not guarantee more homogeneous registers. Another factor is the list of features which were counted. In the original version of MD-analysis the input to the factor analysis comprised 67 features, which are meticulously listed and explained (Biber 1988: 221245). Neither Atkinson nor Biber & Finegan specify explicitly which features they counted in their diachronic studies. The feature lists by which they illustrate the different dimensions (Biber & Finegan 1997: 258-259; Atkinson 1999: 66) are smaller than the original list, containing only 50 and 43 elements, respectively. My own list of 17 items looks rather poor by comparison. But much more research is needed before search items like general emphatics, discourse particles, time adverbials, or split auxiliaries can be established as input of a factor analysis on 17th- and 18th-century texts. The periods investigated also differ. The relevant periods in Biber & Finegan’s diagrams are given as 1650–1699 and 1700–1799. A closer look at ARCHER reveals, however, that the science texts of the first three 50year periods do not cover half a century each, but a decade at most.3 The medical texts of the first 50-year period date from the years 1685–1688, those of the second sub-period from 1735 and those of the third sub-period from 1785. Which of these data were used for the computation of the mean text dimension scores of the period 1700–1799 is not specified. Although the sub-periods of my corpus represent only 20 years each, their texts cover a wider time span than those in ARCHER (cf. Table 1). This difference is perhaps not crucial, but it may explain some of the discrepancies between Biber & Finegan’s results and mine. A similar argument holds for Atkinson’s periods because most of his texts are identical with those in ARCHER. The most convincing explanation can be derived from a reconsideration of the discourse models presented in Section 1. Halliday et al.’s model is one-dimensional, conceived as a hierarchical taxonomy of language varieties. With its one-one relation between registers and disciplines it cannot explain why in register studies of the same disciplines different results are obtained. In Bhatia’s two-dimensional model one dimension is also defined by the relation of form (= register) and content 3

The texts of the first 50-year period date from the years 1674 and 1675 only, those of the second 50-year period from the years between 1720 and 1725, and those from the third 50-year period from the seventies of the 18th century.

Lilo Moessner

305

(= discipline). The other dimension is defined by purpose (= genre). This model does not solve the problem at hand either because the additional component genre is neither related to register nor to discipline. In Biber & Conrad’s (2009) discourse model each of the three components (register, genre, and style) is directly related to ‘text varieties’, and in my understanding texts belonging to the same discipline constitute a text variety. A relation between register, genre, and style need not be specified because they are “different approaches or perspectives for analyzing text varieties, not … different kinds of texts or different varieties” (Biber & Conrad 2009: 15). There are texts which lend themselves at the same time to register, genre, and style analysis. In the present context two register analyses are compared. One way to account for their different results is to assume that the text varieties were not the same. This is only possible if discipline is not a sufficient characterization of a text variety. Bhatia’s component genre could be considered an attractive additional variable. As a matter of fact, the texts of Biber & Finegan’s corpus were sampled from the Philosophical Transactions and from the Edinburgh Medical Journal. In the terminology of genre studies, they could be described as experimental reports or research articles. The texts of my corpus are extracts from fully-fledged professional treatises. Yet this is not their only difference. The journals had a wide, but mixed readership ranging from interested laymen to specialists in their respective fields. The treatises reached only few, but expert representatives of their disciplines. Especially with regard to the issues of the Philosophical Transactions of the first sub-period it has also to be borne in mind that Henry Oldenburg as their editor exercised a powerful influence on their format. It is only the combination of these and other extralinguistic variables which sufficiently define text varieties. A discourse model which can explain that different register analyses of texts of the same discipline arrive at different results needs to integrate extralinguistic and linguistic aspects. Among the linguistic aspects are its lexicogrammatical structure (= Biber & Conrad’s register), its rhetorical structure (Biber & Conrad’s genre), and its aesthetic structure (Biber & Conrad’s style). Among the extralinguistic aspects are its content (= Bhatia’s discipline, Halliday et al.’s field of discourse), its purpose (= Bhatia’s genre), its degree of formality (= Halliday et al.’s style of discourse), its channel/medium (= Halliday et al.’s mode of discourse), its format, and its audience. Both lists are open-ended; more aspects may prove relevant in future research. The model is visualized in Figure 6.

306

Diverging Registers of the Disciplines Medicine and Science

Linguistic aspects

Extralinguistic aspects

content purpose degree of formality channel/ medium format audience

lexicogrammatical structure rhetorical structure aesthetical structure . . .

. . .

Figure 6. A new discourse model The model implies that linguistic analysis can focus on different aspects of a text/discourse variety and that text/discourse varieties can be defined at different levels of specificity. The more extralinguistic aspects are taken into account, the higher is their degree of specificity. Textbook is a very general text variety, biology textbook is a more specific variety, etextbook for advanced biology students is even more specific. This explains why text varieties defined only via their content (i.e. representing the same discipline) can have very different lexicogrammatical structures, i.e. registers.

Works Cited Primary Sources Andree, John. 1746. Cases of Epilepsy, Hysteric Fits, and St. Vitus Dance, with the Process of Cure: Interspersed with Practical Observations, To which are added Cases of the Bite of mad Creatures, Treated after the Method of Dr. Desault, with some additional Means of my own; Observations upon the same, and Remedies proposed for the Cure of Hydrophobia. London: W. Meadows and C. Clarke. Baker, Henry. 1742. The Microscope Made Easy. London: R. Dodsley. Boyle, Robert. 1693. Medicinal Experiments: Or, a Collection of Choice and Safe Remedies, for the most part Simple and easily prepared:

Lilo Moessner

307

Useful in Families, and very Serviceble to Country People. London: S. Smith and B. Walford. Cheshire, John. 1735. A Treatise upon the Rheumatism, as well Acute as Chronical: With Observations upon the various Causes that may produce them, And A Method of Cure, subjoin’d to every distinct Species of that Distemper. London: C. Rivington. Colbatch, John. 1704. Novum Lumen Chirurgicum: Or, A New Light of Chirurgery. Wherein is Discover’d A much more Safe and Speedy way of Curing Wounds than has heretofore bin usually practis’d. London: Dan. Brown. Cotes, Roger. 1738. Hydrostatical and Pneumatical Lectures by Roger Cotes A. M. Cambridge: S. Austen. Early English Books Online: Accessed at http://eebo.chadwyck.com. Eighteenth Century Collections Online. Accessed at http://galenet.galegroup.com. Elkes, Richard. 1651. Approved medicines of little cost, to preserve health and also to cure those that are sick. Provided to the souldiers knapsack and, the country mans closet. London: Robert Ibbitson. French, John. 1651. The Art of Distillation. London: Richard Cotes. Gregg, Hugh. 1691. Curiosities in Chymistry: Being New Experiments and Observations Concerning the Principles of Natural Bodies. London: Stafford Anson. Maynwaringe, Everard. 1668. Useful discoveries and practical observations, in some late remarkable cures of the scurvy. London: T. Basset. MEMT = Middle English Medical Texts compiled by Irma Taavitsainen, Päivi Pahta & Martti Mäkinen. 2005. CD-ROM with MEMT Presenter software by Raymond Hickey. Amsterdam & Philadelphia: John Benjamins. OED = Oxford English Dictionary Online. Accessed at http://www.oed.com. Pecquet, John. 1653. New Anatomical Experiments. London: Octavian Pulleyn. Pitt, Robert. 1704. The antidote. London: John Nutt. Power, Henry. 1664. Experimental philosophy, in three books: Containing new experiments microscopical, mercurial, magnetical. Hall, Marie Boas (ed.). New York and London: Johnson Reprint Corporation. Ray, John. 1692. Miscellaneous Discourses Concerning the Dissolution and Changes of the World. London: Samuel Smith. Shaw, Peter. 1747. A Philosophical and Chymical Analysis of Antimony. London: Joseph Davidson. Tennent, John. 1745. Physical Disquisitions. London: W. Payne.

308

Diverging Registers of the Disciplines Medicine and Science

Turner, Robert. 1654. Mikrokosmographa. A description of the littleworld, or, body of man, exactly delineating all the parts according to the best anatomists. With the severall diseases thereof. Also their particular and most approved cures. London: Edward Archer. Wilson, George. 1700. A Compleat Course of Chymistry. London: W. Turner.

Secondary Sources Atkinson, Dwight. 1992. “The Evolution of Medical Research Writing from 1735 to 1985: The Case of the Edinburgh Medical Journal”. Applied Linguistics 13:4.337-374. —. 1996. “The Philosophical Transactions of the Royal Society of London, 1675–1975: A Sociohistorical Discourse Analysis”. Language in Society 25.333-371. —. 1999. Scientific Discourse in Sociohistorical Context. The Philosophical Transactions of the Royal Society of London, 1675– 1975. London & Mahwah, N.J.: Laurence Erlbaum. Bhatia, Vijay K. 2004. Worlds of Written Discourse. London: Continuum. Biber, Douglas. 1988. Variation across Speech and Writing. Cambridge: Cambridge University Press. —. 2001. “Dimensions of Variation among 18th-century Speech-based and Written Registers”. Towards a History of English as a History of Genres ed. by Hans-Jürgen Diller & Manfred Görlach, 89-109. Heidelberg: Carl Winter. —. 2006. University Language. A Corpus-based Study of Spoken and Written Registers. Amsterdam & Philadelphia: John Benjamins. Biber, Douglas & Susan Conrad. 2009. Register, Genre, and Style. Cambridge: Cambridge University Press. Biber, Douglas & Edward Finegan. 1997. “Diachronic Relations among Speech-based and Written Registers in English”. To Explain the Present: Studies in the Changing English Language in Honour of Matti Rissanen ed. by Terttu Nevalainen & Leena Kahlas-Tarkka, 253-276. Helsinki: Société Néophilologique. González-Álvarez, Dolores & Javier Pérez-Guerra. 1998. “Texting the Written Evidence: On Register Analysis in Late Middle English and Early Modern English”. Text 18:3.321-48. Gotti, Maurizio. 1996. Robert Boyle and the Language of Science. Milano: Guerini. Halliday, Michael A. K., Angus McIntosh & Peter Strevens. 1964. The Linguistic Sciences and Language Teaching. London: The English Language Book Society & Longman.

Lilo Moessner

309

Hyland, Ken. 2000. Disciplinary Discourses: Social Interactions in Academic Writing. Harlow: Longman. [2004. Ann Arbor: University of Michigan Press] Kaplan, Barbara Beigun. 1994. “Divulging of Useful Truths in Physick”: The Medical Agenda of Robert Boyle. Baltimore & London: Johns Hopkins University Press. Méndez-Naya, Belèn & Päivi Pahta. 2010. “Intensifiers in Competition. The Picture from Early English Medical Writing”. Early Modern English Medical Texts ed. by Irma Taavitsainen & Päivi Pahta, 191213. Amsterdam & Philadelphia: John Benjamins. Moessner, Lilo. 2006. “The Birth of the Experimental Essay”. Explorations in Specialized Genres ed. by Vijay K. Bhatia & Maurizio Gotti, 59-77. Bern: Peter Lang. —. 2008. “Variation and Change in the Writings of 17th-century Scientists”. English Historical Linguistics 2006. Volume II: Lexical and Semantic Change ed. by Richard Dury, Maurizio Gotti & Marina Dossena, 75-93. Amsterdam & Philadelphia: John Benjamins. —. 2009a. “How Representative are the ‘Philosophical Transactions of the Royal Society’ of 17th-century Scientific Writing?” Corpus Linguistics: Refinements and Reassessments ed. by Antoinette Renouf & Andrew Kehoe, 221-237. Amsterdam & New York: Rodopi. —. 2009b. “The Influence of the Royal Society on 17th-century Scientific Writing”. ICAME Journal 33.65-87. Pahta, Päivi. 2006. “This is very important: A Corpus Study of Amplifiers in Medical Writing”. Advances in Medical Discourse Analysis: Oral and Written Contexts ed. by Maurizio Gotti & Françoise SalagerMeyer, 357-382. Bern: Peter Lang. Taavitsainen, Irma. 1994. “On the Evolution of Scientific Writings from 1375 to 1675: Repertoire of Emotive Features”. English Historical Linguistics 1992 ed. by Francisco Fernández, Miguel Fuster & Juan José Calvo, 329-342. Amsterdam & Philadelphia: John Benjamins. Taavitsainen, Irma & Päivi Pahta (eds.). 2010. Early Modern English Medical Texts. Corpus Description and Studies. Amsterdam & Philadelphia: John Benjamins. Tyrkkö, Jukka & Turo Hiltunen. 2009. “Frequency of Nominalization in Early Modern English Medical Writing”. Corpora: Pragmatics and Discourse. Papers from the 29th International Conference on English Language Research on Computerized Corpora (ICAME 29) ed. by Andreas H. Jucker, Daniel Schreier & Marianne Hundt, 297-320. Amsterdam & New York: Rodopi.

WHITE KENNETT’S ETYMOLOGICAL COLLECTIONS OF ENGLISH WORDS AND PROVINCIAL EXPRESSIONS: AN UNPRINTED SOURCE OF REGIONAL VOCABULARY JAVIER RUANO-GARCÍA

1. Introduction Our knowledge of provincial words during the early modern period is scarce. No comprehensive survey in the tradition of medieval atlases or of the impressive Survey of English Dialects (1962–1971) has hitherto been undertaken. The dearth of regionally-anchored material comparable to ME documents, and the small number of localised literary works such as Meriton’s A Yorkshire Dialogue (1683, 1685) and the anonymous “A Lancashire Tale” (c1690–1730) (Ruano-García 2010a) have rendered attempts to envisage a map of early modern regionalisms virtually unfeasible. John Ray’s collections of regionalisms remain the most significant repositories of early modern localised terms, along with the information recorded in Meriton’s “Clavis” to his 1685 Dialogue and the data contained in Bishop White Kennett’s glossary to Parochial Antiquities (1695), A Glossary to Explain the Original, the Acceptation, and Obsoleteness of Words and Phrases; and to Shew the Rise, Practice, and Acceptation of Customs, Laws and Manners. There is another valuable wordlist compiled in the 1690s which has gone largely unnoticed by scholars: Bishop White Kennett’s Etymological Collections of English Words and Provincial Expressions (MS Lansd. 1033) supplements the available regional data of the period. To my knowledge, Fox (2000) and Griffiths (2005) are amongst the few scholars who have paid to this work the attention it deserves, although a linguistic evaluation of the document remains to be made; such is the aim of this paper.

Javier Ruano-García

311

2. Dialect Material in White Kennett’s Works Born at St. Mary (Dover, Kent) in 1660, White Kennett was educated at Westminster School and St. Edmund’s Hall, Oxford. In 1685 he was appointed Vicar of Ambrosden (Oxfordshire) where he made his living until he became Rector of St. Botolph’s, Aldgate (London) in 1700. In 1701 he became Archdeacon of Huntingdon (Cambridgeshire), and later Dean of Peterborough. He was made Bishop of this city in 1718. He died in Westminster in 1728. White Kennett published 57 works, leaving behind several manuscripts that are held at The British Library as the Kennett Collection, Lansdowne MSS 935-1041. As far as is known, there are at least two of Kennett’s works that contain dialect material. Alongside the manuscript under analysis, Bishop White Kennett annexed a glossary to Parochial Antiquities (1695) in which regional words were introduced for etymological purposes. In keeping with the antiquarian scholarship of the period, Kennett endorsed the view that regional varieties were rich and valuable repositories of data which illuminated the origins of the English language. In fact, Parochial Antiquities (1695) echoes the archaeological spirit characteristic of contemporary works such as Laurence Nowell’s Vocabularium Saxonicum (c1567) (see Marckwardt 1947), William Somner’s Dictionarium SaxonicoLatino-Anglicum (1659), Stephen Skinner’s Etymologicon Linguae Anglicanae (1671), and Bishop William Nicolson’s Glossarium Brigantinum (1677) (see Shorrocks 2001: 1554). According to Skeat (1879: 1), “The original edition [of Parochial Antiquities] appeared in 1695, and is a rare book. The edition of 1818 is the standard one, having been «greatly enlarged from the author’s MS. notes»”. It is worth noting that the glossary annexed to Kennett’s work was “reprinted, separately, in 1816” (Skeat 1879: 1) with the title A Glossary to Explain the Original, the Acceptation, and Obsoleteness of Words and Phrases; and to Shew the Rise, Practice, and Acceptation of Customs, Laws and Manners. Unlike other contemporary lists such as John Ray’s (1674, 1691), Kennett’s glossary was not primarily conceived as an inventory of provincialisms. It was originally thought of as a collection of Latin terms listed to explain words found in Parochial Antiquities (1695) and to shed light upon their origins. English words were introduced by way of etymological illustrations which, according to Skeat (1879: 2), are erroneous in virtually every instance.1 1

In his edition of the glossary appended to Parochial Antiquities (1695), Skeat (1879) provides detailed information on the type of material included, on the reliability of the etymological comments made by Kennett, etc.

312

White Kennett’s Etymological Collections of English Words

I have suggested elsewhere (see Ruano-García 2009) that many of the regionalisms collated by Kennett were taken from other contemporary sources, namely Skinner’s Etymologicon Linguae Anglicanae (1671), Ray’s lists (1674, 1691) and Meriton’s “Clavis” (1685). It is worth noting that most of the terms copied, especially those from Ray, were restricted to the North. Along with the words taken from other contemporary works, Kennett listed 79 regional terms which had not been included in those sources. A careful analysis of the data indicates that some of Kennett’s dialect attributions added information to that provided by Ray’s collections, as Kennett on occasion indicated specific geographical information. For example, aver ‘a sluggish horse or lazy beast’ (s.v. averia) is marked by Ray as generally northern, whilst Kennett refers to its use in Northumberland.2 It is also worth noting that Kennett glossed northern terms that were not listed by Ray. For example, we could mention nouns such as bing ‘The cistern into which they throw the crystallized allum, for the water to drain from it’ (s.v. abunda) and slot ‘The bolt (of a door)’ (s.v. slade) that are ascribed to Whitby (Yorkshire) and Northumberland, respectively. In like manner, Kennett’s glossary gives valuable additional information about eastern and southern varieties. Some terms and expressions are marked explicitly as Kentish, e.g., guzzle ‘a gutter’ (s.v. carecta); as natural to Wiltshire: ear ‘to plough’ (s.v. arura); or as belonging to Sussex: esh ‘The stubble after the corn is cut’ (s.v. ernes) (see further Ruano-García 2009; 2010b: 135-136). In line with this glossary, White Kennett compiled Etymological Collections of English Words and Provincial Expressions. This work contains over 6,700 entries that include regionalisms and expressions from all over England, along with vocabulary in widespread use. A large amount of the dialect material included in this unpublished dictionary is not found in contemporary lists, which makes it an interesting document for dialect research.

2

Unless otherwise indicated, the definitions of the words included in the glossary to Parochial Antiquities (1695) are from the 1816 reprint. For the sake of clarity, reference is made to the lemmas under which regional terms are included.

Javier Ruano-García

313

3. Etymological Collections of English Words and Provincial Expressions (MS Lansd. 1033) 3.1. The Manuscript This is a collection of English words listed for etymological purposes. Also referred to as “Etymol[ogia] Anglican[a]” (fo. 3r), the document remained in manuscript on Kennett’s death, and is still kept at The British Library as part of the Lansdowne MSS. A Catalogue of the Lansdowne Manuscripts in the British Museum (Ellis 1819: 263) explains that it is MS 1033 of the Bibliothecae Lansdownianae, and volume XCIX of Bishop White Kennett’s manuscripts. The entry for MS 1033 in the catalogue indicates that the document includes Kennett’s Etymological Collections of English Words and Provincial Expressions, with some papers at the end containing other etymological collections by George Hickes (1642–1715) that appear to have been added to the MS by Dr. West. The MS is signed, as the name of Kennett appears in folio 5v: “The Rev. Mr White Kennett Vicar of Ambrosden”. No specific date is provided, but the collection of words it contains might have been compiled in the late 1690s and the early eighteenth century (see below). To my knowledge, little scholarly attention has been paid to Kennett’s work, Fox (2000) and Griffiths (2005) being isolated examples that have referenced it. The document (330 x 225 mm) comprises 484 folios that include over 6,700 entries. The leaves of the MS are now mounted on guards, which makes it difficult to determine the quiring. The order of the folios is marked by means of numbers that appear on the upper right-hand corner of the recto sides. It is worth noting that from folio 21 onwards two different numbers are found on the recto sides of the folios, one placed above the other. The one at the top is crossed out, so that the number below marks the order from folio 22 onwards. The sides of the folios are also marked up in alphabetical blocks with the headings AA, AB, AC, etc. that are for the most part placed in the middle top. They are in turn numbered consecutively – AC1, AC2, AC3, etc. – when the alphabetical blocks comprise more than one page. The entries for the words listed are generally arranged into two columns though not in strict alphabetical order. MS entries under SC1 (fo. 339r) may illustrate this: 3 3

Examples from the manuscript have been transcribed by myself. My transcription is intended to reproduce faithfully what White Kennett wrote; no capitals have been modified and punctuation has been respected. The underlining of headwords has not been kept. Spellings are presented as they appear in the original, although

314

White Kennett’s Etymological Collections of English Words

(1) to Scale Scales a Scale a Scabb a Scallion Scarpe Scandalous Scantlin to Scatter Scathe Scade Scaddle Scafe It is worth noting that some folios contain just one or two entries, and that there are quite a few pages that are left blank, but marked with an alphabetical heading (fo. 166v is marked with GU2, and fo. 286r is marked with PA1, for example). It is also worthy of remark that there are pages in which the words listed do not correspond with the alphabetical heading that appears. That is the case of the three entries under CE6 (fo. 58v): chafts, chalice, chalk. Each entry contains different kinds of data representative of different fields, namely the lemma or headword that is underlined, followed by the meaning and etymology. The next examples illustrate the basic entry structure in Kennett’s manuscript: (2) To abide, expect or wait for. Sax. Bidan, Abidan, expectare. Island, Bidia, bid, expectare, mora. Goth. Beidan, expectare (fo. 6v) (3) a Lighter, a Boat or Vessel to carry coals, wood &c. on the river Thames. Sax. Liðan, navigare, Liðe, classis (fo. 227r)

their opaqueness has made it sometimes difficult to decide what they represent. In those cases, square brackets are used to indicate a doubtful sequence. In what follows, the definitions are taken from the manuscript. For the sake of clarity, the folio reference is given in brackets.

Javier Ruano-García

315

The amount of information that can be found in the entries as regards these fields is variable. In some cases, Kennett gave alternative spelling variants for the headword, as in brake: (4) Brake or Broke. An instrument to break hemp. Sax. bracan, conterere (fo. 47r) In other cases, several meanings are given for the term, at the same time as various etymological connexions are indicated to explain the origin of a word. Remarks on French, Scandinavian, Anglo-Saxon, Hebrew or Celtic etymons are found, these being generally defined in Latin. The entry for arles exemplifies this: (5) Arles or Earles, Earnest as give me earles an Arles-penny. Yorksh. Servants vails or gifts to servants are likewise called Arles. It seems deriv’d from the Brit. Arddelro wch the South Wales men write Ardhel, to vouch, to vindicate, to bear witness (...) Otherwise in Islandick or Scandick Ar is principium, unde Ardagis principium diei. From Ar comes the diminutive Arla, whence Arladas primo mane, very Early or as commonly pronouncd very Arly, and Early or Arly daies. From Arle probably Arles, (…) what is given at the beginning of a bargain to seal and confirm it (fo. 20r) Whilst this represents the basic information that is included in the great majority of the entries, data of other kinds may also be found. In particular, we may find (a) geographical information, (b) reference to the source in which a word is used, (c) citation(s) that illustrate(s) a word in context, and (d) cross-references to other entries in the dictionary. Example (6) clearly illustrates these kinds of information: (6) a Carl-Cat, a boar-cat. Bor. from Sax. Carl, masculus. Thence in Westmorl. Carl-hemp is the rough stronger hemp as it were, the he-hemp. Carl is now in Scotland the familiar word for any Man, as the House-Carl, the old Carl, whence our Churl, vid. Karl.

316

White Kennett’s Etymological Collections of English Words

Trim hemp to serve at a need the fimble to spin, the carle for her seed. T. Tusser, p.86. (fo. 56r) The title of the manuscript suggests that there is abundant regional information. Interestingly, as examples (5) and (6) show, this is sometimes indicated by means of abbreviations which remind us of the system of labels of early dictionaries such as Coles’s An English Dictionary (1676). Kennett used the Latin Bor. (Borealis), Dunelm. (Dunelmensis), and Ebor. (Eboracensis) for northern, Durham, and Yorkshire, respectively. Other abbreviations found are Chesh. for Cheshire, Lancash. for Lancashire, Northumb. for Northumberland, Oxfordsh. for Oxfordshire, Suss. for Sussex, or Westmorl. for Westmorland. These regional abbreviations are not however deployed systematically, as the full name of counties is sometimes provided, and some regional words are not assigned any geographical comment. The citations that illustrate the usage and meaning of certain words deserve special attention. Unlike other contemporary collections of regional terms, such as Ray’s, it is worth noting that Kennett frequently quotes from other sources to exemplify the meaning of a word in context. The examples were for the most part extracted from literary works upon which the author seemingly drew for some of his terms. Amongst them, references are found to Meriton’s Yorkshire Dialogue (1685), to the anonymous Scottish ballad “Christs Kirk on the Greene”, and to Edward Manlove’s poem on the Derbyshire mines (1653). Needless to say, examples from these works are generally quoted to illustrate regionalisms, as (7) and (8) show: (7) Agye, to look agye, to look awry, to look on one side. Else he’l leauk agye, when he comes in. Yorksh. Dial. p. 39. (fo. 11r) (8) Feed, Feud, quarrle or enmity. He trow’d it had been for old feed. Xl Kirk Cimb. faide, Sax. fahð. Lat. barb. faide, feid inimicitia, simultas. Goth. (…) (fo. 123v) Also, Kennett used isolated examples from popular songs or from the works of Ben Jonson, quotations from husbandry books such as Tusser’s Five Hundred Points of Good Husbandry (1557) (see (6) above), and from topographical writings such as Robert Plot’s natural histories of

Javier Ruano-García

317

Oxfordshire (1677) and Staffordshire (1686). Additionally, Kennett resorted to seventeenth-century dictionaries such as Blount’s Glossographia (1656) and William Somner’s Dictionarium Saxonico-Latino-Anglicum (1659). Likewise important are the references to Scottish literary works. There are numerous examples of Scottish terms that are illustrated by examples from or references to the works of some Scottish authors like Gavin Douglas and David Lyndsay: (9) a Rilling, a shoe. The Scotch now call a shoe that hath a sole wth the hair to the ground a Rough Rilling, which is the very term used by G. Dougl. f. 185. [T]are left fute and al thare leg was bare Ane rouch rylling of raw hyde & of hare The tothir fut coverit wele (fo. 323v) (10) Lake, disgrace, scandal.- with shame and lake D. Lyndsay. (fo. 219v) In line with this, it is worth indicating that there are a great many entries that represent cases of Scottish spelling variants of standard words that for the most part include abbreviated cross-references to indicate where a word is found or has been extracted from; for example, D. and L., as in (11) and (12). It is hard to know with confidence what these two abbreviations might stand for, although it seems reasonable to assume that Kennett was referring to the works that contained the variants listed. (11) Affrayit, affrighted. D. 52, a. (fo. 10v) (12) Althocht, although. L.3 (fo. 13v) It is also difficult to know if Kennett benefited from the help of someone else in writing his dictionary. Griffiths (2005: xv) notes that Kennett probably received information from the well-known AngloSaxonist George Hickes (1642–1715), a native of Yorkshire. In fact, the manuscript (fos. 453-484) contains fragments of the correspondence between Kennett and Hickes in which it is said that Hickes sent Kennett information about 60 words with their etymologies and a Yorkshire book that Kennett did not apparently receive. Reference to the work of Manlove

318

White Kennett’s Etymological Collections of English Words

and to Scottish proverbs is also found in one letter written by Hickes.4 Along with the correspondence, there are also shorter lists of words which are not arranged alphabetically and are mostly defined in Latin. These were probably compiled by George Hickes; regional material can be found in them too. It is unclear when the compilation of words was finished. The dates of the letters included in the manuscript suggest that the dictionary was arranged during the last years of the 1690s and the early eighteenth century.

3.2. Lexical Survey For the purpose of this paper, the vocabulary contained in Kennett’s manuscript has been broadly classified into two different groups on account of the types of words that are present. As the title of the work indicates, English words and provincial expressions are listed in it. As for the former, in this collection one can find grammatical words – above, back, in, into –, words in common use of everyday life – to bake, breakfast, inne, iron –, the paradigm of to be – “I am, Thou art, He is” (fo. 14r, s.v. am) –, or words which had apparently fallen into disuse: agrise ‘to dread or greatly fear’ (fo. 11v), behest ‘promise’ (fo. 33r).5 Whilst words of this kind, that is non-regional terms, constitute an important part of the dictionary material, regional words are more abundant. Terms and expressions from many different parts of Britain are included in this document. As previously indicated, Kennet agreed with many of his contemporaries in that regional varieties of English were fruitful sources of lexicalisms to illuminate the history of the language. In what follows, my analysis will concentrate on the lexical data representative of northern, Midland and southern varieties, along with the incidental but interesting remarks given on Scotland, Wales and Ireland. It 4

One of the letters reads as follows: “Sir, I have have [sic!] sent you 60 words, wth their etymologies. I hope you received the little Yorkshire book, and those inclosed in it, wch are much more valuable than these. I will use all money to get Manlove, and the Scotica Proverbia, certify me of the receipt of these, and the former, if you have got them. The cold has so nummd me I can write no more than to give my humble service to Mr Kennet and to wish you an happy Christmasse. Adieu” (fo. 463r). 5 The OED (s.v. agrise v.1) marks agrise as “Obs.”, with 8 quotations from a1000 to 1598. Similarly, behest is marked as “Obs.”, with 10 quotations from a1200 to 1634 (OED behest n.1). Kennett indicates in the entries for these two terms that they were used by Chaucer.

Javier Ruano-García

319

is worth indicating that I have taken the county divisions of the Survey of English Dialects (Orton et al. 1962–1971) for my classification of the data. No specific distinctions have been made between East and West Midlands, though. 3.2.1. Northern Words Over 1450 words that are marked as belonging to northern English dialects have been counted in the document. They are generally ascribed to the North, although Kennett also provided localised attributions that give information on very specific localities and counties that are not reported by other sources. As Fig. 1 illustrates, information on each of the six northern counties and the Isle of Man is given, along with a few references to the areas beyond the Trent and the Humber.

Figure 1. Distribution of the northern words recorded6 Examples of these northernisms are boor ‘the parlor, bed chamber, or any inner room’ (fo. 44r) that is localised to Cumberland, clud-nut ‘when two nuts grow into one’ (fo. 66v) that is said to be used in Westmorland, deemsters ‘Judges chose among themselves by the inhabitants (…) to decide causes’ (fo. 82r) that is given as a word characteristic of the Isle of Man, coggles ‘a sort of small vessels’ (fo. 68r, s.v. coggs) that is marked as natural to Scarborough (Yorkshire), and fax ‘hair’ that was commonly used “on the other side” of the Trent (fo. 122v, s.v. fax). 6

In this and other graphs below the name of counties is indicated by means of the traditional abbreviations that are used in Joseph Wright’s English Dialect Dictionary (1898–1905) (henceforth EDD).

320

White Kennett’s Etymological Collections of English Words

Much of the information provided by Kennett’s dictionary regarding northern words extends that given by Ray’s attributions for two reasons. Firstly, he often reported on the specific counties or localities in which the words were used. In this sense, Kennett’s data improve the geographical ascriptions found in Ray’s glossaries, since Ray often marked terms as generally northern. We could mention words such as burtle ‘a sweating or sweet apple’ (fo. 51v) and maug ‘a wive’s brother’ (fo. 241v), which are specifically given by Kennett as Yorkshire and Durham terms, respectively. Ray included both of them in his list of northernisms with no further indication.7 Secondly, Kennett recorded a large number of northern words that were not included in Ray’s lists. To name but a couple of them, asherland ‘wood land grubd and ploughed up’ (fo. 22r) and layerly fellow ‘an idler scoundrel scandalous rogue’ (fo. 221v), which are marked as northern, and natural to Kendal (Westmorland), respectively. Quantification of the northern words recorded and comparison with the terms listed by Ray indicate that Kennett apparently relied on him for a significant number of entries. In particular, 32.2% of the words collected (481 words) are common to both lists; those that are not listed by Ray are greater in number, though (see Fig. 2). Additionally, some of the northern ascriptions provided appear to have been based on Meriton’s Yorkshire regionalisms, as a great many of the Yorkshire attributions given by Kennett are illustrated by examples from Meriton’s Dialogue (1685). Also, there are words that are exemplified by literary material other than Meriton’s work and are marked as northern. For example, bab ‘to bob down, to bow often’ (fo. 28r) is illustrated by an example taken from the anonymous “Christs Kirk on the Greene”. This naturally suggests that other printed documents helped Kennett in his collection of northernisms. Further research is needed to ascertain with precision those other printed sources.

7

Burtle is defined by Ray (1691: 12) as ‘a sweeting’; maug is glossed as ‘my Wives Brother, or Sisters Husband’ (Ray 1691: 46, s.v. meaugh).

Javier Ruano-García

321

Figure 2. Some sources behind Kennett’s northern attributions (with rounded figures) The northern data included in the manuscript comprise words from different fields of the lexicon, various expressions and grammatical words. By way of illustration, we can find terms related to farming vocabulary, e.g., framal ‘a with or wood-band wth which the cattle are tied to their stall or standing’8 (fo. 138r) that is marked as a Lancashire term, and nowtgeld ‘money paid for the agistment or pasture of cattle’ (fo. 272v) that is given as a general northern word. Also listed are nouns representative of measures – boll of salt ‘two bushels’ (fo. 43r) that is marked as a Yorkshire term, of household objects – fible ‘a sort of spatula of wood wth wch they stir the water and oatmeal in making milk-porrige’ (fo. 127v) that is said to belong to Yorkshire, examples of adverbs – liever ‘rather’ (fo. 228r) that is marked as natural to Northumberland, and expressions like mow a woman ‘rem cum muliere habere’ (fo. 252v) [‘to copulate’], which is marked as generally northern. Interestingly, the document gives insights into contemporary cultural practices that inform the reader about aspects of common everyday life in the northern provinces. For instance, feut, which is glossed ‘a darling, as in a family of children the best beloved is called a feut’ (fo. 126r). Other examples are hand-festing ‘a contract made 8

The OED defines with as follows: “A band, tie, or shackle consisting of a tough flexible twig or branch, or of several twisted together; such as twig or branch, as of willow or osier, used for binding or tying, and sometimes for plaiting” (s.v. withe, with n. 1).

322

White Kennett’s Etymological Collections of English Words

for marriage’ (fo. 170r) and rush-bearing ‘the custom of carrying rushes to adorn the Church’ (fo. 331v), marked as natural to Northumberland and the West Riding of Yorkshire, respectively, that refer to practices connected with religion and marriage. Likewise significant are the names denoting games for children: bad ‘to play at bad wth a bad-stick, to play at cat or trippet wth a cat stick or trippet stick’ (fo. 28v), bitty-base ‘Boys play in Yorksh. the same as Prison-base in Southern parts’ (fo. 39r), penniston ‘to play at quoits wth stones or hors shoes’ (fo. 294r) or to play at knur and knor ‘A game among the boys in Yorksh. wth a little round chessball (wch they call a knur) struck from one to another wth little [c]andy sticks calld knur-sticks or knurl-sticks from knor a not in a tree’ (fo. 213v, s.v. to knur). All of these are marked as characteristic of the county of Yorkshire. Side by side with the northern words counted, the document contains some words that are not marked, but that appear to be northern because they are used in contemporary northern dialect literature. In particular, 129 unmarked northern terms have been counted, most of which are exemplified by quotations taken from Meriton’s Dialogue (1685). In addition, the manuscript includes some terms representative of spelling variants that suggest northern sounds. They are marked as generally northern or as natural to specific counties, from which phonetic and orthographical data might be retrieved. There are examples of northern open vowels: amang ‘among’ (fo. 14r) or lang ‘long’ (fo. 219v), among others. Also, the document contains cases illustrative of the northern lack of rounding: ath ‘oath’ (fo. 23r), bath ‘both’ (fo. 31v) or stane / stean ‘stone’ (fo. 372r), instances of the so-called ‘oo-fronting’, e.g., beauk ‘book’ (fo. 32r), examples representative of the lack of assibilation, e.g., to belk ‘to belch’ (fo. 33v), and words showing the northern development of ME /ܼ+çt/: neet ‘night’ (fo. 261v), reet ‘right’ (fo. 319r). Similarly, the dictionary lists examples of spelling variants for the same words, which attests to variation in pronunciation across the North. For example, bargham “the collar of a working horse” (OED) is quoted as brechim (fo. 48r) in Durham, barwham (fo. 30r, s.v. barf-hame) in Northumberland, and barf-hame (fo. 30r) in Yorkshire. In like manner, neife (fo. 261r), neif (fo. 262r) and neaf (fo. 261r) are marked as variants of nieve “the fist” (OED) in Durham, Durham and Westmorland, and the North, respectively. 3.2.2. Midland Words Midland words also constitute an important section of the regionalisms included in the dictionary. Over 550 terms representative of the East and

Javier Ruano-García

323

West Midlands have been counted, provincialisms from 21 different counties having been listed, alongside general attributions to the Midlands, eastern and western counties.

Figure 3. Distribution of the Midland words recorded As with the northernisms so far described, the Midland words recorded by Kennett are an important contribution to the lexical history of England. The dictionary provides localised information which improves that contained in other sources, especially in Ray’s collections. Actually, Kennett remarked on the specific counties in which words were used, thus extending Ray’s data. In this connection, we can find nouns such as dorr ‘the common great round bodied black beetle’ (fo. 90r) that is marked as natural to Essex, and verbs like to glafer or glaver ‘to flatter’ (fo. 157r, s.v. glavering) that is ascribed to Cheshire.9 Additionally, Kennett glossed terms about which Ray did not give any information. For example, he listed terms belonging to the counties of Derbyshire: gor-cock ‘red game, distinguisht from grous or Black game’ (fo. 162r), skellard ‘warpt, cast, given’ (fo. 351v); Norfolk: bunny ‘a swelling or tumour arising from a 9

Note that dorr is listed by Ray (1691: 96) in his collection of South and East Country words with no specific indication of the county or counties in which it was used. On the contrary, and despite being a Cheshire word according to Kennett, glaver is included in Ray’s collection of northernisms (1691: 31). See Ihalainen (1994: 201) who argues that John Ray “regarded the Walsh-Bristol Channel Line as the main linguistic divider in England rather than the Humber-Ribble line”.

324

White Kennett’s Etymological Collections of English Words

stroke or blow’ (fo. 51v), cothish ‘morose’ (fo. 69v); Oxfordshire: cotterel ‘a small round iron plate wth an hole in the middle (…)’ (fo. 69v), cuckle ‘to cuddle applied to children’ (fo. 74r); or Staffordshire: denshirring ‘to cast pairings of earth turf and stubble into heaps’ (fo. 83r), omfry-floor ‘the 4th parting or laming’ (fo. 280r). Kennett’s Midland data are especially noteworthy on account of the comments he made on words used in counties about which little information is known from the period. Amongst them, we may refer to glat ‘hole or gap in the hedge’ (fo. 157r) and to take vere ‘to step backward in a running jump for the better advantage of leaping forward’ (fo. 411v, s.v. vere) that are given as characteristic of Herefordshire. Also significant are grig ‘heath’ (fo. 164r), which is said to be natural to Shropshire, haspinfull ‘a handfull’ (fo. 173r) and wong ‘a croft or little parcel of land’ (fo. 436r) that are marked as Nottingham terms, or yemmouth ‘after-math’ (fo. 445r) that is ascribed to the county of Gloucestershire. Quantification of the data and comparison with Ray’s lists show that a great many words are not included in Ray’s collections. In particular, 417 of the 558 words counted are quoted by Kennett, but not by Ray, which amounts to 74.7% of the terms recorded. Fig. 4 shows that some of Kennett’s attributions are based on the evidence given by sources other than Ray’s lists. Indeed, there are some entries in which Kennett acknowledged his indebtedness to Tusser’s husbandry book (1557), Plot’s Natural History of Oxfordshire (1677), and Manlove’s Derbyshire poem (1653). A careful analysis of the Midland data further hints at the possibility that Kennett drew upon other works that he did not acknowledge in the corresponding entries. By way of illustration, anconies-end ‘a Bloom of iron is a square bar in the middle, wth two square knobs at the end, (...), the smaller being calld the Ancony-end’ (fo. 16r) is marked as a Staffordshire term, and was plausibly taken from Plot’s Natural History of Staffordshire (1686).10 It would be necessary to make a thorough comparison of sources, in order to ascertain the exact proportion of words that are original to Kennett’s document. Whatever might have been the sources consulted, Kennett’s is probably one of the few known lexicographical accounts in which Midland words of the period are listed. 10

In his description of Staffordshire minerals and stones, Plot (1686: 163) indicates that “they work it [the hammer] into a bloom, which is a square barr in the middle, and two square knobs at the ends, one much less then the other, the smaller being call'd the Ancony end, and the greater the Mocket head”. Kennett’s entry for anconies-end reads as follows: “a Bloom of iron is a square bar in the middle, wth two square knobs at the end, one much less than the other, the smaller being calld the Ancony-end, and the greater the mocket-head” (fo. 16r).

Javier Ruano-García

325

80 70 60 50 40

74.7

30 20

23.8

10

1.43

0

Recorded by Kennett

Recorded by Ray

Recorded by others

Figure 4. Some sources behind Kennett’s Midland attributions (with rounded figures) The Midland vocabulary recorded belongs to different lexical domains. Attention should be paid to farming words – bagging ‘the cutting up of the H[e]lm or wheat Stubble’ (fo. 28v) that is marked as an Oxfordshire noun, colpices ‘samplers or young standers left for trees upon cutting down the under wood’ (fo. 68v) that is quoted as natural to Warwickshire; terms related to animals – copps of a cart ‘that part wch hangs over the fillar [hor]se’ (fo. 69r) that is ascribed to Bedforshire; types of drink – cherry curd milk ‘the milk wch a Cow first gives after calving’ (fo. 59r) that is attributed to Oxfordshire; grout ‘the liquor wth malt infused for Ale or Beer before it is fully boiled’ (fo. 165r) that is given as a Leicestershire term; and to mining terms: charks ‘pit coat charrd’ (fo. 59r), which is ascribed to Worcestershire, and jack ‘In Coal Mines they have two engines to draw up the water, the greater calld a Gin turnd allwaies by horses, the latter a Jack turnd commonly by Men the small engine turnd commonly by men in coal mines’ (fo. 190r) that is marked as characteristic of Staffordshire. As is the case with the northern words listed, information might be retrieved as regards spelling variants suggestive of Midland sounds; 21 cases have been counted. Examples are girse ‘grass’ (fo. 154v), marked as an Essex term, which represents a clear example of metathesis, and haw ‘hall’ (fo. 173v), quoted as natural to Staffordshire, that shows a case of /l/vocalisation. Also, we find smeeth ‘smooth’ (fo. 359r) that stands for vowel alternation of /u:/ in Derbyshire, and cases such as zullough ‘a sullow’, showing the voiced pronunciation of /s/ in the county of Gloucestershire: “in Gloucest. a Sullow pronounc’d wth liquefaction a Zullough” (fo. 375v, s.v. sull). Significantly, Kennett commented on the

326

White Kennett’s Etymological Collections of English Words

Midland forms of the third person singular feminine subject pronoun she. He glossed he and hoo as Staffordshire variants (fo. 174r), ho as the Worcestershire form (fo. 183r), and hoo as characteristic of the NorthWest (fo. 185v), also recorded by Ray (1691: 38). 3.2.3. Southern Words In line with the above said about northern and Midland words, Kennett’s manuscript contains a substantial amount of southern data: over 500 terms have been counted. Specific attributions are usually given to the words, to the extent that only 24 of them are labelled as generally southern. Fig. 5 shows that information about 11 different counties is documented in the dictionary, data from Kent, Sussex and Wiltshire clearly predominating over those from Somerset or Berkshire.

Figure 5. Distribution of the southern words recorded Kennett’s biographical links with southern counties, as a native of Dover, might explain the profuse incorporation of words from these provinces, especially from Kent. He gave precise location to a considerable number of terms that were also listed by John Ray but were given a general southern restriction. Some clear examples are attar ‘matter or corruption of a sore or wound’ (fo. 23r), which is quoted by Kennett as a Sussex term, flasket ‘a long shallow wicker basket to carry linnen cloaths to the hedge’ (fo. 131r) that is localised to Kent, and goche ‘a pot or pitcher of earth or stone for Ale or Beer’ (fo. 160r) that is ascribed to the counties of Wiltshire and Suffolk. Ray (1691: 88, 98, 100) includes all of them in his collection of East and South country words with no indication as to the county or counties in which they were used. Hand in hand with

Javier Ruano-García

327

the improved geographical attributions, Kennett reported on vocabulary that is not found in Ray’s lists. As an illustration of this, we may find words used in Somerset: berrithatch ‘litter for horses’ (fo. 34v); terms current in Berkshire: moyn ‘a maxer or dunghill’ (fo. 252v); and nouns used in Surrey: pightel of ground ‘a small croft or enclosure’ (fo. 297r). From this it appears that Kennett’s is probably one of the first word collections in which remarks on terms used in these regions, as well as in others such as the Isle of Wight: hoblers ‘runners who presently give inteligence to the governour’ (fo. 184r) and Hampshire: croud ‘an Appels Pasty’ (fo. 71r), are included before Joseph Wright’s EDD.11 This makes Kennett’s dictionary a pivotal one as regards the inclusion and labelling of terms ascribed to these southern counties. Quantification of the southern data collected and comparison with Ray’s lists indicate that almost 74% of the southern material is only included by Kennett, with 25.8% of words being common to both works. As Fig. 6 shows, Kennett appears to have based some of his ascriptions on other printed material too. By way of illustration, Somner’s etymological dictionary (1659) is acknowledged as a source for the Kentish tigh ‘close or croft of land’ (fo. 397r). As is the case with the Midland words, further comparison with other contemporary sources would probably allow us to ascertain the input provided by other works that are not mentioned in the corresponding entries.

80 70 60 50 40

73.9

30 20

25.8

10

0.2

0 Recorded by Kennett

Recorded by Ray

Recorded by others

Figure 6. Some sources behind Kennett’s southern attributions (with rounded figures) 11 Note that Kennett has another entry for croud with the meaning ‘A course apple pasty’ (fo. 71r) that is marked as characteristic of Wiltshire.

328

White Kennett’s Etymological Collections of English Words

The vocabulary recorded comprises terms from different fields. For example, there are words related to animals: broc ‘an ordinary horse or jade’ (fo. 49r) that is marked as Kentish; stut ‘a gnat’ (fo. 374r) that is ascribed to Somerset; household objects, e.g., crock ‘a porridge pot’ (fo. 71r) that is marked as natural to Wiltshire; mining terms mostly attributed to the county of Cornwall: load ‘the vein of tin’ (fo. 230r), loob ‘a pit or vessel into which the trough delivers’ (fo. 230r); farming practices, e.g., titching ‘setting up turfes in order to dry them by the sun and wind & make them fitt to burn upon the land’ (fo. 398r) that is said to be used in the counties of Devonshire and Cornwall; or verbs of everyday speech like bounse ‘to tell a great and incredible lie’ (fo. 46r) and brustle ‘to crackle, to make a noise like straw or small wood in burning’ (fo. 50v), both marked as Kentish. Incidental but valuable remarks are given on linguistic issues other than lexical. Kennett indicates the Kentish /d/ pronunciation of rathe ‘early’ (rade, rader) in the expression rade ripe fruit (fo. 317r, s.v. rathe). Also, reference is made to the fronted pronunciation of rush, rushes in Kent: “In Kent truly pronounct Rish, Rishes” (fo. 332r, s.v. rush). 3.2.4. Scottish, Welsh and Irish Terms Although less in number, Kennett’s dictionary includes some terms that improve our knowledge about other British varieties, namely those of Scotland, Wales and Ireland. As Fig. 7 shows, information about Scottish lexis predominates, whilst data documented as regards Wales and Ireland are scarce, with 15 and 2 terms having been counted, respectively, for each. As for the Scottish words, most of them are given as natural to Scotland, with gar ‘to doe’ (fo. 147r), mikel ‘magnus’ (fo. 247v) and snod ‘a hare-lace’ (fo. 364v), among others, being marked as common both to the North of England and Scotland. Examples representative of general Scottish attributions are okerins ‘usurers’ (fo. 278r, s.v. oker), roughrilling ‘a shoe that hath a sole wth the hair to the ground’ (fo. 329v, s.v. rough) and spay ‘to divine, or foretell’ (fo. 369v, s.v. spayman). Remarkably, Kennett gave specific locations to terms such as mull ‘the promontory or neck of land’ (fo. 254r), which is ascribed to the county of Galloway, and lawes ‘great heaps of stone wch were a sort of rude monument for the dead’ (fo. 221v) that is localised to the border districts.

Javier Ruano-García

329

Figure 7. Distribution of terms belonging to other British varieties Similarly, some of the Welsh words are ascribed to specific counties. For instance, stemples ‘At the silver wires (...) travers pieces of wood’ (fo. 372v) is marked as a word used in Cardiganshire, and slag ‘the black slat wch lies commonly abov the coal in sinking their pits’ (fo. 352r) is given as a term used in Flintshire. As a rule, the words are, however, given a general Welsh distribution; such is the case with bragget ‘a compound drink of malt and honey’ (fo. 47r) and tranters ‘ripiers who carry fish or dorsers’ (fo. 404r), among others. Finally, the terms of Irish distribution – lough ‘a pool, pond’ (fo. 233r) and plaids ‘striped mantles’ (fo. 300r) – are marked as Scottish too, which indicates that they are of Gaelic origin.

4. Concluding Remarks This paper has endeavoured to provide a descriptive analysis of the regional material contained in White Kennett’s unpublished Etymological Collections of English Words and Provincial Expressions. It has been my concern to show that this document extends our knowledge about regional varieties of the late seventeenth century. The dictionary gives a considerable amount of lexical data from all over England that are specifically ascribed to northern, southern and Midland counties. Some remarks on words used in Scotland, Wales and Ireland are included too. Quantification of the data shows that northern words prevail over southern and Midland ones. Comparison with the data listed by John Ray’s glossaries, well-known specimens of early dialect lexicography, points to the fact that the amount of data that Ray did not list is considerably higher with regard to southern and Midland words (Fig. 8). In this connection, Kennett’s manuscript should be considered an important testimony to the

White Kennett’s Etymological Collections of English Words

330

lexical history of Midland and southern counties, and a valuable document that bridges some of the gaps existing in the historical record of regional varieties of British English. Clearly, Kennett’s early days in Kent and his pastoral work in Midland counties might account for the profuse incorporation of Kentish and Oxfordshire material, for example. 9.2

1.4

0.2

23.8

25.8

100% 80%

32.2

Recorded by others Recorded by Ray

60% 40%

74.7

73.9

Recorded by Kennett

58.5

20% 0% North

Midlands

South

Figure 8. Distribution of the regionalisms recorded per sources (with rounded figures) It is worth highlighting that Kennett partly drew upon the information given by other sources. As indicated, he seems to have relied on printed material that includes both literary and non-literary works, amongst which John Ray’s glossaries, Meriton’s “Clavis” to A Yorkshire Dialogue (1685) and Tusser’s husbandry book (1557) are worth mentioning. This clearly exemplifies that his method of word gathering was different from Ray’s original research. Indeed, the use of quotations as part of the method of semantic illustration not only confirms Kennett’s reliance on a variety of printed works, but also points to the fact that his dictionary is an innovative one in the history of English regional lexicography. Actually, later dictionaries undertaking a similar endeavour looked at Kennett’s data to furnish the historical profile of a considerable number of entries. To name but a couple of them, in Halliwell’s A Dictionary of Archaic and Provincial Words (1847) there are over 300 entries which contain references to Kennett’s work, and in Joseph Wright’s EDD there are over 1600 words which are illustrated by means of Kennett’s data. All in all, Kennett’s unpublished work means an important contribution to the history of English regional vocabularies. The information contained in this dictionary should be taken seriously and be referred to in historical accounts of English regional varieties, as it considerably supplements the data provided by other early modern regional wordlists.

Javier Ruano-García

331

Works Cited Primary Sources Coles, Elisha. 1676. An English Dictionary. London: Printed by Samuel Crouch. Halliwell, James O. 1847. A Dictionary of Archaic and Provincial Words, Obsolete Phrases, Proverbs, and Ancient Customs, from the Fourteenth Century. London: John Russell Smith. Kennett, Bishop White. 1690s. Etymological Collections of English Words and Provincial Expressions. MS Lansd. 1033. Kennett, Bishop White. 1816 (1695). A Glossary to Explain the Original, the Acceptation, and Obsoleteness of Words and Phrases; and to Shew the Rise, Practice, and Acceptation of Customs, Laws and Manners. London: Printed by Richard & Arthur Taylor, Shoe-Lane. And sold by J. & A. Arch, Cornhill. Meriton, George. 1685. “An Alphabetical Clavis, Unfolding the Meaning of All the Yorkshire Words Made Use in the Aforegoing Dialogue”. A Yorkshire Dialogue ed. by George Meriton, 79-113. York: Printed by John White. OED = Simpson, John, ed. 2009 (1989). Oxford English Dictionary. Second edition on CD-ROM (v. 4.0.0.3). Oxford: Oxford University Press. Plot, Robert. 1688. The Natural History of Staffordshire. Oxford: Printed at the Theater. Ray, John. 1674. A Collection of English Words not Generally Used. London: Printed by H. Bruges for Tho. Barrell. Ray, John. 1691 (1674). A Collection of English Words not Generally Used. London: Printed for Christopher Wilkinson. Tusser, Thomas. 1557. Fiue Hundred Good Pointes of Husbandrie. Ed. (with introduction, notes, and glossary) by W. Payne, Esq. & Sidney J. Herrtage, Esq., B.A. London: Published for the English Dialect Society by Trübner & Co., 1878. Wright, Joseph. 1981 (1898–1905). English Dialect Dictionary, 6 Vols. Oxford: Oxford University Press.

332

White Kennett’s Etymological Collections of English Words

Secondary Sources Ellis, Henry. 1819. A Catalogue of the Lansdowne Manuscripts in the British Museum with Indexes of Persons, Places and Matters. London: British Museum. Fox, Adam. 2000. Oral and Literate Culture in England 1500–1700. Oxford: Oxford University Press. Griffiths, Bill. 2005. A Dictionary of the North-East Dialect. Newcastle upon Tyne: Northumbria University Press. Halliwell, James O. 1847. A Dictionary of Archaic and Provincial Words, Obsolete Phrases, Proverbs, and Ancient Customs, from the Fourteenth Century. 2 Vols. London: John Russell Smith. Ihalainen, Ossi. 1994. “The Dialects of England since 1776”. The Cambridge History of the English Language Vol. 5: English in Britain and Overseas. Origins and Development ed. by Robert Burchfield, 197274. Cambridge: Cambridge University Press. Marckwardt, Albert H. 1947. “An Unnoted Source of English Dialect Vocabulary”. Journal of English and Germanic Philology 46.177-182. Orton, Harold, Stewart Sanderson & John Widdowson. 1962–1971. Survey of English Dialects (B). The Basic Material, 4 vols. Leeds: A. J. Arnold & Son. Ruano-García, Javier. 2009. “Looking for Regional Words in Late Seventeenth-Century England: Bishop White Kennett’s and his Glossary to Parochial Antiquities (1695)”. Sederi 19.151-173. —. 2010a. “I’ll tell o how Gilbert Scott sowd is mere Berry: ‘A Lancashire Tale’ as a Source for Lancashire Speech in the Late Seventeenth- and Early Eighteenth-Century”. Proceedings of Methods XIII: Papers from the Thirteenth International Conference on Methods in Dialectology ed. by Barry Heselwood & Clive Upton, 53-66. Frankfurt am Main: Peter Lang. —. 2010b. Early Modern Northern English Lexis: A Literary CorpusBased Study. (Linguistic Insights, 105) Bern: Peter Lang. Shorrocks, Graham. 2001. “The Dialectology of English in the British Isles”. An International Handbook on the Evolution of the Study of Language from the Beginnings to the Present ed. by Sylvain Auroux, E. F. K. Koerner, Hans-Josef Niederehe & Kees Vesteegh, 1553-1562. Berlin & New York: Mouton. Skeat, Walter W. 1879. “Dialectal Words from ‘Kennett’s Parochial Antiquities’”. Reprinted Glossaries XVIII-XXII ed. by the English Dialect Society, 1-22. London: Published for the English Dialect Society by Trübner & Co.

CONTRIBUTORS

John Anderson, emeritus professor of English Language, University of Edinburgh, Scotland ([email protected]) His research interests are in grammmatical theory and English and its history. He has contributed to the development of dependency syntax, phonology, and morphology, of localist approaches to case, and of notional syntax. Recent books include Modern Grammars of Case, The Grammar of Names, and The Substance of Language (all OUP). His most recent publication is “Parasitic passives of intransitives in English”, Token: A Journal of English Linguistics 1. He has also (co-)edited various volumes, and with Colin Ewen founded the journal Phonology, as well as being for a time co-editor of Lingua. Artur Bartnik, assistant professor at the Department of English Historical and Applied Linguistics, John Paul II Catholic University of Lublin, Poland ([email protected]) His research interests include theoretical and diachronic syntax with special focus on the structure of the noun phrase, quantifiers, relative clauses and resumption. He is author of Noun Phrase Structure in Old English, Lublin: KUL, 2011; “Discontinuous quantificational structures in Old English” in English Historical Linguistics 2008: The history of English verbal and nominal constructions ed. by Ursula Lenker, Judith Huber & Robert Mailhammer, 185-196. Amsterdam: Benjamins. He has also published several articles on syntactic aspects of resumption and relative clauses. Anna Cichosz, lecturer at the Department of English and Applied Linguistics, University of àódĨ, Poland ([email protected]) Her research interests focus on the history of English and other Germanic languages, especially on the syntax of Old English and Old High German. In her Ph.D. thesis she analyzed the word order of these two languages in different text types (The Influence of Text Type on Word Order of Old Germanic Languages: A Corpus-Based Contrastive Study of Old English and Old High German. Frankfurt am Main: Peter Lang, 2010). At the moment she is involved in the project “The influence of

334

Contributors

Latin on word order of selected Old English and Old High German translations”. Octav Eugen DeLazero, PhD in Linguistics at Cornell University, Ithaca, NY, USA ([email protected]) His scholarly interests lie mainly in the areas of historical linguistics (Indo-European, particularly Germanic, Slavic, and Baltic), syntax, and semantics: argument structure, verbal aspect, and the morphology/syntaxsemantics interface in adjectives and nouns. One of his recent publications is “On the semantics of modal adjectives”, University of Pennsylvania Working Papers in Linguistics 17.1, Article 11 (2011), available at: http://repository.upenn.edu/pwpl/vol17/iss1/11 Irén HegedĦs, associate professor at the Department of English Linguistics, University of Pécs, Hungary ([email protected]) Her research interest focuses on historical comparative linguistics and linguistic reconstruction. She has published more than 50 articles, the most recent one is “English historical linguistics in East-Central and Eastern Europe” in Historical Linguistics of English. An International Handbook, Volume II., ed. by Alexander Bergs & Laurel Brinton, 1375-1396. [HSK 34.2] Berlin: Mouton de Gruyter. She has co-edited 4 volumes, the latest one is English Historical Linguistics 2010. Selected Papers from the 16th International Conference on English Historical Linguistics, Pécs, 23-27 August 2010. [CILT 325] Amsterdam: Benjamins (co-editor: Alexandra Fodor). Lucía Loureiro-Porto, lecturer in English Linguistics at the University of the Balearic Islands, Palma de Mallorca, Spain ([email protected], http://www.usc-vlcg.es/LLP.htm) Her main research interests are the study of grammaticalization processes in the history of English, English historical syntax and semantics, and sociolinguistic variation from both a synchronic and a diachronic perspective. Her monograph The Semantic Predecessors of need in the History of English (c750–1710), Wiley-Blackwell, 2009, received the ESSE 2010 Book Award in the category ‘English Language and Linguistics’. Her most recent article is “On the relationship between subjectification, grammaticalization and constructions. Evidence from the history of English”, Journal of Historical Pragmatics 13.2.

Periphrasis, Replacement and Renewal: Studies in English Historical Linguistics

335

Brian Lowrey, Maître de conférences at the Université de Picardie Jules Verne, Amiens, France ([email protected]) He teaches historical linguistics and the history of English. His principal research interests include causative and perception verb constructions, and complex complement structures, as well as language change and grammaticalization from a more general perspective. His most recent publication is “Early English causative constructions and the second agent factor” in Outposts of Historical Corpus Linguistics: From the Helsinki Corpus to a Proliferation of Resources ed. by J. Tyrkkö, M. Kilpiö, T. Nevalainen & M. Rissanen. [Studies in Variation, Contacts and Change in English 10, http://www.helsinki.fi/varieng/journal/volumes/10/] Lilo Moessner, professor emerita of English Historical Linguistics and Medieval Studies at RWTH Aachen University, Germany (moessner@ anglistik.rwth-aachen.de) Her research interests include various aspects of synchronic and diachronic English grammar (e.g. relative clauses, passive constructions, the indicative/subjunctive dichotomy), regional and functional varieties, translation studies, and lexicogrammatical properties of specialized discourse. Her approach is empirical, mostly based on corpus data, and she combines quantitative and qualitative analyses. Her recent publications cover the following topics: Early Modern English standardization processes, Middle English mandative constructions, and historical discourse analysis. She is the author of the textbook Diachronic English Linguistics. An Introduction (Tübingen: Narr, 2003). Fuyo Osawa, professor at the Department of English, Hosei University, Tokyo, Japan ([email protected]) Her main academic interest is explaining syntactic changes, mainly in English within the framework of generative grammar. Her work has appeared in the proceedings of various international conferences, periodicals, and books, and her publications include “Syntactic parallels between ontogeny and phylogeny”, Lingua 113:1.3-47, “The emergence of DP from a perspective of ontogeny and phylogeny” in Nominal Determination ed. by E. Stark, E. Leiss & W. Abraham, 311-337 (Benjamins, 2007), “The emergence of DP in the history of English: the role of the mysterious genitive” in Historical Linguistics 2007 ed. by M. DuFresne et al., 135-147 (Amsterdam: Benjamins 2009).

336

Contributors

Víctor Parra-Guinaldo, adjunct faculty member in the Department of English, Arizona State University, Phoenix, Arizona, USA ([email protected]) He teaches courses in sociolinguistics, introductory linguistics, and composition. His scholarly interest lies in the area of diachronic linguistics, with a special interest in morpho-syntactic changes in the history of English. His article “Typological overview of comparatives of inequality with an emphasis on Old English more than expressions” has appeared in Anuario de Estudios Filológicos (http://dialnet.unirioja.es/ servlet/articulo?codigo=3850008). In addition, his work on Romance Historical Linguistics has led to The Diminutives Project, which explores the relexification of diminutives as a process of lexical renewal. Dóra PĘdör, associate professor at the Department of English Linguistics, Károli Gáspár University of the Reformed Church in Hungary, Budapest, Hungary ([email protected]) Her research interests include the history of the English and the Irish languages, varieties of English, terminology, and lexicography. Her most recent article is “A skóciai angol megjelenése az Akadémiai Kiadó angolmagyar kéziszótáraiban” [The representation of Scottish English in the English-Hungarian concise dictionaries of Akadémiai Publishers] in Szavak pásztora [A Shepherd of Words] (Szeged: Grimm Publishers, 2012), the bilingual Festschrift for Tamás Magay, the outstanding lexicographer, which she co-edited with Katalin P. Márkus and Tibor Pintér. Paula Rodríguez-Puente, assistant lecturer at the Department of Philology, University of Cantabria, Santander, Spain (paula.rodriguez@ unican.es) She is a member of the Research Group Variation, Linguistic Change and Grammaticalization (University of Santiago de Compostela, http://www.usc-vlcg.es/PRP.htm). Her research interests include historical linguistics and sociolinguistics and grammaticalization processes. Her most recent publication is “Talking ‘private’ with phrasal verbs: a corpusbased study of the use of phrasal verbs in diaries, journals and private letters” in Outposts of Historical Corpus Linguistics: From the Helsinki Corpus to a Proliferation of Resources ed. by J. Tyrkkö, M. Kilpiö, T. Nevalainen & M. Rissanen. [Studies in Variation, Contacts and Change in English 10, http://www.helsinki.fi/varieng/journal/volumes/10/]

Periphrasis, Replacement and Renewal: Studies in English Historical Linguistics

337

Justyna Rogos, assistant professor in the Department of the History of English at the Faculty of English, Adam Mickiewicz University, PoznaĔ, Poland ([email protected], http://wa.amu.edu.pl/wa/rogos_justyna) Her research interests concern the interactions between paleographic and orthographic features of late medieval English manuscripts and their implications for the study of linguistic variability in the Middle English period. Javier Ruano-García, lecturer of English at the University of Salamanca, Salamanca, Spain ([email protected]) He received his PhD (with Extraordinary Award) from the University of Salamanca in 2008. His main research interests include English diachronic dialectology, lexical variation, dialect lexicography, corpus linguistics, literary dialects and dialect literature. He has published articles and book chapters on these fields and the monograph Early Modern Northern English Lexis: A Literary Corpus-Based Study (Bern: Peter Lang, 2010). He is also a member of The Salamanca Corpus project (http://salamancacorpus.usal.es/SC/Team.html) headed by Dr. María F. García-Bermejo Giner. Lynn D. Sims, assistant professor in the Department of Languages and Literature at Austin Peay State University, Clarksville, Tennessee, USA (simsl@ apsu.edu) In addition to teaching linguistics courses, she directs the Linguistics Minor. Her research interests include the history of early Germanic languages, morpho-syntactic changes in the English language, aspectual constructions, and grammaticalization and language change from a generative perspective. Her previous publication is “Do the verbmovement patterns in the Peterborough Chronicle reflect the influence of a northern dialect?” in The Language of the Peterborough Chronicle ed. by Alexander Bergs & Janne Skaffari, 111-134. Frankfurt am Main: Peter Lang, 2007. Daisuke Suzuki, Global Centers of Excellence Researcher at the Graduate School of Letters, Kyoto University, Kyoto, Japan (suzuki0213 @gmail.com) His research interests include modal adverbs, corpus linguistics, and the history of the English language. His forthcoming publications include “A corpus-based study of modal adverbs in English from the viewpoint of grammaticalization”, Token: A Journal of English Linguistics 1, and “A functional approach to the modal adverbs no doubt, doubtless, and

338

Contributors

undoubtedly” in Thinking Modality: English and Contrastive Studies on Modality ed. by Juan Rafael Zamorano-Mansilla, Carmen Maíz, Elena Domínguez & Ma Victoria Martín de la Rosa, Newcastle upon Tyne: Cambridge Scholars Publishing. Fumiko Yoshikawa, professor at the Faculty of Commercial Sciences, Hiroshima Shudo University, Japan ([email protected]) She started to study English historical linguistics at Tsuda College, Tokyo and received a doctorate from Osaka University in 2000. Over the past decade, most of her research has been on the pragmatic analysis of Middle English religious prose. One of her more recent publications is “Politeness strategies in Late Middle English women’s mystical writing” in English Historical Linguistics 2008, Vol. 2: Words, Texts and Genres, ed. by Hans Sauer & Gaby Waxenberger, pp. 209-222. Amsterdam: Benjamins, 2012.

INDEX

A adjectival sentences 180 adjectival modification 204 adverb 7, 155, 212, 244, 245, 321 intensive ~ 256 locative ~ 272 no sooner 7, 240 adverbial 6-7, 126, 134, 249, 251 ~ of place 8, 268, 269, 270, 292 ~ of time 292, 304 locative ~ 271, 273-279, 282, 283 adverbial clause/phrase 21, 251, 270, 271, 272, 274, 279, 293, 300 adverbial test (in/for) 249, 257 agentivity 4, 105, 113, 115, 117, 118, 119, 120 ~ of the causee 4, 112, 114 ~ of the causer 4, 112, 113, 114 agreement 66, 153, 179 Aktionsart 7, 57, 243, 244, 246, 248-251, 252, 255, 259 American English 76 atelic 249 See also telic auxiliary 59, 61, 66-68, 118 modal ~ 34 subject and ~ inversion 228

B Borer–Chomsky conjecture 154 borrowing ~ from Latin 192, 194, 260 ~ from Old Norse 192

C calque syntactic ~ 5, 130, 144, 147, 148

case ~ hierarchy 5, 181-182 change of ~ system 176 Exceptional Case Marking 174, 176, 177 Old English ~ system 171-174 separation of ~ from thematic roles 5, 177 causative 4, 100, 102, 119-120 analytical ~ in Middle English 92, 93, 117, 119 ~ aspect 259, 250, 251, 252, 254 ~ complements in Middle English 94-96, 112-113, 116 ~ complements with/without to 101, 104, 105, 106 ~ do 4, 94, 96, 97, 99, 100, 113-115, 118-119, 120 ~ gar 4, 94, 96, 97, 99, 100, 113, 114, 115, 118, 120 ~ let 4, 94, 96, 97, 98, 99, 103, 118, 119, 120 ~ make 4, 93, 94, 96, 98, 99, 100, 103, 113, 114, 115-119, 120 ~ verbs in Middle English 93-94, 97-99, 104, 106, 112-113, 116 dialect variation of ~ 93, 96, 99, 107-112, 113-118 semantic characteristics of ~ 97, 98 clause complement ~ 100 conditional ~ 293, 298, 300 infinitive ~ 100, 103 non-finite ~ 134, 137 no sooner ~ 7 ‘small clause’ complement 115, 116 subjectless complement ~ 98

340 subordinate ~ 4, 14, 27, 134 V-1 (verb-first) clauses 133135, 137, 139-146, 148 V-1 (verb-first) clause in OE declarative ~ 126, 129, 135-136, 140, 144, 145, 148 V-2 (verb-second) clauses 128 See also adverbial clause, adverbial phrase, complementizer phrase complementizer 162, 164 ~ phrase 154-155 Complementizer Phrase cycle (= CP cycle) 5, 157 conditional See clause conjunction 159, 161-163, 239-240 conversion 45, 252 cycle grammaticalization ~ 163 ~ of low frequency and productivity 38 ~ of aspectual loss and renewal 59

Index ~ particles 292, 304 non-narrative ~ 268 discourse strategy locative ~ 268, 269, 274, 275, 277-279, 280, 282 participant/topic-oriented ~ 268, 278, 280 temporal ~ 268, 269, 277, 278, 280, 282 disjunctive question 158, 162, 163 dynamic model of variation 105-106

E Early Modern English ~ dialect vocabulary 310-330 ~ ingressives 72-73, 75, 79 ~ texts 286, 287 ~ travelogues 268, 274 no sooner … than/but in ~ 229231 Eastern dialect ~ in the 18th century 312, 323 economy principles 5, 21, 152, 156-157, 163

D desemanticization, See semantic bleaching dialect variation 67, 70, 75 ~ in Chaucer 110, 111, 113, 120 ~ in Gower 109, 113, 114, 119, 120 ~ in Langland 96, 99, 112, 113, 114, 115, 118 ~ in Richard Rolle 107, 113 ~ of causatives 93, 96, 97, 99, 107-112, 113-118 ~ of vocabulary 311-312, 319329 See also Eastern dialect, Kentish, Mercian, Midland dialect, Northern dialect, Southern dialect, West Saxon discourse ~-based approach to OHG syntax 128-129, 145, 146 ~ models 288-289, 304-306

F factor analysis 291-292, 304 French 18, 192, 193, 200, 270, 315 frequency 35, 37-41, 46-53, 68, 74, 77, 79, 237, ~ of adverbial variants (before) 216 ~ of features and factor analysis 291 ~ of inversion 235-236 ~ of past perfect with no sooner 236 ~ of pronouns 287 ~ of the V-1/V-2 pattern 132133, 135, 148, 164 See also parsability, productivity

G German 192, 200

Periphrasis, Replacement and Renewal: Studies in English Historical Linguistics Germanic 59, 60, 74, 125-126, 128-130, 147, 148, 158, 164, 195, 259 See also Proto-Germanic, West Germanic Gothic 60, 195, 197-198 grammaticalization 58-59, 64, 6668, 78, 152, 155-157, 229, 238 adverb > conjunction 240 ~ cycle 163 ~ of have 2 ~ of question markers 164 unidirectionality of ~ 156, 163 Greek 14, 158, 193-199

H hapax legomena 36, 41-44, 47-48, 50, 52, 53 ~ and neologisms 42, 52

I imperfect 17, 21 imperfective 22, 249 impersonal constructions 65-66, 78, 127, 167, 168, 172, 175, 178 infinitive (= bare infinitive) ~ in causatives 96, 98, 100, 103, 104, 115 ~ with ingressives 60, 61, 62, 63, 64, 67, 68, 69, 70, 79 variation with to-infinitive 100 See also to-infinitive infinitive complement, See infinitive, to-infinitive ingressive ~ aspect 58-59, 61, 62, 64, 66, 68, 69, 73-79, 249 differences between ingressives 60 inversion 26, 30, 147, 228, 235236, 240

K Kentish 312, 327-328, 330

341

L Latin 18, 20-23, 60, 158, 193, 195196, 198, 199, 260, 311, 315, 318 borrowing from ~ 192, 194, 260 ~-based verbs 259, 262 ~ influence on word order in Old English 125, 127, 128, 132, 136-140, 142-148 ~ original text 63, 125, 129, 130-131 ~ perfect passive 17 lexicalization 244 loss cycle of aspectual ~ and renewal 59 ~ of causative do and let 119, 120 ~ of causative gar 120 ~ of impersonal constructions 167, 176 ~ impersonal passives 176 ~ of lexical case 167 ~ of semantic features 157

M mental lexicon 3, 35, 36, 46-47, 52 activation of a morpheme in the ~ 51, 54 storing complex words in the ~ 37-38 Mercian 63-64 Middle English case assignment in ~ 172, 176 ingressives in ~ 59, 67-73, 79 periphrasis in ~ 14 phonological coalescence in ~ 168 phrasal verbs 247, 259 spelling in ~ 83-84, 89 See also Midland dialect, Northern dialect, Southern dialect Midland dialect 224 East ~ in Middle English 223

342 ~ in the 18th century 318, 319, 322-326, 327, 329-330 South-East ~ in Middle English 109-111 South-West ~ in Middle English 112, 113 West ~ in Middle English 67, 93, 96, 97, 99, 223 Minimalist Program (= MP) 5, 60, 152, 153-154, 167, 178 modal adjectives denotation of ~ 204-207 ~ as a class 187-188 ~ in Greek 195 ~ with entity nouns 189, 190, 193, 196, 197, 199, 201, 202, 208 ~ with situation nouns 189, 200, 202, 208 origin of ~ in English 192, 194 multidimensional analysis 287, 291

N neologism 38, 41, 43, 52 nominal subject 141, 144 See also pronominal subject Northern dialect ~ in Old English 63-64 ~ in Middle English 67, 70, 96, 97, 99, 107, 108, 118 ~ in the 18th century 312, 316, 318, 319-322, 323, 325, 326, 329

O Old (Church) Slavonic 60, 195, 197-198 Old English ingressives in ~ 58-60 lexical-thematic nature of ~ 174-176 ~ case system 171-174, 180182 ~ subjunctive 162 ~ syntax 98, 125-149, See also clause, syntactic variation

Index productivity of morphemes in ~ 35-36, 42-44, 47, 54 transitive and intransitive verbs in ~ 168-169 whether in ~ 158-163 word order in ~ 126, 132, 148, 159, 160, 162, 163, 164 See also Latin, Mercian, modal adjectives, Northern dialect, prefix(es), syntactic, West Saxon Old High German 60 ~ syntax 125-149 ~ word order 129, 132, 135, 137, 139, 147, 148 Old Norse 193, 194, 196, 198 borrowing from ~ 192

P paradigm analytic ~ 24-25 multidimensional ~ 25 ~ symmetry 17-18 unidimensional ~ 23 parsability 37 ~ and frequency 39, 46, 51-53 ~ and productivity 40 particle discourse ~ 292, 304 emphatic ~ 256, 258, 259, 261, 262 interrogative ~ 161-162, 163 ~ as colloquializer 258, 259, 261 ~ expressing aspect 245-246, 248-252, 255, 257 ~ functioning as a suffix 254 ~ up 244, 245, 248, 251, 255, 256, 258, 261, 262 post-verbal ~ 243 See also phrasal verbs passive 3, 17, 24, 25, 101, 169, 171, 178, 293, 300, 301, 302 ~ deverbal adjectives 196, 198 indirect ~ 175-176 impersonal ~ 175-176, 178 Latin perfect ~ 3, 17, 21, 23-25

Periphrasis, Replacement and Renewal: Studies in English Historical Linguistics perfect 23-24, 29, 57, 295 past ~ 228, 236, 237 ~ passive in Latin 17, 21, 23 present ~ 237 See also perfective perfective 61, 62, 68, 71, 78, 249, 251, 293 ~ in Latin 21, 22 periphrasis categorial ~ 18, 19 definition of ~ 15-16 lexical ~ 16 subjunctive ~ 14-15, 18, 26, 28, 31 periphrastic aspect expressed periphrastically 16, 57-58 definition of ~ 15 ~ construction 57 ~ construction with start 74 ~ do 31 ~ replacement 2 ~ use of -ginnan 61, 63 phrasal verbs adjective-derived ~ 248, 252-254 noun-derived ~ 248, 252-254 phrasal-verb particles 248-249, 258, 262 semantic groups of ~ 243-247 prefix(es) be- 45, 64 decay of Old English prefixes 168 derivation by prefixes 45, 5960, 168 ge- 45-46, 168 on- 60-61, 64, 67 transitivity expressed by ~ 5 prefixed verbs 59-60 replaced by phrasal verbs 261 preterite-present verb 34, 51 Principles and Parameters Model (= P&P Model) 153-154 productivity 35-36, 37-38, 43, 5254, 167

343

factors influencing morpheme ~ 39-40, 46-51 measuring ~ 41-42 ~ and hapax legomena 43-44 ~ and neologisms 52 ~ of morphemes in Old English 35-36, 42-44, 47, 54 ~ of transitive constructions 167 progressive 3, 21, 22, 23, 24, 25, 57, 71 pronominal subject 135, 139, 141, 144, 145, 146, 147 See also nominal subject pronoun case features of personal ~ 178180 interrogative ~ 157, 159-161, 163 personal ~ 287, 293, 294, 295, 296, 297 3rd pers. sing. fem., Midland form 326 relative ~ 297 wh-pronouns 292-293 See also pronominal subject Proto-Germanic 125

R register 43, 257, 286-287 diachronic ~ comparison 287, 288 hypothesis of the rise of separate registers 298 impersonal style 300 narrativity in science ~ 296 replacement ~ and productivity 34-54 ~ failure due to CP cycle interruption 157 ~ of agentive structures by nonagentive forms 119 ~ of begin by start 77 ~ of but by than 230 ~ of hortative subjunctive by let construction 2 ~ of neod by þearf 34-54

Index

344 ~ of niman by take 239 ~ of no sooner by as soon as 240 ~ of optative subjunctive by may periphrasis 2 ~ of prefixed verbs by phrasal verbs 261 ~ of sell by give 239 ~ preceded by synchronic variation 2

S semantic ~-based case system 175, 177 ~ categories/features of phrasal verbs 243-248, 255-259, 261 ~ change in no sooner 238-240 ~ characteristics of causatives 98 ~ emptiness of auxiliaries 67 ~ emptiness of used 26 ~ features of modal adjectives 187, 192-193, 196, 202 ~ restrictions on begin 78-79 ~ roles 65, 169, 173-175, 177178, 183, 269 (fn.3) ~ transparency 37, 39-40, 46, 51, 52-53, 243 semantic bleaching 26, 30-31, 58, 61, 68, 156 ~ of modals 14-15, 26, 27, 29, 30 Southern dialect ~ in Middle English 67, 97, 98, 119 ~ in the 18th century 312, 318, 326-328, 329, 330 spelling ~ variation in 15th-century manuscripts 83-90 subjunctive hortative ~ 2 non-factive vs. counterfactual ~ 29 Old English ~ 162 optative ~ 2 preterite ~ 28

~ periphrasis 14-15, 18, 26, 28, 31 synthetic ~ 14, 26, 27, 28, 29, 30 subjectless complement clause 98 subjectless construction 173, 175, 198 syntax Old High German ~ 127-130, 147-148 relationship between morphology and ~ 15, 16, 19 ~ and Old English whether 152-164 ~ of subjectless complement clauses 98 See also clause, Old English, word order syntactic ~ calque 5, 130, 144, 147, 148 ~ case assignment 172, 176, 183 ~ change 154, 167, 176 ~ models from Latin 127, 148 ~ processes 153 ~ properties of modal adjectives 193 syntactic variation ~ in Middle English causatives 92-120 ~ in Old English word order 125-149 violation of syntactic rules 133

T telic, telicity 98, 249, 250, 257 See also atelic telicity test 251, 252 to-infinitive forto infinitive 70, 96, 97 ~ in causatives 97, 98, 100, 103, 120 ~ with ingressives 60, 61, 62, 63, 64, 70, 73, 75, 77, 79 See also infinitive transitivity ~ in Old English 5, 103, 169171, 177, 181-182

Periphrasis, Replacement and Renewal: Studies in English Historical Linguistics transparency phonological ~ 37, 39, 40, 46, 51, 52-53 See also semantic transparency

V variation dynamic model of ~ 105-106 free ~ 3, 89, 97, 100, ~ of spelling in 15th-century manuscripts 83-90 See also dialect variation, syntactic variation variational space 4, 92, 105, 106, 113-119, 120 verb combination with particles 7

345

preterite-present ~ 3, 34 relative position of finite ~ 4

W West Germanic 129-130, 148 West Midland, See Midland dialects West Saxon 63 word order 132, 137, 139, 148 ~ in Old English 126, 132, 148, 159, 160, 162, 163, 164 ~ with before(n) in Middle English 221 See also Latin, Old High German

X X-bar theory 154