Studies in the History of the English Language V: Variation and Change in English Grammar and Lexicon: Contemporary Approaches 9783110220339, 9783110220322

Citation preview

Studies in the History of the English Language V

Topics in English Linguistics 68

Editors

Elizabeth Closs Traugott Bernd Kortmann

De Gruyter Mouton

Studies in the History of the English Language V Variation and Change in English Grammar and Lexicon: Contemporary Approaches Edited by

Robert A. Cloutier Anne Marie Hamilton-Brehm William A. Kretzschmar, Jr.

De Gruyter Mouton

ISBN 978-3-11-022032-2 e-ISBN 978-3-11-022033-9 ISSN 1434-3452 Library of Congress Cataloging-in-Publication Data Studies in the history of the English language V : variation and change in English grammar and lexicon : contemporary approaches / edited by Robert A. Cloutier, Anne Marie Hamilton-Brehm, William A. Kretzschmar. p. cm. ⫺ (Topics in English linguistics ; 68) Includes bibliographical references and index. ISBN 978-3-11-022032-2 (hardcover : alk. paper) 1. English language ⫺ History. 2. English language ⫺ Grammar, Historical. I. Cloutier, Robert A., 1979⫺ II. Hamilton-Brehm, Anne Marie, 1970⫺ III. Kretzschmar, William A. IV. Title: Studies in the history of the English language 5. V. Title: Studies in the history of the English language five. VI. Title: Variation and change in English grammar and lexicon : contemporary approaches. PE1075.S885 2010 420.9⫺dc22 2010020072

Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available in the Internet at http://dnb.d-nb.de. ” 2010 Walter de Gruyter GmbH & Co. KG, 10785 Berlin/New York Cover image: Brian Stablyk/Photographer’s Choice RF/Getty Images Typesetting: RoyalStandard, Hong Kong Printing: Hubert & Co. GmbH & Co. KG, Göttingen ⬁ Printed on acid-free paper Printed in Germany www.degruyter.com

Table of Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1

English Grammar Dialogic Contexts as Motivations for Syntactic Change . . . . . . . . . Elizabeth Traugott

11

Commentary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Akiko Nagano Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Elizabeth Closs Traugott

28

Whatever Happened to English Sluicing . . . . . . . . . . . . . . . . . . . . . Joanna Nykiel

37

Commentary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Elizabeth Closs Traugott

60

Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Joanna Nykiel

65

Notion of Direction and Old English Prepositional Phrases. . . . . . . Olga Thomason

67

Commentary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Joanna Nykiel Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Olga Thomason

81

Survival of the Strongest: Strong Verb Inflection from Old to Modern English . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sherrylyn Branchaw Commentary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Markku Filppula and Juhani Klemola Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sherrylyn Branchaw

34

85

87 105 107

Subject Compounding and a Functional Change of the Derivational Su‰x -ing in the History of English . . . . . . . . . . . . . . . . . . . . . . . . 111 Akiko Nagano

vi

Table of Contents

Commentary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Olga Thomason Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Akiko Nagano

132

Bad Ideas in the History of English Usage . . . . . . . . . . . . . . . . . . . Don Chapman

141

Commentary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Stefanie Kuzmack

156

Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Don Chapman

158

136

English Lexicon The State of English Etymology (A Few Personal Observations) . . . Anatoly Liberman

161

Commentary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ann-Marie Svensson

183

Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Anatoly Liberman

186

From Germanic ‘fence’ to ‘urban settlement’: On the Semantic Development of English town. . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ann-Marie Svensson and Ju¨rgen Hering

187

Commentary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Don Chapman Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ann-Marie Svensson

202

Celtic Influence on English: A Re-Evaluation. . . . . . . . . . . . . . . . . Markku Filppula and Juhani Klemola

207

Commentary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Elisabeth Tacho

226

Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Markku Filppula and Juhani Klemola

229

When arı #ven Came to England: Tracing Lexical Re-Structuring by Borrowing in Middle and Early Modern English. A Case Study . . . Elizabeth Tacho

206

231

Table of Contents

Commentary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Emily Runde Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Elisabeth Tacho Reexamining Orthographic Practice in the Auchinleck Manuscript Through Study of Complete Scribal Corpora . . . . . . . . . . . . . . . . . Emily Runde Commentary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sherrylyn Branchaw Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Emily Runde How Medium Shapes Language Development: The Emergence of Quotative Re Online . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Stefanie Kuzmack

vii 257 261

265 288 291

293

Commentary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Anatoly Liberman Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Stefanie Kuzmack

311

Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Subject Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

317 322

314

Introduction The conversation these days about the history of the English language (HEL) has changed from what it used to be. The historical linguistics (viz. internal history) and cultural studies (viz. external history) that have marked traditional research on HEL are alive and well, but they have been improved now by methods from corpus linguistics and sociolinguistics. This collection shows how historical studies of English are increasingly engaged with these contemporary trends in linguistics, and the volume demonstrates how empirical and other methods can bring classical philology fully into the sphere of contemporary linguistics without abandoning its traditional concerns. This volume has two sections, the first on grammar and syntax and the following section on word-based studies. Of course grammar and lexicon cannot be entirely segregated. Both sections highlight the contributions that strong empirical research can make to our knowledge of the development of English grammar, especially as realized in lexical development. And both sections pay serious attention to the frequencies and discourse characteristics with which particular words have been used at di¤erent times. Each essay will be followed immediately by commentary from another of the authors in the cluster of papers, and then the author will have the opportunity for a response to the commentary. In this way the collection will show the kind of discussion currently obtaining in the field, and more specifically in the section of the field in which the pairs of authors find themselves. The essays in this volume thus portray current research in HEL in the sort of conversations that in fact actually characterize the field today. As Anatoly Liberman, known for his classic work on historical etymology, writes in a commentary in this volume on Kuzmack’s essay about development of the word re on the Internet in the last two decades, regarding such new additions to the house we have known as HEL,‘‘Welcome to the housewarming party.’’

English Grammar The first section challenges researchers to examine and re-examine historical developments in English syntax from new perspectives and multiple methods, including quantitative studies. Traditional analyses of historical developments in English have focused on phonological, morphological,

2

Introduction

and syntactic motivations for change. The authors here expand the approach to include consideration of pragmatic and semantic motivations in qualitative and quantitative studies, complementing theoretical approaches rather than competing with them. The evident benefits shown here encourage viewing the historical development of English with a multidisciplinary perspective. In the lead article for the first section, Elizabeth Traugott demonstrates the potential for new syntactic constructions to arise in dialogic contexts by examining diachronic changes in the use of all- and wh- pseudo-clefts. In a review of approaches to motivations for language change, Traugott explains that invited inferencing motivates speakers to exploit languageinternal implicatures, which may become conventionalized. Traugott identifies several linguistic expressions that function dialogically, such as the concessives although and however, which convey dissonance or incompatibility between two eventualities. Tracing the history of all- and whpseudo-clefts, Traugott provides textual evidence that they were initially shaped in English by their use in argument refutation, progressing to non-dialogic contexts after about fifty years in the historical record. Based on the evolution of all- and wh- pseudo-clefts in dialogic contexts, Traugott argues for an interactional approach to the study of language change, and challenges scholars to reconsider oversimplified monologic perspectives and generalized notions of motivation in diachronic syntactic research. In her commentary on Traugott, Akiko Nagano suggests three constructions that would benefit from the interactional approach because they involve speaker evaluation of an utterance or its context, including conversion in retorts, superlative adjectives, and speech-act conditionals. Responding to Nagano’s discussion of conversions, Traugott agrees that some may have arisen in dialogic and dialogual contexts of the type Nagano suggests, and explains that because the histories of individual constructions di¤er, it is important to consider the full range of interactional contexts in which they arose and distinguish dialogic from dialogual contexts. Joanna Nykiel addresses the problem of syntactic, semantic, and pragmatic involvement in English sluicing by examining the evolution of this structure in the language’s history. Sluicing is defined as a surface anaphor with a full underlying structure that goes unpronounced, which Nykiel exemplifies in the title of her article, ‘‘Whatever happened to English sluicing.’’ This structure was initially associated with a full underlying representation, an analysis that has persisted in later work. Through her diachronic approach, Nykiel shows that sluicing is not such a purely syntactic operation. In fact, despite drastic changes in the syntax and

Introduction

3

morphology of English, her data indicate stability in this structure over time, with syntax being statistically much less of a factor than has been assumed, a fact that in turn speaks against an internal divide between surface and deep anaphora. In its place, Nykiel proposes that an anaphor’s features fall out from the number of syntactic and semantic clues it contains that lead to successful resolution. Elizabeth Traugott then comments on two points, what she calls the ‘‘Recency Illusion’’ and the stability of constructions over a millennium, and raises the further research question of whether the discovery of stable variation requires the investigator to posit a prior stage without such variation. Nykiel responds by suggesting the necessity to go beyond individual languages under investigation, when stability or change in the history of a given construction seems surprising, to consider typological similarities and di¤erences. Olga Thomason examines Old English prepositional phrases that denote the general notion of direction, namely ‘to, toward,’ to gain greater insight into the factors that contribute to the representation of this meaning. This particular group of prepositions is peculiar because of the number of prepositions designating this concept (four of which are examined in this study: to#, wiþ, toge#an, and onge#an) and because of the variety of cases most of these prepositions can govern while maintaining a directional reading (up to three). This contrasts with the other, more specific directional notions of ‘into,’ ‘onto,’ and ‘up to,’ which generally have fewer prepositions designating the concept, and which are generally limited to governing accusative case. Thomason explores each of the four prepositions individually and demonstrates just how complicated their semantics can be. She finds that a combination of the original semantics of the preposition, the semantics of the di¤erent cases, and the specific verbs with which these prepositional phrases combine contributes to the variation in this group. Joanna Nykiel’s commentary focuses on the puzzling variation in distribution of cases. Thomason agrees with Nykiel’s suggestion to present a chronological organization and statistical analysis of the data, which anticipated the next step of research, to add a diachronic spin to the primarily synchronic study. Sherrylyn Branchaw examines seventy-eight Modern English verbs that to some extent retain strong verbal inflection to see which factors already evident in Old English may have influenced such an outcome. She focuses on four factors: the number of verbs with the same ablaut pattern (type frequency), the number of occurrences of a particular verb in Old English (token frequency), the shape of the root, and the e‰ciency of the ablaut pattern. She assumes a correlation between a verb’s being fully attested

4

Introduction

(the vocalism of each of its four principal parts is attested) and its frequency and hypothesizes that strong verbs that are not fully attested in Old English, if they survive, will be weak in Modern English. Her results suggest that to varying degrees, all four of the factors investigated influence the outcome of strong verbs. A verb’s ablaut pattern seems to be a very important factor in determining whether it will remain strong or become weak: in almost all cases in Branchaw’s study, verbs with ablaut patterns where the vowels are less distinct mostly shift to the weak category or more rarely select very distinct vowels for the present and preterite. Token frequency also has an e¤ect: more frequent verbs are more likely to remain strong. Type frequency, on the other hand, does not seem to play as great a role – only if the number of verbs with a particular ablaut pattern is extremely high or low is there any sort of predictive power because the organization into classes changed dramatically in early Middle English. Even at these extremes, however, type frequencies are easily overridden by token frequency and ablaut pattern e¤ects. Commentary on Branchaw by Markku Filppula and Juhani Klemola o¤ers the caveats that the nature of the surviving OE textual evidence suggests caution regarding any quantitative generalizations based on the corpus, and that non-standard varieties need to be considered as well as the standard. Branchaw responds especially to the second point, on which she had written elsewhere, and suggests that her practice helps to explain the origin of standard forms. Akiko Nagano challenges the view that modern synthetic compounds, such as city planning and housekeeping, can embody only the verb-object relationship. Nagano traces the diachronic development of -ing compound nouns, showing that Old and Middle English -ing compound nouns allowed a subject-verb relationship, and that some types of modern -ing compound nouns still do. An example from Nagano’s corpus-based research is the compound noun artery hardening, which has the subjectverb interpretation ‘‘arteries harden.’’ Further, Nagano provides evidence that the main function of the derivational su‰x -ing has shifted from naming in Old and Middle English to recategorization, where -ing nominals are event nominals and inherit the argument structure of the base verb, while the naming function has remained the same since Old English and has invariably produced result nominals. Nagano establishes a connection between historical changes in the possibility of subject compounding in -ing and the function of -ing nominalization, asserting that the possibility of subject compounding depends on the function of the nominalization. In her commentary on Nagano, Olga Thomason questions the constitution

Introduction

5

of compounds of the ‘‘subject-compounding’’ type, the validity of the distinction between event and result nominals, and the morphological status of the su‰x -ing. Thus, rather than accept the subject-verb interpretation of fruit-ripening as ‘fruit ripens,’ Thomason explains the construction along nominal lines such as ‘the ripening of fruit,’ where attributive, rather than verbal, semantics apply. Nagano responds by clarifying that her research addresses the derivational su‰x -ing rather than the participial or gerundive -ing, which derives from a di¤erent inflectional su‰x in Old English, arguing that the functional properties cited by Thomason do not impair her claim. In the final paper of this section, Don Chapman considers prescriptive language rules that appear in usage books only once, which he terms ‘‘one-o¤s.’’ Collecting one-o¤s from a number of popular usage guides dating from 1770 to 2007, he shows that one-o¤s constitute a very high percentage of the total number of prescriptions. Chapman categorizes one-o¤s, according to the probable reason for their failure, such as rare constructions which go unnoticed, rules lacking su‰cient justification, and rules having arbitrary justification. He considers why rules with justification that appears to be less arbitrary, such as logic, may also fail. In her commentary on Chapman, Stefanie Kuzmack supports the treatment of usage rules as a form of etiquette for the purpose of explaining them to the public in terms other than truth and falsity. She suggests that a factor worth considering in future research is that the e¤ectiveness of a type of justification may vary with the audience in terms of era and region. Kuzmack notes that while prescriptivism is viewed as a conservative practice, the high percentage of one-o¤s in guides suggests that prescriptivists are actually not conservative. Chapman responds favorably to Kuzmack’s suggestions regarding the quantified aspects of his study, and agrees that the prescriptive tradition will continue to refresh its content in response to new developments, noting that two recent usage manuals not included in his study also have high percentages of one-o¤s. He observes that usage manual editors, rather than being mere conservators of tradition, propose new prescriptions on their own.

English Lexicon The section on word-based studies invites readers to consider alternate approaches to diachronic study of lexical variants based on new discoveries in linguistics and elsewhere and newly available technology. The research

6

Introduction

presented here encourages consultation of research written in languages other than English and draws from both linguistic and non-linguistic fields, such as socioeconomic and geographic history, archaeology, and language-contact theory. In particular, the authors demonstrate the power of computers and the Internet to enhance research and influence the development of English. At the same time, the essays in the section do not abandon traditional lexical research, such as the study of etymology in the lead essay by Anatoly Liberman. In his signature style that combines sharp wit with meticulous scholarship, Liberman raises awareness of both historical problems in etymological science and modern di‰culties in obtaining funding. Tracing the evolution of English etymology from Minsheu’s 1617 dictionary, he illuminates the technological challenges which have limited accuracy and comprehensive investigation of word histories. Pointing out the limitations of etymological works, including the canonical Oxford English Dictionary, and lauding the benefits of computerized storage, Liberman argues strongly for more rigorous research of word histories that have been ignored or inaccurately derived from assumptions passed along from dictionary to dictionary. Crucially, he points out the usefulness of etymological studies printed in languages other than English, which are often missed. AnnMarie Svensson draws attention to the importance of Liberman’s distinction between present-day etymological lexicology and the science of English etymology, noting Liberman’s characterization of the OED as a historical rather than an etymological dictionary. In his response, Liberman notes the disparity between the etymological information found in dictionaries and the rich material in the accumulating corpus of articles and books on Indo-European, Germanic, and English etymology and looks forward to the establishment of an international center for English etymology. Ann Marie Svensson and Ju¨rgen Hering examine the evolution of the word town in English throughout the Middle English period, tracing its development from the meaning ‘fence’ to ‘urban settlement’ in ninety texts between 1100 and 1500. By considering the size, importance, and location (British or non-British) of the localities receiving the designation town, they are able to pinpoint the meaning of the word at various points in time. They attribute the shift in meaning to various social, economic, and political changes evident during the Middle English period, namely the growth in population and importance of places originally designated as town and the restriction of the designation borough, which was previously used for urban settlements, to places with representatives summoned to

Introduction

7

Parliament. In his commentary, Don Chapman aptly observes, ‘‘Who would have suspected that town would be such an interesting word?’’ Chapman then focuses on context, and suggests that its importance suggests a core mechanism of semantic change: not only do words mean as they are used, they change meaning as they are used in di¤erent contexts. Chapman hopes that study of interesting words like town can help us to better understand the mechanisms of semantic change. Svensson responds in agreement, that only context can guide the modern reader to understand the developing shades of meaning in the semantic field. Markku Filppula and Juhani Klemola challenge the traditional claim of Germanic genocide upon the British population, with the consequence that Celtic had little influence on English. As a case study, they apply archaeological, demographic and historical evidence, language-contact theory, and areal evidence from modern dialect research to argue for a Celtic origin for periphrastic do. Archeological and genetic evidence supports a process of acculturation lasting two centuries after the arrival of Germanic tribes, during which linguistic contact influences were highly likely. Filppula and Klemola suggest that during a period of extensive bilingualism, the Britons shifted to English and were assimilated culturally and linguistically into the Anglo-Saxon population. They point out that the dearth of Celtic loanwords in English should not be viewed as evidence against Celtic influence, because it is predicted by language-contact theory. They also list a number of non-Germanic features shared by English and Celtic that are di‰cult to dismiss as coincidental, including periphrastic do, which is characterized by properties which sets English apart from other Germanic languages. Tacho commends Filppula and Klemola on their interdisciplinary approach to the debate on Celtic influence. She raises the possibility that Anglo Norman and French loanwords may have contributed to the increased number of periphrastic do constructions in Middle English. On the other hand, Filppula and Klemola provide evidence that the earliest examples of periphrastic do found in thirteenth century southwestern verse show it was used with native verbs rather than French loans. They further suggest a simple quantitative study calculating the percentage of French loans associated with periphrastic do. Elizabeth Tacho examines the process of the borrowing of the AngloNorman loan word arıven # ‘to come ashore.’ By approaching this word and meaning from both an onomasiological and a semasiological perspective, she traces the lexical and semantic changes during the verb’s transition period from Middle English to Early Modern English, examining a number of written and speech-based text types from the mid-twelfth to

8

Introduction

the fifteenth centuries. Arıven # was borrowed in the course of the thirteenth century, eventually rivaling and replacing synonymous ME le#nden ‘to land.’ After spreading rapidly and extending its meaning beyond ‘to come ashore’ in the first half of the fifteenth century, ME arıven # has remained largely stable henceforth. Emily Runde’s commentary generally approves of what she calls a ‘‘meticulous semasiological study.’’ Runde points out, however, that the paucity of early speech-based documents may call into question Tacho’s assertion that arıven # was used first in more literary texts, and that the views of one contemporary writer, Mannyng, may contradict some of her findings. Tacho responds that, as regards the scarcity of early speech-based texts, we are indeed limited by what survives (‘‘the bad data problem’’), and that the Mannyng issue is one that deserves further study. Focusing on the Auchinleck Manuscript, Emily Runde tackles the problem of scribal intervention in the transmission of medieval manuscripts. She narrows her study on the complete bodies of work of the two most prolific Auchinleck scribes, commonly known as Scribes 1 and 3. She traces the orthography of a number of words in the work of each of these two scribes in order to determine their internal linguistic consistency as well as the extent to which their practices throughout their complete corpora reflect their linguistic profiles in LALME. She examines the consistency with which initial is written, the spelling of words usually used to distinguish Types II and III (Samuels 1989), and the spelling of words that usually used to distinguish various Middle English dialects. Her study shows greater consistency of spelling than is generally attributed to Middle English scribes and fluctuation between Type II and Type III spelling (challenging Samuels’ classification of these scribes as Type II) and suggests a possible northern origin of two texts, namely Sir Tristem and Horn Childe & Maiden Rimnild, due to the prevalence of northern spellings of certain words not present in the Scribe 1’s other texts. Sherrylyn Branchaw’s commentary approves of Runde’s innovation that, instead of looking at di¤erent scribes for the same text, she looks at di¤erent texts prepared by the same scribe. This practice permits her to estimate the overall consistency of a single scribe’s habits so that, for instance, it is possible to say that Scribe 1 was unlikely to introduce northern forms in one text when he had not introduced them in twenty-eight other texts. Runde suggests in reply that she intends to extend her practice of using complete scribal corpora to additional sets of manuscripts. Stefanie Kuzmack considers the potential of the Internet as an environment for language change using the example of the quotative complemen-

Introduction

9

tizer re, which evolved c. 1990 in threaded discussions to set up response to earlier discourse. The use of quotative re derives from and is semantically related to the English preposition re ‘about, regarding,’ from Latin in re ( ‘see’/‘self-selection marker’, Diciamo ‘(let’s) say.’ He argues that speakers self-select by using attention-getters in ‘‘illegitimate’’ ways, e.g., Guarda! when there is nothing to look at, or Diciamo when interlocutors

14

Elizabeth Closs Traugott

are not engaged in simultaneous talk. Waltereit and Detges (2007: 79) propose that the kinds of interaction that precede subjectification, pragmaticalization (of discourse markers), and grammaticalization, can be specified in more detailed ways than has been usual in the past by appealing to argumentation to a conclusion, and negotiation of viewpoints. They propose that the development of modal particles like French bien ‘indeed’ as in (1a) out of the manner adverb bien ‘well’ derives from ‘‘stereotypical argumentational moves negotiating common ground (‘‘What do I believe that you believe concerning the felicity of my speech act?’’).’’ On the other hand, discourse markers such as Spanish bien ‘well’ as in (1b) arise out of the negotiating strategy of ‘‘further[ing] verbal interaction (‘‘What are we going to do next’’)’’ (Waltereit and Detges 2007: 79). (1) a.

Vous avez bien rec¸u mon message? ‘You did receive my message, didn’t you?’ (Ibid.: 63)

b. A. . . . todo ciudadano . . . tiene derecho a eas legı´tima defensa ‘. . . every citizen . . . has the right to this self-defense’ B.

Bien. Eh . . . creo que. . . ‘Well. Eh . . . I think there. . .’ (Ibid.: 62)

A di¤erent motivation, that of presupposition accommodation, is proposed by Schwenter and Waltereit (forthcoming) to account for such developments as use of additive too as a refutation marker. An early example is: (2) ‘‘Surely you can’t be thinking of marrying a man who wasn’t in the army, who jeered at men who did enlist?’’ ‘‘He was, too, in the army. He was in the army eight months.’’ (1936 Mitchell, Gone with the Wind [Schwenter and Waltereit, forthcoming]) As Waltereit and Detges show, in the case of Spanish bien the interaction is often one of disagreement (see (1b)). The same is true of the use of too in (2). In other words, the context for their use is one in which multiple viewpoints are expressed using a strategy that is contesting and refutational, oriented toward an alternative conclusion. In other words it is ‘‘dialogic.’’ 3. Dialogic interactions For over twenty years there has been considerable discussion, especially in Europe, of the distinction between the number of speakers and the number of points of view invoked (see Roulet 1984, Ducrot 1984, 1996; more

Dialogic Contexts as Motivations for Syntactic Change

15

recently Schwenter 2000, 2007, Nølke 2006).4 A distinction is made between ‘‘monologual’’ – ‘‘dialogual’’ interaction and ‘‘monologic’’ – ‘‘dialogic’’ interaction.5 The first, ‘‘monologual’’ – ‘‘dialogual,’’ refers to the number of speakers (simplistically, one or two) and concerns absence or presence of turn-taking. The second, ‘‘monologic’’ – ‘‘dialogic,’’ refers to number of view points invoked (simplistically, one or two). Monologic orientation concerns the extent to which speakers share common ground and build their argument toward the same or similar conclusions (e.g., and, which signals agreement or addition). Dialogic orientation concerns the extent to which speakers contest, refute, or build an argument toward alternative or di¤erent conclusions (e.g., but, modal in fact). Monologicity and dialogicity are on a continuum (Schwenter 2000) – very little language use is purely monologic (Taavitsainen, Ha¨rma¨, and Korhonen 2006: 1). There are many linguistic expressions that index some degree of dialogicity. Among them are: a)

Adversatives: these ‘‘[signal] a confrontation of ‘incompatible’ viewpoints’’ (Schwenter 2000: 261), e.g., but, Spanish si. b) Concessives: these convey the ‘‘implicature that there is a dissonance or incompatibility between two eventualities’’ (Ko¨nig 1991: 134), e.g., although, however. c) Negation: this has been conceptualized as denying or correcting the ‘‘truth’’ of a prior proposition or utterance (Givo´n 1978), or of a presupposition, implicature, etc. (Geurts 1998). While the extent to which canonical negation is used this way has been challenged (e.g., Tottie 1991, Thompson 1998), non-canonical negatives target a salient a‰rmative proposition in the ongoing discourse record, e.g., not . . . either, OE na . . . wiht ‘no . . . thing’ (> not), Fr. ne . . . pas ‘no . . . step’ (Schwenter 2007) and are refutational. d) Epistemic modal adverbs: these invoke alternative worlds (Lyons 1977) and therefore doubt, e.g., surely, possibly. e) Focus particles: these exclude alternatives and ‘‘carry an implication of dissonance or incompatibility’’ (Ko¨nig 1991: 131; also Traugott 2006), e.g., even, only. f ) Scalars in general since they invoke alternatives (Ko¨nig 1991). 4. Much of this work originates with Bakhtin (see Holquist 1981). Similar issues are also central to much work on stylistics and types of ‘‘indirect speech’’ (e.g., Leech and Short 1981). 5. Note, however, that ‘‘dialogic’’ is sometimes used to refer to dyadic interaction, i.e. what is here called ‘‘dialogual’’ (e.g., Taavitsainen, Ha¨rma¨, and Korhonen 2006).

16

Elizabeth Closs Traugott

While synchronic studies of dialogicity have focused not only on expressions of dialogicity but also on interactional stance, largely in conversation (e.g., Mann and Thompson 1992, Ford 1994, Couper-Kuhlen and Kortmann 2000), most diachronic work has been devoted to the development of expressions indexing dialogicity as instances of grammaticalization or of subjectification. Dialogic expressions typically derive from nondialogic ones, e.g.: but < ‘except’ < butan ‘on the outside’ (Nevalainen 1991), only ‘adversative conjunction’ (‘‘denotes the opposite of the consequence or conclusion expected from the first,’’ Poutsma 1904–05: 385, cited in Brinton 1998) < focus marker < anlic ‘singly’, instead < in stede ‘in place of ’ (Schwenter and Traugott 1995) to name only a few. Relatively little attention has been paid to dialogic contexts for changes. There is, however, some mention in Detges’ and Waltereit’s work, and Schwenter and Traugott argue that the dialogicity of adverbials like epistemic in fact arises out of the semanticization of dialogic contexts such as are illustrated by (3): (3) You were pleased before to make some reflexions on this custom, and laugh at the irresolution of our free-thinkers: but I can aver for matter of fact, that they have often recommended it by their example as well as arguments . . . In whatever light you may consider it, this is in fact a solid benefit. But the best e¤ect of our principles is that light and truth so visibly spread abroad in the world. (1732 Berkeley, Alciphron ii. sect. 24, p. 105 [Schwenter and Traugott 2000: 16]) Note here the prior context of alternative points of view ( you were pleased . . . but I can aver. . . , in whatever light you may consider it), as well as the following one (But the best e¤ect. . .). Here I argue in greater detail for the importance of paying attention to evidence in texts for interactional goals involving contesting of prior claims or introduction of alternative points of view, i.e. of dialogic contexts, in coming to grips with micro-changes with the example of the development of pseudo-clefts. In earlier work (Traugott, 2008) I tested two hypotheses: i) that they might have arisen in primarily dialogual interaction, given that analyses of WH-clefts based on constructed data suggest they are responses to questions (e.g., Higgins 1979 and many studies building on his work), ii) that they might have arisen at turns, given that other studies based on spontaneous conversation suggest they are used to delay an assertion at a turn (e.g., Hopper 2001). I found that the textual evidence gave little support for turn-taking as a motivation for the development of pseudo-clefts. Rather, dialogic contexts appear to have played

Dialogic Contexts as Motivations for Syntactic Change

17

an important role, most especially in the case of ALL-clefts. This is consistent with Kim’s (1995) finding that WH-clefts are used in conversation mainly to signal a counter-active, i.e. dialogic, stance in spontaneous conversation.

4. Contexts for the development of pseudo-clefts Pseudo-clefts are constructions like: (4) a.

What Bruce ate was the crab. (WH-cleft)

b. What Bruce did was (to) peel the potatoes. (WH-cleft) c. All/*Everything Bruce ate was the crab. (ALL-cleft) d. All/*Everything that you have to do is (to) close the window. (ALL-cleft) They involve a string of the type WHAT/ALL – NP – V – BE – X (see Prince 1978, Higgins 1979, Collins 1991, Lambrecht 2001, Delin and Oberlander 2006, among many others) and: Two clauses, one of which is a relative,6 one of which involves a copula. b) Givenness: some part of the construction (typically the relative) must be given or at least recoverable. c) Uniqueness and contrastiveness: the focus constituent is construed as an exhaustive, exclusive listing (Bruce ate only the crab, not the shrimp, squid, etc.). d) Specificational/identifying focus: the complement of the copula is specific and referential (not ascriptive or non-referential) (see Patten 2007 for detailed discussion of specificationality).

a)

Ball (1994) analyzed the history of IT-clefts, and showed that antecedents of what Prince (1978) called ‘‘stressed focus’’ IT-clefts (but without it) are attested in Old English; these require X to be given or at least inferrable and salient in the discourse. What Prince called ‘‘informative-

6. However, den Dikken (2006) argues that they are interrogative. Carlson (1983) argues that in terms of use in dialogue, they ‘‘serve the purpose of articulating a sentence as an answer to a particular question’’ (p. 222) but are structurally free relatives (thanks to Markku Filppula for this reference).

18

Elizabeth Closs Traugott

presupposition’’ IT-clefts arose around 1400; in this type X may be new.7 ALL-pseudo-clefts arose around 1600, and WH-pseudo-cleft around 1660. All three clefts are examples of ‘‘grammaticalization without lexical bleaching’’ (Lehmann, 2008), and of constructionalization (Traugott, 2008). 4.1. Early examples of ALL-pseudo-clefts Early examples in the data base8 with the string ALL – NP – V – BE – X are ascriptive (5a; ‘everything I said was tricky/designed to trick’) or purposive (5b). Here all means ‘everything’: (5) a.

I haue made him happie by training you forth: In a word, all I said was but a traine to draw you from your vow: Nay, there’s no going backe. ‘I have made him happy by drawing you forth: in a word, everything I said was only a trick to draw you from your vow. No, there is no going back.’ (1606 Chapman, Monsieur D’Oliue [LION: EEBO]

b. I loue thee dearer then I doe my life, And all I did, was to aduance thy state, To sunne bright beames of shining happinesse. (1601 Yarrington, Two Lamentable Tragedies [Ibid.]) In (5a) the prior context is not obviously dialogic, but the following context is contesting (Nay, there’s no going back). In (5b) one might infer that the speaker is making such a strong claim because he fears he has been misunderstood. Indeed, everything one person says or does may not be enough for some other person or may be interpreted as mistaken or at best inadequate (due to the quantificational meaning of all ), as is poignantly expressed by Henry V in (6): 7. Interestingly, in Present Day English ALL-pseudo-clefts and WH-clefts in some contexts can be of either type. The pseudo-clefts suggest that at first X was given, but that as the constructions became conventionalized this restriction was relaxed. 8. I searched the Middle English Dictionary, LION: EEBO, LION: Early English Drama (Jacobean and Caroline (1603–1660), and Restoration (1660–1700) periods), and trials as represented by the Old Bailey Proceedings Online from 1678 to 1743, and the Old Bailey Speech Set from 1732–1743. Each file was searched for all/what (that) I/ you/he/she/it/we/they and said/ did was (the latter to determine what subjects were selected in the early period).

Dialogic Contexts as Motivations for Syntactic Change

19

(6) More will I do; Though all that I can do is nothing worth, Since that my penitence comes after all, Imploring pardon. (1599 Shakespeare, Henry V IV.i.319 [UVa]) Ca.1600, strings of type ALL – NP – V – BE – X appear with the pseudocleft meaning: all can be interpreted as ‘only’, not ‘everything’, and the focus may be understood as exhaustive and specificational.9 Since the focus is a clause in all early ALL-cleft examples, with a verb of speaking, usually say (7a), or do plus infinitive marker (7b), the criterion used for contemporary English that the focused NP should be definite does not apply. There are, however, examples like (7a) in which the focus of an ALL-cleft with say is this, followed by a clause. (7) a.

[A ‘‘confutation’’ between a Jesuit (S.R.) and Bell] Our slanderous and rayling Iesuite, reporteth my wordes in this manner; for saith Bell ) (sic), it is a thinge proper to God, to make something of nothing in al cases, and at al times. So then, all that I said was this; (viz) That though man can at sometime in some cases, make one thing of another; yet to make of nothing something, is proper to GOD alone, neither is man able to performe the same. (1608 Bell, The Jesuits Antepast [LION: EEBO]) b. I was desir’d to put a stop to the Sedition of the People. I answered, That all that I could do, was to give no Encouragement to it, but God only could appease it. (1693 Du Pin, History of Ecclesiastical Writers [Ibid.])

Both examples are highly dialogic. In (7a) Bell draws attention to what he actually said (though a man can at sometime. . .), as opposed to what others construed him as saying (saith Bell: it is a thinge proper to God. . .). In (7b) Naylor foresees the impossibility of his addressee having any political success, and proposes that he simply do good. Note that whereas the purposive construction (5b) (all I did was for the purpose of Xing in the future) is future-oriented, the ALL-cleft with do is present-oriented (or, more specifically, is oriented to the event time). This is presumably what allowed for the loss of to after do as in (8a) and the verbal gerund in (8b): 9. An anonymous reviewer asked whether there might have been influence of French or Latin texts. Some of the earlier examples of ALL-clefts are in translations from French, but whether or not there was direct influence remains to be studied. Examples (7b) and (8b) are translations from French, but relatively late in the development of ALL-clefts.

20

Elizabeth Closs Traugott

(8) a.

When any bow’d to me with Congees (¼ ceremonious bow) trim, All I could do, was stand and laugh at him. (1681 Baxter, Poetical fragments [LION: EEBO]) b. These words so resolute and kind, pierced my very heart, and turned me into a Statue, leaving me without sense or motion. All I could do, was embracing my dear Sultaness for a final Adieu. (1686 Bre´mond, The Happy Slave, Part III [Ibid.])

(8) unequivocally shows that a new construction had come into being. Not only does do not require to, but the context is no longer dialogic. The construction in itself signals dialogicity. It puts the focus on a scale and signals that it is the only alternative; it also signals that the speaker/writer regards the focus as less than adequate (all is ‘‘downward inferential’’ in Horn’s (1996: 18) analysis), a meaning derived from interaction of the quantifier all with exclusivity and negation (see also albeit, all the same, after all ). We may say that dialogicity has been semanticized into the construction. This means it is understood as dialogic in non-dialogic contexts, though such contexts tend to continue to be used. 4.2. Early examples of WH-clefts As in the case of the ALL-constructions, prior to about 1660 the only examples in data base with the structure WHAT – NP – V – BE – X are ascriptive (9a), or purposive (9b): (9) a.

Then Sostratus taking the occasion to speake, said: ‘‘what I did was of no great valour, and therfore not worthy the rehearsal’’ (1597 Tatius, Clitiphon [LION: EEBO]) b. Mistake mee not faire Knight, . . . what I did, was to deceiue the Pagans, who are waking Dragons that neuer sleepe about mee (1612 Markham, Meruine [Ibid.])

These examples have dialogic contexts, but not the specificational structure of WH-pseudo-clefts, i.e. X is not a definite description. It was only a short step to the pseudo-cleft construction, in which the dialogic contexts were retained. As in the case of ALL-clefts, WH-clefts with do are oriented to event-time, not future (10b). (10) a.

I write not out of a designe to advance the repute of our WestIndy Commodities in the making Chocolata. What I say is the Assertion of others, who did not intend by their Writings to serve the English Interest in Iamaica. (1662 Stubbe, Indian Nectar [LION: EEBO])

Dialogic Contexts as Motivations for Syntactic Change

21

b. If it be objected that I preached to separate Congregations; my Answer is, That I preach’d only to some of many Thousands that cannot come into the Temples, many of which never heard a Sermon of many years. And what I did, was only to preach to such as could not come to our Churches. (1697 Baxter, Mr. Richard Baxter’s Last Legacy [LION: EEBO]) Like ALL-clefts, WH-clefts came within about fifty years to be used in non-dialogic contexts. Here again we can say that the dialogic context has been semanticized in the new construction. However, in the case of WH-clefts there is no downward entailment:10 (11) I heard a Noise, and came down Stairs, but all the Things were gone: I wash Linnen, and what I lost was the Property of Mr. Gold. (May 1736, Trial of Christopher Freeman and Samuel Ellard [BAILEY: s17360505-463173605050]) 5. Conclusion and further work I have added to arguments that by bringing an interactional approach to the study of change, and attempting to go beyond very general ideas about motivations, we can reach a better understanding of how specific microchanges come about. In particular I have suggested that dialogic contexts deserve special attention. Among future research questions is whether all expressions that inherently code dialogic meaning, such as those cited in section 3, arise in dialogic contexts. Other questions that deserve attention include how best to refine the continuum from monologic to dialogic contexts and meanings (Schwenter 2000), and further, how best to define the continuum within dialogicity (e.g., from quotation and scalarity to refutation). Since semanticizing dialogicity involves semanticizing stance, a further question is how degrees of subjectification intersect with these continua. And since contesting strategies are partly governed by conventions of interaction in specific discourse contexts, we need also to understand how these factors relate to register and genre and how there may be di¤erences over time due to shifts in cultural norms (Biber 2004). 10. Carlson (1983: 223) points out that What David wants is his wallet ‘‘implies that his wallet is all David wants. This . . . is due to the fact . . . that the character of free relatives is left open between existential and universal force.’’ Here we must understand Carlson to be thinking of the exclusivity rather than downward entailment of all.

22

Elizabeth Closs Traugott

Sources BAILEY

The Proceedings of the Old Bailey, London 1674 to 1834. Edited by Tim Hitchcock and Robert Shoemaker. Used by permission of Tim Hitchcock and Magnus Huber. www.oldbaileyonline.org (accessed March–September 2007).

LION

Chadwyck Healey website, http://lion.chadwyck.com.

UVa

University of Virginia, Electronic Text Center, Modern English Collection, http://etext.lib.virginia.edu/modeng/modeng0.browse.html.

References Ball, Catherine N. 1994 The Origins of the Informative-presupposition It-cleft. Journal of Pragmatics 22: 603–628. Biber, Douglas 2004 Historical Patterns for the Grammatical Marking of Stance. Journal of Historical Pragmatics 5: 107–136. Brinton, Laurel J. 1998 ‘The flowers are lovely; only, they have no scent’: The Evolution of a Pragmatic Marker. In Anglistentag 1997 Giessen, edited by Raimund Borgmeier and Andreas H. Jucker, 9–18. Trier: Wissenschaftlicher Verlag. Butler, Christopher S. 2006 On Functionalism and Formalism: A Reply to Newmeyer. Functions of Language 13: 197–227. Carlson, Lauri 1983 Dialogue Games: An Approach to Discourse Analysis. Dordrecht/ London: Reidel. Chomsky, Noam 1988 Language and Problems of Knowledge: The Managua Lectures. Cambridge, MA: The MIT Press. Clark, Herbert H. 1996 Using Language. Cambridge: Cambridge University Press. Collins, Peter C. 1991 Cleft and Pseudo-cleft Constructions in English. London: Routledge. Couper-Kuhlen, Elizabeth and Bernd Kortmann, eds. 2000 Cause – Condition – Concession – Contrast: Cognitive and Discourse Perspectives. Berlin: Mouton de Gruyter. Croft, William 2000 Explaining Language Change: An Evolutionary Approach. Harlow, Essex: Longman, Pearson Education. Cuyckens, Hubert, Thomas Berg, Rene´ Dirven, Klaus-Uwe Panther, eds.

Dialogic Contexts as Motivations for Syntactic Change 2003

23

Motivation in Language: Studies in Honor of Gu¨nter Radden. Amsterdam: Benjamins. Delin, Judy and Jon Oberlander 2006 Cleft Constructions in Context: Some Suggestions for Research Methodology. http://www.fb10.uni-bremen.de/anglistik/langpro/ projects/gem/delin-publications.html. Den Dikken, Marcel 2006 Specificational Copular Sentences and Pseudoclefts: A Case Study. In The Blackwell Companion to Syntax, edited by Martin Everaert and Henk van Riemsdijk, 62–91. Oxford/Malden, MA: Blackwell. Detges, Ulrich 2006 From Speaker to Subject. The Obligatorization of the Old French Subject Pronouns. In La Linguistique au Coeur. Valence verbale, grammaticalisation et corpus. University of Southern Denmark Studies in Literature, edited by Hanne Leth Andersen, Merete Birkelund, Maj-Britt Mosegaard Hansen, 48: 75–103. Du Bois, John W. 1985 Competing Motivations. In Iconicity in Syntax, edited by John Haiman, 343–365. Amsterdam: Benjamins. Ducrot, Oswald 1984 Le dire et le dit. Paris: Minuit. Ducrot, Oswald 1996 Slovenian Lectures/Confe´rences Slove`nes: Argumentative Semantics/ Se´mantique argumentative. Lubljana: ISH. Fischer, Olga 2007 Morphosyntactic Change: Functional and Formal Perspectives. Oxford: Oxford University Press. Ford, Cecilia E. 1994 Dialogic Aspects of Talk. Text 14: 531–554. Gerritsen, Marinel and Dieter Stein, eds. 1992 Internal and External Factors in Syntactic Change. Berlin: Mouton de Gruyter. Geurts, Bart 1998 The Mechanisms of Denial. Language 74: 274–307. Givo´n, Talmy 1978 Negation in Language: Pragmatics, Function, Ontology. In Syntax and Semantics 9: Pragmatics, edited by Peter Cole, 69–112. New York: Academic Press. Grice, H. Paul 1989 [1975] Logic and Conversation. Studies in the Way of Words, 22–40. Cambridge, MA/London: Harvard University Press. Haiman, John 1980 The Iconicity of Grammar: Isomorphism and Motivation. Language 56: 515–540.

24

Elizabeth Closs Traugott

Haspelmath, Martin 1998 Does Grammaticalization Need Reanalysis? Studies in Language 22: 315–351. Higgins, Francis Roger 1979 The Pseudo-cleft Construction in English. New York: Garland. Holquist, Michael 1981 The Dialogic Imagination: Four Essays by M. M. Bakhtin. Trans. by Caryl Emerson and Michael Holquist. Austin: University of Texas Press. Hopper, Paul J. 2001 Grammatical Constructions and Their Discourse Origins: Prototype or Family Resemblance? In Applied Cognitive Linguistics I: Theory and Language Acquisition, edited by Martin Pu¨tz, Susanne Niemeier, Rene´ Dirven, 109–129. Berlin: Mouton de Gruyter. Hopper, Paul J. and Elizabeth Closs Traugott 2003 [1993] Grammaticalization. Cambridge: Cambridge University Press, 2nd revised ed. Horn, Laurence R. 1996 Exclusive Company: Only and the Dynamics of Vertical Inference. Journal of Semantics 13: 1–40. Joseph, Brian D. 1992 Diachronic Explanation: Putting Speakers Back into the Picture. In Explanation in Historical Linguistics, edited by Garry W. Davis and Gregory K. Iverson, 123–144. Amsterdam: Benjamins. Keller, Rudi 1994 [1990] On Language Change: The Invisible Hand in Language. Translated by Brigitte Nerlich. London: Routledge. Kim, Kyu-hyun 1995 WH-clefts and Left-dislocation in English Conversation: Cases of Topicalization. In Word Order in Discourse, edited by Pamela Downing and Michael Noonan, 247–296. Amsterdam: Benjamins. Kiparsky, Paul 1968 Linguistic Universals and Linguistic Change. In Universals in Linguistic Theory, edited by Emmon Bach and Robert T. Harms, 171–202. New York: Holt, Rinehart and Winston. Ko¨nig, Ekkehard 1991 The Meaning of Focus Particles: A Comparative Approach. London: Routledge. Lambrecht, Knud 2001 A Framework of the Analysis of Cleft-constructions. Linguistics 39: 463–516. Langacker, Ronald W. 1977 Syntactic Reanalysis. In Mechanisms of Syntactic Change, edited by Charles N. Li, 57–139. Austin: University of Texas Press.

Dialogic Contexts as Motivations for Syntactic Change

25

Leech, Geo¤rey N. and Michael H. Short 1981 Style in Fiction: A Linguistic Introduction to Fictional Prose. London: Longman. Lehmann, Christian 1995 [1982] Thoughts on Grammaticalization. Munich: LINCOM EUROPA. Lehmann, Christian 2008 Information Structure and Grammaticalization. In Theoretical and Empirical Issues in Grammaticalization, edited by Elena Seoane and Marı´a Jose´ Lo´pez-Couso, in collaboration with Teresa Fanego, 207–229. Amsterdam: Benjamins. Lightfoot, David 2003 Grammaticalisation: Cause or E¤ect? In Motives for Language Change, edited by Raymond Hickey, 99–123. Cambridge: Cambridge University Press. Lyons, John 1977 Semantics, Vol. 2. Cambridge: Cambridge University Press. McMahon, April 2006 Restructuring Renaissance English. In The Oxford History of English, edited by Lynda Mugglestone, 147–177. Oxford: Oxford University Press. Mann, William C. and Sandra A. Thompson 1992 Relational Discourse Structure: A Comparison of Approaches to Structuring Text by ‘Contrast.’ In Language in Context: Essays for Robert E. Longacre, edited by Shin Ja J. Hwang and William R. Merrifield, 19–45. Dallas: Summer Institute of Linguistics and University of Texas at Arlington. Milroy, James 1992 Linguistic Variation and Change: On the Historical Sociolinguistics of English. Oxford: Blackwell. Nevalainen, Terttu 1991 BUT, ONLY, JUST: Focusing on Adverbial Change in Modern English 1500–1900. Helsinki: Socie´te´ Ne´ophilologique. Nølke, Henning 2006 The Semantics of Polyphony (and the Pragmatics of Realization). Acta Linguistica Hafniensia 38: 137–160. Patten, Amanda 2007 How Specificational are Cleft Sentences? Paper Presented at the Second International Conference on the Linguistics of Contemporary English (ICLCE), Toulouse, July 2nd–4th. Poutsma, Henry 1904–05 A Grammar of Late Modern English, Part I: The Sentence. Groningen: Noordho¤. Prince, Ellen F. 1978 A Comparison of WH-clefts and it-clefts in Discourse. Language 54: 883–906.

26

Elizabeth Closs Traugott

Radden, Gu¨nter and Klaus-Uwe Panther, eds. 2004 Studies in Linguistic Motivation. Berlin: Mouton de Gruyter. Roulet, Eddy 1984 Speech Acts, Discourse Structure, and Pragmatic Connectives. Journal of Pragmatics 8: 31–47. Schwenter, Scott A. 2000 Viewpoints and Polysemy: Linking Adversative and Causal Meanings of Discourse Markers. In Couper-Kuhlen and Kortmann, 257–281. Schwenter, Scott A. 2007 Dialogicity and Negation. Paper presented at Stanford University, May 25th. Schwenter, Scott A. and Elizabeth Closs Traugott 1995 The Semantic and Pragmatic Development of Substitutive Complex Prepositions in English. In Historical Pragmatics, edited by Andreas H. Jucker, 243–273. Amsterdam: Benjamins. Schwenter, Scott A. and Elizabeth Closs Traugott 2000 Invoking Scalarity: The Development of In Fact. Journal of Historical Pragmatics 1: 7–25. Schwenter, Scott A. and Richard Waltereit Forthcoming Presupposition Accommodation and Language Change: From Additivity to Speech-act Marking. In Subjectification, Intersubjectification and Grammaticalization, edited by Hubert Cuyckens, Kristin Davidse, Lieven Vandelanotte. Berlin: Mouton de Gruyter. Slobin, Dan I. 1977 Language Change in Childhood and in History. In Language Learning and Thought, edited by John MacNamara, 185–214. New York: Academic Press. Taavitsainen, Irma, Juhani Ha¨rma¨, Jarmo Korhonen, eds. 2006 Dialogic Language Use. Helsinki: Socie´te´ Ne´ophilologique. Taylor, John R. 2006 Motivation. Proceedings of the Sixth Annual Meeting of the Japanese Cognitive Linguistics Association 6: 486–504. Thompson, Sandra A. 1998 A Discourse Explanation for the Cross-linguistic Di¤erences in the Grammar of Incorporation and Negation. In Case, Typology, and Grammar: In Honor of Barry Blake, edited by Anna Siewierska and Jae Jung Song, 307–340. Amsterdam: Benjamins. Tottie, Gunnel 1991 Negation in Speech and Writing. New York: Academic Press. Traugott, Elizabeth Closs 2006 The Semantic Development of Scalar Focus Markers. In The Handbook of the History of English, edited by Ans van Kemenade and Bettelou Los, 335–359. Oxford/Malden, MA: Blackwell.

Dialogic Contexts as Motivations for Syntactic Change

27

Traugott, Elizabeth Closs 2007 On the Development of ALL-pseudo-clefts in English. 10th International Pragmatics Conference (IPra), Go¨teborg, Sweden, July. Traugott, Elizabeth Closs 2008 ‘‘All that he endeavoured to prove was . . .’’: On the Emergence of Grammatical Constructions in Dialogual and Dialogic Contexts. In Language in Flux: Dialogue Coordination, Language Variation, Change and Evolution, edited by Ruth Kempson and Robin Cooper. London: Kings College Publications. Traugott, Elizabeth Closs and Richard B. Dasher 2002 Regularity in Semantic Change. Cambridge: Cambridge University Press. Traugott, Elizabeth Closs and Ekkehard Ko¨nig 1991 The Semantics-pragmatics of Grammaticalization Revisited. In Approaches to Grammaticalization, edited by Elizabeth Closs Traugott and Bernd Heine, Vol. I: 189–218. Amsterdam: Benjamins. Waltereit, Richard 2006 The Rise of Discourse Markers in Italian: A Specific Type of Language Change. In Approaches to Discourse Markers, edited by Kirsten Fischer, 61–76. Oxford: Elsevier. Waltereit, Richard and Ulrich Detges 2007 Di¤erent Functions, Di¤erent Histories. Modal Particles and Discourse Markers from a Diachronic Point of View. In Special Issue on Discourse Markers, edited by Maria Josep Cuenca. Journal of Catalan Linguistics 6: 61–82.

Commentary on Traugott, Dialogic Contexts as Motivations for Syntactic Change11 Akiko Nagano

Traugott’s paper shows how the shift of viewpoint and the existence of di¤erent viewpoints expressed by a dialogic context can be semantically incorporated into a specific linguistic expression, and how the expression is ‘‘constructionized’’ as a dialogic expression. The contribution of this paper for the study of linguistic change lies not only in proving the significance of interactional factors as vital motivations for change, but also in providing a general format for locating a contextual source of a specific construction. Thus, in addition to the pseudo-cleft construction examined in her paper, several other constructions are likely to benefit from her interactional approach; a detailed examination of texts in terms of how a particular interactional goal is expressed linguistically will help to delineate the process of a diachronic development and to account for why such a development has occurred. I would like to point out three constructions that deserve study along this line. The first case comes from the domain of word-formation. It is well known that conversion (or zero-derivation) has been productive since Old English (Biese 1941: 18–49, Marchand 1969: 359–378, Pennanen 1971) and in Present-day English it enjoys particularly high productivity (Bauer 1983: 226–227, Plag 1999: 93–118). Unlike V-deriving su‰xes (e.g., -ate, -ify, -ize), whose base is restricted to N or A of certain phonological patterns (Plag 1999: 119–218), V-forming conversion accepts categorially, morphophonologically, and semantically various types of base, and the following data show that even ‘‘utterances’’ can be the input to V-forming conversion: (1) a.

‘‘I was explaining the Golden Bull to his Royal Highness.’’ ‘‘I’ll Golden Bull you, you rascal!’’ roared the Majesty of Prussia. (Jespersen 1942: 106)

11. In writing this commentary, I have greatly benefited from discussion of the material with Professor Masaharu Shimada. Of course, responsibility for any inadequacies is my own.

Commentary on Traugott

29

b. ‘‘Honey’’ ‘‘Don’t honey me,’’ she said. (Raymond Carver, ‘‘Vitamins’’) c.

Nell: ‘‘Ling!’’ Ling: ‘‘Don’t Ling me. I’m tired of being Linged around here.’’ (TV show, Ally McBeal)

Jespersen (1942: 105–107) calls this usage of conversion ‘‘retort,’’ because in these instances ‘‘in anger one simply seizes one word or phrase in what was said by the other party, and repeats it as a verb in a scornful tone of voice’’ (Jespersen 1942: 105). The retort usage of conversion is possible only in a dialogue, and the meaning of the converted verb is not simply ‘to say base utterance’ but includes the strong connotation of contesting such as ‘How can you use such a word?’ or ‘How dare you use such a word to me?’ This means that retort is not a mere categorial change for word formation but a form of refutation or ‘‘interactional contesting’’ (Traugott, this volume). Therefore, it can be seen as a dialogic construction in a broad sense. Since converted verbs in general do not have any dialogic function, the conjecture is possible that this construction should have developed in a dialogic context; a converted verb used in a dialogic context has semantically incorporated the dialogicity. To put it di¤erently, the usage of conversion has changed (or extended) in such a way to deal with a shift of viewpoint. If this conjecture is on the right track, Traugott’s paper suggests that there could be a stage in the history of English where the dialogicity of converted verbs such as those in (1) had a separate expression in a neighboring context and the verbs themselves expressed only the non-dialogic, transparent meaning ‘to say base utterance,’ as in Though you honey me, I don’t like it, or You have just Golden Bulled me, but it is inappropriate. In fact, some converted verbs based on an utterance do seem to be free from any tone of anger or contestation and simply denote the act of making that utterance, as follows: (2) a.

‘‘Poor Robin.’’ ‘‘What are you two girls poor Robining about?’’ (Jespersen 1942: 107)

b. We don’t Mr. each other here. (ibid.) The second candidate for a dialogic construction also involves morphology: the superlative form of an adjective. Absolute superlatives are

30

Akiko Nagano

often concerned with possible maximum or minimum degree (Huddleston and Pullum 2002: 1167) and can be used as a construction semantically incorporating the focus particle even, as the following instances show: (3) a.

The ground was so soft that the lightest step made a deep imprint. (Huddleston and Pullum 2002: 1167) b. The slightest touch will break a soap bubble. c. The best musician is liable to make a mistake when he is tired.

Traugott’s paper provides a new perspective on this type of grammaticalization and opens up a new research possibility for its motivation; this usage of superlative could be preceded by the use of a superlative form in a dialogic context that includes the focus particle even (e.g., Even the slightest touch will break a soap bubble) or a concessive adverbial clause. Lastly, Traugott’s research format could be applied to the development of the (so-called) speech-act conditional sentences (Rutherford 1970: 109– 110, Sweetser 1990: 113–144). The adverbial clauses of the following sentences modify an unexpressed speech-act verb underlying the main clause (see the underlined part of the parenthesized paraphrase): (4) a.

If you’re interested, Dick’s coming to the party too. (Huddleston and Pullum 2002: 740) [If you’re interested, it is worth telling you that Dick’s coming to the party too.]

b. Where did your parents go, if you know? (Quirk et al. 1985: 1097) [If you know, tell me where your parents go.] c.

While we’re on the subject, why didn’t you send your children to a public school? (Quirk et al. 1985: 1073) [While we’re on the subject, tell me why you didn’t send your children to a public school.]

The italicized adverbial clauses express the condition under which the speaker makes the utterance, so they relate not to the proposition of the main clause itself but to the speech act performed in uttering the clause. As a closely-related usage, the italicized if-clause in the following sentence relates to an unexpressed epistemic modality:

Commentary on Traugott

31

(5) If the key is not in my pocket, I have left it in the door. (Huddleston and Pullum 2002: 740) [If the key is not in my pocket, it must be the case that I have left it in the door.] This usage of an adverbial clause in the ‘‘speech-act domain’’ and the ‘‘epistemic domain’’ (Sweetser 1990: 113–121) is exhibited also by an adverbial clause of reasoning (ibid.: 76–86), as shown below. (6) a.

What are you doing tonight, because there’s a good movie on. (Sweetser 1990: 77) [I ask you what you are doing tonight, because there’s a good movie on.]

b. He’s not coming to class, because he just called from San Diego. (Rutherford 1970: 97) [It must be the case that he’s not coming to class, because he just called from San Diego.] c.

Since you don’t seem to know, all further negotiations have been suspended. (Quirk et al. 1985: 1073) [Since you don’t seem to know, it is worth telling you that all further negotiations have been suspended.]

d. Since John isn’t here, he has (evidently) gone home. (Sweetser 1990: 78) [Since John isn’t here, it must be the case that he has gone home.] To the extent that expressions of speech-act and epistemic modality concern the dialogicity, it must be worthwhile entertaining the possibility that dialogic contexts played a role for the development of the usage of adverbial clauses discussed above.12 In fact, the speech-act and epistemic-modality expressions covert in English adverbial clauses (e.g., the underlined parts of the parenthesized paraphrases in (4), (5), and (6)) have to be realized overtly in Japanese (Sakahara 1985: 154, Nakau 1994: 103–106, Shizawa 2008). This crosslinguistic di¤erence cannot be ignored in delineating the diachronic process of grammaticalization of speech-act and epistemic modality, and 12. According to Traugott and Ko¨nig (1991) and Hopper and Traugott (1993: 75–77), the causal meaning of since is developed from its temporal meaning.

32

Akiko Nagano

o¤ers an interesting question of how such a ‘‘parametric’’ variation can be accounted for by the interactional approach.

References Bauer, Laurie 1983 Biese, Y. M. 1941

English Word-Formation. Cambridge: Cambridge University Press.

Origin and Development of Conversions in English. Annales Academiae Scientiarum Fennicae, B XLV, Helsinki. Hopper, J. Paul and Elizabeth Closs Traugott 1993 Grammaticalization. Cambridge: Cambridge University Press. Huddleston, Rodney and Geo¤rey K. Pullum 2002 The Cambridge Grammar of the English Language. Cambridge: Cambridge University Press. Jespersen, Otto 1942 A Modern English Grammar on Historical Principles, Part 6: Morphology. Copenhagen: Ejnar Munksgaard. Marchand, Hans 1969 The Categories and Types of Present-Day English Word-Formation: A Synchronic-Diachronic Approach. 2nd ed. Munich: C. H. Beck. Nakau, Minoru 1994 Ninchi Imiron no Genri (Principles of Cognitive Semantics). Tokyo: Taishukan. Pennanen, Esko V. 1971 Conversion and Zero-Derivation in English. Tampere: Tampereen Yliopisto. Plag, Ingo 1999 Morphological Productivity: Structural Constraints on English Derivation. Berlin: Mouton de Gruyter. Quirk, Randolph, Sidney Greenbaum, Geo¤rey Leech, and Jan Svartvik 1985 A Comprehensive Grammar of the English Language. London: Longman. Rutherford, William E. 1970 Some Observations Concerning Subordinate Clauses in English. Language 46: 97–115. Sakahara, Shigeru 1985 Nichijo Gengo no Suiron (Inferences in Everyday Speech). Tokyo: Tokyo University Press. Shizawa, Takashi 2008 Conditionals Giving Reasons for Utterances: A Contrastive Study of Japanese and English from the Viewpoint of AddresseeOrientedness. A paper read at the 26th Conference of the English Linguistic Society of Japan.

Commentary on Traugott

33

Sweetser, Eve E. 1990 From Etymology to Pragmatics: Metaphorical and Cultural Aspects of Semantic Structure. Cambridge: Cambridge University Press. Traugott, Elizabeth C. and Ekkehard Ko¨nig 1991 The Semantics-Pragmatics of Grammaticalization Revisited. In Approaches to Grammaticalization, edited by Elizabeth C. Traugott and Bernd Heine, 189–218. Amsterdam: John Benjamins Publishing Company.

Response to Commentary by Nagano Elizabeth Closs Traugott

Akiko Nagano presents three constructions that might fruitfully be investigated to determine whether they arose in dialogic contexts: conversion in retorts, use of [Definite Article Adjective-superlative N] constructions, and speech act conditionals. All three involve speaker evaluation of some element in the utterance or its context: the first evaluates the validity of what someone else has said, the second situates the complement on the extreme end of a scale, and the third evaluates the upcoming utterance with respect to its relevance to the addressee or the speech situation. Here I will consider only the first type, conversions. The conversions Nagano cites exemplify ‘‘metalinguistic negation’’ in the sense of Ducrot (1972) and Horn (1985), as they are devices ‘‘for objecting to a previous utterance on any grounds whatever – including its conventional or conversational implicata, its morphology, its style or register, or its phonetic realization’’ (Horn 1985: 121). Most conversions appear to be on-the-fly nonce uses that are not conventionalized as part of a community’s linguistic system (i.e. they are innovations, not changes). However, it is clear that in some cases conversions do become conventionalized, e.g. V-forming conversions of T/V pronoun distinctions such as French tutoyer ‘to use/say ‘T’’, French vouvoyer ‘to use/say ‘V’’, English to thou, as in: (1) None of hyghnesse schal thou another in spekynge, but eche schal speke reuenrently to other ‘none of high position shall ‘thou’ another in speaking, but each shall speak reverently to the other’ (c1450 Aungier, Syon Monastery 287 [MED thouen, OED thou, v.]) (1) is dialogic, but not dialogual, as it does not involve turn-taking. Indeed most citations in the MED and OED appear not to be dialogual retorts, but regulatory, like (1), or invectives. Other conversions that quote utterances include delocutive verbs such as Latin negare ‘to say nec ‘no’ ’, English hail ‘to say ‘Hail!’ ’, French (re)mercier ‘to say merci ‘thanks’ ’ (Benveniste 1971 [1958]). Some may well have arisen in dialogic and dialogual contexts of the type Nagano suggests, e.g. negare. However, as she

Response to Commentary by Nagano

35

notes, not all V-forming conversions contest. By hypothesis, hail, (re)mercier did not. The histories of individual micro-constructions within the same set may be very di¤erent, e.g. a shred of is likely to occur in negative polarity contexts, a bit (of) considerably less likely. While a shred of favors positive complements (honor, truth), there is no such semantic prosody (Stubbs 2001) for a bit (of). Therefore, to the extent that macro(parametric) changes may be said to be involved, the micro-changes and micro-contexts that lead up to them may be very varied, as may also the structure of any individual construction selected in a particular language. In investigating the construction-types Nagano discusses, it will be important to consider the full range of interactional contexts in which they arose, and to distinguish dialogic from dialogual contexts.

References Benveniste, Emile 1971 [1958] Delocutive Verbs. In Problems in General Linguistics, 239–246. Trans. by Mary Elizabeth Meek. Coral Gables, FL: University of Miami Press. (Les verbes de´locutifs, Proble`mes de Linguistique Ge´ne´rale, 277–285. Paris: Gallimard 1966; orig. publ. in A. G. Hatcher and K. L. Selig, eds., Studia Philologica et Litteraria in Honorem L. Spitzer, 57–63. Bern, 1958.) Ducrot, Oswald 1972 Dire et ne pas dire. Paris: Hermann. Horn, Laurence R. 1985 Metalinguistic Negation and Pragmatic Ambiguity. Language 61: 121–174. Stubbs, Michael 2001 Words and Phrases: Corpus Studies of Lexical Semantics. Oxford/Malden, MA: Blackwell.

Whatever Happened to English Sluicing* Joanna Nykiel

1. Introduction We know from experience that an anaphoric construction usually allows more than one kind of licensing context. Because we refer to entities in the surrounding discourse, we expect to be able to do so whether the selected entity is discourse- and hearer-old (linguistic antecedents) or discourse-initial and hearer-old (situational antecedents). Hankamer and Sag (1976), however, propose a distinction between surface and deep anaphors. They define surface anaphors as derived via deletion under identity with the antecedent surface structure; surface anaphors are thus opposed to deep ones, which arise directly in the base. While this view necessitates that surface anaphors accept only linguistic antecedents, it places no such restrictions on deep ones. The distinction has been influential ever since it was proposed, but it is now well established that surface anaphors are freer in their choice of antecedent than Hankamer and Sag originally assumed (cf. Webber 1978, Hardt 1993, 2005, Ginzburg and Sag 2000, Kehler 2002, Culicover and Jackendo¤ 2005, Sag 2006, Stainton 2006). In this paper I o¤er insight about the history of sluicing. Since the timehonored definition of sluicing classifies it as a surface anaphor with a full underlying structure that goes unpronounced (Ross 1967, Hankamer and Sag 1976, Sag and Hankamer 1984), it makes sense to verify whether this elliptical construction has ever required structurally identical antecedents. A fact of Present-day English (PDE) is that it does not place this constraint on sluicing. This fact is evaluated against the data coming from four periods: Old English (OE), Middle English (ME), Early Modern English (ENE) and Late Modern English (LNE). When so cast into an empirical mode, sluicing proves to accept at least two di¤erent rela* I wish to thank Elizabeth Traugott, Ivan A. Sag, and two anonymous reviewers for comments on earlier drafts of this paper. The research was made possible by a grant from the Kosciuszko Foundation and by Stanford University, which was the host institution.

38

Joanna Nykiel

tionships with its antecedents at any given point in time, with completely or partially unidentical syntactic structure having a more pronounced presence in the majority of the corpora than any other relationship. The intuitive idea is, therefore, that sluicing is a case of apparent stability. How much stability there really is in languages remains an open question. Some linguists have argued for immediate change (cf. Lightfoot 1979); others support an initial reanalysis whose actualization is gradual and thus gives an impression of stability (cf. Timberlake 1977, Langacker 1977). Whether the use of sluicing in fact is a stable feature or the first small steps on the way of an on-going, and indeed very gradual, change is yet to be determined. At this point, I show records that reveal hardly any drift. Further, I demonstrate that because the relaxation of structural identity found in modern sluicing has persisted ever since OE, sluicing cannot be so far removed from deep anaphors or subject to a purely derivational strategy. Instead, such data seem to reduce sluices, along with other anaphors, to directly generated fragments. To take these ideas a step further, I propose that for any sluice, it is only required that the link with its antecedent be based on coherence and that to tease apart surface and deep anaphors was misleading because di¤erences between them have independent motivation.

2. Basics The story begins with Ross (1967), who o¤ers a strategy for the resolution of sluicing – a stranded wh-phrase illustrated in (1)–(3). (1) I was afraid of something that day, but I didn’t know of what. (2) A: You want a massage? B: By who? (3) Pssst. Wanna copy contacts over to yahoo!? Here’s how. (www.yahoo.com) Following the convention established in the (linguistic) literature, I will henceforth refer to a stranded wh-phrase as a sluice and the preceding material that supports its interpretation as an antecedent. Ross shows that clear syntactic e¤ects cluster around sluicing, pointing to a pre-deletion structure beyond what is visible. For one thing, a casemarking language like German or Old English will require that the case of a wh-phrase be in correspondence with its counterpart in the antecedent, as illustrated in (4) and (5).

Whatever Happened to English Sluicing

39

(4) Er will jemandem schmeicheln, aber sie wissen nicht, he wants someone.DAT flatter but they know not *wer / *wen / wem who.NOM who.ACC who.DAT ‘He wants to flatter someone, but they don’t know who.’ (5) Er will jemandem loben, aber sie wissen nicht, he wants someone.ACC praise but they know not *wer / wen / *wem who.NOM who.ACC who.DAT ‘He wants to praise someone, but they don’t know who.’ (Lasnik 2007: 144) OE records an analogous e¤ect, as seen in (6). Here the case of the whphrase (hwilcum deaþe) corresponds to that of an implied adjunct in the antecedent.1 (6) Ða befran Iohannes færlice, and cwæð: Hu ys he la dead oððe then asked John quickly and said: How, is he dead and hwilcum deaþe? what death.INSTR? ‘Then John asked quickly: What! Is he dead and (by) what kind of death?’2 (c1000 Ælfric: Letter to Sigeweard [On the Old and New Testament]) To this, Merchant (2004, 2006, 2007) adds another observation. Prepositions can apparently be omitted under sluicing only if preposition stranding produced by wh-movement is a feature of the language, e.g., English but not German. (7) Peter was talking with someone, but I don’t know (with) who. Who was he talking with? (8) Anna hat mit jemanden gesprochen, aber ich weiss nicht Anna has with someone spoken but I know not *(mit) wem. with whom. *Wem hat sie mit gesprochen? (Merchant 2006: 666–667) 1. This example is the only illustration of case dependencies in the OE corpus. 2. The interpretation of Hu as a pragmatic marker was suggested to me as the more plausible by Matti Kilpio¨. If taken to be a question word, Hu would be the explicit referent of hwilcum deaþe.

40

Joanna Nykiel

This correlation, if correct, provides strong support for pre-deletion structures. Early generative grammar has a straightforward means of handling these e¤ects: phonological deletion under identity with the antecedent.3 A sluice thus represents a full syntactic structure whose interpretation is recoverable from the antecedent. The idea carries over into later work, notably that by Hankamer and Sag (1976) and Sag and Hankamer (1984). Both additionally note a fine line between what surface anaphors, like sluicing, can do as opposed to deep anaphors, like do it or Null Complement Anaphora (NCA). In short, the approach makes a strong commitment to the fact that only linguistic antecedents may license surface anaphors while deep anaphors accept linguistic and non-linguistic ones alike. This initial solution has long since been questioned – most recently by Ginzburg and Sag (2000), Culicover and Jackendo¤ (2005) and Stainton (2006) – not least because modern sluicing seems to be growing less and less like a surface anaphor, if it has ever been one. Consider the naturally occurring examples in (9)–(13). (9)

A: I’d like to take Katie on weekends. B: Why?

(10) My first impulse was to run away, up or down stairs, I wasn’t quite sure which. (11) [Cab driver to passenger on their way to the airport:] ‘Which airline?’ (12) [Someone stands before the scene of some awful event and exclaims:] ‘Why, oh why? I dunno.’ (Ivan Sag, personal communication) (13) A: Tell me! B: What? Because in each case the relationship between the sluice and the antecedent is partially or entirely non-syntactic, additional factors have been posited beyond structural identity (structural identity will henceforth be referred to as merger, a term due to Chung et al. 1995). Specifically, the recent literature (cf. Ginzburg and Sag 2000, Merchant 2001, Kehler 2002, Culicover and Jackendo¤ 2005) lists a semantic relationship that allows syntactic mismatches of two kinds: 3. This view has been rivaled by an alternative in which pre-deletion structures are made available at the level of LF; see Williams (1977), Kitagawa (1991), Fiengo and May (1994).

Whatever Happened to English Sluicing

41

– sprouting, in which a sluice builds on the argument structure of the predicate embedded in the antecedent, pointing to an implied argument (13). Beyond implied arguments, implied adjuncts may be pointed to as well (2, 9). – how-mismatch, where an antecedent does not provide the to required by the structure how þ to-infinitive understood at ellipsis site (3). Further, sluicing is licensed when a pragmatic relationship is operative based on inference (10). Here the pragmatic status of a sluice is attributed to the fact that no element of the structure of the antecedent, whether overt or implied, motivates its use. Along pragmatic lines again, a sluice may be used discourse-initially and so may be situationally-controlled (11, 12). Now the questions are: Are the above passages due to changes in English sluicing, or have they always been around? Is a sluice a directly generated fragment (cf. Ginzburg and Sag 2000, Culicover and Jackendo¤ 2005, Stainton 2006) or a product of derivation (cf. Merchant 2004, 2006, 2007)? I turn to these questions in the next section.

3. What the history has to say One negative aspect of a historical analysis is that, although its frequency data may reflect processing costs associated with particular structures, it often su¤ers from incompleteness or reduced reliability. Historical records – by definition written – have yet another disadvantage when it comes to ellipsis. It can reasonably be expected that spoken language will bear more characteristics of elliptical expression or, at a minimum, that it will host a greater variety of antecedents. Therefore, even the most detailed results are best treated as (strong) tendencies though, of course, the more diverse the records, the closer we come to the ideal. I now o¤er statistical data from OE, ME, ENE and LNE, with text types addressed as required. For clarity of presentation, I reiterate that sluicing can accommodate the following relationships with apparently varying degrees of felicity: – merger, as in (1) – syntactic mismatch (sprouting, as in (2), (9) and (13) above, and howmismatch, as in (3)) – pragmatic control/situational context, as in (11) and (12) – inference, as in (10)

42

Joanna Nykiel

Table 1. Timeframe: Distribution of sluicing in the Dictionary of Old English (DOE) based on relationship between sluice and antecedent Period

Merger

Sprouting

Total

Early OE

27 (81.82%)

6 (18.18%)

33 (100%)

Late OE

107 (81.06%)

25 (18.94%)

132 (100%)

Total

134 (81.21%)

31 (18.79%)

165 (100%)

For each of the relationships, if present in the data, I give its number of occurrences relative to the total number of sluices in each corpus examined. These figures then serve as a basis for identifying diachronic developments in sluicing, if any. An initial answer to this question may be fashioned from the material in Tables 1–3. In Table 1, the OE data signal a heavy reliance on syntax in terms of merger (14). Sprouting (15–17), with its availability of simultaneous syntactic and semantic e¤ects, could be equated with evidence against predeletion structures. However, 18% throughout the period is not much to build on yet. (16) and (6), repeated as (17) below, illustrate syntactic e¤ects, that is, case and preposition choice, next to implied referents that sprouting always picks up. Such syntactic e¤ects are discernible in each corpus I have analyzed, and I return to them later in this section. (14) Canst þu ænig þing? ænne cræft ic cann. Hwylcne? Can you anything? one thing.ACC I can. What? ‘Can you do anything? I can do one thing. What?’ (Ælfric, Colloquy: Garmonsway 1939, 18–49) (15) Gea, butan nettum huntian ic mæg. Hu? Yeah, without net hunt I can. How? ‘Yeah, I can hunt without a net. How?’ (Ælfric, Colloquy: Garmonsway 1939, 18–49) (16) Stranguilio cwæð: Hwa fordemde þe? Stranguilio said: Who condemned you? Apollonius cwæð: Antiochus se cyngc. Apollonius said: Antiochus the king. Stranguilio cwæð: For hwilcum intingum? Stranguilio said: For what reasons? (Vision of Leofric: Napier 1907–10, 182–6)

Whatever Happened to English Sluicing

43

(17) Ða befran Iohannes færlice, and cwæð: Hu ys he la dead then asked John quickly and said: How, is he dead oððe hwilcum deaþe? and what death.INSTR? ‘Then John asked quickly: What! Is he dead and (by) what kind of death?’ (Ælfric: Letter to Sigeweard [‘On the Old and New Testament’]) Cumulatively, these facts are perhaps not totally unexpected and seem to point in the direction of a change along the way insofar as we are missing three further licensers.4 At the same time, the strength of this conclusion is limited by a bias toward an exclusively written style, though, as Allen (1980) argues, this style is not quite out of touch with patterns found in speech. It is possible to minimize the potential incompleteness of these data if they are addressed together with the results for ME because ME contributes rich resources of formal and informal varieties, including personal correspondence. For its temporal adjacency to OE, early ME (1150– 1350) in particular provides a fine insight about sluicing. Table 2 gives the figures for sluicing in early and late ME, and in the transition between the two periods (third row). Sprouting in early ME (and later on) shows a dramatic rise in frequency over merger (69% vs. 21%), which suggests that in OE, syntactic relationships too may have played a less important role than indicated by the data in Table 1. Further, the sprouting column includes nine mismatches. These are simply

Table 2. Timeframe: Distribution of sluicing in the Middle English Compendium (MEC) based on relationship between sluice and antecedent Period

Merger

Sprouting including 9 (Late ME) mismatches

Merger/ sprouting

Total

Early ME

34 (21.8%)

108 (69.23%)

14 (9.00%)

156 (100%)

8 (17.02%)

35 (74.5%)

4 (8.51%)

47 (100%)

Late ME

50 (11.82%)

326 (77.06%)

46 (10.9%)

423 (100%)

Total

92 (14.7%)

469 (74.9%)

64 (10.22%)

626 (100%)

Early/late ME

4. For a full review of OE sluicing, see Nykiel (2007).

44

Joanna Nykiel

how-mismatches that always co-occur with sprouting in late ME; hence, I collapsed both categories. Since these mismatches do not have any e¤ect on the statistics at this point, I discuss them later in this section where they may be treated as an independent category. That ME sluicing tolerates non-syntactic relationships other than sprouting is suggested by one instance of inference given in (18). (18) And he shal han Custaunce in mariage And he shall have Custanuce in marriage And certeyn gold. I noot what quantitee And certain gold. I know not what quantity. ‘And he shall have Custaunce in marriage, and certain gold. I do not know what quantity’ (14c. Hengwrt Manuscript of Chaucer’s Canterbury Tales) This single occurrence is statistically insignificant although it allows us to hypothesize that the conditions on sluicing are, and always have been, tied to the participation of the various relationships in it rather than to their presence or absence. As we saw in section 1, inference, unlike sprouting, only indirectly relies on linguistically overt antecedents for resolution. Therefore, it is a relationship relatively di‰cult to interpret, thus somewhat less frequent than sprouting. Note that there is another new, unique to ME, relationship between antecedent and ellipsis here, which I dubbed merger/sprouting. The relationship is partly semantic and partly syntactic in that an ellipsis has a correlate in its antecedent, of which it is a semantic paraphrase. A paraphrase is usually richer in information content than the correlate, as illustrated by (19). (19) Of whens y am and what men clepe me of whence I am and what men call me And where y was borne and in what cuntre. and where I was born and in what country ‘Where I am from and what I am called, and where I was born and in what country’ (15c. The Romance of Guy of Warwick) Merger/sprouting does not occur outside of the ME corpus and correlates with verse more often than it does with prose (4% di¤erence, see Table 3). If the purpose of merger/sprouting was only to satisfy metrical considerations, however, we might expect more of a bias toward verse. With the

Whatever Happened to English Sluicing

45

Table 3. Text types: Distribution of sluicing in the Middle English Compendium (MEC) based on relationship between sluice and antecedent Genre

Merger

Sprouting including 2 (Prose), 6 (Verse), 1 (Drama) mismatches

Merger/ sprouting

Inference

Total

Prose

35 (13.4%)

206 (78.6%)

21 (8.01%)

0

262 (100%)

Verse

56 (15.9%)

253 (71.7%)

43 (12.2%)

1 (0.3%)

353 (100%)

1 (9.1%)

10 (90.9%)

Drama

0

0

11 (100%)

Table 4. Text types: Distribution of sluicing in the ENE part of the Helsinki Corpus based on relationship between sluice and antecedent Genre

Merger

Sprouting including 1 (Prose) mismatch

Howmismatch

Inference

Total

Prose

27 (38.02%)

42 (59.15%)

0

2 (2.81%)

71 (100%)

1 (12.5%)

0

Drama

3 (37.5%)

4 (50%)

8 (100%)

actual distribution, the purpose could be tied to discourse instead. For example, forty-seven out of sixty-four instances appear in foregrounded clauses, that is, in dialogue and narrative. I leave open the question of what exactly supports this relationship. The interest in Table 3 for this discussion is that it points to an almost even distribution of the relationships under sluicing in both prose and verse. The latter genre is poorly represented in OE while in ME it hosts the majority of sluices. This fact together with the distribution predicts that the results in Table 3 impart reliability to the OE data. Whether text types influence sluicing in the later periods of English will be addressed again as more data are investigated. If, as we move on, we could find some marked departure from the OE and ME distribution, we would stand a good chance of strengthening our assumptions about what development there has been. This is not the case, however. Consider the ENE data in Tables 4–7. Table 4 gives a general

46

Joanna Nykiel

Table 5. Distribution of sluicing in the corpus of William Shakespeare’s plays based on relationship between sluice and antecedent Merger

Sprouting including 6 mismatches

Howmismatch

Inference

Total

148 (44.18%)

180 (53.73%)

2 (0.6%)

5 (1.49%)

335 (100%)

Table 6. Distribution of sluicing in the Corpus of Early English Correspondence Sampler (CEECS) based on relationship between sluice and antecedent Merger

Sprouting

How-mismatch

Inference

Total

8 (34.78%)

13 (56.52%)

1 (4.35%)

1 (4.35%)

23 (100%)

statement of the data, with only a distinction between prose and drama. Clearly, sprouting remains the dominant relationship in both genres. Tables 5–7 separate out very specific text types, ranging from drama (Table 5) to correspondence (Table 6). This is to ensure a fair assessment of the phenomenon under discussion. Among the available records, Shakespeare’s drama (Table 5), under the standard assumptions, is fairly close to speech and richest in dialogue, with an economy of expression specific to it. So, as predicted, the varying frequency of sluicing in the corpora is due to text type and corpus size because the Shakespeare corpus is the largest and four times the size of the CEECS (Table 6). Interestingly, the participation of the di¤erent antecedents in sluicing varies somewhat unexpectedly between the tables. Note that sprouting remains constant at about 50% in Shakespeare’s (Table 5) and other drama (Table 4). In written correspondence (Table 6), however, it reaches 56% and in prose 59% (Table 4). The percentage of merger only exceeds 40% in Shakespeare’s drama, otherwise remaining at about 30%. Thus compared to drama, the percentage of sprouting rises at the expense of merger in prose. Within this genre, fiction reaches a high of 80%, a significant di¤erence if we remember that fiction is also speech-related (Table 7). In the same table, there is another representative record of speech – the State-Trials.5 5. This is A Complete Collection of State-Trials and Proceedings for Hightreason, and Other Crimes and Misdemeanors; From the Reign of King Richard II to the End of the Reign of King George I.

Whatever Happened to English Sluicing

47

Table 7. Prose: Distribution of sluicing in the ENE part of the Helsinki Corpus based on relationship between sluice and antecedent Genre

Merger

Sprouting including 1 (Prose) mismatch

Howmismatch

Inference

Total

Fiction

1 (16.66%)

5 (83.33%)

0

0

6 (100%)

Trials

10 (58.82%)

6 (35.29%)

0

1 (5.88%)

17 (100%)

Other Prose

16 (33.33%)

31 (64.58%)

0

1 (2.08%)

48 (100%)

Here we have a reversal of frequencies: sprouting covers 35% of the total and merger 58%, which seems highly inconsistent compared to the results we obtained for fiction. Hence, a clear development is doubtful. Thus far, the only safe conclusion is that sluicing minimally admitted merger and sprouting early on, as indicated by the temporal factor included in Table 1.6 On a theoretical note, even in the absence of other non-syntactic relationships sprouting still shows that syntax and semantics are both involved because only part of the antecedent structure is visible to and made use of by the hearer. This in turn suggests the possibility that no pre-deletion structure underlies a sluice. Another candidate for signaling a potential change is how-mismatch and inference. If lack of the former in OE and early ME is no surprise because the how þ infinitive structure was non-existent until the twelfth century (cf. Oxford English Dictionary), in late ME we find a single instance of inference.7 I leave the detailed discussion of these two relationships until later. At this point, just notice that their numbers are conspicuously low and that how-mismatch sometimes overlaps with sprouting (indicated as required in the tables). Next I turn to the ARCHER corpus, which brings data from LNE with some PDE attestations, all supplemented with yet more evidence from 6. The introduction of sluicing into English is very much unclear inasmuch as we cannot rule out that Latin played a part in it. OE texts translated from Latin do not in fact reveal much of a departure from genuine ones and thus stop short of enabling valid conclusions (cf. Nykiel 2007). 7. An interesting consequence of the rise of the how þ infinitive structure is that syntax itself introduced a new – problematic for syntactic accounts – relationship into sluicing.

48

Joanna Nykiel

Table 8. Text types: Distribution of sluicing in the ARCHER corpus based on relationship between sluice and antecedent Genre

Merger

Sprouting including 4 (Prose), 2 (Fiction), 2 (Drama) mismatches

HowInference mismatch

Pragmatic Total control

Prose

25 (37.88%)

36 (54.54%) 1 (1.52%)

4 (6.06%) 0

66 (100%)

Fiction

41 (39.05%)

53 (50.48%) 1 (0.95%)

7 (6.66%) 3 (2.86%)

105 (100%)

Drama 129 (44.33%) 135 (46.39%) 1 (0.34%) 23 (7.90%) 1 (0.34%)

291 (100%)

ENE. The results come in Tables 8–9 addressing the relevant criteria of timeframe and text type. For ease of interpretation, I separate the ENE data out from that of LNE and the twentieth century in Table 9. First, English seems to have shifted in a surprising direction. Prose shows a drop in the percentage of sprouting from the previous 56–59% (Table 8) as does drama, remaining now at 46%. Fiction, separated out from the rest, drops to 50% while merger remains stable in both prose and drama. It is sprouting, then, that gives way to how-mismatch, inference, and a new antecedent – pragmatic control. The application of pragmatic control seems somewhat limited, even to speakers of PDE. This fact straightforwardly falls into place if we remember that the former surface anaphors typically accommodate few discourse-initial positions. Thus both Verb Phrase Ellipsis (VPE) (They might) and sluicing (Guess who!) would be less felicitous in response or reaction to non-linguistic antecedents than to linguistic ones (cf. Hankamer 1978, Pullum 2000, Stainton 2006). The low frequency of pragmatic control is expected although it is di‰cult to explain its apparent non-appearance until ENE.8 Pragmatically controlled sluices could have existed in speech without being preserved in written records though this 8. This prediction is consistent with the absence of pragmatically controlled sluices in the Dictionary of Old English and Middle English Compendium. A cautionary note is in order, however. There are phrases in ME that look like sluices, e.g., What chere?, spoken discourse-initially. Given their meaning of How do you feel?/How are you? (cf. Middle English Dictionary), they should be viewed as formulas instead.

Whatever Happened to English Sluicing

49

seems unlikely, given the variety of the available data. Alternatively, they could have evolved as an option secondary to linguistically controlled sluices. The latter prediction is supported by a similarly late arrival onto the scene that a¤ects pragmatically controlled VPE (cf. Gergel 2004, Nykiel 2006) despite syntactic di¤erences between the two constructions. I return to the problem of pragmatic control and a possible explanation for it in the next section. From Table 8, it emerges that inference comes to license more sluices than either how-mismatch or pragmatic control. However, sprouting and how-mismatch can go together, and so, where it is the case, mismatch is included in the Sprouting column in the tables. Consider the examples below: (20) In all instances he knows what is best to decree, and what is best to command, and what is best to do; and in all instances he decrees, he commands, and he conducts, as well as he knows how. (1762. Bellamy, Joseph. An Election Sermon. In B. Kuklick (ed.), the Works of Joseph Bellamy, vol. I) (21) and therfore, whan they be drye, they muste be sore brused and broken, the wiues knowe howe, and than winowed and kepte drye, tyll yere-tyme come (Fitzherbert. The Book of Husbandry (1534). English Dialect Society, 37. Ed. W. W. Skeat. Vaduz: Kraus Reprint LTD., 1965 (1882)) Upon closer inspection, therefore, how-mismatch and inference have comparable incidences. Is there any movement, though? ENE gives the highest numbers for inference in correspondence and the State Trials (Tables 6 and 7); later on drama becomes dominant (Table 8), but fiction and other prose are not far behind. Table 9 in fact gives a neat statement of how the rate of inference-based sluices rises by about 1% over the period of some 400 years. How-mismatch, if we include the cases listed in the Sprouting column, is strongest in Period II (3.3%) and then drops to 1.35%, a rather non-directional change. There is a di¤erence of less than 1% across text types except for prose (Table 8).9 Checked against the ENE tables, howmismatch su¤ers a considerable blow in LNE and the twentieth century. (Recall that the rate of how-mismatch reaches 12% in ENE drama).

9. Including the four instances listed in the Sprouting column.

50

Joanna Nykiel

Table 9. Timeframe: Distribution of sluicing in the ARCHER corpus based on relationship between sluice and antecedent Period

I 1600– 1700 II 1701– 1900

Merger

9 (27.27%)

Sprouting HowInference including mismatch 1 (Period I), 5 (Period II), 2 (Period III) mismatches 21 (63.63%) 0

Pragmatic Total control

2 (6.06%) 1 (3.03 %)

33 (100%)

84 (40.57%) 103 (49.75%) 2 (0.97%) 15 (7.24%) 2 (0.97%) 207 (100%)

III 1901– 102 (45.94%) 100 (45.04%) 1 (0.45%) 17 (7.65%) 1 (0.45%) 222 (100%)

For lack of space, I have not included more detailed tables showing the distribution of sluicing in the ARCHER corpus, but even from Tables 8–9, it is evident that the previously identified preference for sprouting weakens with time, even in fiction, which seemed to favor it most. In sum, sprouting has an initial rising tendency only to eventually fall to 40–50%; merger has an initial falling tendency but then continues at a stable rate of about 40%. Of the minor relationships, only inference gradually rises uniformly in all the genres considered, as already noted. Upon closer inspection then, applying text type and temporal criteria points to little directional movement within the range of the antecedents or to consistent variety between written and speech-related genres. Perhaps this in turn indicates that we should not worry too much about the diminished accuracy of the extant texts with respect to the genres and time periods they represent. 3.1. Syntactic e¤ects in sprouting Syntactic and semantic involvement in sprouting was emphasized in the previous section. I reiterate that syntactic e¤ects connect a sluice to its antecedent via identity of preposition, case, or both. Such e¤ects are also operative under merger, but the more interesting of the two, and potentially problematic, is sprouting. (22) Thanks, Si. {¼ m SIMON} What for? {¼ m STEPHEN} Sharing my triumph with me. (Gray, Simon. 1975, Otherwise engaged: A play)

Whatever Happened to English Sluicing

51

There is nothing in the antecedent structure that licenses the use of the preposition. Therefore, an ellipsist account like Ross’s runs into trouble. And yet, among the corpora, inclusion of ‘‘unlicensed’’ prepositions covers 4% of the OE data, 42% of the ME data, 25% of the ENE data, and 12% of the LNE data. Viewed as a directly generated fragment instead, a sluice that needs to point to a salient, though implied, argument PP, cannot do so without su‰cient content. In (22), the preposition is thus licensed by virtue of being part of the predicate’s (thank) argument structure. I take these sprouting facts to be a compensation strategy, a necessity that follows from prior manipulation of antecedent structure, as discussed in the next section.

4. Where to go next Much more remains to be said about the history of sluicing, not because it seems ‘‘uneventful’’ so far while a linguist’s ambition is to look for ‘‘action.’’ If there were big changes, there would also be explanations for them; if there are few changes but a problem exists, it is even more tempting to find out why. Below, I sketch a potential solution that draws upon the proposals by Ginzburg and Sag (2000), Kehler et al. (2006, 2007), Culicover and Jackendo¤ (2005), and Stainton (2006a, b). The data point to a diachronically stable link that ties sluices to their antecedents. While the link can be anything from syntactic to pragmatic, it obviously needs to remain based on coherence. Imagine now that the only requirement we place on a sluice is that it be coherently related to some salient antecedent without having a pre-deletion structure.10 This is where my analysis meets Kehler et al’s coherence-driven approach to pronoun resolution in which use of referring expressions is dictated by expectations about what can coherently come next in a given speech situation. In fact, Merchant (2004) too talks about anaphors and antecedents being ‘‘appropriately related’’ if somewhat vaguely. Once coherence has been established between sluice and antecedent, the former will undergo integration into the proposition carried by the antecedent. It will have access to such corresponding syntactic features of the antecedent as case and preposition, together with any arguments or adjuncts implied by the antecedent, that is, it will be indirectly licensed by the antecedent (cf. Ginzburg and Sag 2000, Culicover and 10. What I mean by salient here is recently brought into focus (cf. Ward et al. 1991, Ginzburg and Sag 2000).

52

Joanna Nykiel

Jackendo¤ 2005). This now reduces to the issue of why the various links have the distribution that we saw in the corpora. A principle that is likely to be operative is one of ease of integration. The more clues a sluice o¤ers, the faster and more e‰ciently it finds its place in the antecedent proposition. What kind of clues are they? They are necessarily semantic but even more importantly syntactic. It is no big surprise that a sluice preserving the antecedent structure, or at least part of it, is more frequent. But the antecedent structure can be manipulated – just as long as coherence is maintained. This move comes at a cost, though, in that fewer clues will be available and so lower frequencies of use follow for mismatches and inferences. Sprouting seems to su¤er less from loss of clues. It has a way of repairing the loss, however. Consider how speakers compensate for manipulating the antecedent structure in (2) repeated as (23). (23) A: You want a massage? B: By who? (24) A: You want a massage? B: Who? For lack of the preposition, the sluice in (24) can only refer to the subject while that in (23) refers to whoever does the massage. Thus, speakers have to add the relevant preposition as an extra clue that helps trigger the intended interpretation. Of course, ambiguities may result where no preposition is available as in (25), which is potentially ambiguous between subject and implied object interpretation. (25) A: She called. B: Who? In this approach, deep anaphora, like do it or NCA, are by definition richer in clues, so however we manipulate an antecedent structure, interpretation will still be relatively easy. There is, therefore, no surface/deep dichotomy. Rather, the apparent di¤erence straightforwardly falls out from how much actual semantic and syntactic information an anaphor contains. Certainly, this solution should also explain why sluicing and VPE are taken to mostly resist discourse-initial use. Merchant (2004) notices this ever-problematic e¤ect and so does Stainton (2006). But the answer may be simpler than expected. VPE stands out as a recalcitrant anaphor and with good reason. If the modals began to lose their independent predicate status already in late OE (cf. Nagle 1989, Harris and

Whatever Happened to English Sluicing

53

Campbell 1995), speakers could not freely use them discourse-initially for their degraded semantic content. As the modals progress into replacing the OE morphological subjunctive, some remnants of their former meanings linger on. It is no accident that among the primary auxiliaries – marked just for tense and aspect – only do is attested discourse-initially (cf. Pullum 2000). Modal use is conceivably more informative if not su‰cient in itself. It is only after the auxiliaries become associated with precise meanings that they can refer to situational contexts. Note also that for an anaphor to so refer, we need appropriate clues. This process could well have taken some time, delaying the appearance of pragmatic control as a legitimate relationship in VPE and fixing its formulaic status. Now, is sluicing like VPE? Not quite. Other e¤ects that Merchant and Stainton point to in connection with sluicing are considerably less constrained use of stripping, Bare Argument Ellipsis, or simply fragments. We naturally evoke them discourse-initially to identify, explain, or request things, or to indicate location, destination, and manner. (26) [(From The Hudsucker Proxy.) Barnes gets o¤ an elevator, looking for Mr. Musburger’s o‰ce. He addresses Al, who is working in the hallway:] Barnes: Mr. Musburger’s o‰ce? Al: [Points] Al: Not that way. Through the door. (Stainton 2006a: 109) Predictably, there is a preference for PPs and NPs over bare nouns – if only because they o¤er more clues. Sluices can be viewed as interrogative counterparts of such fragments. If so, we would expect more informative wh-phrases, i.e., a pattern along the lines of (P) þ wh-phrase þ N.11 Indeed, of the four instances I identified in the ARCHER corpus, three follow this pattern; the fourth uses where.

11. Here and throughout, I adopt the terminology used in Hofmeister et al. (2007), Arnon et al. (in press) and Sag et al. (2008) in place of D-linked phrases (Pesetsky 1987). It has been demonstrated that D-linking is irrelevant to processing di‰culty (Hofmeister et al. 2007). The distinction between D-linked and non-D-linked phrases is just as irrelevant to sluicing in that discourse-initial use admits more informative sluices, which are clearly not D-linked; see example (28).

54

Joanna Nykiel

(27) I left her with a porter on the Pennsylvania side. [‘‘Good-by, sid, dear,’’] she said, and though I was startled when she called me ‘‘dear,’’ I felt that I had known her for a long, long time. [‘‘Where to now, sir?’’] Williams asked me. It was ten minutes before twelve but Dottie Peale had asked me to meet her at the o‰ce early. It was always a part of Dottie’s stage e¤ect to show o¤ that o‰ce of hers at Peale House. (1951. Marquand, John Phillips Melville Goodwin, U.S.A.) (28) There was a sense of Christmas about the travellers and the people who were at the terminus to meet them. The porter who came to the carriage door reminded Trefusis by his manner and voice that the season was one at which it becomes a gentleman to be festive and liberal. [‘‘What luggage, sir? Hansom or fourwheel, sir?’’] (1887. Shaw, George Bernard. An Unsocial Socialist. 7/3/91.) Far from being a troublesome exception to my view, where, and possibly why, have solid enough meanings of destination/location and explanation, respectively, that we can use them unambiguously. You might now ask why the number of discourse-initial sluices is so low if they are reasonably productive. The productivity of sluicing and Bare Argument Ellipsis should be checked against that of do it and NCA. The former are productive for as few clues as they make available. The approach suggested is well supported by the data in that at least 60% of the relationships are totally or partially non-syntactic in all the periods except for OE. This fact leaves little reason to insist on any more structure in a sluice than is visible. Instead, what seems promising is an inquiry into both the distribution of bare wh-phrases versus the (P) þ whphrase þ N patterns and the relationships that support their use in order to better evaluate the claims made here. 5. Conclusion I have reported on a phenomenon that may turn out to show a distinct stability throughout its history. So far it has. Such evidence requires a revision of the prior notions of anaphora and the dichotomies therein. To the extent that anaphors vary in the amount of information that they make available, they cannot be expected to show the exact same patterns of

Whatever Happened to English Sluicing

55

behavior across the set. Nothing more is needed by way of explanation; a forced distinction would indeed cloud the picture. The second argument presented in this paper is more challenging. The derivational/non-sentential debate has a long history. Both Ross and Merchant o¤er compelling counter-evidence to my analysis, but wherever a sluice copies the case that the corresponding argument in the antecedent carries, it does not necessarily pick it up from its pre-deletion structure. It is even di‰cult to imagine that, in coherent discourse, speakers would alter the case when referring back to an argument embedded in a certain proposition unless they wish to introduce another proposition. But that would not be sluicing. My solution correctly captures this, and as long as an antecedent can license the case of a sluice, there is no diachronically justified need for positing a full syntactic structure. There remains one problem to be solved. Recall Merchant’s insistence on a connection between omission of prepositions in sluicing and availability of preposition stranding. By my rough estimates, these e¤ects do not go together. Rather, presence of a preposition is a conscious move to reduce confusion; its absence – definitely also possible in non-prepositionstranding languages – removes redundant clues.

References Dictionary of Old English 2003 Edited by Antonette diPaolo Healy, Joan Holland, Dorothy Haines, Ian McDougall, and David McDougall. Toronto: The University of Toronto. . Penn-Helsinki Parsed Corpus of Middle English 2000 Edited by Anthony Kroch and Ann Taylor. 2nd ed. Philadelphia, PA: University of Pennsylvania, Helsinki: University of Helsinki. Middle English Compendium 1954–2000 Edited by Frances McSparran, Paul Scha¤ner, and John Latta. Michigan: The University of Michigan Press. . ICAME Collection of English Language Corpora. Corpus of Early English Correspondence Sampler CEECS 1999 Edited by Knut Hofland, Anne Lindebjerg, and Jørn Thunestvedt. Bergen: The HIT Centre. University of Bergen. Shakespeare, William 1989–1991 The Complete Works. Portions copyright 6 Creative Multimedia Corp.

56

Joanna Nykiel

Helsinki Corpus of English Texts 1996 3rd ed. Helsinki: University of Helsinki. ARCHER-3. A Representative Corpus of Historical English Registers 3 1990–1993/ Northern Arizona University, University of Southern California, 2002/2007 University of Freiburg, University of Heidelberg, University of Helsinki, Uppsala University, University of Michigan, and University of Manchester. Oxford English Dictionary 2000–2008 Edited by John Simpson. Oxford: Oxford University Press. .

References Allen, Cynthia 1980

Movement and deletion in Old English. Linguistic Inquiry 11(2): 261–323. Arnon, Inbal, Neal Snider, Philip Hofmeister, T. Florian Jaeger, and Ivan A. Sag In press Cross-linguistic variation in a processing account: The case of multiple wh-questions. Proceedings of Berkley Linguistic Society 32 (2006). Chung, Sandra, William Ladusaw, and James McCloskey 1995 Sluicing and Logical Form. Natural Language Semantics 3: 239– 282. Culicover, Peter, and Ray Jackendo¤ 2005 Simpler Syntax. New York, NY: Oxford University Press. Everaert, Martin and Henk van Riemsdijk (eds) 2006 The Syntax Companion. Blackwell: London. Featherston, Sam, and Wolfgang Sternefeld (eds) 2007 Roots: Linguistics in Search of its Evidential Base. Berlin: de Gruyter. Fiengo, Robert, and Robert May 1994 Indices and Identity. Cambridge, Mass: The MIT Press. Gergel, Remus 2004 Short-distance reanalysis of Middle English modals: Evidence from ellipsis. Studia Linguistica 58(2): 53–87. Ginzburg, Jonathan and Ivan A. Sag 2000 Interrogative Investigations. The Form, Meaning and Use of English Interrogatives. Stanford, California: CSLI Publications. Hankamer, Jorge, and Ivan A. Sag 1976 Deep and surface anaphora. Linguistic Inquiry 7(3): 391–426. Hankamer, Jorge 1978 On the nontransformational derivation of some null VP anaphors. Linguistic Inquiry 9(1): 66–74.

Whatever Happened to English Sluicing Hardt, Daniel 1993

57

Verb phrase ellipsis: Form, meaning and processing. Doctoral dissertation, University of Pennsylvania, Philadelphia, PA.

Hardt, Daniel 2005 Inference, ellipsis and deaccenting. Amsterdam Colloquium 2005. Harris, Alice, and Lyle Campbell 1995 Historical Syntax in Cross-linguistic Perspective. Cambridge: Cambridge University Press. Hofmeister, Philip, T. Florian Jaeger, Ivan A. Sag, Inbal Arnon, and Neal Snider 2007 Locality and accessibility in wh-questions. In Roots: Linguistics in Search of its Evidential Base, edited by Sam Featherston and Wolfgang Sternefeld. Berlin: de Gruyter. Kehler, Andrew 2002 Coherence, Reference, and the Theory of Grammar. Stanford, CA: CSLI Publications. Kehler, Andrew, Laura Kertz, Hannah Rohde, and Je¤rey L. Elman 2008 Coherence and coreference revisited. Journal of Semantics 25: 1–44. Kertz, Laura, Andrew Kehler, and Je¤rey L. Elman 2006 Grammatical and coherence-based factors in pronoun interpretation. Proceedings of the 28th Annual Conference of the Cognitive Science Society, 1605–1610. Kitagawa, Yoshihisha 1991 Copying identity. Natural Language & Linguistic Theory 9: 497– 536. Langacker, Ronald 1977 Syntactic reanalysis. In Mechanisms of Syntactic Change, edited by Charles N. Li, 57–139. Austin: University of Texas Press. Lasnik, Howard 2007 On ellipsis: The PF approach to missing constituents. University of Maryland Working Papers in Linguistics 15: 143–153. College Park, MD: UMWPiL. Li, Charles N. (ed) 1977 Mechanisms of Syntactic Change, Austin: University of Texas Press. Lightfoot, David 1979 Principles of Diachronic Syntax. Cambridge: Cambridge University Press. Merchant, Jason 2001 The Syntax of Silence: Sluicing, Islands, and the Theory of Ellipsis. Oxford: Oxford University Press. Merchant, Jason 2004 Fragments and ellipsis. Linguistics and Philosophy 27: 661–738. Merchant, Jason 2006 ‘‘Small structures’’: A sententialist perspective. In The Syntax of Nonsententials, edited by Kate Paesani, Ljiliana Progovac,

58

Joanna Nykiel Eugenia Casielles, and Ellen Barton, 73–91. Amsterdam: John Benjamins Publishing Company.

Merchant, Jason 2006 Sluicing. In The Syntax Companion, edited by Martin Everaert and Henk van Riemsdijk, 269–289. Blackwell: London. Merchant, Jason 2007 Three kinds of ellipsis: Syntactic, semantic, pragmatic? Prepared for the Erniefest, 4–5 Oct 2008, Rutgers. Nagle, Stephen 1989 Inferential change and syntactic modality in English. Frankfurt am Main: Verlag Peter Lang. Nykiel, Joanna 2006 Ellipsis in Shakespeare’s syntax. Doctoral dissertation. University of Silesia, Poland. Nykiel, Joanna 2007 Resolution of ellipsis: evidence from Old English sluicing. Studia Anglica Posnaniensia 43: 111–125. Paesani, Kate, Ljiliana Progovac, Eugenia Casielles, Ellen Barton (eds) 2006 The Syntax of Nonsententials. Amsterdam: John Benjamins Publishing Company. Pesetsky, David 1987 Wh-in-situ: Movement and unselective binding. In The Representation of (In)definiteness, edited by Eric J. Reuland and Alice G. B. ter Meulen, 98–129. Cambridge, Mass: The MIT Press. Pullum, Geo¤rey 2000 Hankamer was! In Sandy Chung, Jim McCloskey, and Nathan Sanders, (eds). Jorge Hanmaker WebFest. http://ling.ucsc.edu/Jorge/ Reuland, Eric J. and Alice G. B. ter Meulen (eds) 1987 The Representation of (In)definiteness. Cambridge, Mass: The MIT Press. Rohde, Hannah, Andrew Kehler, and Je¤rey L. Elman 2006 Event structure and discourse coherence biases in pronoun interpretation. Proceedings of the 29th Annual Meeting of the Cognitive Science Society. Ross, John 1967 Constraints on variables in syntax. Doctoral dissertation, Massachusetts Institute of Technology. At http://hdl.handle.net/1721.1/ 15166). Sag, Ivan A., and Jorge Hankamer 1984 Toward a theory of anaphoric processing. Linguistics and Philosophy 7: 325–345. Sag, Ivan A. 2006 What’s LF got to do with it? Workshop on Ellipsis. Annual Meeting of the Linguistic Society of America, Albuquerque, NM.

Whatever Happened to English Sluicing

59

Sag, Ivan A., Philip Hofmeister, Inbal Arnon, T. Florian Jaeger, and Neal Snider 2008 Processing accounts for superiority e¤ects. Manuscript, Stanford University. Stainton, Robert 2006a Neither fragments nor ellipsis. In The Syntax of Nonsententials, edited by Kate Paesani, Ljiliana Progovac, Eugenia Casielles, and Ellen Barton, 93–116. Philadelphia: John Benjamins Publishing Company. Stainton, Robert 2006b Words and Thoughts. Oxford: Oxford University Press. Timberlake, Alan 1977 Reanalysis and actualization in syntactic change. In Mechanisms of Syntactic Change, edited by Charles N. Li, 141–80. Austin: University of Texas Press. Ward, Gregory, Richard Sproat, and Gail McKoon 1991 A pragmatic analysis of so-called anaphoric islands. Language 67: 439–474. Webber, Bonnie 1978 A formal approach to discourse anaphora. Doctoral dissertation. Harvard University. Williams, Edwin 1977 Discourse and Logical Form. Linguistic Inquiry 8: 101–139.

Commentary on Nykiel, Whatever Happened to English Sluicing Elizabeth Closs Traugott

My comments will focus on two issues raised in the introduction to Joanna Nykiel’s very rich paper on the history of sluicing since they frame her interpretation of the findings. The first, which I touch on only briefly, is the ‘‘Recency Illusion’’; the second is stability over a millennium. It appears that the title ‘‘Whatever happened to English sluicing’’ and some of the assumptions in the introduction arise out of a conflation of two things. One is that in the history of research on sluicing in contemporary English and other languages, there has been a shift away from viewing it as a strictly syntactic phenomenon to viewing it as a partially pragmatic one. This is the kind of development that often occurs as researchers elaborate on the larger picture. The other is that this history of research results is projected onto the history of the linguistic structure itself so that there is an assumption that a ‘‘relaxation of structural identity’’ will be found in the historical data. The outcomes of research should without question guide hypotheses about the history of the construction under consideration. Assuming that they will be parallel is, however, what Zwicky (2005) has called the ‘‘Recency Illusion,’’ whereby people believe that ‘‘things YOU have noticed only recently are in fact recent.’’ This assumption is often mistaken, e.g., the claim that adverbial all as in She was all sad is recent (Waksler 2001) as discussed in Buchstaller and Traugott (2006). Any such assumption should be converted into a hypothesis to be tested, and test the hypothesis is what Nykiel’s paper essentially does in the rest of the paper. Nykiel’s prime objective is to show that the choice of antecedents under sluicing is not, and has never been, as constrained as has been assumed by generative grammarians and that sluicing has more parallels with other kinds of anaphors than has usually be recognized. She demonstrates that there is corpus evidence from Old English that sluicing has had both syntactic and pragmatic characteristics, specifically ‘‘merger’’ and ‘‘sprouting.’’ The ratio of merger to sprouting is very di¤erent in Old English and later periods. In Old English, over 80% of the examples are of merger, the rest of sprouting, whereas the ratio is almost reversed in Middle

Commentary on Nykiel

61

English, and somewhat leveled out in the Early Modern period. It appears therefore that merger and sprouting have always been available in English, and, from the perspective of English as a whole, there has been relative stability from Middle English on. Nykiel raises the question how much stability there is in language. It appears that there is in fact quite a lot if one cares to look for it (for early investigations beyond Timberlake and Langacker, which she cites, see Romaine 1982, Milroy 1992). Of stable sociolinguistic variation Labov has said: Stable, long-term variation that persists over many centuries in much the same form is perhaps even more common than changes which go to completion . . . [F]rom the sociolinguistic point of view . . . the absence of change has the most important consequences for our understanding of linguistic structure. (Labov 2001: 75, 85)

The stability that Nykiel has uncovered in the history of sluicing is a case of persistence (Milroy 1992). She did not investigate age, social class, or style (nor indeed would the data allow this), so we do not have a case of stable variation in the traditional sociolinguistic sense. In the latter the expectation is that: each age cohort of the same class, gender, ethnic background, and other social characteristics, will be similar to older and younger groups in the use of variants and the amount of style-shifting. (Chambers 1995: 107, cited in Raumolin-Brunberg 2002: 102)

Raumolin-Brunberg (2002) tests claims like these, which are based on agegrading, against evidence from real time provided by historical data, and finds that they can be supported, despite the many problems attendant on historical corpora (written rather than spoken, often only partially representative of language-users of the time, etc.). Studies of sociolinguistic variables have tended to highlight regular stratification (see Labov 2001). By contrast, Nykiel’s study of sluicing suggests that variation is much less regular when measured against genre. This is no doubt in part because very di¤erent conventions arise at different times with respect to genres (see e.g., Culpeper and Kyto¨ 2000, Fitzmaurice and Taavitsainen 2007). In Nykiel’s data, fiction favors merger least (Tables 7 and 8). This seems hardly surprising considering that fiction often includes dialogue and is generally considered closer to spoken language and, therefore, more likely to evidence pragmatic factors than many other genres (e.g., Biber 1988, discussing twentieth century

62

Elizabeth Closs Traugott

fiction). However, Culpeper and Kyto¨ (2000) found that fiction tended to have fewer oral characteristics than trials in their data (which is limited to the period 1600–1720). They speculated that this might be because fiction, at least at that period, tends to require a significant amount of information about character and plot to be conveyed (p. 195). In future work it would be worthwhile to develop diagnostics for, and characterizations of, the kinds of stable variation that can be correlated with genres where expressions like sluicing that operate at the interface of syntax and pragmatics are concerned. What seems striking to me is the fact that there is so little sluicing in the fiction represented in the Early Modern English part of the Helsinki corpus (in Table 7 16.66% merger and 83.33% sprouting disguise the raw figures of 1 and 5 respectively). Possibly sluicing was simply not regarded as a feature to be exploited – as Culpeper and Kyto¨ point out (2000: 195), authors select di¤erent features to convey ‘‘spokenness’’ in di¤erent genres. A further research question is whether the discovery of stable variation requires the investigator to posit a prior stage without such variation. Absent data prior to Old English, it does not appear necessary in the case of sluicing to posit a stage in which merger alone can be identified, even though over 80% of the examples show this use in Old English. Likewise, although object-clause zero complementizer (‘‘that-deletion’’) is very rare in Old English, its presence in texts nevertheless suggests it was always in variation with that. Its dramatic increase in frequency in later Middle English is highly dependent on factors such as the verb, personal pronoun, and text type (Rissanen 1991). Therefore, issues of saltation (macrochange) or gradualness (micro-change) that Nykiel raises in the introduction appear not to be particularly relevant. What is relevant is that, when correlated with genre, stability may show considerable frequency fluctuations over time. Various types of more highly pragmatic sluicing appear during the history of English. Inference without any syntactic antecedent appears first in Late Middle English, how-mismatch and pragmatic control in the Early Modern period, as evidenced by ARCHER. All three are of low frequency. They may perhaps be seen as competing with sprouting since the percentages of sprouting are lower than in the earlier periods. Nykiel comments, ‘‘English seems to have shifted in a surprising direction,’’ without giving reasons for the surprise. Perhaps she has in mind arguments from Givo´n (1979: 209) on developments from ‘‘discourse > syntax. . . .’’ It is true that structurally English became more ‘‘syntacticized’’ in Middle English, if we measure syntacticization by such phenomena as the shift

Commentary on Nykiel

63

from topic-focus structuring to obligatory use of a subject or subject-slot filler. However, it is also the case that there has been a tendency in writing for certain sub-domains of grammar to become less formal over time (see Biber and Finegan 1989, Mair 2006). It might be worth investigating whether the modest rise of more pragmatic uses of sluicing is correlated with changing practices associated with writing or with specific genres.

References Biber, Douglas 1988

Variation across Speech and Writing. Cambridge: Cambridge University Press. Biber, Douglas and Edward Finegan 1989 Drift and the evolution of English style: A history of three genres. Language 65: 487–517. Buchstaller, Isabelle and Elizabeth Closs Traugott 2006 ‘‘The lady was demonyak: Historical aspects of adverb ALL’’. English Language and Linguistics 10: 345–370. Chambers, J. K. 1995 Sociolinguistic Theory: Linguistic Variation and its Social Significance. Oxford: Blackwell. Culpeper, Jonathan and Merja Kyto¨ 2000 Data in historical pragmatics: Spoken interaction (re)cast as writing. Journal of Historical Pragmatics 1: 175–199. Fitzmaurice, Susan M. and Irma Taavitsainen, eds 2007 Methods in Historical Pragmatics. Berlin: Mouton de Gruyter. Givo´n, Talmy 1979 On Understanding Grammar. New York: Academic Press. Labov, William 2002 Principles of Linguistic Change, Vol. 2: Social Factors. Malden, MA: Blackwell. Mair, Christian 2006 Twentieth-century English: History, Variation and Standardization. Cambridge: Cambridge University Press. Raumolin-Brunberg, Helena 2002 Stable variation and historical linguistics. In Helena RaumolinBrunberg, Minna Nevala, Arja Nurmi, and Matti Rissanen, eds., Variation Past and Present. VARIENG Studies on English for Terttu Nevalainen, 101–116. Helsinki: Socie´te´ Ne´ophilologique. Rissanen, Matti 1991 On the history of that/zero as object clause links in English. In Karin Aijmer and Bengt Altenberg, eds., English Corpus Linguistics: Studies in Honour of Jan Svartvik, 272–289. London: Longman.

64

Elizabeth Closs Traugott

Romaine, Suzanne 1982 Socio-historical Linguistics: Its Status and Methodology. Cambridge: Cambridge University Press. Waksler, Rachelle 2001 A new ALL construction. American Speech 76: 128–138. Zwicky, Arnold 2005 More illusions. Language Log, August 7th. .

Response to Commentary by Traugott Joanna Nykiel

In her commentary, Elizabeth Traugott first notes that the recognition of a hitherto unnoticed phenomenon may lead linguists to misinterpret the phenomenon as a recent one. To prevent such misinterpretation we ought to test every hypothesis formulated in this way. But beyond exploring the hypothesis that sluicing has over time become a less syntactic construction, my research seeks to answer a rather di¤erent need. Analyses of sluicing, and indeed of elliptical constructions, have always su¤ered from lack of an empirically adequate account of the facts.1 It comes as little surprise that without statistics on what kinds of sluicing appear in actual use, a few scattered examples could be easily dismissed as infrequent and insignificant or used to counter a previously established line of argument. In bridging this gap, my purpose was three-fold: (i) to provide statistical evidence for and an account of the distribution of sluicing, (ii) to verify whether there ever was any basis for teasing apart surface and deep anaphora, and (iii) to indeed test the hypothesis mentioned above. I have no quarrel with Traugott’s second comment that the stability in the behavior of sluicing could be investigated via analysis of sociolinguistic variables and writing conventions. This kind of endeavor would, however, encounter the usual problems that beset investigation of the oldest sources: a bias toward the written variety, male authors and sometimes toward prose, as is the case with Old English. My analysis of sluicing shows that with the transition from Old English to Middle English comes a drastic change in the ratio of merger to sprouting, clearly the most intriguing of the few shifts. An explanation may lie in the fact that social strata and genres are poorly represented by the available Old English material as compared to the relative richness of the Middle English data. Unfortunately, a truly sociolinguistic comparison of the two periods remains problematic. 1. Beecher’s (2006, 2007) investigation of sluiced prepositional phrases in presentday American English is the closest I know of to a corpus-based analysis.

66

Joanna Nykiel

What issues like this illustrate is the necessity to go beyond English, and any individual language under investigation, whenever stability or change in the history of a given construction seems surprising. A discovery of what typological similarities and di¤erences there are can help us better understand the historical conditions and discourse contexts which might favor a particular construction. Some researchers (cf. Bybee 2008, Hopper 2008) already emphasize that connections between constructions and the discourse contexts in which they arise should be explored across a variety of languages. Typological research in turn avoids the danger of modeling the architecture of grammar on English; this danger is real since English exhibits a high degree of exceptionality (Traugott 2008).

References Beecher, Henry 2006 Pragmatic licensing of sluiced prepositional phrases. Research paper II. Available at:

Beecher, Henry 2007 Pragmatic inference in the interpretation of sluiced prepositional phrases. In Proceedings of the Fifth Cambridge Postgraduate Conference in Language Research, 9–16. Bybee, Joan 2008 Formal universals as emergent phenomena: the origins of structure preservation. In Linguistic Universals and Language Change, edited by Je¤ Good, 108–124. Oxford: Oxford University Press. Good, Je¤, ed. 2008 Linguistic Universals and Language Change. Oxford: Oxford University Press. Hopper, Paul J. 2008 Emergent serialization in English: pragmatics and typology. In Linguistic Universals and Language Change, edited by Je¤ Good, 253–286. Oxford: Oxford University Press. Traugott, Elizabeth Closs 2008 Presidential address. International Society for the Linguistics of English Inaugural Meeting. Freiburg, Germany.

Notion of Direction and Old English Prepositional Phrases Olga Thomason

Since prepositions present a versatile lexical category, interest in them and especially their semantics is steadily growing (Luraghi 2003, Tyler and Evans 2003, Segen 2001). In the present paper, we take a closer look at Old English constructions that designate direction. There are several prepositional phrases that mark direction in Old English: ‘into’ in þ Acc, on þ Acc, binnan þ Acc, innan þ Acc ‘onto’ on þ Acc, ofer þ Acc ‘up to’ æt þ Acc or oþ þ Acc (rare oþ þ Dat) ‘to, toward’ to# þ Acc/Dat/Gen, wiþ þ Acc/Dat/Gen, onge#an þ Acc/Dat, to#ge#an þ Dat. It is the last group of prepositions, designating a more general notion ‘to, toward,’ that is the scope of this study. The questions that this investigation will try to answer are: why does Old English reserve so many prepositional phrases to mark such a general concept as ‘to, toward’ and what are the possible reasons for such diversity in case governance of these prepositions? The vast majority of research analyzing prepositional semantics is conducted using Present Day English data as a base. Semantic investigations that examine materials of earlier stages of language development (e.g., Old English or Middle English) are less common. Textbooks on Old and Middle English often give a brief list of prepositions with some short notes on their usage, referring readers to glossaries and dictionaries for further information simply because the study of prepositional semantics is not seen as being necessary when one introduces a new language (Baker 2003: 101–2). These facts are unfortunate since detailed and well-balanced synchronic and diachronic explorations of data from earlier periods of language development (and of the English language in particular) could shed more light on such a complex notion as prepositional semantics. For example, diachronic study of prepositional meanings shows that spatial

68

Olga Thomason

roles develop earlier than other semantic functions (Luraghi 2003: 18). The present study joins those that are making every e¤ort to fill this gap and add to the knowledge of prepositional semantics and usage. Spatial meanings deserve special attention not only because they tend to be the original semantics of prepositions but also because they frequently present a background for structuring many other conceptual domains. It has already been pointed out that speakers use spatial expressions in constructions that designate temporal concepts, kinship relations, body parts, and some others (Levinson 2003: 16–8). For example, the English prepositions in and on are often used to mark spatial and temporal concepts: cf. in the classroom versus in the evening and on the plate versus on Sunday. Linguistic discussions of prepositional semantics often focus on relations between prepositional meanings and human physical experience of the surrounding world. Researchers attempt to find a cognitive stimulus for the use of prepositions. In the course of their everyday life, humans interrelate in and with the surrounding world. Recurring events, structures, arrangements, etc. become encoded in the human mind as abstract notions and concepts. As a consequence, several linguists argue that the semantic structure of prepositional phrases arises from human conceptual structure or mirrors it (Cienki 1989, Cuyckens 1997, Kemmerer 2005, Tyler and Evans 2001). Some linguists employ geometric or functional specifics of constructions in which spatial prepositional phrases are used in order to pinpoint certain semantic particulars (Coventry 2001). Throughout our discussion, we will strive to show in which manner various cognitive stimuli trigger or support particular semantic developments of the prepositional constructions designating the notion ‘to, toward’. We will pay particular attention to the functional and syntagmatic specifics of these prepositions in our attempt to show that even though in Old English spatial relations are primarily coded in the semantics of prepositions, some spatial information is derived from the meanings of verbs with which these prepositions are used and from the functions of cases which these prepositions govern. The domain of spatial prepositional semantics is characterized by a high level of variation found in means of expressing spatial concepts (Coventry 2001, Rudanko 1995, Van der Zee and Slack 2003). It is common for the same spatial notion to be expressed by a variety of prepositional phrases. On the other hand, many languages demonstrate that the same prepositional phrase can designate a number of spatial concepts. Despite the fact that languages have a finite number of spatial prepositions, it is often problematic to isolate exact rules for their usage. Old

Notion of Direction and Old English Prepositional Phrases

69

English is not an exception in this case; variation of prepositional phrases is expected in this language. However, it does demonstrate a particularly high level of variation among prepositional constructions denoting the meaning ‘to, toward’. This study o¤ers a detailed examination of this variation and attempts to find potential reasons for such diversity. Even if we do not take into consideration more abstract extensions of the directional concept, such as purpose, order/command, instruction, management, result, etc., and concentrate only on spatial characteristics of this notion, they are still exceptionally complex and di‰cult to describe precisely. Direction is considered to be one of the basic spatial notions (in line with location and source). Direction is commonly thought of and represented as a line or an arrow along which something moves, but it is also a point, area or region toward which some entity is heading. In this light, it is not surprising that many prepositional phrases chosen to mark direction also designate a static location. In fact, it is sometimes difficult to draw a definite border between these two concepts since in many languages, the same prepositional phrase can designate both direction and location. Cases of such syncretism have already been documented. Silvia Luraghi o¤ers a good summary of investigations that depict loss or preservation of distinction among the three concepts – direction, location, and source (Luraghi 2003: 20–2). The notion of direction is multi-layered and closely connected with the concept of containment (the meaning ‘into’), the concept of surface (the meaning ‘onto’), and the concept of proximity (the meaning ‘up to’): cf. jumping into the box, jumping onto the box, jumping up to the box. The first two notions presuppose the presence of contact while the third one marks instances when a moving object does not come into a full contact with its final goal. All these notions are distinct in Old English. In the case where the concept of direction is connected with the notions of containment and surface, Old English responds with a well established set of prepositional phrases where the di¤erence between location and direction is marked with the help of the dative and accusative cases (with minor variations in case forms): cf. ‘in’ in þ Dat, on þ Dat, binnan þ Dat, innan þ Dat versus ‘into’ in þ Acc, on þ Acc, binnan þ Acc, innan þ Acc and also ‘on’ on þ Dat, ofer þ Dat versus ‘onto’ on þ Acc, ofer þ Acc. These are not the only prepositional phrases that are used to mark the stated notions (e.g., to# þ Dat can also designate the locations ‘in’ and ‘on’), but they are the primary ones. Such a way of marking is expected since it reflects not only the primary meanings of the prepositions per se but also the original semantics of cases: the designation of destination for the accusative case and the designation of static location for the

70

Olga Thomason

dative case. Due to syncretism in Old English (as in all other Germanic languages), the dative case has a variety of functions, reflecting the semantics of the original Indo-European cases dative, ablative, locative, and instrumental. The meaning ‘up to,’ usually marked in Old English with æt þ Acc or oþ þ Acc (rare oþ þ Dat), presupposes the absence of contact, the concepts of containment and surface do not play any salient role and, thus, are not marked. These prepositions govern the expected accusative case, which rea‰rms the designation of direction. In this light, the more general direction notion ‘to, toward’ is problematic since these prepositions govern a diversity of cases: to# þ Acc/Dat/Gen, wiþ þ Acc/Dat/ Gen, onge#an þ Acc/Dat, to#ge#an þ Dat. But the three concepts discussed above may not be the only ones that a¤ect the semantics of the prepositional phrases denoting direction. The pragmatic nature of the directional concept is more multifaceted. Considering that this notion is associated with the motion in the direction of something which is in the area or vicinity of, near, close to, or facing something, the spatial notions ‘in,’ ‘near, next to,’ ‘opposite to, against,’ and ‘before, in front of ’ should be salient and have influence on the choice of the prepositional phrases. Let’s take a closer look at the prepositional phrases that denote the meaning ‘to, toward’ in Old English, focusing only on literal spatial usages of these constructions. The prepositional phrase to# þ Dat is the one that is used primarily to mark a general direction ‘to, toward’. It commonly occurs in combination with motions of coming, running, falling, etc. as well as with verbs of bringing, taking, sending, etc. (in these instances the directional semantics of the prepositional phrase are reinstated by a verb). (1) God him com to# (Gen 20, 3 Mk. Skt. 5, 21) ‘God came to him’ Sende se Fæder his sunu to# cwa#le (Homl. Th. ii. 6, 17) ‘Father sent his son to death’ In the spatial usages of to# þ Dat, the object toward which a motion/action is directed could be reached or not: (2) Bryne stigeþ to# heofonum (Exon. Th. 233, 7; Ph. 521) ‘Flame reaches heaven.’ Hie onhnigon to# ðam herige (Cd. Th. 227, 3; Dan. 181) ‘They bowed to the altar.’

Notion of Direction and Old English Prepositional Phrases

71

There are also instances where to# þ Dat, designating direction, is combined not with motion verbs but with verbs whose semantics imply a certain motion. For example, we find cases like this in instances where to# þ Dat becomes a complement to verbs that denote looking or listening. (3) Beseoh to# me# (Ps. Th. 12, 3.) ‘Look at me.’ These verbs do not mark a motion per se, but it is implied that in order for somebody to look in a certain direction or listen to somebody, one has to turn toward this direction/this person. The question to answer here is why the preposition to# governs the dative instead of the expected accusative. It is plausible that the semantics of to# þ Dat receive the extension to a directional meaning only because of the meaning of these verbs since the designation of directionality is inherent to their semantics. Propositions of this kind have been made earlier by those researchers who believe that it is the semantic and syntactic qualities of verbs that dictate what kind of meaning a nominal or a prepositional phrase would have (Fillmore 1968: 10–3). Another way to explain the extension to a directional meaning is to derive it from the meaning of the dative case that the preposition to# governs in such passages. It has been already mentioned that the Old English dative is an amalgamation of a number of Indo-European cases, the locative being one of them. There are known instances when the locative case expresses direction (without a preposition): cf. Skt. papa#ta medinya#m (MBh.) ‘he fell to (so as to be upon) the earth’ (Whitney 1993: 103). Thus, it is the semantics of the Old English dative case that could condition the location-todirection semantic development of to# þ Dat. On the other hand, the meaning ‘toward’ seems to be original for the preposition to# as it is preserved and evidenced by several Indo-European cognates of this preposition: cf. Germ. zu ‘to’, L. do-nec ‘as long as’, Gk. -de, O.Ir. do, Lith. da- ‘to, toward’ (Holthausen 1963: 350). Therefore, one can argue (and we are in favor of this argument) that there was no location-to-direction extension here and the notion of direction is already at the core of the semantics of the preposition itself and this concept is simply reinforced by other members of collocations (i.e. verbs and case forms). Any proposition made seems to be speculative at this point; however, one thing is certain: directionality is the primary function of to# þ Dat be it due to the original meaning of the preposition itself or due to a mixture of factors. There are many instances where to# þ Dat designates static location specifying such meanings as ‘near, next to,’ ‘by,’ and even ‘in,’ ‘on’:

72

Olga Thomason

(4) Hı # man bebyrigde to# hyre were (Homl. Th. i. 318, 1: ii. 188, 5) ‘She was buried next to her husband.’ He# gesette Iudas to# bisceope to# Godes temple (Elen. Kmbl. 2114; El 1058) ‘He made Judas to sit down next to the high priest in the God’s temple.’ To# horse (Exon. Th. 298, 7; Cra¨. 81) ‘on horseback’ Since in the majority of cases where to# þ Dat marks a static location notion it specifies the meaning ‘near, next to’, one can talk about saliency of the proximity notion for the semantics of this prepositional phrase. Even though we do find examples where it marks the location of an object/person within or on top of the boundaries of some locations (see the last two passages in (4)), such instances are rare and could be considered the result of secondary development. In some cases, the semantics of to# þ Dat are extended even to the designation of source where it is combined with verbs denoting seeking, expectation, or attainment: (5) To# eorðan æ#tes tilian (Cd. Th. 94, 5; Gen. 1557: 59, 31; Gen. 972) ‘to obtain from the earth of food’ This function is probably secondary, acquired on the basis of the location usages of to# þ Dat creating, consequentially, the following chain of extensions: ‘next to the area’ > ‘at, in the area’ > ‘from the area’. The semantics of to# þ Dat are an interesting example demonstrating syncretism of all three basic spatial roles – direction, location, and source. The construction to# þ Gen can also mark the direction ‘to, toward’ although not as frequently as to# þ Dat and only in combination with a demonstrative or interrogative pronoun: (6) To# ðæs gingran þider ealle urnon ðæ#r se e#ca wæs (Cd. Th. 298, II, Sat. 531) ‘Thither all the disciples ran, to the place where the Eternal were.’ To# hwæs hı# gearwe bæron (Cd. Th. 190, I; Exod. 192.) ‘Whither they should bear their arms.’ We also find several instances where to# þ Gen designates location. However, in the majority of such instances, to# þ Gen is part of a bigger adverbial construction to# middes ‘in the midst’:

Notion of Direction and Old English Prepositional Phrases

73

(7) He# wæs to# middes wætres (Homl. Skt. ii. 30, 176) ‘He was in the midst of the stream.’ In instances such as these, to# þ Gen becomes a part of a set construction, and it is hard to make judgments about the original location meaning of this phrase and its conceptual connections since its semantics are already fossilized in the semantics of the whole construction. The only spatial meaning that to# þ Acc has is the designation of the direction ‘to, toward’. Such instances are notably less frequent than similar directional occurrences of to# þ Dat. This situation is surprising since the semantics of the accusative are regularly connected with the designation of direction, and it is probably for this reason that we do not find any passages where to# þ Acc comes to mark a static location: (8) He# lea´t to# ðæs ca#seres ea´re (Homl. Th. i. 376, 28) ‘He leaned toward the ear of the emperor.’ To# is an excellent example that demonstrates the complexity of prepositional semantics at its best. The explored material shows that to# þ Dat is the primary construction among those with to# that is used to denote the direction ‘to, toward’ due to a combination of factors: the original meaning of the preposition and the semantic support from motion verbs and the dative that also express directionality. The main static location meaning of to# þ Dat – ‘next to’ – suggests that the notion of proximity is a salient one for the speakers of Old English. This prepositional phrase exemplifies syncretism of the three basic local semantic roles – designation of direction, location, and source. To# þ Acc, the phrase that is expected to be a primary one for the denotation of direction, is in fact a less frequent variant even though its semantics are limited to the designation of this notion. To# þ Acc and to# þ Dat are probably in free variation at this stage of language development. Despite the fact that to# þ Gen can mark location and direction, its usage is restricted to a few syntactical constructions. The preposition wiþ is similar to the preposition to# as far as its governance and some semantic functions are concerned. It governs the dative, the accusative, and the genitive. In combination with all of these cases, wiþ can express directionality just like to#. Let’s take a closer look at the semantic specifics of wiþ and try to establish reasons why all three phrases, namely wiþ þ Dat, wiþ þ Gen, and wiþ þ Acc, came to designate the direction ‘to, toward,’ and why Old English preserves these prepositional phrases even though it already has to# þ Dat/Acc/Gen, which is versatile in itself.

74

Olga Thomason

Once again, governance of the dative case greatly enriches the prepositional semantics (compare examples (1)–(5) of to# þ Dat discussed earlier). Just like in the case with to# þ Dat, wiþ þ Dat can designate spatial relations specifying the notion ‘near’: (9) Hire lı c# hama resteþ wið Ro#mebirig on ðam wege ðe man nemneþ Latina (Shrn. 31, 28) ‘Her body rests near the city of Rome on the way which one calls Latin.’ But in addition to this meaning, wiþ þ Dat can also denote the location notion ‘against, opposite to’: (10) Sæweall uplang gesto#d wið Israhe#lum (Cd. Th. 197, 8; Exod. 303) ‘A high sea-wall stood against Israel.’ The meaning of wiþ þ Dat in this passage is not connected with hostility but rather once again with the concept of proximity as if stating that whatever is against/opposite to someone/something is necessarily next to/ near someone/something. Therefore, this passage could easily be understood as ‘a high sea-wall stood next to/near Israel’. But compare this passage with the following sentence: (11) Se wið mongum sto#d (Exon. Th. 121, 26; Gu# 294) ‘That one stood against many.’ In this instance, wiþ þ Dat does have a meaning that has a close association with the notion of hostility. The question is: is the designation of opposition an inherent feature of the semantics of this preposition, or is it a secondary development? Before we answer this question, let’s look at those instances where wiþ þ Dat expresses direction: (12) Scearp cymeþ sceo´ wiþ o#þrum, ecg wið ecge (Exon. Th. 385, 8; Ra¨. 4, 41) ‘A pointed cloud comes toward/against another, an edge toward/ against an edge.’ Streamas wundon sund wið sande (Beo. Th. 431; B. 213) ‘Streams whirled water toward/against the sand (¼ the sandy shore).’ Ongan ic steppan forð a#na wið englum (Cd. Th. 280, 1; Sat 249) ‘I alone stepped forth against the angels.’ Bordrand onswa#f wið ðam gryregieste (Beo. Th. 5113; B. 2560) ‘Shield swung forward against the terrible stranger.’

Notion of Direction and Old English Prepositional Phrases

75

In the first two examples, wiþ þ Dat could be rendered as both ‘toward’ and ‘against’ whereas in the second two passages, the interpretation is clearly ‘against’. It is important to note that in the last two sentences, there are other members (aside from the analyzed prepositional phrases) that also express the meaning ‘against’: ongan ‘towards, against’ and onswifan ‘turn against’ (the preposition/prefix on can also express the meaning ‘against’ in line with its more frequent senses ‘on’, ‘upon’). It is possible that it is these elements that ensure the correct reading of the construction as ‘against’ as if the usage of wiþ þ Dat alone is not enough to take care of this meaning. On the other hand, multiple cognates of wiþ preserve the meaning ‘against,’ thus showing that the notion of opposition (and proximity) is salient for the semantics of this preposition: cf. Goth. wiþra ‘against, opposite,’ OCS. vu˘toru˘ ‘other, second,’ Skt. vi ‘asunder,’ Avestan vi- ‘asunder’ (Holthausen 1963: 401). Since the primary function of wiþ þ Dat is to mark a static notion, namely opposition, the locationto-direction semantic development is most likely made possible by the syntagmatic features of this preposition, namely by its repeated occurrence in combination with motion verbs and by the inherited location function of the dative case that this preposition governs (see the discussion of the directional usages of to# þ Dat above). Taking into consideration all of the discussed directional and location usages of wiþ þ Dat, one can suppose that the notion of opposition (separation) in connection with the concept of proximity is original for the semantics of wiþ þ Dat in Old English. But the notion of hostility is secondary and often reinstated in passages by elements other than wiþ þ Dat. Such instances demonstrate a complex interface between the semantic and syntactic characteristics of various members of a sentence in general and prepositional phrases in particular. Compare also the occurrences of to# þ Dat that also mark hostility but only in combination with words that designate hostility: (13) Monige ðe to# me# feohtaþ (Ps. Th. 55, 3) ‘Many who fought against me.’ Wiþ þ Gen can also mark an object toward which a motion or an action is directed. The meaning ‘toward’ is its primary function. (14) Fleo´gan wið ðæs holtes (Byrht. Th. 131, 14; By. 18) ‘To fly to the forest’ There is also a rare case where this construction designates location that could be understood as both ‘next to, near’ and ‘opposite to’:

76

Olga Thomason

(15) Sætt se Hæ#lend wið ðæs dores (Mk. Skt. Lind. 12, 41) ‘The Savior sat next to/opposite to the gate.’ However, this passage does not hold a great value for our discussion since it is a translation from the Latin version of the Bible and might simply mirror the corresponding construction of the original: Et sedens Jesus contra gazophylacium Mk.12, 41 (Itala 1970, II: 119). Wiþ þ Acc unexpectedly shows a particular productivity in the designation of location (despite the fact that the accusative case usually marks direction). It is probable that the semantics of the preposition itself prevail here. Since the concept of proximity is prominent for this prepositional phrase, it allows wiþ þ Acc to denote the location ‘near’ repeatedly: (16) Wið ðæt do#msetl ic sitte (Æfc. Gr. 47; Zup. 269, 16) ‘I sit near the tribunal.’ He# gesto#d wið stea´pne rond (Beo. Th. 5126; B. 2566) ‘He stood near the high edge.’ Just like in the instance with wiþ þ Dat, wiþ þ Acc expresses the meaning ‘against’ where it is combined with words marking hostility: (17) Ic eom fa#h wið God (Cd. Th. 270, 28; Sat 97; Beo. Th 1627; B. 811) ‘I am hostile toward/against God.’ In the instances where wiþ þ Acc expresses direction, it merely specifies the meaning ‘to, toward’: (18) Se Hæ#lend eode wið ða sæ# (Mt. Kmbl. 4, 18) ‘The Savior went to the sea.’ This prepositional phrase is rarely associated with a contrary motion or action: (19) Æ# r ge# sceonde wið gesceapu fremmen (Cd. Th. 149, 4; Gen. 2469) ‘Before you injure (¼ drive against) the creatures, they will do (so).’ All of the discussed passages show that in Old English, wiþ þ Dat, wiþ þ Acc and wiþ þ Gen often share the same or similar functions. On several occasions these constructions express the same meaning within one clause, therefore presenting a case of free variation: (20) Breo´stnet wið ord and wið ecge ingang forsto#d (Beo. Th. 3102; B. 1549) ‘Coat of mail defended the entrance against a spear-point and against a sword.’

Notion of Direction and Old English Prepositional Phrases

77

The semantics of wiþ þ Dat, wiþ þ Acc and wiþ þ Gen emphasize the salience of the two senses ‘near’ and ‘against’, thus designating the conceptual importance of these notions for the concept of directionality. It is because of this double-sided nature (together with the frequent occurrence of these prepositions in combination with words denoting opposite entities/motions/actions) that it becomes possible for the semantics of wiþ þ Dat, wiþ þ Acc, and wiþ þ Gen to be extended to designate such notions as hostility versus protection, exchange, address, etc. Directional usages of these constructions are made possible by syntagmatic properties of prepositions – their regular occurrence in combination with motion verbs. To#ge#an þ Dat and onge#an þ Acc/Dat are also found to express directions ‘to, toward’ and ‘against’. This is expected for these prepositional phrases since they are compounds in which one of the components is a preposition whose rich semantics already encompass a variety of directional notions including the meaning ‘against’ (the second unit is a form of ge#n, which is a rare word that is sometimes used as an adverb with the meanings ‘yet,’ ‘still,’ ‘again,’ ‘further’ and sometimes used as an adjective denoting ‘direct’). Onge#an þ Dat and onge#an þ Acc are both found to designate static and dynamic spatial notions (location and direction). In instances where these constructions mark location, they specify only the meaning ‘opposite, against’: (21) Ða# arn he# and gesto#d ongea´n ðam le#ge (Gen. 221, 11) ‘The he ran and stood against the bed.’ He# sæt ðæ#r him getæ# ht wæs ongea´n ðone cyngc (Ap. Th. 14, 13) ‘He sat there where it was instructed to him, against the king.’ Furthermore, both these phrases can express direction, marking the meanings ‘to, toward’ and ‘against’: (22) Him com seo´ menio ongea´n (Jn. Skt. 12, 18) ‘The host came toward him.’ Fe#rdon ongea´n ðæ#m he#ðnum (Blickl. Homl. 203, 2) ‘They marched against heavens.’ Hı # fe#rdon onge#n ðone bry#dguman (Mt. Kmbl. 25, I, 6) ‘They marched toward the bridegroom.’ Ongea´n stream (Cod. Dip. B. i. 502, 3: ii. 374, 10) ‘Against the stream (¼ against the direction of the stream flow)’

78

Olga Thomason

It is interesting to see how the meaning ‘against’ is promoted in the semantics of onge#an þ Dat/Acc since we do not find any location meanings for these phrases other than the ones designating ‘against’ even though they both still express the directional meanings ‘towards’ and ‘against’ with similar frequencies. Since these prepositional constructions occur in combination with similar (and sometimes the same) words (see examples above), they are most likely in free variation. The semantics of to#ge#an þ Dat are limited to directional values and are not extended to any other spatial notions (e.g., location). It can designate both directional values ‘to, toward’ and ‘against’: (23) Foerdon to#gægnes him (Jn. Skt. Lind. Rush. 12, 13) ‘They marched toward him.’ Hı # fe#rdon to#gea´nes ðæ#m he#ðenum (Homl. Th. i. 504, 27) ‘They marched against heavens.’ The concept of meeting plays an important role for the semantics of to#ge#an þ Dat since it is frequently found in combination with verbs meaning ‘to march, go (to meet)’. It is understood that when two people/objects move toward each other to meet at a certain point, they will be opposed to each other (with or without hostility toward each other). As the present study demonstrates, there is a great deal of overlap in semantic fields that these prepositional phrases cover. Practically all analyzed prepositional phrases are used to designate location and direction values. Only the semantics of to#ge#an þ Dat and to# þ Acc are limited to the denotation of direction. The semantics of to# þ Dat are extended even further, resulting in constructions where this phrase marks source. Such divisions of semantic space create favorable conditions for a high level of prepositional variation. As far as prepositional governance and its influence on prepositional semantics goes, we find a variety of instances ranging from those where the meaning of a prepositional phrase is an expected result (e.g., to# þ Acc designating the direction ‘to, toward’) to those where it is an outcome opposite of the expected one (e.g., wiþ þ Acc marking the location ‘near’). These results are due to interplay among the semantics of a preposition itself, meanings of verbs in collocation, and functions of cases which these prepositions govern. Thus, in the case with to# þ Dat it is the semantics of the preposition that promote the directional meaning. On the other hand, the examples with wiþ þ Dat show that the directional meaning is primarily derived from the semantics of motion verbs.

Notion of Direction and Old English Prepositional Phrases

79

Some combinatory properties of prepositional phrases create limitations for these constructions. Thus, to# þ Gen has restricted usage occurring primarily in combination with a demonstrative or interrogative pronoun. As a consequence, it has a semantic range more limited than to# þ Dat, for example. The semantics of the investigated Old English data show a continuous interplay among three concepts: directionality (the meaning ‘to, toward’), proximity (the meaning ‘next to, near’) and opposition (the meaning ‘against, opposite to’). It is obvious that for speakers of Old English, the notions of proximity and opposition have close links with the notion of direction. We can even talk about di¤erent degrees of salience of the notions of proximity and opposition for the Old English prepositional phrases. Thus, we find prepositional phrases of three kinds: those for which the concept ‘against’ is the most important one (e.g., to#ge#an þ Dat), those that present the salience of the notion ‘next to, near’ (e.g., to# þ Dat), and, finally, those whose semantics comprise both of these meanings (e.g., wiþ þ Dat). It is important to note that the salience of a given concept (or both) is retained by a preposition no matter what case it governs (cf. both wiþ þ Dat and wiþ þ Acc can designate the meanings ‘near’ and ‘against’).

References An Anglo-Saxon Dictionary, Based on the Manuscript Collection of the Late Joseph Bosworth 1972 Edited and enlarged by T. Northcote Toller and Alistair Campbell. Oxford: Oxford University Press. An Anglo-Saxon Dictionary, Based on the Manuscript Collections of the Late Joseph Bosworth: Supplement 1920 Edited by T. Northcote Toller and Alistair Campbell. Oxford: Oxford University Press. Cienki, Alan 1989 Spatial Cognition and the Semantics of Prepositions in English, Polish, and Russian. Munich: Otto Sagner. Coventry, Kenny R. 2001 The Interplay between Geometry and Function in the Comprehension of Over, Under, Above, and Below. Journal of Memory & Language 44: 376–398. Cuyckens, Hubert 1997 Prepositions in Cognitive Lexical Semantics. In Lexikalische und grammatische Eigenschaften Elemente, edited by Dagmar Haumann and S.J. Schierholz, 63–82. Tu¨bingen: Niemeyer.

80

Olga Thomason

Fillmore, Charles 1968 The Case for Case. In Universals in Linguistic Theory, edited by Emmon Bach and R. T. Harms, 1–88. London: Holt, Rinehart and Winston. A Microfiche Concordance to Old English 1980 Edited by Antonette diPaolo Healey and Richard L. Venezky Toronto: Dictionary of Old English Project. Holthausen, Ferdinand 1963 Altenglisches etymologisches Wo¨rtebuch. 2nd edition. Heidelberg: Carl Winter. ¨ berlieferung Itala; das Neue Testament in altlateinischer U 1970– Nach den Handschriften hrsg. von Adolf Ju¨licher. Durchgesehen und zum Druck besorgt von Walter Matzkow und Kurt Aland. Kemmerer, David 2005 The Spatial and Temporal Meanings of English Prepositions Can Be Independently Impaired. Neuropsychologia 43: 797–806. Levinson, Stephen C. 2003 Space in Language and Cognition. Cambridge: Cambridge University Press. Luraghi, Silvia 2003 On the Meaning of Prepositions and Cases: the Expression of Semantic Roles in Ancient Greek. Amsterdam/Philadelphia: John Benjamins Publishing Company. Rudanko, Juhani 1995 Balking at and Working at: On Verb Governing at -ing in Present-day English. English Studies 3: 264–281. Segen, Bazyl 2001 Prynazounik u Sisteme Adnoj Usxodneslavjanskaj Gavorki Belastochchyny. Belastok. Tyler, Andrea, and Vivian Evans 2003 The Semantics of English Prepositions: Spatial Scenes, Embodied Meaning and Cognition. Cambridge: Cambridge University Press. Tyler, Andrea, and Vivian Evans 2001 Reconsidering Prepositional Polysemy Networks: The Case of Over. Language 77, No. 4: 724–765. Van der Zee, Emile and Jon Slack (eds.) 2003 Representing Direction in Language and Space. Oxford: University Press.

Commentary on Thomason, Notion of Direction and Old English Prepositional Phrases Joanna Nykiel

This paper suggests a wider landscape in which to locate Thomason’s discussion of English prepositional phrases (PPs). Thomason points out that there has been little work on the semantics of PPs beyond present-day English. Her data are intended to redress this imbalance by o¤ering insight into the phrases expressing direction in the Old English (OE) period. The study shows that the traditional correlation between the accusative case and the sense of direction does not always apply in OE. The accusative may in fact carry a location meaning when in a construction with preposition wiþ. Further, the dative, normally associated with location, is also found in phrases that denote direction. This e¤ect is strongest in sequences headed by preposition to#, where the accusative is an option too, though a ‘‘notably less frequent’’ one. Assuming that it is possible to identify OE cases by nominal su‰xes, it would have been helpful to see numerical data along with a chronological organization of the results. Allen (1995) and more recently Krygier (2002) have argued that the case system breaks down already in OE, sometimes leaving the dative and accusative forms identical. The data seem particularly dubious to me where one relies on final -e as an indicator of the dative because it might just as well be the accusative, as is the case with strong feminine nouns, e.g., cwalu in example (1). The inclusion of the temporal factor in particular would have imparted more precision to Thomason’s findings and helped warrant a more detailed analysis of them. Suggesting an approach is, inasmuch as the results allow it, my purpose in what follows. Thomason’s data raise two general issues: repeated use leading to entrenchment and a possible fit of the data with the constructional framework. I address them in this order. I find it indeed puzzling that the distribution of cases is so variable across phrases. But perhaps there is an explanation for the variety. Central to Thomason’s discussion is the contrast between the behavior of Ps in, on, binnan, innan and ofer on the one hand and that of to#, wiþ, onge#an and to#ge#an on the other. She notes that the former set selects the dative for

82

Joanna Nykiel

location and the accusative for direction in the context of containment and surface. The latter set appears to somewhat randomly select the dative or accusative with respect to both direction and location. Part of the explanation may lie in frequency of occurrence, to which I now turn. It has been recognized that grammar is shaped by speakers’ linguistic experience: strings of words experienced more often will have stronger mental representations than those that are only sporadically encountered. This recognition is formalized in usage-based models with an emphasis on how particular exemplars may guide semantic change (Barlow and Kemmer 2000, Langacker 2000, Bybee 2001, 2006). Because speakers are believed to keep a mental record of all the contexts associated with a given string, that string’s representation is variable to the extent that it may accommodate new contexts and meanings or lose previous ones if they become too infrequent. It could be that the division of labor between the dative and accusative in surface/containment contexts arose through repetition (entrenchment); if there was prior variation, it might have been eliminated due to insu‰cient frequency. With respect to other contexts, Thomason argues that directional uses of the dative may be supported by aspects of the string it is part of: the semantics of the verb or that of the governing preposition. For example, motion verbs impose a direction meaning on the following PP. If the preposition is to#, a direction meaning receives additional support. But the sequence to# þ dative can indicate location and source as well – are these its basic meanings? It is possible that the meaning of direction was inferred for to# þ dative and added to its representation as a result of frequent use with motion verbs. The accusative could then be the case originally associated with direction, and continue to be used so, presumably at a rate that still allows it to exist as an alternative. The preference for the string to# þ dative with the meaning of direction, however, does not yet solve the issue of whether to# þ accusative antedates it, leading to a later extension to the dative, or whether the reverse is true, or whether both phrases occurred in parallel. What would help us decide between these scenarios is considering the temporal factor because a rise in frequency may often reflect a new use developing out of an earlier one, pushing it into a secondary role (see, for example, Bybee (2006) on the development of the be going to construction). Whatever the scenario, it emerges for the data that directional meanings are best taken to derive from accessing an entire verb phrase as a unit and not as individual constituents, which enables a constructional treatment in the sense of Croft (2001), Goldberg (2006) and De Smet and

Commentary on Thomason

83

Cuyckens (2007). In the formalism of construction grammar, strings form micro-constructions (‘‘individual construction-types’’) based on their contexts of use; these may then align to form higher-level structures: mesoand macro-constructions (Traugott 2008a,b, Trousdale 2008). OE may have seen formations along these lines. As long as the accusative is selected, a directional meaning is arguably constructed out of the constituents of a verb phrase; an extension to the dative, though, signals a partial loss of constituent transparency. If this were the case (cf. first scenario), a micro- (or meso-) construction formed in OE. In fact, such a construction would have formed even if the second and third scenarios turned out to be true, only it must have formed prior to OE. It would be interesting to know whether Old English PPs may be recast as part of direction (or location) constructions. This possibility would bring a solution to the otherwise puzzling variation.

References Allen, Cynthia 1995

Case-Marking and Reanalysis: Grammatical Relations from Old to Early Modern English. Oxford: Clarendon Press. Barlow, Michael, and Suzanne Kemmer, eds. 2000 Usage-based Models of Language. Stanford, CA: CSLI Publications. Bybee, Joan 2006 From Usage to Grammar: The Mind’s Response to Repetition. Language 82(4): 711–733. Bybee, Joan 2001 Phonology and Language Use. Cambridge: Cambridge University Press. Croft, William 2001 Radical Construction Grammar: Syntactic Theory in Typological Perspective. Oxford: Oxford University Press. De Smet, Hendrik, and Hubert Cuyckens 2007 Diachronic Aspects of Complementation: Constructions, Entrenchment, and the Matching Problem. In Studies in the History of the English Language III: Managing Chaos: Strategies for Identifying Change in English, edited by Christopher M. Cain and Geo¤rey Russom, 187–213. Berlin/New York: Mouton de Gruyter. Goldberg. Adele 2006 Constructions at Work: The Nature of Generalization in Language. Oxford: Oxford University Press.

84

Joanna Nykiel

Krygier, Marcin 2002 A Re-classification of Old English Nouns. Studia Anglica Posnaniensia 38: 311–319. Langacker, Ronald 2000 Grammar and Conceptualization. Berlin/New York: Mouton de Gruyter. Traugott, Elizabeth Closs 2008a ‘All that he endeavoured to prove was’: On the Emergence of Grammatical Constructions in Dialogual and Dialogic Contexts. In Language in Flux. Dialogue Coordination, Language Variation, Change and Evolution Volume 1, edited by Robin Cooper and Ruth Kempson. London: College Publications. Traugott, Elizabeth Closs 2008b Grammaticalization, Constructions and the Incremental Development of Language: Suggestions from the Development of Degree Modifiers in English. In Variation, Selection, Development. Probing the Evolutionary Model of Language Change, edited by Gerhard Jaeger, 219–250. Berlin/New York: Mouton de Gruyter. Trousdale, Graeme 2008 Constructions in Grammaticalization and Lexicalization: Evidence from the History of a Composite Predicate Construction in English. In Constructional Approaches to English Grammar, edited by Graeme Trousdale and Nikolas Gisborne, 33–67. Berlin/New York: Mouton de Gruyter.

Response to Commentary by Nykiel Olga Thomason

Nykiel o¤ers a thorough and insightful commentary to the article. This response is organized around the main points which were brought up in her discussion. The first argument is connected with the syncretism of the case system in Old English and grammatical and semantic ambiguity of words (like strong feminine nouns) that have identical forms in the dative and accusative. This argument is reasonable, however, not all nominals have lost their distinctions in these cases (see example (2)), and some preserve it in plural even though they lost it in singular (like strong feminine nouns). Preserved dative/accusative distinction in third-person singular pronouns (example (1)) and demonstrative pronouns makes the posited problem even more tangible (example (2)). Thus, it is still plausible to talk about the designation of direction by to# þ Dat. Answering Nykiel’s question about the basic meaning of to# þ Dat, we would like to stress that it is precisely the fact the basic meaning of to# þ Dat is direction (and not location or source, which this construction also marks albeit less frequently) that initially triggered this study. This meaning could be a result of frequent use with motion verbs (as suggested by the reviewer and mentioned in the article). It could also be due to the case syncretism and broadening of semantic range of the Old English dative (see the discussion in the article). However, the opinion is maintained that the denotation of direction is inherent to the semantics of to# since there is an abundance of Indo-European cognates of this preposition that retain the directional value (see examples in the article). It is agreed that any semantic analysis should be derived from an entire construction including its verbal and nominal components (as we hoped to show in the article). However, we are not eager to treat them as a one-way relationship where a verbal head dictates the choice of a preposition and a preposition, in turn, rules the selection of case. The relationship is much more complex. There are instances where the semantics of a preposition has a decisive value (directional usages of to# þ Dat, in our view) and examples where it is the meaning of a motion verb that prevails (directional usages of wiþ þ Dat). The role of the accusative case whose primary func-

86

Olga Thomason

tions are connected with the designation of direction are exemplified by means that are used in Old English to express the meanings ‘into’ (in þ Acc, on þ Acc, binnan þ Acc, innan þ Acc) and ‘onto’ (on þ Acc, ofer þ Acc). In addition, one has to keep check with conceptional specifics. In our case the notion of direction is connected with the concepts of proximity and opposition as exemplified by the Old English data. This cognitive complexity adds to the level of variation found in Old English texts. Nykiel’s perceptive suggestion to present a chronological organization and statistical analysis of the data anticipates the next step of our research, which intends to add a diachronic spin to a currently primarily synchronic study and hopes to mark the exact course and stages of semantic development of prepositional phrases denoting direction in Old English. This, of course, should make possible the explanation of the distinction between direction and location usages with the help of the frequency of occurrence suggested by the reviewer.

Survival of the Strongest: Strong Verb Inflection from Old to Modern English Sherrylyn Branchaw

Old English had a relatively robust system of strong verbs, 367 that the extant corpus allows us to count. In the twelfth and thirteenth centuries, the system began to break down in a serious way, leaving us in Modern English with 72 of these inflected as purely weak and 78 inflected with at least some strong forms.1 The forces behind the weakening process have been treated in detail by Marcin Krygier (1994) among others; the question I address here is why those specific 78 verbs, and not others, remained strong to the present day. Because the eleventh century, immediately preceding the breakdown, is one of the worst-attested periods in English history, I find it most useful to look at the verb system of Old English to see what factors already present then allowed certain verbs to resist the regularizing tendencies of Middle English. Where the Old English data are insu‰cient to explain the modern outcome of a verb, I examine the state of the language at the relevant later period. Since the weakening process is still ongoing, the status of some verbs in the present day is uncertain. For example, dive has as a preterite only dove for some speakers, only dived for other speakers, and variation for other speakers. I have, therefore, included three categories in defining the modern outcome of a verb: strong, weak, and uncertain. With the understanding that usage will vary slightly from speaker to speaker, assignment of verbs to the categories has proceeded according to my own usage, which is American and is rather conservative, retaining e.g., sank, trod, and throve. The major factors I examine are type frequency, by which I mean the number of verbs with the same ablaut pattern; token frequency, by which I mean the number of occurrences of that specific verb in the Old English corpus; the shape of the root, not including the vowel; and the e‰ciency of the ablaut pattern in making distinctions among principal parts.

1. I count using the list in Quirk et al. (1985).

88

Sherrylyn Branchaw

A cautionary note on token frequency: due to the nature of the searches I was able to perform on the online Dictionary of Old English Corpus, these numbers should be viewed as ballpark figures only, and no statistical analysis should be attempted on them. They are solely for the purpose of determining which verbs were more or less common than others, not how much more or less common. Further research will be done to obtain figures with higher accuracy and precision so that more detailed conclusions can be drawn. For some forms, I was unable to obtain even ballpark figures in the time available due to homography with much more common words, such as the past tense of etan, æ´t, meaning ‘ate,’ and the far more common preposition æt, meaning ‘at.’ These forms simply have no number beside them in the tables. The absence of any number is to be distinguished from the presence of a 0 in the following way: no number means that the search was unsuccessful due to homography. A 0 means that a search was successfully carried out, and no instances of the form in question were found. The size of the data set can be reduced somewhat by eliminating a group of certain infrequent verbs in the following manner. Looking at Table 1, I define a ‘‘fully attested’’ verb as one for which the vocalism of each of its four principal parts is attested in Old English, according to the compilation of data in Krygier (1994), with modern surviving verbs checked using a search on the electronic corpus. In the first row, Old English had 367 strong verbs, of which 224 were fully attested, which

Table 1 # verbs

# fully attested

% fully attested

Old English strong

367

224

61

Modern English descendants

150

120

81

MdE weak

72

52

72

MdE strong

62

55

89

MdE uncertain

16

13

81

MdE at least some strong (uncertain þ strong)

78

68

87

Survival of the Strongest

89

comes to 61%. Of those 367 verbs, 150 survive in Modern English, and 120 of those, or 80%, were fully attested in Old English. Of those 150 surviving verbs, 72 are now inflected as weak, and 62 as strong. Of those 62 strong verbs, 89% were fully attested in Old English. If we assume a correlation between ‘‘fully attested’’ and ‘‘frequent,’’ we can conclude that in the very broadest sense, the frequency of the verb was a significant factor not only in determining its lexical fate in the language – in other words, whether the word survived at all – but also in determining the survival of its strong inflection. My conclusion is therefore that any verb not fully attested, if it survives at all, will survive as weak, unless special conditions prevail, namely that it can be easily fit into one of the most productive strong patterns, as in the case of slink. Therefore, I reduce the set of verbs I am considering to the fully attested verbs as well as to those handful of surviving strong verbs that were not fully attested. To investigate the e¤ects of type frequency, I sorted these verbs into groups with identical ablaut. I refer to these groups as ‘‘series’’ to avoid confusion with the traditional grouping into ‘‘classes.’’ The tables in this paper are grouped in series according to Old English type frequency, from highest to lowest, with some mergers I will talk about. The first column contains the modern descendant of the Old English verb, which may or may not have the same meaning, and the second through fifth columns contain the four principal parts – infinitive, preterite first and third singular, preterite plural and second singular, and past participle – along with the ballpark token frequency counts of each form where obtainable from the corpus search. For the first two series, I have given only lists for the outcomes of verbs, which seemed the most e‰cient method of presentation given the minimal impact of token frequency and the large number of surviving verbs. Including each verb on a table line would have added very little information in a great deal of space. The Old English type frequencies are given in parentheses in the heading of each series, e.g., Series I (28). Modern English verbs that are italicized were not fully attested in Old English but survive as strong nonetheless. Series I (28) Strong: bind, drink, find, grind, begin, run, shrink, spring, spin, sting, sing, sink, swim, swing, wind, win, wring, stink, cling, slink Weak: burn, climb

90

Sherrylyn Branchaw

Series II (28) Strong: bite, drive, ride, rise, shrive, smite, write, shine, stride, strike (transferred to Series I) Weak: glide, gripe, slide, writhe Uncertain: bide, shit, cleave Series III (26) Table 2. Strong Outcome Modern English

Infinitive

Pret. sg.

Pret. pl.

Past participle

fly1

fle#ogan 63

fle#ah 190

flugon 118

flogen 13

choose

ce#osan 42

ce#as 322

curon 79

coren 859

freeze

fre#osan 2

fre#as 2

fruron 4

froren 6

shoot

sce#otan 30

sce#at

scuton 19

scoten 31

1 Three

principal parts of fly and flee are identical.

Table 3. Weak Outcome Modern English

Infinitive

Pret. sg.

Pret. pl.

Past participle

chew

ce#owan 8

ce#aw 5

cuwon 2

cowen 5

creep

cre#opan 3

cre#ap 5

crupon 5

cropen 6

flee2

fle#on 229

fle#ah 190

flugon 118

flogen 13

lie

le#ogan 19

le#ah 12

lugon 27

logen 19

reek

re#ocan 14

re#ac 2

rucon 0

rocen 1

seethe

se#oðan 3

se#að 4

sudon 4

soden 162

2 No

figures were obtainable for mete due to excessive homophony.

From the first three series, it emerges that type frequency was not the most important factor. Here we have three series with nearly identical frequency counts, yet the percentage of verbs surviving as strong is dramatically lower for Series III. To explain this phenomenon, we must remember that the nature of the strong verb is to indicate grammatical changes through a change in the root vowel. If the vowels in the principal parts of a verb are not easily distinguishable from one another, then the grammatical changes

Survival of the Strongest

91

are indicated only weakly and, from a perceptual point of view, such an ablaut pattern will be disfavored. The robustness of the ablaut pattern correlates well with the outcomes of the first three series. [i]P[a]P[u], the well-known vowel triangle, are maximally distinct and [i:]P[a:]P[i] moderately so. In Series III, however, [eo] spelled and [æa] spelled underwent a near merger in early Middle English to [e:] and [:] (Lass 1992: 42–45) and in some dialects had become homophonous already in Old English. Because [e:] and [:] are so similar to each other, as a pair they do a poor job of distinguishing a past tense from a present. I conclude that Series III verbs were more susceptible to the use of the dental preterite than they would have been had the two vowels been more distinct from each other. Taking into account this greater (relative to Series I and II) tendency to become weak, the verbs from Series III survive as weak or strong according to their relative token frequency in Old English, in general. I must leave freeze, with its low token frequency, for further research. At least three pieces of evidence make it clear that root shape had some impact in Series I. Series I had a high survival rate of verbs without fully attested vocalism in Old English, such as stink, cling, and slink. As mentioned above, when the concept of ‘‘fully attested’’ was introduced, verbs not fully attested in English continued as strong to the present day only if they could be fit into a very productive series such as this one. Furthermore, both denominal verbs such as ring and string and Norse verbs such as fling were transferred into this series in Middle English. By the sixteenth century (OED), a critical mass had been reached, and the final velar became the most salient part of the root without requiring the presence of a nasal. An initial consonant cluster, especially one beginning with [s], was helpful but not necessary. Evidence for the importance of the velar in defining this series is found in the reassignment of certain verbs into this category (Bybee 2001: 126). The verbs stick and dig ought to be weak, but because of their final velars, they became conjugated like slink in Middle English even though they had no nasal. Similarly, strike ought to be conjugated strikePstrokePstricken but because of its final velar was susceptible to being conjugated strikeP struck. In the case of stick and strike, the consonant cluster beginning with [s] made them even more similar to many verbs in this series, and thus they transferred more easily. From strike especially we see the importance of root shape since the word properly belongs to Series II, which has always had relatively high type frequency, and the vowel of strike does not even match the usual vowel of Series I, which is [I]. In a product-

92

Sherrylyn Branchaw

Table 4. Factors A¤ecting Series I-III Type Frequency

Consistent Root Shape

Vowel distinctness

Series I

þ

þ

þ

Series II

þ



þ

Series III

þ





oriented schema, this mismatch is not as surprising as it might be in other theories because inputs need not be identical in form. So if the present form allows for the creation of a preterite that follows the type of Series I, the di¤erences in the vowels are not problematic. In Modern English, Bybee points out that the e¤ects of the root shape are felt even more strongly, as seen in the nonstandard preterites snuck and drug of sneak and drag, respectively (2001: 126). Table 4 shows the factors that influenced the outcome of verbs in the first three series. Series I was favored by high type frequency, optimal vowel distinctness, and easily definable root shape, and accordingly it has the largest percentage of verbs surviving as strong. Series II was favored by type frequency and by vowel distinctness, and accordingly it has a lower percentage of verbs surviving as strong compared to Series I but relatively high compared to other series. Series III was favored by type frequency alone, and accordingly it began to break down both earlier and more thoroughly than the first two series, leaving Modern English with a still lower percentage of surviving strong verbs. In Series IV and IX, the importance of the function of ablaut over type frequency also appears clearly. Here the two series, with identical vocalism in the first three principal parts, merged their past participles in favor of the vowel /o/ from Series IX. This process began in Old English (Krygier 1994: 54), and I therefore categorize them into a single series by the time of the great shift to weak verbs in early Middle English. The [o] participles in Series IX had a lower type frequency, but [o] is more distinct from [e] and [æ] than is [e], and the more distinctive participle was extended, not the more frequent one. Furthermore, all verbs from these series that remain strong to the present day now have preterites in [o] taken from the past participle. This pattern reinforces the conclusion that distinctness of vowel quality heavily favors survival. Due to the scarcity of surviving weak verbs from this class, it is di‰cult to analyze the e¤ects of token frequency, but wreak looks oddly frequent for a weak verb. It may have been

Survival of the Strongest

93

Table 5. Strong Outcome Modern English

Infinitive

Pret. sg.

Pret. pl.

Past participle

speak

sprecan 276

spræc

spræ´con 310

sprecen 53

tread

tredan 19

træd 28

træ´don 14

treden 21

weave

wefan 10

wæf 1

wæ´fon 1

wefen 26

bear

beran 333

bær 336

bæ´ron 127

boren 532

break

brecan 142

bræc 221

bræ´con 96

brocen

steal

stelan 43

stæl 47

stæ´lon 17

stolen 48

tear

teran 6

tær 17

tæ´ron 17

toren 15

Table 6. Weak Outcome Modern English

Infinitive

Pret. sg.

Pret. pl.

Past participle

mete2

metan

mæt

mæ´ton

meten

wreak

wrecan 112

wræc 80

wræ´con 7

wrecen 68

quell

cwelan 5

cwæl 1

cwæ´lon 4

cwelen 1

2 No

figures were obtainable for mete due to excessive homophony.

due to analogy with work if metathesis is applied often enough to the [wr] cluster2. Back-up support for the survival of strong verbs in IV and IX is found in the root structure. Krygier finds a 90% correlation between preservation of strong verbs in the 12th century and the presence of a single root-final sonorant (1994: 248–249). The transfer of wear from the weak system to the strong on the analogy of bear and tear (OED) is also evidence for the e¤ects of root structure here. Also transferred to this series was wake from Series VI based on the analogy of break and speak (OED), all of which sounded a great deal more alike in the sixteenth century than they do now. In Table 7, I use Leith’s table showing the sociolinguistic distribution of vowels for mate, meat, and meet in 1600 to show what the possibilities for our verbs are. The reader is advised to keep in mind that as their spell2. Suggested to me by Donka Minkova.

94

Sherrylyn Branchaw

Table 7. Wake, break, speak c. 1600. (Leith 1983: 148–9) Aristocracy

Bourgeoisie

Lower Class

meet [i:]

meet [i:]

meet [i:]

break/speak

meat [:]/[e:]

meat [e:]

meat [i:]

wake

mate [æ:]/[:]

mate [e:]

mate [e:]

ings suggest, break and speak were once pronounced identically, and their modern pronunciation reflects the selection of an unshifted break from a dialect that did not undergo the stage of the Great Vowel Shift that raised the vowel of speak (Lass 1999: 96–98). The same product-oriented schema that allows us to expect dug, struck, and even snuck to match swung and stung allows woke to match broke despite di¤erences in the infinitive. Wake, much like shit, had only two principal parts attested in Old English and now has a preterite from another series and an unexpected infinitive (OED). In the case of wake, the infinitive comes from the weak verb. Shit we will return to later. In Series IV, like VI and IX, the survival of verbs as strong or weak correlates reasonably well with their token frequency. These three series have in common relatively high type frequency and good vowel distribution. For the figures of Series VI, see Tables 8 through 10. The outstanding exceptions to survival based on token frequency are wreak and fare. The OED attributes the weakening of fare to confusion with its weak counterpart, and the weakening of wreak, as mentioned above, may have been due to confusion with work. It still remains to be asked, of course, why these particular verbs were more susceptible to confusion with their weak counterparts than other similar verbs such as steal, and that will require further investigation. Few verbs survive from Series VII. None retains the original strong inflection; three are unequivocally weak, and one, namely dive, hesitates between strong and weak, but the strong inflection is that of another series. The figures appear in Tables 11 and 12, and individual verbs are discussed following the tables. Bow, according to the OED, began to absorb the meanings of its weak causative bey in the 13th and 14th centuries, and simultaneously it became weak. Its weak outcome is therefore like fare and possibly wreak, which as seen above also became confused with weak counterparts. Brook has no attested strong forms in Middle English (Krygier 1994: 257), so we can conclude that it was not robustly strong even in Old English. Dive, the one verb from this series with strong

Survival of the Strongest

95

Table 8. Strong Outcome Modern English

Infinitive

Pret. sg.

Pret. pl.

Past participle

(for)sake

sacan 130

so#c 139

so#con 29

sacen 22

shake

sceacan 21

sceo#c 80

sceo#con 3

sceacen 25

stand

standan 473

sto#d 766

sto#don 296

standen 80

draw

dragan 7

dro#h 2

dro#gon 11

dragen 21

wo#c 41

wo#con 14

wake

Table 9. Weak Outcome Modern English

Infinitive

Pret. sg.

Pret. pl.

Past participle

fare

faran 573

fo#r

fo#ron 291

faren 291 [sic]

wade

wadan 13

wo#d

wo#don 2

waden 4

wash

wascan 6

wo#x 2

wo#xon 1

waxen 2

Table 10. Uncertain Outcome Modern English

Infinitive

Pret. sg.

Pret. pl.

Past participle

grave

grafan 8

gro#f 9

gro#fon 2

grafen 28

lade

hladan 12

hlo#d 9

hlo#don 8

hladen 27

shave

scafan 3

sco#f

sco#fon 0

scafen 4

forms, takes its infinitive from the weak system and its strong preterite and participle from Series VII to create a paradigm that fits nicely into Series II (OED). The fact that so few verbs from this series survive into Modern English should not, of course, be taken to say anything about the type frequency of the series in Middle English. The Middle English type frequency numbers will need to be computed separately, relative to those of all the other series. I point this out because of the disparity between the similar Old English type frequencies of Series VI and VII and the number of modern survivals. The disparities are presented in Table 13. There were two other series, V and VIII, that merged the last principal part that kept them separate, this time for a phonological reason. In the early part of the Middle Ages, [a:w] and [o:w] regularly became [ou]. I

96

Sherrylyn Branchaw

Table 11. Weak Outcome Modern English

Infinitive

Pret. sg.

Pret. pl.

Past participle

brook

bru#can 189

bre#ac 50

brucon 13

brocen

bow

bu#gan 271

be#ah 138

bugon 111

bogen 43

shove

scu#fan 30

sce#af

scufon 14

scofen 40

Table 12. Uncertain Outcome Modern English

Infinitive

Pret. sg.

Pret. pl.

Past participle

dive

du#fan 12

de#af

dufon 1

dofen 3

Table 13 Old English type frequency

# of Modern English survivals

Series VI

11

11

Series VII

10

4

therefore think it fair to consider these as one class with high type frequency by the time of the breakdown of the strong verb system. However, unlike in most series, including the otherwise similar IV/IX, the survival of strong versus weak verbs in V/VIII does not evidently follow token frequency lines. Figures are presented in Tables 14, 15, and 16. Grow and throw have inexplicably low token frequencies to survive as strong although with the proviso that homophony prevented counts of the preterite of throw. If these two survive as strong, one would expect flow to do so as well with the same root shape, ablaut, and type frequency, and significantly higher token frequency, which is not the case. The verbs of this series remain a mystery to me. All verbs in Series X and above have low type frequencies, and several have frequencies as low as 1, meaning they are unique patterns. Accordingly, they have a low survival rate of strong verbs, and generally only verbs with very high token frequencies survive. Verbs belonging to a series higher than X have type frequencies of no more than 7, and it is at this point that I have grouped all the remaining verbs together rather than giving separate tables for each. Because of the high numbers of verbs surviving as weak, I give only a selection of those in Table 23.

Survival of the Strongest

97

Table 14. Strong Outcome Modern English

Infinitive

Pret. sg.

Pret. pl.

Past participle

grow

gro#wan 15

gre#ow 10

gre#owon 3

gro#wen 3

blow

bla#wan 16

ble#ow 57

ble#owon 2

bla#wen 40

know

cna#wan 242

cne#ow 341

cne#owon 96

cna#wen 82

throw

ðra#wan 2

ðre#ow

ðre#owon

ðra#wen 15

Table 15. Weak Outcome Modern English

Infinitive

Pret. sg.

Pret. pl.

Past participle

flow

flo#wan 24

fle#ow 51

fle#owon 24

flo#wen 22

row

ro#wan 9

re#ow 8

re#owon 3

ro#wen 3

mow

ma#wan 2

me#ow

me#owon 2

ma#wen 2

Table 16. Uncertain Outcome Modern English

Infinitive

Pret. sg.

Pret. pl.

Past participle

sow

sa#wan

se#ow 26

se#owon 9

sa#wen 6

Table 17. Strong Outcome Modern English

Infinitive

Pret. sg.

Pret. pl.

Past participle

fall

feallan 91

fe#oll 209

fe#ollon 184

feallen 86

hold

healdan 828

he#old 1047

he#oldon 256

healden 473

Table 18. Weak Outcome Modern English

Infinitive

Pret. sg.

Pret. pl.

Past participle

fold

fealdan 3

fe#old 19

fe#oldon 2

fealden 21

wield3

wealdan

we#old 67

we#oldon 8

wealden 25

wax

weaxan 62

we#ox 126

we#oxon 25

weaxen 53

3 Wield

is a result of the merger of strong wealdan and weak wieldan according to the OED.

98

Sherrylyn Branchaw

Series XI and greater Table 19. Strong Outcome Modern English

Infinitive

Pret. sg.

Pret. pl.

Past participle

slay

sle#an 281

slo#h >365

slo#gon 325

slagen 355

lie

licgan 76

læg 440

læ´gon 11

legen

sit

sittan 164

sæt 663

sæ´ton 309

seten

give4

giefan 16

geaf 785

ge#afon 50

giefen 20

get4

gietan 75

geat >345

ge#aton 170

gieten 65

eat

etan 206

æ´t

æ´ton 146

eten 9

swear

swerian 40

swo#r 174

swo#ron 30

sworen 28

come

cuman 826

co#m 6050 cwo#m 425

co#mon 1023 cwo#mon 65

cumen 580

fight

feohtan 89

feaht 269

fuhton 230

fohten 19

see

se#on 803

seah 1691

sa#won 728

sewen 613

4 The Old Norse borrowings account for the initial [g] in Modern English, which would have been [ j] if from Old English.

Sorting out the relative impacts of root shape, token frequency, and vowel distinctness is interesting. In general, verbs with high token frequency, in particular of the preterite singular, survive as strong. As we expect, the lower the type frequency, the higher the token frequency needs to be for a verb to come out as strong. Consider for example step as compared to weave in Series IV above. There are no root shape e¤ects, nearly similar ablaut pattern once the participial [o] of weave was transferred to the preterite in the fourteenth and fifteenth centuries, and a much more frequent preterite for step, yet the preterite of step has been stepped since the Middle Ages (OED; Krygier 1994). The lower type frequency of step was obviously what disfavored it for survival. In many cases, however, a type frequency of less than 10 overlaps with the presence of a liquidconsonant cluster, which Krygier finds a significant factor in the weakening process during the thirteenth and fourteenth centuries (1994: 248).

Survival of the Strongest

99

Table 20. Weak Outcome A selection of carve, starve, warp, help, melt, swallow, leap, flay, sleep, read, let, burst, braid, ban, span, step, yield, yelp, weep, shed, shape, delve, spurn, thresh/thrash Modern English

Infinitive

Pret. sg.

Pret. pl.

Past participle

carve

ceorfan 59

cearf 47

curfon 19

corfen 69

help

helpan 106

healp 9

hulpon 2

holpen 36

melt

meltan 13

mealt 6

multon 5

molten 13

swallow

swelgan 33

swealh 45

swulgon 15

swolgen 10

leap

hle#apan 6

hle#op 17

hle#opon 3

hle#apen 1

sleep

slæ´pan 2

sle#p 56

sle#pon 31

slæ´pen 1

read

ræ´dan 159

re#d

re#don 2

ræ´den

let

læ´tan 512

le#t

le#ton

læ´ten 360

burst

berstan 25

bærst 79

burston 36

borsten 16

Table 21. Uncertain Outcome Modern English

Infinitive

Pret. sg.

Pret. pl.

Past participle

shear

scieran 1

scear

scearon

scoren 31

heave

hebban 104

ho#f

ho#fon 14

hafen 7

hang

ho#n

he#ng 3

he#ngon 11

hangen 1

swell

swellan 2

swealh 43

swullon

swollen 38

beat

be#atan 18

be#ot

be#oton 22

be#aten 31

hew

he#awan 19

he#ow

he#owon 11

he#awen 29

bid

biddan 579

bæd 1068

bæ´don 330

beden 105

Two cases where verbs with high token frequency failed to remain strong are outstanding: let and read. Both have low type frequency, high token frequency – unfortunately due to homophony I was unable to obtain figures for the preterite singular of either – limited vowel distinctness, especially after [æ:] became [:], and both end in dentals.

100

Sherrylyn Branchaw

Now, Krygier finds a strong enough correlation between dental-final verbs and the shift to weak in the twelfth century for him to conclude that root-final dentals triggered the disintegration. His explanation for the phenomenon is that the Norman French speakers, using English as a second language, reinterpreted the root-final dental as the dental preterite. Anglo-Saxons, with native-speaker intuitions about which verbs were weak and which were strong, interpreted that as the speakers of the prestige dialect inflecting strong verbs as weak, and they generalized it to the weakening but the most salient strong verbs (1994: 148). Although relatively little e¤ect by the Normans on the English grammar has been demonstrated3, weakening of strong verbs is a move in the direction of simplifying the grammar, and therefore I consider it at least possible that once the Normans triggered the change, the Anglo-Saxons might have continued to carry it out. I am not committed to his explanation, but I would like to point out that no matter who or what triggered the change, if the shift from strong to weak indeed began among dentalfinal verbs, several of these verbs ablauted with vowels of a low degree of perceptual distinctness. Examples include let, read, shed, and beat. In each case, the vowel of the preterite was a front mid vowel or diphthong that became a front mid vowel in Early Middle English, and in each case the vowel of the present tense was a front mid vowel or diphthong that became a front mid vowel. If Norman speakers, or even native speakers, could hardly hear the di¤erence between the root of the strong past and of the present, and if the preterite already had a dental at the end, they might all the more easily have been reinterpreted as weak. In contrast, verbs ending in dentals with distinctive vowels in the present and preterite or participle, such as writePwrotePwritten and sitPsat survive as strong. Their survival indicates that the dental alone did not fate a verb to become weak and that additional explanation, such as vowel distinctness, is called for. Sit is perhaps the strongest of the strong verbs. Despite its dental and low type frequency, sit had a high enough token frequency to continue to 3. There is more evidence for Norse influence on the grammar of English, but in most cases where the OE verb is strong, the Norse cognate is also strong. In fact, in some cases where the Norse verb is weak, e.g., fling, the OE verb is strong and remains so. Whether the interaction between two languages with robust strong verb systems but with di¤ering vowel qualities in many preterites – e.g., OE bær and ON bar – would be enough to trigger a general shift toward weakening verbs, I cannot answer definitively, but I doubt it.

Survival of the Strongest

101

be inflected as strong, which itself is not surprising. What may come as a surprise, depending on the framework in which one operates, is that sit must have been largely responsible for the preterite shat of shit and probably the preterite spat of spit. Shit should conjugate shitePshote, and spit was originally weak (OED). The problem is that the process of analogy is not well understood by linguists. In a nearest neighbor model, the most similar form already present in the lexicon, in this case sat, provides the template for the new form being produced, in this case the preterites of shit and spit. Other findings, though, including those of Bybee (2001: 124), suggest that a type frequency of more than three verbs should be necessary for productivity. Furthermore, even if one counters that other verbs, such as bidPbad, display the same pattern, Moder (1992) finds that ‘‘as high token frequency leads to greater autonomy, items with high token frequency have weaker connections to related forms and thus are more likely to become independent and less likely to contribute to the formation of productive classes’’ (Bybee 2001: 136). According to this principle, a type including sitPsat and bidPbad, for example, should not take in new members like spit and shit because sitPsat should be treated as an isolate. A resolution may lie in the use of a framework that allows for analogies to multiple neighbors. If in the sixteenth century, spit was perceived as closely resembling both sit and spin in di¤erent respects, then the preterites sat and span might both have contributed to the formation of the preterite spat. Then, by the nineteenth century, speakers of the language have sitP sat, spitPspat, and spinPspan, which would all tend to contribute toward shat as one of the options for the preterite of shit. However, the jury is still out on whether this sort of multiple analogy e¤ect is possible in grammars: Albright and Hayes (2003: 152) are rather emphatic that it is not. From the results obtained in this paper, I conclude that the single most important factor in preserving the strong inflection of a verb was the perceptual ease of distinguishing the vowels of the principal parts. Token frequency and root shape are hard to rank with respect to one another. Within series of all type frequencies, with isolated exceptions, token frequency was responsible for selecting which survived as weak and which as strong. Root shape was definitely responsible for the transfer of a verb from one series to another or from outside the strong verb system into it. Root shape played a significant role in holding existing strong verbs in the system, which is especially visible when the verb was poorly attested in Old English. Type frequency seems less important than the other factors as type of high frequencies might have low survival rates, and isolated

102

Sherrylyn Branchaw

type frequencies might have robust strong inflection, depending on the other factors. It is certainly not unimportant, however, as reflected in the organization of this paper. The lower the type frequency, the greater the other factors, especially token frequency, must be in order to keep a verb strong. Regarding the more general theoretical implications, the strong verbs support a product-oriented schema in which a preterite of a desired shape, such as struck and dug is derived from inputs of di¤erent shapes – strike and dig. There is also some evidence for multiple analogies in which one new form may be produced by analogy with multiple existing forms. Much work, however, remains to be done on the development of the strong verbs. I intend to do a more accurate and precise token frequency count of forms in Old English using the searchable corpus and to obtain frequency counts from Middle English and perhaps Early Modern English. Toward that end, there exist searchable parsed corpora from those periods from the Penn Helsinki project, as well as other resources, such as Long’s dissertation The English Strong Verb: From Chaucer to Caxton. Once the numbers are more reliable, statistical analyses can be carried out to find the answer to the question ‘‘How frequent is frequent enough?’’ The verbs that are historically weak but synchronically strong, verbs of the leadPled type, and historically weak but synchronically partially strong verbs of the keepPkept type, must be included in the study. Such verbs both provide further examples in the language of indicating grammatical change through a change in the vowel, which must have been important in the preservation of the historically strong verbs, and they provide analogies for specific strong verbs, such as readPread. Di¤erent models of analogy will be tested to see how well they predict the outcomes of these data. Hayes and Albright (2003), among others, have done a study with modern native speakers to see how they handle nonce verbs and have produced a model based on their study, but to my knowledge, no one has investigated the success of di¤erent analogical models in explaining the data of Middle English. I intend to quantify statements about vowel distribution with as much precision as possible. Flemming’s dissertation Auditory Representations in Phonology explores the spread of vowels within the space of the mouth and the number of distinctions that can be made and perceived. I will see how his model applies to the data of verbs losing ablaut because their vowels were too close together. I also intend to quantify the ‘‘critical

Survival of the Strongest

103

mass’’ reached in Middle English when Series I verbs no longer required a nasal, allowing the past tense of strike to be struck.

Acknowledgements I am indebted to my advisor, Donka Minkova, for guidance in writing this paper and to the students and professors in the Program in IndoEuropean Studies at UCLA for feedback on drafts.

References Albright, Adam and Bruce Hayes 2003 ‘‘Rules vs. Analogy in English Past Tenses: A Computational/ Experimental Study.’’ Cognition 90: 119–161. Bybee, Joan 2001 Phonology and Language Use. Cambridge: Cambridge University Press, ‘‘Dictionary of Old English Corpus in Electronic Form.’’ February 11, 2005. Antonette di Paolo Healy Centre for Medieval Studies, University of Toronto. Flemming, Edward S. 2002 Auditory Representations in Phonology. New York: Routledge. Krygier, Marcin 1994 The Disintegration of the English Strong Verb System. University of Bamberg Studies in English Linguistics, vol. 34. Frankfurt am Main: Peter Lang GmbH. Lass, Roger 1994 Old English: A Historical Linguistic Companion. Cambridge University Press. Lass, Roger 1992 ‘‘Phonology and Morphology.’’ In The Cambridge History of the English Language, Volume II: 1066–1476, edited by Norman Blake, 23–155. Cambridge: Cambridge University Press. Lass, Roger 1999 ‘‘Phonology and Morphology.’’ In The Cambridge History of the English Language, Volume III, 1476–1776, edited by Roger Lass, 56–186. Cambridge: Cambridge University Press. Leith, Dick 1983 A Social History of English. London: Routledge and Kegan Paul.

104

Sherrylyn Branchaw

Long, Mary McDonald 1994 The English Strong Verb from Chaucer to Caxton. Menasha, Wisc.: Banta. Moder, Carol Lynn 1992 Productivity and Categorization in Morphological Classes. Bu¤alo, NY: SUNY dissertation. Oxford English Dictionary September John Simpson. Oxford University Press. 2007 Quirk, Randolph et al. 1985 A Comprehensive Grammar of the English Language. London, New York: Longman. Welna, Jerzy 1996 English Historical Morphology. Wydawnictwa Uniwersytetu Warsawskiego.

Commentary on Branchaw, Survival of the Strongest: Strong Verb Inflection from Old to Modern English Markku Filppula and Juhani Klemola

Sherrylyn Branchaw’s article focuses on the diachronic development of strong verb inflection from Old English to Modern English and, specifically, on the question of why some verbs have remained strong to the present day while others have become weak. The method Branchaw has chosen for her study is sound: she uses the OE verb system as her starting point and sets out to find what kind of factors could explain survival of certain kinds of strong verbs into the ME period and beyond. The factors examined include type frequency (the number of verbs with the same Ablaut pattern), token frequency (the number of occurrences of a specific verb in the Dictionary of Old English Corpus), shape of the root of the verb, and e‰ciency of the Ablaut pattern in making the distinctions. All these are relevant and yield interesting results despite the fact that the statistics are not yet completely accurate as the author herself points out. On the other hand, it must be borne in mind that the nature of the surviving OE textual evidence is such that any quantitative generalizations based on the corpus must be treated with caution. The method used in the study has another potentially major limitation: Branchaw compares her OE data only with standard present-day English and totally disregards the evidence o¤ered by non-standard dialects. The English Dialect Dictionary (EDD), for example, provides ample evidence of extensive variation in the past and past participle forms in most if not all of the ‘series’ of strong verbs discussed by Branchaw. To give but some examples, the following verbs from Branchaw’s Series I, which she considers to survive as strong only, are attested in nineteenth-century dialects with weak outcomes: bind, drink, run, spring, swim, swing, wind, slink. The preterite and past participle forms for drink serve as an illustration of the considerable range of variability in the dialect data: pret. drak, drenk, drenked, drinked, dronk, druck, drunk pp. dhrunken, drank, drinked, dronken, drucken, druckin, druken, drukken, drunk, drunken

106

Markku Filppula and Juhani Klemola

In Branchaw’s Series III, in turn, all of the four verbs with strong outcomes ( fly, choose, freeze, shoot) are also found with weak outcomes in the EDD. To give an idea of the broad range of variability found in the dialectal data, we list in the following the di¤erent preterite and past participle forms recorded for the verb freeze in the EDD: pret. fraaz, fraaze, fraze, frez, friz, frore, fruize, fruz, vreezed, vriz, vrore pp. frawn, freezen, frez, friz, frizzent, froan, froar, froaz, froozed, fror, frore, froren, frorn, frown, froze, frozed, frozzan, frozzen, fruozen, fruz, fruzzen, vraur, vreezed, vriz, vroar, vror, vrore Taking these dialectal data into consideration would not necessarily have a¤ected the overall conclusions reached by Branchaw, but it would have brought to light some of the complexities involved in assessing the relative weight of each of the factors examined here. The otherwise clear exposition of the data and the argumentation suffers somewhat from the abrupt transition from Series I to III to IV and IX in the description of the types of strong verbs. This is aggravated by the lack of explanation in the previous text about the number of the ‘series’ – as Branchaw calls them – and their distinguishing features. The reader would also have benefited from some kind of an introduction to Tables 5 and 6. Other shortcomings include a rather heavy reliance on work by others (especially Krygier 1994 and Bybee 2001) in matters relating to factors influencing survival of (types of ) strong verbs. ‘Multiple analogy’ is suggested as an explanation for some individual verbs (shit, spit), but the mentioned counterarguments to multiple analogies by Albright and Hayes (2003) remain unexplained. Despite the mentioned shortcomings, Branchaw has made an important contribution to the field by showing, first, that a multiplicity of factors need to be considered in trying to explain why some strong verb forms have survived up to the present day, and, second, that some of these factors have been more influential than others. According to her results, the perceptual ease of distinguishing the vowels of the principal parts turned out to be the most important factor in preserving the strong inflection of a verb. Token frequency and shape of the root were next in the hierarchy whereas type frequency had less of an e¤ect on the survival rates than the others. Branchaw’s results are persuasive, notwithstanding the aforementioned limitation concerning the evidence from dialectal forms of English.

Response to Filppula and Klemola Sherrylyn Branchaw

I would like to thank the reviewer of my paper for the helpful comments provided, and here I will respond to points made about dialect forms and about multiple analogy. Data forms from dialects other than what is now the American standard are of course invaluable. Their importance is hinted at in my paper in this volume in the use of the forms drug and snuck, and they are used more crucially in my talk at ICEHL 15 (Munich, August 2008). It is not my goal to explain all forms in any dialect of the English language, though that is a valuable endeavor, for that would make the task of this particular project too large, but to use them both to explain the origin of standard forms and as evidence for linguistic phenomena. As an example of the former, the OED explains the preterite drew for draw as a form that arose in the Middle English dialect of the north of England where a phonological change that occurred in the south did not take place, and therefore in the north draw and blow continued to resemble each other closely. The analogy blow : blew :: draw : X, where X is drew, was therefore more apt in the north than in the south, and the form drew spread until it entered what would become the standard dialect. The north, therefore, is indirectly responsible for other preterites such as slew and flew. Those forms probably arose when the past tense morpheme became associated with present tenses in both and thus making it a product-oriented schema. Because tense form no longer occurred only with present tenses of a single form, it was free to become more productive and expand to verbs of still other vowels such as slay and fly. Nonstandard forms in modern American English such as drug and snuck provide still more evidence for product-oriented schemas since drag and sneak do not share the vowel of sting and the other members of its class. Examples such as these abound. Furthermore, as my reviewers point out, one can imagine that Old English might predict the outcomes of some dialects better than it does standard American English, or, for that matter, standard British English.

108

Sherrylyn Branchaw

In that case, the dialects will form a backdrop against which to evaluate other factors that have been at work in the standard language, one of which might be the influence of grammarians. As for multiple analogy, the debate is still open, and further research will be required before I can assert a belief in its existence or nonexistence. Hayes and Albright test their learner models against predictions that would be made if multiple analogy, which they call variegated analogy, held true. They identify the kinds of patterns in English that their model, programmed not to use variegated analogy, would fail to identify, but they do not find their model failing to identify these patterns. They conclude that variegated analogy adds nothing to their model and therefore that it is not at work in the grammar of speakers. Still other scholars, such as Bybee and Hay, have models where each word has phonological, morphological, and semantic connections of varying strengths with other words. Such models would allow spit to have phonological connections with the -it of sit and with the spi- of spin. Because the ablaut of sat and span is a morphological connection between them, the morphological pattern could be extended analogically to create spat. I am interested in evaluating this possibility because the rest of the verbs4 that form their past tense with /æ/ end in nasals, velars, or nasal velars. Much work has been devoted to characterizing this class, to which swimPswam and singPsang belong. It is hard for me to see how spit could have been generated without the existence of sat since spit does not end in a nasal or velar, and the grammar of modern English by Quirk et al. (1985) places sit and spit into a subcategory to which no other verbs belong. Yet a popular claim asserts that high frequency words with unique morphology remain idiosyncratic and do not form the basis for analogy. If this assertion is true, sit alone makes a poor model for spit. This claim itself, however, has also been called into question. Hare and Elman (1995) report an experiment in which speakers pronounced the nonce-word vone to rhyme with the idiosyncratic gone rather than bone. Nevertheless, it remains to be seen if indubitable examples in natural language can be found. The case of spat, then, will make an interesting test case for the claims about the existence of multiple analogy and about the existence of analogizing based on a single, high-frequency item. If neither of those phenomena exists, then it is necessary to re-evaluate the criteria of the class of sing 4. With the exception of (for)bidP(for)bad(e), for speakers who have /æ/ rather than /ei/.

Response to Filppula and Klemola

109

and swim to admit a verb ending in a dental stop. In any case, the English strong verb preterites, both of the standard American English and of other dialects, form a fertile testing ground for morphological theory.

References Hare, M. & J.L. Elman 1995 Learning and Morphological Change. Cognition 56.1: 61–98. Quirk, Randolph et al. 1985 A Comprehensive Grammar of the English Language. London, New York: Longman.

Subject Compounding and a Functional Change of the Derivational Su‰x -ing in the History of English* Akiko Nagano

1. Introduction In the literature on Present-day English (PE) word formation, compound nouns of the form [Noun þ Verb-ing] (e.g., city planning, housekeeping, letter writing) and compound nouns of the form [Noun þ Verb-er] (e.g., dish washer, taxi driver, watchmaker) are often called ‘‘synthetic compound nouns.’’ The possible grammatical relation between the first Noun and the second Verb in these constructions has constituted an important topic of discussion. For example, Bloomfield (1933: 231–232) claims that synthetic compounds embody the verb-object relationship, and Marchand (1969: 15–19) also defines synthetic compounds in terms of the verb-object relationship. To state simply the most generally held view, PE synthetic compounds are based on the verb-object relationship and exclude the subject-verb relationship (Adams 2001: 78–79, Lieber 2005: 381). Against this background, this paper will examine -ing compound nouns diachronically and will argue that -ing compound nouns in Old English (OE) and Middle English (ME) allowed the subject-verb relationship and that certain types of PE -ing compound nouns do as well. The aim of this paper lies in elucidating the relationship between the possibility of ‘‘subject compounding’’ (SC hereafter) and a functional change of the derivational su‰x -ing. In short, I will argue that the possibility of SC in -ing changes throughout the history of English and that this change can be accounted for in terms of a functional change of the derivational su‰x -ing. * I would like to thank the audience at the 5th meeting of the Studies of the History of English Language (SHEL) for their helpful comments. I am also indebted to Elizabeth Traugott and two anonymous reviewers for their valuable comments and suggestions. Thanks also go to Molly Bassett and Kurt Spurlock for stylish improvements. Needless to say, responsibility for any errors is my own. This work is financially supported by Grant-in-Aid for Young Scientists (B), No. 19720115, from the Ministry of Education, Culture, Sports, Science and Technology in Japan.

112

Akiko Nagano

2. Subject Compounding in PE Consider the representative examples of PE -ing compound nouns cited in (1). The derivational su‰x -ing forms a deverbal compound noun. The acceptability di¤erence between (1a) and (1b) indicates that, unlike an object, the subject of a head verb cannot be compounded as the non-head of an -ing deverbal compound. For instance, the non-head noun of the acceptable compound noun car-driving is interpreted as an object of the head verb, while the compound noun *girl-swimming is unacceptable if the non-head is interpreted as a subject of the head verb. (1) a.

car-driving, carol-singing, fruit-devouring, pasta-eating e.g., Bicycle-repairing went on in the back room. (Adams 2001: 78) Taxi-driving by John can be dangerous. (Di Sciullo 1992: 65) Flower-arranging by experts is preferable to do-it-yourself. (ibid.)

b. *child-devouring (of fruit) ‘a child devours fruit,’ *dog-running, *girl-swimming, *rain-falling, *sunrising, *weather changing e.g., *Unexpected guest-arriving is a nuisance. (Adams 2001: 78) *Man-sleeping is sometimes noisy. (Di Sciullo 1992: 65) *Sun-rising is nice to watch. (ibid.) This fact is considered to be a general property of PE -ing compound nouns, as Adams (2001: 78–79) shows: Compounded process nominalizations cannot easily exhibit relationships other than that of verb-object: compare the subject-verb expressions in *‘unexpected guest-arriving is a nuisance’, *‘frequent dog-barking disturbs the neighbors’, *‘mechanic-repairing of bicycles’, *‘tycoon-evasion of taxes’. Adjuncts and complements of the verb other than direct objects also appear strange when compounded with process nominalizations: *‘guest-cooking of meals’, *‘council-sending of letters’.

Unacceptable -ing compound nouns such as (1b) and those found in the above citation led some researchers to propose general principles to exclude SC in -ing. Roeper and Siegel’s (1978) First Sister Principle and

Subject Compounding and a Functional Change

113

Selkirk’s (1982) First Projection Principle are probably the most famous examples of such general principles on compounding.1 Interestingly, PE allows SC when the head verb is nominalized by a suffix other than -ing. For example, the non-heads of the deverbal compound nouns in (2a) and (2b) below can be interpreted as subjects of the head verbs. In these cases, the head verbs are nominalized not by the su‰x -ing but by conversion or by nominalization su‰xes of lesser productivity like -al. (2) a.

baby-step, bee sting, earthquake, heart-ache, lion attack, the Mitterrand visit, mouse-squeak, rainfall, sunset, weather change e.g., One of the best-known episodes of repeated lion attack occurred in Kenya in 1898. Rail construction was going on about 40 miles north of Tsavo National Park when two lions began killing the workers. Twenty-six Indians and a similar number of Africans lost their lives before the lions were shot. (www.webcorp.org.uk) There is concern that a weather change in South Australia may do little to help the fight against big bushfires on Kangaroo Island. (www.webcorp.org.uk)

b. consumer choice,2 dust accumulation, population growth, train-arrival e.g., Consumers choose products based on various tangible and intangible attributes. Previous research has shown that there is a di¤erence between appearance-based and word-based evaluations of wood species. However, little research has been done on how this di¤erence a¤ects consumer choice. (www.webcorp.org.uk) As Bauer and Renouf (2001: 117–120) claim, conversion gives rise to numerous examples of SC. Along with instances of object compounding (e.g., handshake, pay raise, tax cut), there are many instances, such as bee sting and weather change, where the subject of an intransitive verb is compounded. Additionally, there are some instances, such as lion 1. See Lieber (2005: 380–383) for an up-to-date survey of the development of this discussion in generative-linguistic word-formation theories. 2. Strictly speaking, the head noun of this deverbal compound is nominalized not by su‰xation but by vowel alternation (choosePchoice).

114

Akiko Nagano

attack, where the subject – rather than the object – of a transitive verb is compounded. In the literature, the acceptability di¤erence between (1a) and (1b) seems to be taken as a universal fact about -ing deverbal compounds in PE; however, my corpus-based research (Nagano 2007) has found that PE does allow SC in -ing to a limited extent, as the -ing compound nouns in (3) below show. For example, the compound noun fruit-ripening in the sample sentence in (3a) has the subject-verb interpretation ‘‘fruit ripens.’’ One might argue that the compounds in (3a, b) are object compounds and their category is adjective. Of course, as PE -ing compounds, they might be used as object-incorporating compound adjectives (e.g., a fruitripening factor, milk-souring bacteria), but the sample sentences cited in (3a, b) clearly show that they can also be used as subject-incorporating compound nouns. In fact, since these -ing compounds all describe a natural phenomenon that is not caused volitionally by an external agent, the subject-verb reading is much more frequent and easier to induce than the verb-object reading. (3) a.

artery-hardening, fruit-ripening, gap-widening, muscle softening, poverty deepening, skin-darkening, world-flattening e.g., Lower blood sugar (by natural means or with insulin if necessary) according to the condition. This will decrease the blood sugar in the body e¤ectively, control the diabetes and thus prevent or postpone the occurrence of artery hardening indirectly. (www.webcorp.org.uk) Because ethylene is the main trigger for fruit ripening, several genetic engineering strategies involve the reduction or prevention of ethylene production. Tomato fruits that do not produce ethylene develop fully on the plant and then stop before ripening and turning red. (www.webcorp.org.uk)

b. airway-narrowing, hair-thinning, milk-souring, muscle thinning e.g., In exercise-induced asthma, the airway narrowing begins within 5 to 15 minutes after initiating physical exercise. (www.webcorp.org.uk) Unfortunately there is no one simple explanation as to why hair thinning occurs as it may be down to a medical condition or it may be the type of lifestyle a person is leading. In fact the only real way of determining the cause of hair thinning is for a person to visit their doctor. (www.webcorp.org.uk)

Subject Compounding and a Functional Change

115

The absence of these female hormones after menopause may lead to . . . hair loss, skin coarsening, decrease in breast size and support, and bone thinning. (www.webcorp.org.uk) Notice that the head verbs of the -ing compounds in (3a, b) cannot be nominalized by conversion. As Marchand (1969: 276–277) observes, a derived verb, whether a su‰xed one (e.g., to widen) or a converted one (e.g., to thin), cannot be converted into a noun (e.g., *[widen]N, *[thin]N).3 Due to this morphological restriction, the converted versions of the SC compounds in (3a, b) (e.g., *[ gap-widen]N, *[hair-thin]N) are systematically impossible. This is why -ing SC in PE is exceptionally allowed in (3); in PE, SC is realized by conversion or by non-productive a‰xes as discussed above, but the unavailability of these options ‘‘forces’’ the exceptional use of -ing compounding. My observation that conversion and -ing forms constitute an almost complementary distribution in SC, with conversion forms being unmarked options, is confirmed by the contrast between the unacceptable instances in (1b) and the acceptable instances in (2a). For example, compare the unacceptable -ing form *rainfalling with the acceptable conversion form rainfall and the unacceptable *weather-changing with the acceptable weather change. It should be noted that we do not see this distributional fact as conversion forms blocking -ing forms in SC. Rather, we will claim in section 5 that -ing SC is very di‰cult in PE as a result of the ‘‘recategorization’’ function of the su‰x -ing, while conversion allows SC because it does not have that function. See section 5 for details. In sum, PE uses conversion or less-productive su‰xes than -ing for SC, and SC in -ing is exceptionally allowed when these options are unavailable.

3. Subject Compounding in OE and ME In OE and ME, SC in -ing occurred more frequently than in PE, and its possibility was independent of the existence of conversion counterparts.

3. This restriction on the morphological property of a base word seems to apply to conversion in general, independently of syntactic category. A derived noun or adjective cannot be converted into a verb either (e.g., *to arrival, *to freedom, *to guidance, *to idleness, *to piggy, *to spoonful ) (Bauer 1983: 223–227, Marchand 1969: 372–373).

116

Akiko Nagano

For instance, OE compound nouns in (4)4 and ME compound nouns in (5) given below are just a few of many instances of -ing SC. These instances suggest that many of PE conversion instances of SC, such as (2a), originally had an -ing form. For instance, the PE conversion compound earthquake had the -ing form erthe-quakinge in ME. (4) eorþbeofung ‘earthquake,’ feaxfallung ‘shedding of hair,’ sæ-ebbing ‘ebbing of the sea’ (Kastovsky 1992: 367) (5) dai-dauing ‘day-dawning ¼ daybreak,’ drope-falling ‘drop-fall,’ erthe-moving ‘earth movement,’ erthe-quakinge ‘earthquake,’ sun-rising ‘sunrise’ (Kageyama 1985: 14) cok crowynge ‘cock-crow,’ day-springyng ‘day spring,’ hart stangyng ‘heart sting,’ sonne rysyng/arising ‘sunrise,’ son settyng/doun ganging ‘sunset’ (Tajima 1985: 125) SC by conversion was also allowed in OE and ME, and, as in PE, it seems to have been a rather productive process. ME produced instances like (6) below. (6) earthquake, earthquave, herte-bren ‘heartburn,’ nosebleed, sunne sine ‘sunshine,’ sunrise, sunset, toth-ake ‘tooth-ache’ (Marchand 1969: 76) In short, the -ing SC was rather freely allowed in OE and ME, but, in PE, its productivity has dropped. Instead, the rival conversion form has gained productivity.

4. Functional Change of the Derivational Su‰x -ing Kastovsky (1985, 1986) claims that a derivational a‰x has two main functions: ‘‘recategorization’’ and ‘‘naming.’’ Recategorization refers to the fact that a derivational su‰x usually changes the syntactic category of a word as required by its syntactic environment. Naming describes the process through which derivation yields a new name or label for an extralinguistic entity. For instance, compare the meanings of curiousness and 4. The su‰xes -ung and -ing were the same su‰x in OE. Kastovsky (1985: 241) treats -ung as an alternant of his -ing1 and groups the following OE derivatives into the same action noun category: (i) binding ‘binding,’ delfing ‘digging,’ brastlung ‘rustling,’ huntung ‘hunting’

Subject Compounding and a Functional Change

117

curiosity and those of callousness and callosity. The -ness form is a pure transposition from adjective to noun,5 while the -ity form can refer to an entity as a countable noun (e.g., They admired his dress, but only as a curiosity/ Callosities can be so painful as to a¤ect a person’s gait). Thus, in Kastovsky’s terms, -ness tends toward a recategorization su‰x, while -ity tends toward a naming su‰x. In the domain of nominalization, these two derivational functions correspond to Grimshaw’s (1990) distinction between ‘‘event nominals’’ and ‘‘result nominals.’’6 An event nominal is a nominalization that changes only the category of a base verb and inherits its argument structure, as in (7) below. A result nominal is a nominalization that refers to a concrete entity and does not inherit the argument structure of a base verb, as in (8) below. An event nominal has an ‘‘Event argument (Ev)’’ as its external argument and suppresses the original external argument of a base verb (as indicated by ‘x ¼ 0’), while a result nominal only has a non-thematic argument R, which binds a specific LCS-argument of a base verb (as indicated by ‘R ¼ y’).7 (7) assignV ! assignmentN Where do you get this from? Probably a couple of decades of feeding ’em. Why? Do you know something else. . .? > Maybe it is ‘‘just you’’? Or, maybe not. . . In the new use of re, re was added before these block quotes, to make explicit the connection between the quotation and the new material written below. Examples (18–20) provide some typical examples of this usage. (18) re: > Hey that’s major bummer to anyone, and if I found the thiefs I’d turn them in.

298

Stefanie Kuzmack

(19) Re:> I enjoyed the sci-fi excerpt on dolphins, but did not recognize the > source. Where is it from? Its not surprising you dont recognise it Erich, its from a totally unknow and obscure source, namely me. (20) RE: ‘‘But please respect that I DO belive WT to be the best system FOR ME (to fit my needs). I’m not going to appologize for that.’’. I, for one, certainly do respect that and you will not be asked to apologize for it. However, not all uses of re before block quotes are as simple as those above, with re followed by a single quotation. The use of re that led into ‘‘block quote re’’ was one where it could precede any kind of text, as in (21–22). In (21), re is used to quote the entire previous message, including the signature, and in (22), it introduces multiple quotations by di¤erent speakers. (21) Yes I’ve had that same problem. Removing the dust from around the speaker contacts will help. [. . .] RE: > I’ve got a Mac IIsi running system 7.5 (with update 7.5.2). > Sometimes > the sound shuts o¤ for no apparent reason. I can usually get > the sound > to switch back on by cranking the sound level up to 5 or 6. > Anyone had this problem or know how to fix it? – Brian

———————————————— [name and address of sender]

(quoted portion in italics)

How Medium Shapes Language Development

299

(22) re: >>> Are they up to their ‘‘usual tricks’’. . . >> You say you aren’t trolling? > Correct. > Tell me, why can’t MSFT answer a single one of my valid > questions? Are they ‘‘valid’’ questions? The fact that re introduces the entire message or exchange in these examples is not characteristic of a quotative, and illustrates the intermediacy of ‘‘block quote re’’ as a stage in re’s development. Uses like (21–22) are the exception rather than the rule, but they show how speakers interpreted re at the beginning of the block quote stage, not as a quotative complementizer but rather as a preposition that could precede anything. In the more typical examples, like (18–20), ‘‘block quote re’’ functions structurally like a complementizer in that it introduces full clauses, but it is somewhat unusual because non-quotative complementizers introduce either a single clause or conjoined clauses. However, introducing multisentence quotations is not unusual for an established quotative complementizer, as (23–25) illustrate. This ability is not unique to re, nor to written language. The di¤erence with ‘‘block quote re’’ is primarily one of degree; it introduces longer quotations than in spoken language, and it introduces multi-sentence quotations most of the time. Example (23) shows that ‘‘block quote re’’ is also not unique in its ability to introduce a quotation in isolation, without occurring in a full matrix clause. (23) Like, ‘‘Oh my god! You cut your bangs! They look great, Jane! Great, Jane, great!’’ (Singler 2001: 261) (24) And she’s like, she’s like, ‘‘Now I’m not even sure if I like him. Now when I look at him his face is kind of deformed and everything. Like you start seeing little flaws.’’ (Romaine and Lange 1991: 250) (25) He’s like, ‘‘Why, what’s wrong? Why you still sitting up?’’ [. . .] And he was like, ‘‘Well, I just got stuck with the guys. They wouldn’t let me, take me to my car or anything.’’ (Ferrara and Bell 1995: 266) There is no break in the transition from ‘‘block quote re’’ to adjoint complementation; the two are points on a continuum. In some uses of ‘‘block quote re,’’ the quotation is only a single sentence, albeit formatted

300

Stefanie Kuzmack

as a block quote. These cases are essentially no di¤erent from quotative re in adjoint complementation. Only the punctuation and the presence of line breaks show that the writer thought of the quotation as a block quote rather than part of the following sentence. (Examples (26–27) contain ‘‘block quote re’’; (28–29) contain quotative re in adjoint complementation.) (26) RE: ‘‘You will not be able to download the same message to more than one of your clients.’’ When I switched over to an Exchange Server, that was one of the issues I was having. (27) re: I stopped by an appliance store What is your definition of an appliance store? (28) Re, ‘‘Don’t take it out of my pocket,’’ please consider the following: [. . .] (29) However, RE: ‘‘Then again I realize that’s asking way too much of anyone today.’’, asking that is irrelevant if that is not the given martial artist’s objective. In e¤ect, ‘‘block quote re’’ and quotative re overlap; re’s use in adjoint complementation is a straightforward development, emerging from the simplification of ‘‘block quote re.’’

3. Noun phrase complementation vs. verb complementation The use of quotative re in noun phrase complementation is a separate development, prompted by the existence of quotative re in adjoint complementation,6 but not stemming directly from it. This use of re is most likely based on a parallel structure with the preposition re, seen in (30–31). In these examples, the preposition re is distinguished from quotative re in noun complementation (32–33) only by the fact that re’s complements here are noun phrases, not clauses.

6. Even the existence of ‘‘block quote re’’ alone may have been su‰cient to prompt speakers to use re in noun phrase complementation. This is a fine distinction, however, since ‘‘block quote re’’ is not truly distinct from re in adjoint complementation.

How Medium Shapes Language Development

301

(30) His answer re: color support was of course to mention the Pixar board. (31) I would like to give a special thanks to Bridger at RAND-UNIX for his comments re the bugs in the previous release. (32) Your comments re: ‘‘we dont know how to fight in one of these wars’’ is also true. (33) Can you supply the facts to back up your comment Re: ‘‘We wouldn’t even have got this far without him’’ Quotative re’s use in this structure would have been aided by the fact that many of the noun phrases that take the preposition re as a complement are equally capable of taking prepositions and complementizers as complements. (34) His answer that the manual should have the instructions wasn’t very helpful. (35) I would like to thank Bridger for his comments that the bugs can be found by. . . However, this source for the structure raises a question: why is quotative re not used in verb complementation? After all, the other English quotative complementizers, like and all, occur in that structure – in their cases, always as complements of the verb be (see 24–25). Moreover, the preposition re does occur in verb complementation (36–37), as well as in noun phrase complementation (30–31). (36) [. . .] recently one of the MDs and I spoke re: butter and margarine. (37) None of them have so far commented re that book. If speakers could use preposition re as a template7 for noun phrase complementation, then they would be expected to do so for verb complementation as well. Part of the reason that they did not do so is that cases like (36–37) do not function quite as well as (30–31) as templates. The noun phrases that the preposition re occurs with can uniformly take either a preposition or a 7. It should be pointed out that using as templates the structures that the preposition re occurs in is entirely di¤erent from reanalyzing those structures. The preposition’s complements in these cases were always noun phrases, and not simultaneously interpretable as clauses.

302

Stefanie Kuzmack

clause, so that speakers can easily imagine using re as a quotative complementizer where it had previously only been used as a preposition. The verbs that the preposition re occurs with, however, are less flexible in their complements. Some verbs related to speech, such as speak and talk, can take prepositions as complements, but not clauses. As a result, these verbs cannot be used with quotative re, and cannot contribute to the use of quotative re in verb complementation. (38) We spoke/talked about/re butter and margarine. *(39) We spoke/talked that/re margarine isn’t as healthy as people think. Say, the most generic quotative verb in English, also fails to provide a suitable template, although for a di¤erent reason: it cannot have a prepositional phrase as its only complement. As (40–42) demonstrate, it can take a clausal complement, and it can take both a clause and a prepositional phrase as complements simultaneously, but it cannot have a prepositional phrase as its only complement. *(40) You said re Arnold. (41) You said that we couldn’t trust the records. (42) You said re: Arnold that we couldn’t trust the records. Thus, say could theoretically take quotative re as a complement, unlike speak and talk. However, say provides only a flawed template for verb complementation, because it only occurs with the preposition re when there is already a clausal constituent present, highlighting the fact that re is a preposition, not a complementizer. Since say is the most generic quotative verb, and accordingly a commonly used one, it would have had considerable power to promote quotative re in verb complementation. That it does not promote that usage reduces the likelihood that speakers will use it as a complementizer with other verbs. Nevertheless, there are some quotative verbs that can take both clauses and prepositional phrases as complements, individually, and therefore could provide good templates: (43) He wrote/commented that he agreed with the first message in the thread. (44) He wrote/commented on the subject. Although write and comment are less generic than say, their meanings are not so specific as to be a barrier to using them with re. Write is applicable to nearly all online communication, and the noun comment is the most

How Medium Shapes Language Development

303

frequent noun to take quotative re as its complement. These verbs could provide the template for using re in verb complementation, and once speakers became accustomed to seeing quotative re being used with these verbs, extending quotative re’s use to say would be easier. Even with these verbs, however, re is not used as a quotative complementizer. Therefore, while the structures that quotative re is used in are probably influenced by the types of complements that these noun phrases and verbs can take, these syntactic factors do not entirely explain why re is not used in verb complementation. The key factor is the use to which re is put. Both ‘‘block quote re’’ and quotative re are used to bring quotations from earlier in a discussion back into the current discourse in order to set up responding to them, to alleviate the potential confusion caused by the delays between messages that are inherent to the medium. Noun phrase complementation allows the quotation to be embedded within a noun phrase, saving the sentence’s matrix clause for responding to the quote. (45) I disagree with your statement re ‘‘We should start a new, moderated newsgroup.’’ Using verb complementation for this task would require two sentences, one to report what was said and one to include the response. (46) You said/were like, ‘‘We should start a new, moderated newsgroup.’’ I disagree. This is because when speech is reported using verb complementation, the matrix clause is used to indicate what was said, and by whom. That property makes verb complementation appropriate for situations where the primary purpose of a sentence is to report who said what, as in a narrative. Since quotative like is used in telling narratives, verb complementation is a suitable structure for it. In the contexts where re is used, on the other hand, responding to what was said is more important than who said it, and so speakers chose noun phrase complementation over verb complementation because it was better suited to their purpose and to the medium.

4. Direct speech and indirect speech The nature of the online medium also influenced re’s ability to introduce direct and indirect quotations. Re is unusual in that it can introduce both, with no visible di¤erences in sentence structure. In (47), the prono-

304

Stefanie Kuzmack

minal deixis doesn’t change; in (48), the quotation is indirect, and we is changed to you. The deixis is the only clue; without context, it would be impossible to tell whether these quotations were direct or indirect. (47) > YOUi ARE TOO YOUNG TO REMEMBER ANY REAL > CARS ANYWAY [. . .] Now as for yourj comment re: youi are too young . . . yada yada yada. . . (48) Message A: I think you will find wek don’t advise to use the feed and kill insects type of systemic that is put on the ground. [. . .] Message B: And yourk comment re: ‘‘youk don’t use the ground kind’’ only serves further to point out the general and widespread ignorance over their use. . . Direct and indirect quotation in identical constructions creates ambiguity, but re’s function and the medium in which it is used reduce the potential for confusion. Because re is generally used to quote statements from earlier in the same discussion, the addressees were also present for the original utterance. Moreover, there are records online which allow the addressees to confirm the intended reading, if necessary. In spoken language this kind of flexibility could produce confusion, since quoted speech is often entirely new to the addressee, and there are no records to consult. Quotative like, all, and go are all limited to introducing direct quotations. Schourup (1982: 148–9) suggested that quotative go is restricted in this way in order to avoid ambiguity. Most quotative verbs can introduce both direct and indirect quotations, but there the ambiguity can be reduced by adding the complementizer that if the quotation is indirect. (49) Johnj said, ‘‘Ij was responsible for Lauren’s failure.’’ (Schourup 1982: 148) (50) Johnj said Ik was responsible for Lauren’s failure. (Schourup 1982: 148) *(51) Johnj said that ‘‘Ij was responsible for Lauren’s failure.’’ (52) Johnj said that Ik was responsible for Lauren’s failure. Re’s ability to take both direct and indirect speech complements is clearly not due solely to the online environment, since quotative verbs

How Medium Shapes Language Development

305

used in spoken language do have the potential to be similarly ambiguous. At the same time, the nature of the Internet and newsgroups undoubtedly makes it easier for sentences with re to be understood, despite the absence of any signs8 that direct or indirect quotation is intended.

5. Speech and thought Re’s sharply limited ability to quote thoughts is a result both of the structures in which it occurs, and of its function. My sample contains only two instances where re quotes thoughts, as opposed to statements: (53) Also his arrogance re: ‘‘I can be opener, wicketkeeper and captain at the same time, I’m wonderful aren’t I?’’ drives me mad. (54) Your attitude re: ‘‘if you are not for us, then you must be against us’’ is too simplistic a formulation. In the first case, an author describes the general attitude of a cricket player (who is not a participant in the newsgroup). The exaggeration and overall tone of the quotation make it fairly clear that the speaker in (53) is attributing unvoiced thoughts to the cricket player, and a portion of a reply message confirms that reading: (55) I don’t think he has made any great claims to being able to captain, keep, and open all at once. [. . .] I’ll give you that he hasn’t made a lot of public noise saying ‘‘I can’t do all three’’ In the latter example, the speaker in (54) quoted another participant in the discussion, whose actual words are presented in (56). (56) You have been arguing in the past that Maoist METHODS are WRONG. Now you have come to express pretty explicitly your mind saying that the Maoist reasons are WRONG! This clearly leaves us to believe that Government’s REASONS and METHODS of killing innocent civilians in the name of containing Maoist ‘‘terror’’ is right! 8. Although quotation marks could potentially be used for this purpose, Internet users do not use them consistently according to the standard rules of punctuation. Not only may a direct quotation have no quotation marks, as in (47), but an indirect quotation may be contained within quotation marks, as in (48). The use of quotation marks in (48) is not necessarily random or a simple

306

Stefanie Kuzmack

In this case, the quotation conveys the other speaker’s essential idea accurately, but the words are too di¤erent to consider this even an extremely flexible paraphrase (particularly given that the author of (54) had as much access to the exact words as we do), and the fact that the author chose the noun attitude shows that he saw this quotation as representing a mental state, not words.9 However, re’s ability to quote thoughts is restricted by the structures in which it is used. In adjoint complementation, the identity of the person being quoted cannot be specified explicitly. The contexts in which re is used make this construction work when actual statements are quoted, since speakers can either remember or look up the quotations, but it would be challenging at best with quoted thoughts. Noun phrase complementation is more open to quoting thoughts, but the structure is still a complicating factor. When re introduces the complement of a noun phrase, the noun of the noun phrase always characterizes the quotation in some way. The most common noun is comment, seen in (32), (33), (47), and (48), but other nouns are also used, such as statement (5), argument (10), implication (15), etc.10 Choosing a noun to characterize quoted speech is simple, because there are nouns that do no more than characterize the quotation as an utterance (or as a kind of utterance), and calling the quotation an utterance is not controversial. In order to quote thoughts, di¤erent nouns must be chosen, like arrogance and attitude in (53) and (54). Choosing a suitable noun to characterize thought requires error; it may have been motivated by the fact that the quotation is approximate rather than exact. Nevertheless, examples like these show that neither the presence nor the absence of quotation marks can be used to reliably predict whether a quotation is direct or indirect. 9. The choice of noun, as it quite explicitly categorizes the quotation as ‘‘an attitude or a general feeling,’’ indicates that this example meets one of Tagliamonte & Hudson’s (1999: 156) criterions for quoted thought. Their other test for distinguishing between quoted speech and quoted thought is inapplicable to re, since it states that a quotation should be taken to have quoted speech if it ‘‘advanced the story-line, or was part of an utterance to which the protagonists responded’’ (1999: 156), and re is not used in telling narratives. 10. The relative frequency of comment and the other nouns is not the result of the search strings used. In these cases, a noun phrase before re was stipulated in the search with a specific determiner ( your, his, her) followed by a wildcard, allowing any noun to head the phrase. Accordingly, the range of nouns found to be used with quotative re should be representative of the actual patterns of use.

How Medium Shapes Language Development

307

more e¤ort, because the nouns to choose from have more specific meanings than comment or statement, which are neutral in tone and can characterize most quoted speech (unlike arrogance and attitude). Even the generic thought could well be problematic, since it would make explicit the fact that the author is claiming to know another person’s thoughts. The lack of a generic noun to serve as a default in quoting thoughts reduces the likelihood that speakers will quote thoughts using re. Furthermore, quoting thoughts is extraneous to re’s usual function.11 In telling a narrative, including people’s internal reactions and thoughts is valuable, and presenting those thoughts as dialogue is more vivid than describing them, or presenting them as indirect speech. In that context, reporting thoughts and reporting speech are both ways of achieving the goal: telling a story. But when the goal is to bring statements back into the discourse in order to make clear that they are the object of the response, it is necessary to keep the quotation similar enough to the original that it is recognized as a previously made statement. Insofar as quoted thoughts di¤er from what another person actually said, they stand to confuse the issue of what one is responding to, rather than clarifying it. Thus, re is rarely used to quote thoughts, as a direct result of its function and the structures in which it occurs. However, those structures are themselves a result of re’s function, and its function is a result of the online environment. Even though that environment has no direct impact on re’s ability to quote thoughts, it indirectly caused that ability to be limited in two separate ways.

6. Conclusion Every aspect of quotative re has been shaped, either directly or indirectly, by the medium in which it evolved. The driving force behind quotative re’s development has been the need to remind the addressees of previous statements, before responding to those statements. Meeting that need became re’s primary function, and re’s function in turn shaped its other characteristics. Crucially, the need to bring statements back into the current 11. Note that in (53) and (54), the authors did not use re for its usual purpose. In (53), the author was quoting someone outside the discussion entirely. Re’s use in (54) is similar to its usual function, but the author of (54) had already reproduced the previous message in a block quote, making it clear what he was responding to. Re was not needed to do the same.

308

Stefanie Kuzmack

discourse is a direct result of the nature of e-mail and newsgroups. Conversations in e-mail and newsgroups do not occur in real-time, and the delays between messages make it helpful to quote the previous message. Also as an e¤ect of the delays, messages are often much longer than turns in a real-time conversation are. As a result, it is also useful to point out exactly to which part of the previous message a sentence or paragraph is responding. The fact that this is a written medium is also essential, since otherwise the delays between messages would not exist. Of course, correspondence has always involved delays, and the delays were significantly longer, before the Internet. However, there are two factors that favor online correspondence giving rise to a quotative like re, rather than o¿ine correspondence. First, the fact that online messages are electronic makes extensive quoting of previous messages both practical and easy, which encourages people to quote them more often. Second, correspondence on newsgroups involves far more people per message than o¿ine correspondence, allowing linguistic innovations to spread much faster. These factors made it much easier for a quotative like re to develop. Thus, the nature of the medium shaped the uses of the emerging quotative re, which in turn influenced the structures it occurs in. Those structures in their own turn, combined with re’s typical function, form the basis for re’s limited use for quoting thoughts. The written medium also enabled re to take both direct and indirect quotations as complements, with comparatively little confusion. The case of re demonstrates the importance of the role that context of use plays in shaping change: due to the nature of the online environment, the preposition re grammaticalized into a quotative complementizer that di¤ers from all other English quotatives in both structure and function, and rather than competing with them, complements them.

Appendix: Data collection All of the data presented here was collected from Google’s Usenet archives, which date back to 1981. Instances of quotative re can be di‰cult to locate, since a specialized use of re is present in the subject line of every reply message (e.g., ‘‘Re: Tech support question’’), and since the preposition re is used outside of the subject line, as well. Google unfortunately does not provide a way to automatically limit a search to the bodies of messages, but an adequate solution is to use searches that specify at

How Medium Shapes Language Development

309

least one word before re. This technique screens out occurrences of re in subject lines, since re occurs at the beginning of the subject line. Words and phrases such as however, above, and your/his/her (noun) occur before re with some frequency, and can be used in the search string to eliminate instances of subject-line re. Similarly, a higher proportion of hits with quotative re can be obtained when the environment following re is specified. For instance, if the search string specifies a nominative pronoun after re (e.g., ), the constituent following re will be a clause, making that instance of re a complementizer. If the search string specified an accusative or genitive pronoun after re (, ), on the other hand, the constituent following re would be a noun phrase, and that instance of re would be a preposition. (a) [. . .] Mary’s comment re: ‘‘I can’t do this anymore’’ brings back some memories. (b) Ok, so here is Michael Greenberg’s latest post re me, which truly shows his colors [. . .] (c) But the comment re my way of listening to Schoenberg implies I still haven’t made myself clear. The consequence of having to specify portions of re’s environment in this way is that the more that is specified in the search, the less of the full range of re’s uses will appear in the search results. Inevitably, some instances of quotative re are filtered out along with the false positives. Fortunately, though, it is still possible to obtain a wide variety of uses.

References Bresnan, Joan W. 1970 On Complementizers: Toward a Syntactic Theory of Complement Types. Foundations of Language 6: 297–321. Ferrara, Kathleen and Barbara Bell 1995 Sociolinguistic Variation and Discourse Function of Constructed Dialogue Introducers: The Case of be þ like. American Speech 70(3): 265–290. Marchant, J. R. V., and Joseph F. Charles, eds [1952] Cassell’s Latin-English, English-Latin Dictionary. New York: Funk & Wagnalls Company. Oxford English Dictionary, 2nd ed 1989 Prepared by J. A. Simpson and E.S.C. Weiner. http://www.oed.com (accessed August, 2006).

310

Stefanie Kuzmack

Romaine, Suzanne, and Deborah Lange 1991 The Use of Like as a Marker of Reported Speech and Thought: A Case of Grammaticalization in Progress. American Speech 66(3): 227–79. Schourup, Lawrence 1982 Quoting with Go ‘Say.’ American Speech 57(2): 148–149. Singler, John Victor 2001 Why You Can’t Do a VARBRUL Study of Quotatives and What Such a Study Can Show Us. U. Penn Working Papers in Linguistics 7(3): 257–278. Tagliamonte, Sali and Rachel Hudson 1999 Be like et al. beyond America: The Quotative System in British and Canadian youth. Journal of Sociolinguistics 3(2): 147–172.

Commentary on Kuzmack, How Medium Shapes Language Development: The Emergence of Quotative Re Online Anatoly Liberman Stefanie Kuzmack’s paper is devoted to a small event in the development of English. Yet its material poses several general questions. I would like to dwell on three of them. Terms like Old English and Middle English come naturally to language historians. Both periods lasted several centuries and ended long before we were born. Even late, to say nothing of early, Modern English did not begin yesterday. We accept the fact that numerous cataclysms happened between the settlement of Britain and the days of King Alfred, between the Norman Conquest and Chaucer’s epoch, and so on. We express little surprise even at the relatively recent Great Vowel Shift, which gave the vocalic system of English its unique appearance. But how often do we witness the birth of a new construction or a new phoneme? Kuzmack’s case is rare: the object of her investigation is an event that occurred before our eyes, and we know both its environment and its causes. Yet there is a good deal to be said about the phenomenon that attracted her attention. Moreover, we realize that the history of English need not be about the past. Synchrony, as Roman Jakobson never tired of repeating, is dynamic. The distinction between oral and written language is so familiar as to be trivial, and we forget that, although oral language is primary, writing must not be identified with its well-edited transcript. No one has done more for investigating the relations between oral speech and writing than Josef Vachek (see his main works collected in two books: Vachek 1973 and Vachek 1989; both include exhaustive bibliography). In conversation, we lose no time reacting to what is said. The reward for the immediacy of speech is its naturalness; the punishment is its evanescence. Written texts allow people to preserve information ‘‘forever’’ and recall it at any time in the future. Therefore, in literate societies oral speech and writing are equally important. The Internet and the chatroom have partly wiped out the di¤erence between the two media. Emails fly back and forth with the speed of remarks in a telephone exchange. When we speak, we may a¤ord

312

Anatoly Liberman

the luxury of using an imprecise word and shaky syntax (our inaccuracies and incoherence will probably not be noticed). By contrast, good writing is an art, but nowadays most people do not reread their emails and do not mind looking ‘‘spontaneous’’ in them. This type of writing is indeed a transcript of oral speech – a situation unthinkable thirty years ago. Kuzmack has found herself on a linguistic planet whose existence would have surprised Vachek and his contemporaries. The main material of language historians has always been manuscripts and books. Those are sources of the same order, for manuscripts were produced with care, could be very expensive, and, like books, were expected to have a long lifespan. At one time, researchers examined private letters (among other things) for the sake of spelling mistakes and the degree of variation among the literate and semiliterate writers. They knew that they were reading documents meant for one recipient. Today, as this article shows, the contours of privacy have become blurred, and any number of outsiders may claim our innermost thoughts: once the Internet gets hold of our words, they end up in public domain. Anyone may put them to use, and linguists who analyze the written data at their disposal are once again invited to draw conclusions from the chaos that was typical of the private documents of the remote past. Phoneticians have been studying the patterns of individual speech ever since phonetics emerged as a branch of scholarship. Likewise, our knowledge of modern usage depends to a considerable extent on the study of recorded and overheard conversations. Now written samples are being studied by comparable methods. The paper opens with an introduction and an overview, looks at the emergence of quotative re, and goes on to examine the use of re in noun phrase complementation (as opposed to verb complementation) and in direct speech (as opposed to indirect speech). Then Kuzmack concentrates on the ability of re to introduce actual statements, as opposed to thoughts. The paper ends with a conclusion, a note on data collection, a predictably small bibliography, and notes. The most amazing note deals with the folk etymology of re. It would seem that the origin of re is the least controversial question one can imagine. Yet some people believe that re goes back to the first syllable of reply, response, or regarding. This says something about the level of our education. Like every other language phenomenon of this type, quotative re has its syntax and distribution; both have been described in the paper. Above, I have referred to a predictably small bibliography. The phenomenon in question has not existed long enough to become a popular subject of research, but Kuzmack had a few predecessors. The time-honored quota-

Commentary on Kuzmack

313

tive verb has been say (–he says . . . –I says, etc.). Then go appeared (–he goes, –she goes) and like (She like: ‘‘No, that’s not true’’). Neither go nor like will win our admiration, but they bloom on the rich garbage of language creativity indi¤erent to our praise and disdainful of our disapproval. Four articles in Kuzmack’s bibliography are devoted to them; one of them deals with re, the Internet counterpart of like and go. The computer changed our life in many unpredictable ways. It is no wonder that language, which reflects everything in human experience, has reacted to this momentous change. We now have an addition to the house of general and, as it turns out, historical linguistics, namely, ‘‘Language and the Internet.’’ Welcome to the housewarming party.

References Vachek, Josef 1973

Vachek, Josef 1989

Written Language: General Problems and Problems of English. Janua Linguarum. Series critica 14. The Hague [and] Paris: Mouton. Written Language Revisited. Selected, edited and introduced by Philip A. Luelsdor¤. Amsterdam [and] Philadelphia: John Benjamins.

Response to Commentary by Liberman Stefanie Kuzmack

As Liberman says, the Internet has brought written language closer to spoken language. In chatrooms and in text messaging, writing has even become a practical means of real-time communication. And yet, as the case of re shows, the fact that this is written language still has an impact on the utterances that are produced. Much of online communication is still asynchronous, and that fact was crucial to re’s development. If threaded discussions were not asynchronous, and did not involve lengthy delays, re would never have been needed to refresh participants’ memories of what had been said in the last turn. Still, these discussions do also possess some characteristics of oral communication. Participants in these conversations could briefly cite previous messages instead of quoting from them as they do (often extensively), and could force each other to check previous messages for context. This is particularly noticeable since the previous messages are easily accessed; it is easier to locate an earlier message in a thread than it is to follow up on a citation in an academic work. However, since following up on citations is not a normal part of conversation, participants provide the context that the addressees need. Thus, the oral qualities of Internet language are worth taking note of, as Liberman does, although it is still a written register. The fact that many people believe that re has its origins in the first syllable of reply, response, or regarding may in part reflect changes in education, particularly with respect to the role of Latin. However, it may also be due to trends in the use of Latin words in English, and in the use of abbreviation. Latin words and phrases in general are less commonly used than they once were (although that in itself may also be a result of changes in education). Re is still used to mark the subject in memoranda, but the parallel online use of re is distinctly di¤erent, and people who are not familiar with re could overlook the connection. At the same time, new abbreviations are extremely common, especially online. Even speakers who know the Latin might conclude that re is a new abbreviation, given that its use in subject lines is so di¤erent from previous uses.

Response to Commentary by Liberman

315

This paper was devoted specifically to re, but as Liberman noted, the development of new quotatives is ongoing. Re may be the only currently used quotative that developed in a written context, but it is not the first quotative to do so. Moore (2006) discusses a quotative use of Latin videlicet (abbreviated viz.) that developed in legal documents in Early Modern English. The quotative uses of re and videlicet have somewhat similar origins: they both developed in the written medium and in specific genres, and more superficially, both came from Latin. However, the quotative videlicet was essentially restricted to the legal register, and that use has disappeared (2006: 257). Re may share the same fate, since it too appears to be confined to a single genre, but the number of participants in the online register provides hope that even if it does not expand to other contexts, re’s quotative use may stabilize.

Reference Moore, Colette 2006 The Use of Videlicet in Early Modern Slander Depositions: A Case of Genre-specific Grammaticalization. Journal of Historical Pragmatics 7: 245–263.

Author Index Listings do not include authors cited for their linguistic usage or brief exemplary citations of language. Adams, Valerie 111–112, 138 Aitchison, Jean 187 Albright, Adam 101–102, 106, 108 Allen, Cynthia 43, 85 Arnon, Inbal 53 Arono¤, Mark 117 Ayres, Alfred 143–145, 147–148, 150, 152 Baker, Peter 67 Baker, Robert 143–145, 149 Bakhtin, Mikhail 15 Ball, Catherine 17 Barlow, Michael 82 Barnhart, Robert 171 Bassett, Molly 111 Bauer, Laurie 28, 113, 115 Beecher, Henry 65 Bell, Allan 246 Bennett, J. A. W. 245 Benskin, Michael 270, 278, 289 Benveniste, Emile 34 Berg, Thomas 12 Bernstein, Theodore 143–144, 149, 159 Biber, Douglas 21, 61, 63, 232–233 Biese, Y. M. 28 Blake, Norman 196 Bloomfield, Leonard 111 Bo¤ey, Julia 262 Bopp, Franz 170 Borer, Hagit 119 Bosworth, Joseph 166 Braaten, B. 208 Bradley, Henry 167, 169, 171 Branchaw, Sherrylyn 3, 4, 8, 87–109, 288–292 Bresnan, Joan 294

Brians, Paul 143–144 Brinton, Laurel 16 Buchstaller, Isabelle 60 Burchfield, Robert 143–144, 169, 173 Burnley, David 257, 266–268 Burzio, Luigi 124 Butler, Christopher 13 Buyssens, Eric 174 Bybee, Joan 66, 82, 91–92, 101, 106, 108 Campbell, Lyle 53 Capelli, C. 209–210, 226 Carlson, Lauri 17, 21 Casaubon, Meric 162 Chambers, J. K. 61 Chapman, Don 5, 7, 141–159, 202– 206 Charles, Joseph 295 Cherewatuk, Karen 248–249 Chomsky, Noam 12, 137 Chung, Sandra 40 Cienki, Alan 68 Clark, Herbert 12 Coleridge, Samuel 164 Collins, Peter 17 Cooper, Helen 244, 258 Copperud, Roy 143–144 Coseriu, Eugenio 232 Couper-Kuhlen, Elizabeth 16 Coventry, Kenny 68 Craigie, William 169 Croft, William 13, 82 Crystal, David 202 Culicover, Peter 37, 40–41, 51 Culpeper, Jonathan 61–62 Cunningham, I. C. 267 Cuyckens, Hubert 12, 68, 83

318

Author Index

Dal, Ingrid 208, 212, 220 Dalton-Pu¤er, Christiane 125, 127– 128 Dasher, Richard 13, 204 Delin, Judy 17 den Diken, Marcel 17 Denison, David 213–214, 227, 229 De Smet, Hendrik 82 Detges, Ulrich 11, 13–14, 16 Diefenbach, Lorenz 166 Dirven, Rene´ 12 Dolezal, Frederic 164, 171 Downes, William 217 Downing, Pamela 138 Doyle, A. I. 291 Drinka, Bridget 240, 243 Du Bois, John 12–13 Ducrot, Oswald 14, 34 Eckert, Penelope 233–234, 250 Eitler, Tamas 246 Ekwall, Eilert 188 Ellega˚rd, Alvar 213–214, 217–220, 227–229 Ellis, Alexander 217, 227 Elman, J. L. 108 Ettmu¨ller, Ludovicus 166 Evans, Vivian 67–68 Feist, Sigmund 168, 171 Fennell, Barbara 188 Field, Frederic 244–245 Fiengo, Robert 40 Fillmore, Charles 71, 234 Filppula, Markku 4, 7, 17, 105–109, 207–230 Finnegan, Edward 62 Fischer, Olga 11, 228–229 Fitzmaurice, Susan 61 Flemming, Edward 102 Ford, Cecilia 16 Fowler, H. W. 143–144, 152, 159 Freeman, Edward 208, 226 Galloway, Andrew 244–246, 258

Garner, Bryan 145–152, 158 Geckeler, Horst 232 Genee, Inge 208, 212–213, 221 Gergel, Remus 49 German, Gary 212 Gerritsen, Marinel 12 Geurts, Bart 15 Ginzburg, Jonathan 37, 40–41, 51 Givo´n, Talmy 15, 62 Goldberg, Adele 82 Grice, H. Paul 12 Grimm, Jacob 166, 170, 184 Grimshaw, Jane 117–118, 122, 137 Haiman, John 12 Hamerow, Helena 209 Hankamer, Jorge 37, 40, 48 Hanna, Ralph 265, 268 Hardt, Daniel 37 Hare, M. 108 Ha¨rke, Heinrich 209, 226 Ha¨rma¨, Juhani 15 Harris, Alice 52 Haspelmath, Martin 13, 211–212 Hayes, Bruce 101–102, 106, 108 Helquist, Elof 187 Helvigius, Andrae 165–166 Hering, Ju¨rgen 6, 187–206 Hickey, Raymond 208, 226 Higgins, Francis 16–17 Higham, Nick 209, 226 Hoad, T. F. 171 Hodgson, William 143 Hofmeister, Philip 53 Holquist, Michael 15 Holthausen, Ferdinand 71, 75 Hopper, Paul 13, 16, 31, 66 Horn, Laurence 34 Huddleston, Rodney 29–31, 213 Hudson, Rachel 306 Ihalainen, Ossi 218 Ihre, Johan 166 Jackendo¤, Ray 37, 40–41, 51–52

Author Index Jackson, H. J. 164 Ja¨ger, Andreas 229 Jamobson, Roman 311 Jesperson, Otto 28–29, 207, 211, 226 Johnson, Edward 164 Jordan, Richard 279 Joseph, Brian 13 Junius, Franciscus 161, 163, 165, 167, 171 Kageyama, Taro 116 Kaltschmidt, Jakob 167 Kastovsky, Dieter 116–117, 119, 121, 127–128, 136, 232 Kaufman, Terrence 210–211 Kehler, Andrew 37, 40, 51 Keller, Rudi 12 Keller, W. 207, 226 Kemmer, Suzanne 82 Kemmerer, David 68 Kempson, Ruth 11 Kennedy, Arthur 163 Kiliaen, Cornelius 161, 166 Kilpio¨, Matti 39 Kim, Kyu-Hyun 17 Kiparsky, Paul 12 Kiparsky, Valentin 211 Kitigawa, Yoshihisha 40 Klein, Ernst 170–171 Klemola, Juhani 4, 7, 105–109, 207– 230 Kluge, Friedrich 167–168, 170, 174 Knobloch, Johan 174 Koch, Peter 233, 242, 258 Kohnen, Thomas 233 Koma, Osamu 119 Ko¨nig, Ekkehard 13, 15, 31, 212 Korhonen, Jarmo 15 Kortmann, Bernd 16 Krygier, Marcin 81, 87–88, 92–94, 98, 100, 106 Kuzmack, Stephanie 1, 5, 8, 156–159, 293–315 Kyto¨, Merja 61–62

319

Labov, William 61, 219, 231–233, 249, 262 La˚ftman, Emil 173–174, 184 Laing, Jennifer 209, 226 Laing, Lloyd 209, 226 Laing, Margaret 270, 289 Lambrecht, Knud 17 Langacker, Ronald 12, 38, 61, 82 Lange, Deborah 293 Lasnik, Howard 39 Lass, Roger 91, 94 Leech, Geo¤rey 15 Lehmann, Christian 13, 18 Leibnitz, Gottfried 163, 165 Leith, Dick 93–94 Lemon, G. W. 166 Levinson, Stephen 68 Lewis, Henry 215 Liberman, Anatoly 1, 6, 9, 161–186, 311–315 Lieber, Rochelle 111, 113 Lightfoot, David 12, 38 Locke, John 163 Luraghi, Silvia 67–69 Lutz, Angelika 208, 212 Lynch, Jack 143–144 Lyons, John 15, 187, 194, 198 MacDonald, Fiona 231 Mair, Christian 62 Mann, William 16 Marchand, Hans 28, 111, 115–116, 136 Marchant, J. R. V. 295 Ma¨tzner, Eduard 166 May, Robert 40 McMahon, April 13 McNeill, George 282 McWhorter, John 208, 211, 213, 220– 221, 226, 228 Merchant, Jason 39–41, 51–53, 55 Messing, Gordon 174 Meurman-Solin, Anneli 219 Milroy, James 13, 61 Milroy, Lesley 233–234, 250

320

Author Index

Minkova, Donka 93, 103 Minsheu, John 6, 161–163, 165, 167 Mittendorf, Ingo 208 Moder, Carol 101 Mooney, Linne 291 Moore, Colette 315 Mosse´, Fernand 273, 279 Mu¨ller, Eduard 167 Murray, James 169, 171 Mustanoja, Tauno 129, 136, 219, 227

Poutsma, Henry 16 Preusler, Walther 208, 215–216, 220– 221, 226–227 Prince, Ellen 17 Pullum, Geo¤rey 29–31, 48, 53 Putter, Ad 267 Quirk, Randolph 30, 87, 108

Oberlander, Jon 17 Oesterreicher, Wulf 233, 242, 258 Ohler, Norbert 231 Onions, Charles 169 Oppenheimer, Stephen 209–210, 226 Oshita, Hiroyuki 122

Radden, Gu¨nter 11–12 Rask, Rasmus 170, 184 Raumolin-Brunberg, Helena 61, 231– 233 Renouf, Antoinette 113 Reynolds, Susan 191 Richardson, Charles 165 Rissanen, Matti 62 Roeper, Thomas 112, 137 Romaine, Suzanne 61, 231, 233, 293 Room, Adrian 188, 192 Ross, John 37–38, 51, 55 Rothwell, William 236, 257 Roulet, Eddy 14 Rudanko, Juhani 68 Runde, Emily 8, 257–292 Rutherford, William 29, 31 Ryder, Mary Ellen 138

Panther, Klaus-Uwe 11–12 Parkes, Malcolm 268, 291 Partridge, Eric 170–171 Patten, Amanda 17 Paulasto, Heli 212 Pearsall, Derek 265, 267 Pedersen, Holger 215 Pennanen, Esko 28 Perlmutter, David 124 Pesetsky, David 53 Peters, Pam 142, 158 Pitka¨nen, Heli 208 Pfei¤er, Wolfgang 187 Plag, Ingo 28 Plato 161 Poppe, Erich 208, 212 Poussa, Patricia 208, 221, 226

Sag, Ivan 37, 40–41, 51, 53 Sakahara, Shigeru 31 Sandved, Arthur 269 Samuels, M. L. 8, 268, 276, 289 Schendl, Herbert 234, 236, 238, 257 Schourup, Lawrence 304 Schwenck, Konrad 167 Schwenter, Scott 11, 13–16, 21 Segen, Bazyl 67 Seibicke, Wilfried 174 Selkirk, Elizabeth 112 Shimada, Masaharu 28 Shizawa, Takashi 31 Shonk, Timothy 268 Short, Michael 15 Siegel, Mu¤y 112 Sievers, Eduard 166

Nagano, Akiko 2, 4–5, 28–35, 111– 139 Nagle, Stephen 52 Nakau, Minoru 31 Nevalainen, Terttu 16, 187, 202, 204, 231–233, 262 Nunberg, Geo¤rey 142 Nølke, Henning 15 Nykiel, Joanna 2–3, 37–66, 81–86

Author Index Sihler, A. L. 134, 136 Skeat, Walter 161, 167–171, 183 Skinner, Stephen 163, 165, 167 Slack, Jon 68 Slobin, Dan 12 Smith, A. H. 187–188 Snyder, William 137 Sommer, Wililam 166 Sørensen, Knud 240, 243 Spurlock, Kurt 111 Stainton, Robert 37, 40–41, 48, 51–53 Stein, Dieter 12 Stenton, Doris 193 Stokes, Myra 267 Strang, Barbara 207 Stratmann, Francis 166–167 Stubbs, Michael 35 Stubbs, William 208, 226 Svensson, Ann Marie 6–7, 183–206 Sweetser, Eve 29, 31 Sykes, Bryan 210, 226 Szemere´nyi, Oswald 134 Taavitsainen, Irma 15, 61, 233, 241– 242 Tacho, Elizabeth 7–8, 226–264 Tagliamonte, Sali 306 Tajima, Matsuji 116 Taylor, John 12 Taylor, Joseph 175 Thomason, Olga 3–5, 69–86, 132–139 Thomason, Sarah 210–211 Thompson, Sandra 15–16 Timberlake, Alan 38, 61 Timberlake, Henry 231 Tolkien, J. R. R. 208, 212

321

Tooke, Horne 163–164, 167 Tottie, Gunnel 15 Traugott, Elizabeth 2–3, 11–36, 60– 66, 83, 111, 204 Tristram, Hildegard 208, 212, 221 Trousdale, Graeme 83 Tyler, Andrea 67–68 Vachek, Josef 311 van der Auwera, Johan 208, 212–213, 221 van der Wur¤, Wim 228–229 Van der Zee, Emile 68 Vennemann, Theo 208, 212 Verdon, Jean 231 Vezzosi, Letizia 212 Vidrine, Rachel 143–144 Visser, Gerald 208 Vogel, Bertram 282, 290 Wagner, Heinrich 221 Waksler, Rachelle 60 Waltereit, Richard 11, 13–14, 16 Ward, Gregory 51 Webber, Bonnie 37 Webster, Noah 165, 167 Wedgwood, Hensleigh 167–169, 171 Weekley, Ernest 169–171, 175 Whitelock, Dorothy 190–191, 197 Wiggins, Alison 266–268 Williams, Edwin 40 Wilson, Kenneth 143–144 Yonekura, Hiroshi 121 Young, Wendy 142 Zwicky, Arnold 60

Subject Index ablative 295 ablaut 3–4, 87, 89, 91–92, 96, 98, 100, 102, 105, 108, 174 accommodation 246 acculturation 7, 209, 226–227 accusative 3, 69–71, 73, 76, 81–83, 85, 309 ad ‘to’ 236 adjoint complementation 9, 294, 296, 299–300, 306 adripare 236 adventus Saxonum 208, 210, 226 adverbial clauses 30–31 adversatives 15 agentive noun su‰xes 127 ALL-cleft 2, 17, 19 American English 107, 109 analogy 93, 101–102, 106–108, 147 anaphor(a) 3, 37–38, 40, 52, 65 Anglo Norman Dictionary (AND) 234, 236 Anglo-Norman 7, 232, 236, 243–245, 249, 257, 259–260, 271 Anglo-Saxons 7, 100, 166, 207–210, 226, 236, 258, 271 antecedent structure 37, 47, 51–52 antonymy 232 arbitrary justification 5, 146–148, 154, 157 archaeological evidence 7, 208–209, 221, 226 areal-typological evidence 7, 208, 211–213, 221, 227 argument structure 4, 39, 41, 51, 117– 119, 122–124, 134, 136 ariven/arrive 231 arrival 236 arriver 234 asynchronous correspondence 9, 314 Auchinleck Manuscript 8, 265, 266– 268, 271, 276, 278, 283, 288, 291– 292

audience design 246 authorship 259, 279, 282 Avestan 75 bad data problem 8, 262 Bare Argument Ellipsis 53–54 bigot 168 bilingualism 7, 210, 226 biography 233, 242, 261 Blanchardyn and Eglantine 242, 246, 261 block quote (re) 296–300, 303 blow 107 borough/burh 6, 188, 190–194, 198, 206 borrowing 7, 174, 187, 194, 215, 219, 227–228, 232, 234, 236, 241, 249, 257–258 bow 94, 96 Breton 215 British English 107 Britons 7, 207, 209–210, 226 Brythonic languages 208, 215, 221 Canon/canonical 141–143, 149, 153– 154, 156, 158 Canterbury Tales 195, 242 case-assigner 137 causative do 213–214, 227 ceaster 188 Celtic 7, 166, 187, 207–215, 220–221, 226–228 change from above 219, 249 change, diachronic 2–5, 12, 16, 28, 31, 42, 67, 86, 105, 129, 158, 164, 234, 258 chanson de geste 244 chide 167 chilvaric literature 248–249 chronicle 188, 192, 242–245, 247–250, 258–259, 261 city 190, 194–195, 198, 203–204

Subject Index clause 19, 30–31, 62–63, 76, 150, 294, 299, 302–303, 309 cockney 162 cocktail 172–175 collocation 71, 78, 204 comment 301–302, 304, 306–307, 309 complex event nominals 117 compounding 4–5, 111–116, 122–125, 128, 132–134, 136–138, 271, 273 concessives 2, 15 Confessio Amantis 242 conservatism 157 constituent transparency 83 contact 7, 208, 210–211, 220–221, 228 containment 69, 70, 82 context 2, 7, 14–21, 28–31, 34–35, 37, 41, 202–204, 206, 246, 304, 308, 314–315 conversation 16–17, 242, 293, 308, 311, 314 conversion nominals 118–119, 121, 125, 133, 137 conversion 2, 28–29, 34–35, 113, 115– 116, 118–119, 121–122, 124–125, 132–133, 137 copying practice(s) 266, 291 Cornish 215 corpora 8, 265, 283–284, 290 Corpus of Middle English Prose and Verse (CME) 235, 241, 259, 261 Corpus of Older Scots 219 correspondence 43, 46, 49, 243, 258, 261–262, 308 Cratyllus 161 dative case 69–71, 73–75, 81–83, 85 daughter 176 Delta Scribe 291 demographic evidence 7, 208–210, 221 Denmark 209 denominal 91, 236 dental 91, 100, 109 departure 234

323

derivational a‰x 1, 4–5, 111–112, 115–117, 119, 121–122, 125–129, 135–137 deverbal compound nouns 113, 124, 132 dialect 106, 214, 217–220, 227, 291 dialogicity 2, 11, 14–21, 28–31, 34–35 dialogue 17, 29, 45–46, 61, 307 Dictionary of Old English Corpus 88, 105 Dictionnaire du Moyen Franc¸ais (DMF) 234 dig 91, 102 direct speech 303–304, 312 direction(al(ity)) 3, 67, 69–78, 81–83, 85–86 discourse 1, 9, 13–15, 17, 21, 37, 41, 45, 48, 52–55, 62, 66, 149, 293, 303, 307–308 dive 87, 94, 96 document 8, 211, 215, 217–218, 220, 242, 312, 315 donor language 249 drag 92, 107 draw 95, 107 Dutch 163, 165, 176, 178, 187, 213 Early Middle English 4, 91–92, 100, 188, 190, 213, 262 Early Modern English 7, 37, 62, 102, 218–219, 235, 260, 315 Early Old English 187 East Midland 276, 279 economic change 6, 187–188, 198, 206 Edward I 193 Edward IV 248 electronic edition 268 electronic messages 9, 293, 295–296 ellipsis 37, 39, 41, 44, 65 e-mail 9, 297, 308 -ende 136, 269 Epea pteroenta 164 epistemic modality 13, 15, 30–31 etan/aet 86, 98 ethnic cleansing 209, 226

324

Subject Index

etiquette rules 5, 142, 156 etymological dictionary 6, 163–178, 183–186 etymology 1, 6, 9, 146, 150, 154, 157, 159, 161–179, 183–186, 235–236, 257, 295, 312 event nominals 4, 117–118, 121–122, 124, 133–134, 137 exemplar 82, 266–267, 270, 275, 278– 280, 282, 288–289 extra-linguistic evidence 208–221, 226–228 face-to-face interaction 233–234, 250 false positives 296, 309 fare 94–95 fiction 46–50, 61–62, 242 field/ feld 196, 197 final velar 91, 108 Finnish 166–167, 178, 211 First Projection Principle 112–113 First Sister Principle 112–113 flow 96, 97 focus particle 15, 30 folk etymology 9, 295, 312 four principal parts 2, 88, 89 free variation 73, 75, 78 French 7, 13–14, 19, 34, 100, 168– 169, 172–174, 178, 184, 194, 198, 228–229, 236, 240, 242–246, 249, 258–260, 271–273 frequency study 235, 241, 244–245, 261 functional pull 238, 257 galoot 176 gender variation 232 genitive case 73, 133, 309 genre 21, 45–46, 50, 61–63, 65, 175, 227, 232–233, 240, 258–259, 265, 315 gentry 244, 248–249 geographical distribution 214, 216– 219, 227

German 38–39, 161–162, 164–167, 176, 178, 187, 212–213 Germanic 6–7, 70, 125, 127, 165–166, 169, 175–176, 178, 184, 186–188, 206–207, 212–213 gerundive su‰x 128–129, 137 Gloucestershire 276 god 165, 170 Gothic 161, 166, 169, 171, 175, 178 Gothic 161, 166, 169, 171, 175 grammar 1, 100, 207–208, 211, 226 grammaticalization 13–14, 16, 18, 30– 31, 293, 297 Great Britain 249 Greek 161–163, 166, 172, 178 Grice’s Maxims 12 grow 96–97 hand 166, 171, 279–282 have 211, 273–275, 291 Hebrew 161–162, 166 Helsinki Corpus of English Texts (HC) 60, 233, 235, 239–242, 258 Hittite 172, 178 honor 35, 271–273 Horn Childe & Maiden Rimnild 8, 279, 281–283, 290 host 271–273 hostility 74–78 how-mismatch 41, 43–44, 47–49, 62 h-retention 270–272, 274–275 hyponomy 232 Icelandic 161, 166, 178, 212 iconicity 11–12 i-mutation 236 indirect speech 15, 303–304, 307, 312 Indo-European 6, 70–71, 85, 134–136, 162, 166, 169, 174, 176–178, 184, 186, 213 informative wh-phrases 53 -ing 4–5, 111–116, 118–129, 132–138 initial 8, 270–271, 273, 275, 283 interaction 12–16, 21, 28, 32, 208, 232–234, 250

Subject Index internal possessor construction 208, 212, 220 Internet 293, 295–297, 305, 308, 311– 314 Ireland 209–210, 213, 263 Isle of Man 210 it 273–275 Italian 13, 176, 178 IT-cleft 17–18, 212, 220, 226 King Horn 238, 261 lad 162 land 8, 236, 249, 279–282, 291 landen 238, 257 Langtoft’s Chronicle 245, 246–247, 258–259, 261–262 language acquisition 12 language change 2, 8, 12–13, 232, 240, 293, 295 language development 67, 73 language internal shift 233, 236 Late Latin 236 Late Middle English 43, 44, 47, 62, 194–196, 243, 248, 288 Late Modern English 37 Late Old English vowels 279 Late Old English 52, 213 Latin 9, 19, 34, 47, 76, 161–163, 165, 169, 172, 178, 188, 236, 240, 242, 244–246, 249, 258, 260, 293, 295– 296, 314–315 Layamon’s Brut 188, 240, 245, 258, 261 lenden 8, 232, 234, 236–238, 241, 243, 245–249, 257–258, 260, 263 let 99, 100 letters 219, 233, 242, 248, 258–259, 261–262, 264, 312 lexeme 187, 194, 232, 234, 236–238, 243, 245, 248–249 lexical borrowing 234 lexical change 7, 233–234, 250 lexical gap 198, 203 lexical practice(s) 267–268, 291 lexical restructuring 258

325

lexicography 142, 161–162, 164–165, 168, 171 lexicon 1, 5–9, 101, 194, 206, 231, 234, 236, 271 licensing context 37, 41 linguistic aspect 236 Linguistic Atlas of Late Mediaevel English (LALME) 8, 267–268, 273, 276, 278–279, 283, 288–289, 291 linguistic profile(s) 8, 267, 269, 291 liquid-consonant cluster 98 Lithuanian 178 loanword 7, 169, 188, 194, 198, 207, 211, 228, 229, 234, 236, 237, 239, 240, 241, 243–246, 248, 249, 257, 261, 262 location 6, 53–54, 69, 71–78, 81–83, 85–86 location-to-direction semantic development 71, 75 Logic 146, 148–149, 154, 156–157 London 192, 194, 219, 265, 268, 269, 276, 279, 283, 290 man 163, 165, 170, 175 Mannyng’s Chronicle 258–259, 262 manuscript miscellanies 291 matrix clause 294, 299, 303 meaning mappings 125, 127 meaning variation 234, 247 medieval England 236, 258 medium-distinction 233 memoranda 9, 293, 295–296, 314 merchant 231, 244, 249 merger 40–44, 46–48, 50, 60–62, 65 Merlin 246 metalinguistic negation 34 meter 260, 262, 270, 273 Middle Breton 215 Middle Cornish 215 Middle English Dictionary (MED) 197, 234 Middle English 4, 6–8, 37, 61–62, 65, 67, 87, 91–92, 94–95, 100, 102–103, 107, 111, 164, 166, 177, 184, 187–188,

326

Subject Index

190, 195–197, 202, 206, 213–214, 218–219, 228–229, 231–232, 235, 240–241, 243, 248–249, 258, 261– 262, 266, 271, 279, 283, 288, 291, 311 Middle English 43, 45, 61–62, 65, 67, 87, 91–92, 95, 100, 102–103, 107, 111, 115–116, 119–128, 136, 166, 184, 188, 190, 194, 196–197, 202, 206, 213–214, 219, 228–229, 231– 232, 240–241, 243, 248–249, 258, 261–262, 266, 271–275, 277–283, 286–287, 289, 291 Middle High German 176 Middle Persian 211 Middle Scots 219 Middle Welsh 215 Midlands 276, 278–279 mobility 231 Modern English 3, 4, 7, 37, 62, 87, 89, 92, 95, 102, 105, 108, 163, 169, 183, 188, 190, 198, 206, 218–219, 235, 238, 260, 311, 315 Moderna spra¯k 173 monologicity 2, 15 Moroccan Arabic 211 morphological restriction 115 Morte d’Arthur 242, 246–248, 261 motivation 11–14, 16, 38, 228 multiple analogy 101, 106–108 naming su‰x 117, 119, 127–129, 134, 136 naming 4, 116–117, 119, 121–122, 125, 127–129, 134, 136–138 narrative 45, 202, 303, 307 negation 15, 20, 34, 213 Neogrammarians 166 -ness 117, 127 newfangled words 146–147, 157, 159 N-N compounding 123–125, 137–138 nominal gerund 129 nominalization 4, 112–113, 117–119, 122, 124–128, 133, 136–137 non-linguistic antecedent 48 non-productive a‰xes 115

non-standard dialect 4, 105 Norman French speakers 100 Norse 91, 169 North Germany 209 Northern dialect 8, 219, 279, 282–283, 290 Norway 209 noun phrase 9, 294, 296, 300–301, 303, 306, 309, 312 Null Complement Anaphora (NCA) 40 object compounding 113 Of Arthour & of Merlin 281 Old English 3–5, 17, 28, 37, 38, 60, 62, 65, 67–71, 73, 75–76, 79, 81, 83, 85–89, 91–92, 94–95, 101–102, 105, 107, 111, 133, 166, 187–188, 190– 191, 197–198, 206, 210, 213, 236, 257, 279, 311 Old English 3–5, 17, 28, 38, 42, 60, 62, 65, 67–71, 73, 75–76, 79, 83, 85– 86, 87–89, 91–92, 94–96, 98, 101– 102, 107, 111, 115–116, 119–125, 127–128, 133, 136–137, 166, 187– 188, 190–191, 197–198, 206, 212– 213, 236–237, 257, 271, 279 Old French 236, 257 Old French-English Dictionary 234 Old Icelandic 166, 178 Old Irish 178 Old Norse 98, 169 old 276–279 ongean 3, 67, 70, 75, 77–78, 81 online medium 9, 293–295, 303 oral, orality 228, 233, 242, 258 orthographic practice(s) 265, 267–271, 275–276, 279, 282–283, 289–291 Oxford Dictionary of English Etymology 171 Oxford English Dictionary (OED) 5–6, 34, 47, 94, 107, 162, 167, 169, 172–175, 178, 187, 190–191, 197, 202, 234–237, 257, 295 Oxford Genetic Atlas Project 210

Subject Index participles 89, 92, 95, 100, 105–106, 134, 136, 164, 271 Paston family 231, 243, 248–250 perceptual distinctness 100 periphrastic do 7, 208, 211–221, 226–229 phonology 102, 151, 207–208, 210– 211, 217 Phrygian 161 placename 194 Polish 178 political change 6, 188, 198, 206 Polychronicon Ranulphi Higden 245– 247, 258, 259 polysemy 187, 232 pre-deletion structure 38, 40, 47, 51, 55 preposition stranding 39, 55, 212 preposition 2–3, 9, 39, 42, 50–52, 55, 67–68, 70–71, 73–79, 81–83, 85, 88, 148, 156, 212, 236, 293–296, 299– 302, 308–309 prepositional phrase 2, 67–73, 75–79, 83, 86, 302 prescription 5, 141–154, 156–159 Present-Day English 18, 28, 37, 47– 48, 67, 81, 105, 111–116, 118–119, 121–125, 127, 133–134, 136–138, 213, 272–275, 277–278, 280–282, 286–287 preterite 4, 87, 89, 91–92, 94–96, 98– 102, 105–107, 109 private correspondance 242–243, 246, 258, 261–262, 312 productivity 28, 54, 76, 101, 113, 116, 127–128, 132 progressive 207–208, 212, 226 pronominal deixis 304 proximity 69, 72–76, 79, 86 pseudo-cleft 18, 20, 28 qualitative research 2, 158–159, 232 quantitative research 1, 2, 7, 9, 158, 232, 235, 238, 240, 261 quotative all 304

327

quotative complementizer 9, 293–297, 299, 301–303, 308 quotative go 304 quotative like 303–304 quotative re 9, 293–309, 311–312 quotative verb 302 re 1, 9, 293–311, 313–315 read 99, 100, 102 recategorization 4, 115–117, 119, 122, 127–129, 137 Recency Illusion 2, 60 Redundancy 149–150, 157, 296 regarding 9, 293, 296, 312, 314 Renaissance 231 reply 295–296, 305, 308, 312, 314 res 9, 295 response 296, 312, 314 result nominals 4–5, 117–119, 121– 122, 124–125, 133–134, 136–137 retort 2, 29, 34 rhyme 108, 260, 262, 270, 276, 281– 283 ripa ‘shore’ 236 Romance su‰xes 125, 127–128 romance 189, 242–246, 248–250, 258– 259, 261, 265 root 3, 87, 90–93, 96, 98, 100, 101, 105–106 Russian 178, 211 Sanskrit 177, 178 say 302–303 Scandinavian 162, 175 Scotland 210–211, 268, 282 scribal practice(s) 8, 266–268, 270– 271, 283, 291–292 semantic bleaching 9, 294 semantic change 7, 13, 82, 125, 127, 187–188, 198, 202–204, 206, 234, 257 semantic field 7, 194, 196 semantic structure 68 semantically empty auxiliary 214 semantics of case 69

328

Subject Index

semasiological account 237–239, 241, 248 Semitic 162 shit 90, 94, 101, 106 Sir Tristrem 8 situational antecedent 37 situational context 41, 53 sluicing 2, 37–48, 50–55, 60–63, 65 sneak 92, 107 social change 6, 188, 198, 204, 206, 233 social network 233, 250 sociolinguistic variation 61, 93 sociolinguistics 1, 9 Socrates 161 Somerset 216 southern forms 219, 282–283, 290 Spanish 14–15, 178, 211 spatial meaning 68, 73 speak 34, 93–94, 302, 311 speech act 2, 30–31, 34 speech community/-ies 234, 250 spelling repertoire 270, 275 spit 101, 106, 108 spoken language 9, 41, 61, 232, 258, 271, 299, 304–305, 314 sprouting 41–52, 60–62, 65 S-shaped curve 240 stable variation 3, 61–62 stance 11 standard English 228, 268 step 98, 99, 132 stick 91 Stonor family 243, 248, 258, 262 strike 90–91, 102–103 stripping 53 strong verb inflection 4, 87–103, 105– 109 strumpet 162 stubborn 161–162 subject compounding 4–5, 111–116, 122–125, 128, 133, 136–138 subject line 296, 308–309 substrate 211 superlative 2, 29–30, 34

supralocal change 234 surface anaphor 2, 3, 37, 38, 40, 48, 52, 65 Survey of English Dialects 216 Swedish 166, 187 syntax 1–3, 11, 40–42, 47, 62, 115– 117, 132, 207–208, 210, 219, 312 synthetic compounds 4, 111 talk 302 text type 7–8, 41, 45, 46, 48–50, 60, 227, 232, 233, 241–244, 246, 248, 249, 256, 258, 261–262 that-deletion 62 The American Heritage Dictionary of the English Language 143–144, 174 The Century 162 The New Oxford American Dictionary 175 theory of constrained selection 270, 289 threaded discussions 9, 293, 314 throw 96, 97 to# 3, 67, 69–75, 78–79, 81–82, 85 togean 3, 67, 70, 77–79, 81 token frequency 3, 4, 87–89, 91, 94, 96, 98, 99–102, 105, 165, 235, 240, 244, 245, 248, 261, 262 town/tun 6, 7, 187–199 Transactions of the Philological Society 167 translation 76, 190–191, 194, 197– 198, 240, 242–247, 249, 258–261 transposition 117 travelogue 233, 242, 246, 262, 264 turn-taking 11, 13, 15–16, 34 type frequency 3, 4, 87, 89, 90–92, 94–96, 98–102, 105, 106 Type II (LALME) 8, 268, 269, 275– 279, 283, 289 Type III (LALME) 8, 268, 269, 275– 279, 283, 289 unaccusative verbs 124 under the counter change 234

Subject Index usage dictionaries 145 usage rules 5, 142, 156 usage-based theory 29, 82 Usenet 293, 296, 308 utterance 2, 15, 28–30, 34, 304, 306, 314, variation in form 105, 239 variation in meaning 234, 239, 248 verb complementation 302–303, 312 verb forms 19, 28–29, 129 Verb Phrase Ellipsis (VPE) 48 videlicet 315 village 190–191, 195 vowel alternation 113 vowel distinctness 92, 98–100 wake 93–95 Wales 210, 213, 216

329

wear 93 weave 93, 98 Welsh 212, 215–216, 220 West Germanic 175 WH-cleft 2, 17 wh-movement 39 wife 170, 175 Wiltshire 216 wiþ 3, 67, 70, 73–79, 81, 85 word formation 28–29, 111, 113, 238, 257 world 269, 277–279 wreak 92–94 write 302 Year’s Work 163 yeoman 176 zero forms 28, 62, 119