Studies in the history of the english language

The millennial focus is fully apparent in the envoy paper by Richard Bailey University of Michigan; and in thefield-survey papers by Elizabeth Traugott Stanford University; HerbertSchend

≥

Topics in English Linguistics 39

Studies in the History

of the English Language

Mouton de Gruyter (formerly Mouton, The Hague)

is a Division of Walter de Gruyter GmbH & Co KG, Berlin.

앪

앝 Printed on acid-free paper which falls within the guidelines

of the ANSI to ensure permanence and durability.

Library of Congress Cataloging-in-Publication-Data

Studies in the history of the English language : a millennial

perspective / edited by Donka Minkova, Robert Stockwell.

p cm ⫺ (Topics in English linguistics ; 39)

Includes bibliographical references and index.

ISBN 3 11 017368 9

1 English language ⫺ History 2 English language ⫺

Grammar, Historical I Minkova, Donka, 1944 ⫺

II Stockwell, Robert P III Series.

PE1075 S88 2002

420 1.9⫺dc21

2002067795

Die Deutsche Bibliothek ⫺ Cataloging-in-Publication-Data

Studies in the history of the English language : a millennial

perspective / ed by Donka Minkova ; Robert Stockwell ⫺

Ber-lin ; New York : Mouton de Gruyter, 2002

(Topics in English linguistics ; 39)

ISBN 3-11-017368-9

” Copyright 2002 by Walter de Gruyter GmbH & Co KG, 10785 Berlin

All rights reserved, including those of translation into foreign languages No part of this book may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher.

Cover design: Christopher Schneider, Berlin.

Printed in Germany.

Foreword 1

Millennial perspectives

From etymology to historical pragmatics

Mixed-language texts as data and evidence in English historical

Chaucer: Folk poet or littérateur?

A rejoinder to Youmans and Li

Phonology and metrics

On the development of English r

Word order in Old English prose and poetry:

The position of finite verb and adverbs

The “have” perfect in Old English:

How close was it to the Modern English perfect?

This collection contains papers selected from those presented at a ence at UCLA in the Spring of the year 2000 The conference was calledStudies in the History of the English Language, abbreviatedSHEL-1 It wasintended to be the first in what we hope will become a regular biennialseries at various sites in North America;SHEL-2 is being organized at theUniversity of Washington as we are preparing this volume The intention ofthe series is to stimulate research and other scholarly activity in the field ofhistorical English linguistics Our emphasis was deliberately on the history

confer-of English as a discipline: how healthy was it at the end confer-of the millenniumand what if anything needed to be done to maintain its scholarly energy andrelevance? A comparable series of meetings in Europe, the InternationalConferences on English Historical Linguistics (ICEHL), have proved to bevery vigorous since the first conference, in 1979, at Durham

The scholars who work in our field in non-English speaking countrieshave the advantage that English is a popular choice for foreign languagestudy there and often there is generous government and public support forthis kind of enterprise They have a steady student demand based on that fact

In Britain, the 19th and 20th century dedication to historical English studiescreated a robust research tradition and a new generation of outstandingscholars in whose hands the field is thriving The tradition flourished in Brit-ish and American universities in the 19th century and during the first half ofthe 20th century, but during the last two decades only in British universitieshas this kind of scholarship remained in the mainstream of academic life

It has been apparent for some years that in America comparable vigordid not exist in this field In fact, the field has been declining as scholars un-familiar with the linguistic history of English themselves fail to see the rel-evance of a traditional subject in a newly fashioned humanistic curriculum

We believe this to be a misapprehension of the field and its present-day volvement with other disciplines Researching the cognitive and social con-ditions, causes, and consequences of language change should not bebrushed aside as peripheral to the concerns of the new generation of hu-manists in this country We organized the firstSHEL with the conviction thatthe study of the history of English is central to the interpretation of our cul-tural and literary heritage, and both the meeting, and the contributions tothis volume, reaffirmed our conviction

in-2 Foreword

So the first motivation for us to convene this conference was to provide aforum for the presentation of research in English linguistics, specificallyEnglish historical linguistics, to serve as a stimulus for quality and a probefor what more needs to be done Measured by the quality and variety of thepresentations and discussions, 35 speakers and over 80 participants, wehappily report that energy and vigor continue to characterize our field AWorkshop, organized by Anne Curzan (University of Washington), ad-dressed various approaches to the teaching of the History of English TheNorth American representation atSHEL-1 was greatly augmented by trans-atlantic scholars whose participation was deeply appreciated For the rec-ord, a list of paper presenters and session chairs appears at the end of theForeword

The second motivation was to provide, in as many areas as possible, a

sort of millennial stock-taking The millennial focus is fully apparent in

the envoy paper by Richard Bailey (University of Michigan); and in thefield-survey papers by Elizabeth Traugott (Stanford University); HerbertSchendl (University of Vienna); William Kretzschmar (University of Geor-gia); Anatoly Liberman (University of Minnesota); Thomas Cable (Univer-sity of Texas, Austin); and Gilbert Youmans and Xingzhong Li (University

of Missouri and Central Washington University)

The papers addressing individual issues fall into two natural groups:

(1) Phonology and metrics – the papers by Blaine Erickson (Kanazawa

Institute of Technology, Japan); Kristin Hanson (University of California,Berkeley); Betty Phillips (Indiana State University, Terre Haute); GeoffreyRussom (Brown University); Robert Stockwell (University of California,Los Angeles); and David White (University of Texas, Austin)

(2) Morphosyntax/Semantics – the papers by Maurizio Gotti (University

of Bergamo); Edward Keenan (University of California, Los Angeles); Ansvan Kemenade (Katholieke Universiteit, Nijmegen); Jeong-Hoon Lee(University of Texas, Austin); Colette Moore (University of Michigan);and Benji Wald and Lawrence Besserman (Los Angeles and Hebrew Uni-versity, Jerusalem)

Millennial Perspective

Elizabeth Traugott’s contribution is entitled “From etymology to historicalpragmatics” As is well known, she was one of the principal scholars in the

initiation of research into the historical process of grammaticalization, by

means of which new grammatical formatives come into languages In troducing her survey of the field, she notes that many of the major themes

in-of the early twentieth century, e.g the arbitrariness in-of the sign, the dictability of change, and distinctions between structure and use, were de-veloped against the background of historical work primarily on Romanceand Germanic languages Among many themes debated at the end of thecentury have been the extent to which change is predictable and non-arbit-rary, this time against the background of synchronic work primarily onEnglish, as well as many other languages

non-pre-She points out that throughout the twentieth century diachronic theory haslagged behind synchronic Though the seeds of many ideas about morpho-syntactic change that dominated the end of the century are to be found inwork at its beginning, the pragmatics, semantics and syntax were not suffi-ciently far advanced for those ideas to be developed in principled ways Newpossibilities for understanding language change have opened up with ad-vances in the study of the relation between language and use, especially fromthe perspective of work on grammaticalization and historical pragmatics.One area of intense debate arising from these new studies is what status

in linguistic theory the widely attested unidirectionalities in semantic andmorphosyntactic change should have On the one hand there has been anactive research program seeking to identify unidirectionalities (e.g Trau-gott 1982, 1989, Heine, Claudi and Hünnemeyer 1991, Bybee, Pagliucaand Perkins 1994); this has recently been complemented by a research pro-gram seeking to formalize them (e.g Roberts 1993, Kemenade 1999).1 Onthe other it has been argued that since unidirectionalities are tendencies, notabsolute universals, they are epiphenomena and not explanatory (e.g Ro-berts 1993, Newmeyer 1998 on the unidirectionalities identified in gram-maticalization), and even that such searches are a hold-over from nine-teenth century historicism (Lightfoot 1999)

Her paper suggests some ways in which historical pragmatics can shedlight on the unidirectionalities observed in grammaticalization, as the fieldmoves toward an explanatory theory that can account for them

lin-William A Kretzschmar, Jr.’s paper is entitled “Dialectology and thehistory of the English language” In it he argues persuasively that dialectol-ogy has much to offer to historical linguistics and to the history of the Eng-lish language in particular Survey research is the hallmark of the field, inline with the empirical emphasis of the Neogrammarians who were theinitiators of this approach, yet dialectologists have been among the stron-gest critics of the central Neogrammarian position on mechanical sound

change Given this state of affairs, the central role of dialectology is to scribe, for the particular time of a survey, the facts of the complicated lan-guage behavior in the survey area These facts make up a body of raw ma-terial for historical linguistics to use, first as studies of individual words,and second for description of the situation of language variation todaywhich can enlighten our attempts at historical reconstruction Kretzschmarmaintains that contemporary variation is indeed the trace of change for par-ticular words, because the low-frequency responses often reflect historicalprecedents The asymptotic curve (A-curve), the pattern of few commonlinguistic types and many infrequent types which emerges from survey re-search on a single linguistic target, appears to be a basic fact about the dis-tribution of linguistic types and tokens The A-curve gives historical lin-guistics a more realistic model than the all-or-nothing pattern predicted bymechanical sound change The A-curve and the familiar S-curve are thuscomplementary quantitative descriptions of the distributional facts of vari-ant linguistic forms upon which we can develop improved “realism” in ac-counts of linguistic change.

de-Anatoly Liberman’s brilliantly entertaining, yet deeply scholarly essay,

is entitled “Etymology unknown” The author is engaged in the production

of a new etymological dictionary of English that focuses entirely on themysteries of English etymology – the words for which no adequate etymol-

ogy exists in any of the standard works like the Oxford English Dictionary.

He notes that many words are dismissed in etymological dictionaries withthe verdict “origin unknown.” In English, some such words go back toCommon Germanic and even Indo-European, while others are more recent.Sometimes, “origin unknown” means that there is no consensus on aword’s prehistory, but equally often scholars indeed have no clue to thesources and the semantic motivation of words current today Relativelyeasy to etymologize are only onomatopoeias, words from names, somesimplified compounds, and borrowings (if an agreement is reached not topursue the origin of foreign words in the lending languages) In other cases,historical linguistics usually fails to achieve the avowed goal of etymologi-cal research and present an unmotivated sign as motivated The most urgenttask of English etymology as a science, Liberman believes, is to produce adictionary comparable to those by Feist, Vasmer, Frisk, and others with afull critical survey of the existing literature on every word If the idea ofsuch a dictionary materializes, many words of allegedly unknown etymol-

6 Foreword

ogy will be upgraded to words of uncertain or even acceptable etymology,and scholars will be able to pick up where their numerous forgotten, disre-garded, and misunderstood predecessors left off Anatoly Liberman hasbeen working on “an English Feist” for thirteen years, and his suggestionsare the result of this work

Thomas Cable’s paper, “Issues for a new history of English prosody”,proceeds from the observation that an adequate new history of English pros-ody must resolve the main incompatibilities of three rich traditions of pros-odic study: temporal, accentual, and generative He argues that the prob-lems and possible solutions can best be illustrated by focusing on gener-ative metrics (a particular development of the stress tradition) because ofthe explicitness of its goals and rules, a virtue often missing in the other twoapproaches In his critique of the generative tradition, he assigns special rel-evance to three contexts: (1) the diagnostic relevance of the W, or Weak,position, (2) the lack of attention to the metrical pause, and (3) the mislead-ing attention to the caesura He examines and seeks to correct generativeanalyses of the iambic pentameter as used by Chaucer and Shakespeare, and

by the poets between them, in a way that he believes clarifies this pivotalperiod in the history of English prosody He concludes that there is a sig-nificant difference in the internal structure of the line in Chaucer and Shake-speare, and that the most succinct formulation of this difference is to saythat Shakespeare made use of the iambic foot and Chaucer did not.Cable’s original and controversial position on the difference between theChaucerian and the Shakesperean iambic line is disputed in a counterpointpaper entitled “Chaucer: Folk Poet or Littérateur?” by Gilbert Youmans andhis student Xingzhong Li Youmans and Li were not participants in the con-ference but became involved in the exchange at our request Thus we have acomparison of these two views on how the prosody of iambic pentametercan be best described, Cable’s own view and a version of the generativeview represented, in this instance, by Youmans and Li The differences arevast and extremely interesting, and having them aired this way we believe isgreatly to the benefit of us all In a framework consistent with OptimalityTheory, Youmans and Li propose a central prototype for iambic pentameterverse and describe gradient “tension” rules (violable constraints) formeasuring degrees of deviation from this prototype They provide statisticalevidence in support of the traditional view that Chaucer’s decasyllabic verse

is iambic pentameter [WS][WS] rather than an unfooted sequence of

alter-nating syllablesWSWS Chaucer’s verse, they acknowledge, is more regularthan Shakepseare’s and Milton’s, but all three poets share the same centralprototype, a 4//2/4 line structured [WS][WS]//[WS]/[WS][WS]; and vari-ations from prototypical stress patterning are most common at the begin-nings of their lines (and hemistichs) and least common at the ends of theirlines (and hemistichs) Thus, Chaucer’s verse differs stylistically fromShakespeare’s and Milton’s, but it embodies strikingly similar metricalprinciples.

Cable’s response summarizes the differences between his theory of tameter and that of the generativists, as represented by Youmans and Li,under six categories (1) A rigorous insistence on finding five beats to theline – no reading without five beats is acceptable, some beats no doubtweaker than others but still stronger than the adjacent “weak” beats.(2) “Tilting” some weaker syllables toward strong, and some stronger syl-lables toward weak, is both necessary and proper (3) Non-lexical words,i.e function words, not just lexical words, can bear ictus under the rightcontextual circumstances (4) Distinctions between unstressed syllablesthat are cliticized as function words, and lexically unstressed syllables, areirrelevant for the purposes of assessing poetic rhythm (5) Constituentbracketing, except with respect to caesura marking, is irrelevant to the de-termination of metricality (6) Metricality is not the same as tension: What

pen-is relevant to measuring tension may not be relevant to assessing ity What may reasonably be considered a seventh category, though not sonumbered by Cable, is his insightful discussion of trisyllabic feet inChaucer, in particular pointing out that there are no trochaic reversals inwhich the reversal depends on a weak position two syllable consisting only

metrical-of a final -e Though there are epiphenomenal reasons which might explain this gap, the final -e in position two would be an incontrovertible and un-

tiltable trochaic reversal: Cable argues that this gap requires explanation

As editors we would not be willing to take sides and say who won thisexchange, even if we had a firm clear answer (which we don’t) We foundthe exchange extremely enlightening, and we believe our readers will also

We are deeply grateful to all three scholars for having the integrity andcourtesy to present these vigorously conflicting views of the nature of theiambic pentameter as it is (mis)understood at the turn of the century.Finally, among the millennially-oriented papers, Richard W Bailey’spaper, printed at the end of the book because of its envoy tone, “A thousand

8 Foreword

years of the history of English”, is a swashbuckling tour through Englishphilology of the 19th and 20th centuries (“text-centered study of lan-guage”, in Bailey’s words) in support of the view that we need a “renewedphilology” The paper is a vigorous reminder that descriptions of linguisticfacts are not by any means free from assumptions, and that our views of thehistory of English often express biases that are only dimly apparent to us.Bailey compares our own histories of English with those of our prede-cessors: members of the Philological Society squabbling over standards asits New English Dictionary began to appear in 1884; Thomas Percy, theeighteenth-century clergyman who nearly single-handedly invented a dis-tinctive Middle English and demonstrated how he could revive feudalismand place himself within it as a cosseted minstrel even in the midst of theAge of Reason; John Free, writing the first book-length history of English,who managed to revive pride in Germanic languages by demonstrating thatEnglish was one of them; Byrhtferth, in the twelfth-century, who used Eng-lish to demonstrate revealed truth through words His argument concludeswith the claim that we make better histories if we understand, and makepublic, our assumptions, an appropriate appeal to the scholars of the twen-ty-first century

Phonology and Metrics

Blaine Erickson’s paper on “The history of English r” is not a full-scale

his-tory of all varieties of /r/ found in the English language so much as it is aneffort to explain how the phonetic properties of the common AmericanEnglish /r/ came about This topic deserves such special treatment because

it is the odd articulation of /r/ that is the most strikingly different feature ofAmerican as compared with any other variety of English

In American English, retroflex and central approximant articulations are

the most common phones found for /r/ Cross-linguistically, both sounds

are rare, and since they are almost unknown in other Germanic languages,

the r phones of American English are more likely to be innovations than

re-tentions

Many other Germanic languages have a coronal trill as their phone for/r/; evidence suggests that this was also the original phone for English /r/.Erickson assumes that during the development of pre-modern English, the

place of articulation of the coronal consonants changed from dental to veolar This caused the tongue to take on a sulcal shape, which set the stagefor the change from trill to retroflex Retroflex articulation occurred first as

al-a conditioned chal-ange: when /r/ al-appeal-ared before other coronal-als Lal-ater, flex articulation spread to other environments, and it supplanted the trill.After that, the central approximant appeared as an alternate pronunciationfor /r/ This is motivated by the fact that the retroflex and central approxim-ant articulations are nearly identical acoustically, even though the tongueconfigurations for the two are opposite: tongue tip up for the former, anddown for latter

retro-Kristin Hanson’s paper, entitled “Vowel variation in English fullrhymes”, is an effort to account for imperfect rhymes in the history of Eng-lish verse Virtually all modern English poets include some rhymes in theirpractice in which the vowels differ in present day English, such as Shake-

speare’s famous love/prove rhymes in the Sonnets Versification handbooks

and history of English textbooks tend to suggest that English soundchanges explain such rhymes, in that the vowels would have been identical

in the poets’ own dialects But philological studies paint a more complexpicture: Kökeritz (1953), for example, concludes that many of Shake-speare’s rhymes would have involved genuine differences in the vowels,and would have counted as rhymes only in virtue of literary precedents, thepresence of dialect variants in the linguistic milieu, or mere phonetic simi-larity Moreover, such differences in rhyme vowels often seem aestheticallysignificant These considerations invite the question of whether for a givenpoet’s rhyming practice can be formally defined in such a way as to accom-modate expressive variation while still distinguishing possible from im-

possible variations For Shakespeare’s Sonnets at least, the explanations

Kökeritz offers are subtly problematic for this purpose, in that they invoke aconception of language which is social and imitative rather than individualand creative A closely related alternative which would reformulate the role

of history in shaping these practices and locate them within a unified mar rests on the observation that vowels which are paired in such rhymes

gram-are often those which participate in alternations such as reduce/reduction.

Whatever the value of this observation for Shakespeare’s rhymes, it is clearthat the general question of what principles govern the vowel variation inEnglish poets’ full rhymes, and how these are transmitted and changedfrom poet to poet, constitute significant open research questions

10 Foreword

Betty S Phillips’ paper, entitled “Lexical diffusion and competing lyses of sound change”, seeks to show that the current theory of lexical dif-fusion can be used to help choose between competing analyses of a givensound change, specifically the shift of Early Modern English /u:/ to /U/ insuch words asGOOD andSTOOD Ogura (1987) hypothesized that the im-petus behind this change was an attempt to keep the duration of syllabic unitsrelatively constant – following a suggestion by Dobson (1968) that the earlystages of the change complemented lengthening in an open syllable Görlach(1991: 71) suggested that “the short vowel in words likeGOOD andSTOOD

ana-was introduced on the pattern of words in which the occurrence of a short or along vowel was determined by the type of syllable the vowel appeared in(GLAD vs.GLADE).” Yet since both explanations require lexical analysis, asdefined in Phillips (1999), one would expect the least frequent words to havebeen affected first Ogura (1987), however, finds that the change in questionaffected the most frequent words first For that reason, Phillips offers an al-ternative, phonetically motivated explanation for this sound change.Geoffrey Russom’s paper “Purely metrical criteria for dating Old Eng-lish poems”, seeks to provide a new set of dating criteria for early Englishtexts, supplementing standard philological techniques, because improveddating would greatly facilitate research on English linguistic history Heproposes a method for dating Old English poems by reference to frequen-cies of verse types Several strategies are employed to deal with problemsarising in comparable studies First, verses are selected for analysis in away that avoids bias toward the native heroic sphere, which can affect fre-quencies for many variants of types C, D, and E Further, nonstandard fre-quencies due to limited poetic ability are factored out by reference to expertjudgments based in part on the number of verses for which normal scansion

is impossible; and the expert judgments are then cross-checked against quencies of type A3, the verse pattern that would be easiest for a poet of li-mited talent to construct Appropriate relative frequencies are predicted byfundamental principles of the word-foot theory (Russom 1987, 1998),which yield a linguistically-based metrical norm and ways to computedeviation from the norm Deviation from the norm restricts frequency, asexpected; and the decline of the metrical tradition is accompanied by majorshifts in the placement as well as the frequency of deviant verses The re-sults of this study confirm, in striking detail, the results achieved by Fulk(1992), who employs entirely independent dating criteria

fre-Robert Stockwell’s paper entitled “Short shrift for the ‘great’ vowelshift” is an effort to provide an alternative and substantially simplified view

of what might have happened to produce the appearance of vowel shiftingwithout such a complex set of events having actually occurred The EnglishVowel Shift is traditionally thought of as a chain in which the next vowel upthe chain is displaced, in a continuous cycle Stockwell argues that this pic-ture is wrong in two ways: (1) Most stages of the English shift are notchained but actually result in mergers The targets of these mergers arediphthongs, either inherited (like MAIDEN or GROW) or borrowed (like

DAINTY orPRAY from French, orRAISE from Scandinavian) (2) gization of [i:] and [u:] toward [ai] and [au] is not a chain at all, since no-thing is ever displaced: all members of these parts of the shift coexist to thisday (e.g., the misnomer “Canadian raising”) and must be considered allo-phones or diaphones (Kurath’s rarely used but highly appropriate coinage).Combining (1) and (2), Stockwell gets a very different view of the vowelshift, reducing the “shift” to the raising of [e:] and [o:] to [i:] and [u:] (FEEL

Diphthon-andFOOT) and the initial diphthongization, though not the continuing phonic development, noted in (2) This difference of interpretation raisesdifficult problems for the view (e.g Lass 1999) that assumes leveling of the

dia-RAISE vowel to a long monophthong, and similarly theGROW vowel, withthe general diphthongal quality found over the past three centuries taken asinnovative

David White’s paper deals with a highly technical philological questionabout certain sound changes in Old English The paper is entitled “Resto-ration of /a/ revisited” According to the traditional historical phonology ofOld English, all instances of the phoneme /a/ are supposed to have changed

to /æ/, with some /æ/’s later changing back again into /a/ in certain ments where /a/ later appears This last change is known as “restoration of/a/” That it is inelegant is obvious, and we should only believe it for a verygood reason The traditional reason, for /h/ to have acted as a blocker ofback-assimilation, as is implicitly suggested by Moore and Knott (1942), isphonetically implausible But it is possible that the relevant phenomenonwas not blocking but dissimilation /h/ is such a weak consonant that /aho/might as well be /ao/ If stressed back vowels were dissimilated before fol-lowing unstressed back vowels where no “buccal” consonant intervened,then /aho/ would develop into /æho/, from which later /æo/ (or whatever wehold long “ea” to have been) is not problematic Under such a scenario, dis-

words always had /a/.

Morphosyntax and Semantics

Maurizio Gotti’s paper on “Pragmatic uses of shall future constructions in

Early Modern English” carries us forward in the philological tradition ferred to in Richard Bailey’s paper – the “text-centered study of language”.The object of the paper is the analysis of the uses of the future tense with

re-shall in an Early Modern English corpus; the texts examined are those

in-cluded in the third section of the Early Modern English part of The Helsinki

Corpus of English Texts The results of this analysis are compared to the

pragmatic values pointed out in a few grammars of the same period in order

to assess their degree of faithfulness to the real usages in contemporarytexts The period taken into consideration is 1640–1710 and includes thegrammars which first describe the pragmatic uses of this modal auxiliaryfor the formation of the future tense

The analysis of the shall-forms contained in the corpus first identifies all

the quotations referring to futurity and then subdivides them into semantic/pragmatic categories This is done in order to confirm the basic differenti-ation between pragmatic and dynamic uses of this modal suggested by the17th-century grammarians’ adoption of categories labelled “prediction”,

“promising”, “threatening”, “declaration”, “command” The survey alsopoints out the qualitative difference between the deontic and dynamic as-

pects of the shall-forms found in the corpus.

Another aspect which is investigated is the degree of correlation betweenthese modality categories and the text types contained in the corpus; the re-sults indicate that some genres show a very high correlation rate, whileothers are characterised by a much wider range of modal categories.Edward L Keenan, a distinguished logician and semanticist, has re-cently gotten deeply involved in the explanatory aspects of historical phil-ology In this paper he takes on the subject of reflexive pronouns, under thetitle “Explaining the creation of reflexive pronouns in English” He pro-

vides an account for the creation and interpretation of the English reflexives

(himself, herself, …) in terms of two general forces of change, two

univer-sal semantic constraints on language, and the start state of the Old Englishanaphora system His account is based on the study of some 11,000 in-stances of locally bound objects of verbs and prepositions, drawn from over

100 texts dating from c750 to c1750, complementing Keenan (2000) which

is a more extensive presentation of these data than we could accommodatehere He contrasts his account with two other accounts of the creation of

grammatical formatives: grammaticalization and parameter resetting The

creation of reflexives in English falls into neither category though it sharesone feature with grammaticalization A comprehensive presentation in-

cluding a list of Source Materials is to be found in Keenan (2001).

Ans van Kemenade initiated very important work on clitic pronouns inOld English near the end of the 1980’s Her contribution here, “The wordorder of Old English poetry compared with prose once more”, deals withsome non-pronominal aspects of clitic word order in Old English, and hassignificant bearing on the issue of dating the earliest documents The paperhighlights a difference between verse and prose word order in negative rootclauses: in particular, the existence of one main clause word order patternthat is unique to the early poetry, namely negative-initial clauses in whichthe finite verb does not appear adjacent to the negative word Her study isboth a contribution to the description and analysis of the earliest syntacticpatterns in English and, by implication, a syntactician’s confirmation of thedating of poetic texts The same theme, dating the poetic corpus, is address-

ed from a different perspective in this volume by Geoffrey Russom Theword order of Old English poetry has given rise to some considerable de-bate: some hold that the principles governing it represent a genuine depar-ture from those ruling the prose; others argue that it is governed by the sameprinciples as the prose, albeit at differential frequencies dictated, for in-stance, by metrical considerations The special aspect of the word orderthat Kemenade focuses on here is that a negative element is the first consti-tuent, but the finite verb is not attracted to it, as it is in later Old English.The diachronic picture for the history of English suggests that this pattern is

an older one This in turn supports the familiar view that the extant poetry

is, in some respects at least, older than the extant prose

Jeong-Hoon Lee deals with the semantics of Perfect tense in Old English

in his paper “The “have” perfect in Old English: How close was it to the

14 Foreword

Modern English perfect?” He examines the semantics of the have perfect

tense in Old English and the perfect tense in Modern English Lee arguesthat there is no significant semantic difference between these perfect tenses,

contrary to the accepted view that the have perfect tense in Old English had

at best only the so-called “resultative” meaning Based on the three ings of the Modern English perfect, which represents three subcategories byperfect tense forms, i.e., the existential, universal, and resultative meanings,

mean-he shows that, in addition to Old English have perfect constructions with tmean-he resultative meaning, there were also many Old English have perfect con-

structions with either an existential or a universal meaning He also argues

that there were many Old English have perfect constructions that could not have been interpreted as the resultative perfect, namely, have perfect con- structions modified by manner adverbials, and have perfect constructions in negative sentences Additionally, he suggests that many Old English have past perfect constructions had the same ambiguity as that of the Modern

English past perfect; that is, it was not clear whether they had the “perfect”sense, or they expressed just an event/situation before a past time

Colette Moore’s paper is entitled “Reporting direct speech in Early ern slander depositions” Determining how speakers used language in theEarly Modern period presents difficulties due to the deficiencies of collo-quial source material Court records, with their recording of witness testi-mony, provide one source of “speech-based” texts, and they are being morewidely considered in this light after their inclusion in such computer cor-pora as the Helsinki corpus Without dismissing the value of depositions as

Mod-a potentiMod-al source for spoken English, we must Mod-also be Mod-awMod-are thMod-at their fulness comes within a certain context She investigates the context of theserecords and examines the depiction of direct speech in a sample of them.Defamation depositions reveal both code- and style-switching, and re-ported speech plays a central role in the records since it presents the allegedcriminal action She analyses the treatment of direct and indirect speech inthe defamation court records in an attempt to understand the switching be-tween spoken and written language in Early Modern English and the con-ventions that surround the reporting of speech within a written form This,

use-in turn, allows us, she believes, to perceive better the nature of the luse-inguisticconstruction of these depositions and its relation to spoken English, so that

we may make more considered use of court records on their own terms andwithin computer corpora

Benji Wald and Lawrence Bessermann’s paper is entitled “The verb-verbcompound in twentieth century English and twentieth century linguistics”.

It deals with the emergence of the endocentric VV (verb-verb) compound

as a unified productive pattern This pattern poses an interesting set of lems and challenges for twentieth century historical and synchronic lin-guistics Although VV seems to have eluded discussion and even exemp-lification in most major treatments of English throughout the twentieth cen-tury, such studies anticipated a range of relevant historical and synchronicproblems in the analysis of the type The principal problem that they con-sider in detail is that VV compounds frequently seem ambiguous as be-

prob-tween VV and NV analyses How can we tell whether sleep in sleep-walk is

N or V? They refer to this analytical problem as the ambiguous categoryproblem and offer a historical solution to it

After discussing the analytical issues, and recent innovations from whichthey have arisen, including a general constraint on verb compoundingwhich they label “the activity constraint”, they address the longer termquestion: why was the twentieth century a critical period in the develop-ment of the VV compound? Their answer is that the development of VVhas depended on a number of preconditions that have successively accumu-lated since the Old English period They briefly consider the historical evol-ution of VV, first through the development of its preconditions in late OldEnglish and Early Middle English, and then through a quantitative study of

VV and related items listed in the second edition of the Oxford English

Dic-tionary and its later additional supplements.

Acknowledgements

The idea of convening SHEL -1 could not have materialized without the financial backing

of the Center for Medieval and Renaissance Studies, the Department of English, and the Department of Linguistics at UCLA We thank the leaders of these units for their sup- port.

For encouragement and advice in preparing this volume we are grateful to the TiEL series editors, Elizabeth Traugott and Bernd Kortmann, and to Anke Beck, Birgit Siev- ert, and Wolfgang Konwitschny at Mouton de Gruyter.

Above all, we extend our thanks to our authors who were prompt in submitting, tient in awaiting the comments on their papers, and conscientious in revising They have been a terrific group to work with.

pa-16 Foreword

Each paper in this volume was reviewed anonymously by two outside reviewers, and many of the submissions underwent revision several times It is with very special gratitude that we list the names of our colleagues who took time away from their own pursuits to advise us and the authors, often writing critiques that could easily have be- come independent contributions It is a heartening comment on the scholarly dedi- cation to the field that so many people were ready to render first rate professional ser- vice with paper bags over their heads Disclosing their names is the least we can do to acknowledge their help We hereby express our very special thanks to them, with the important further acknowledgement that their criticism was never ignored, though it was not always accepted with agreement either by the authors or the editors, no doubt

to our eventual discomfort, and certainly no one but the authors and editors can be blamed for errors that remain.

Leslie Arnovick, University of British Columbia, Vancouver, Canada

Laurel Brinton, University of British Columbia, Vancouver, Canada

Fran Colman, University of Edinburgh, Scotland

Claire Cowie, University of Sheffield, England

Christanne Dalton-Puffer, University of Vienna, Austria

Hans Jürgen Diller, University of Bochum, Germany

Martin Duffell, Queen Mary and Westfield College, London, England

Nigel Fabb, University of Strathclyde, Scotland

Susan Fitzmaurice, Northern Arizona University, Flagstaff

Rob Fulk, Indiana University, Bloomington

Elly van Gelderen, Arizona State University, Tempe

Michael Getty, University of Toronto, Canada

Robert Howell, University of Wisconsin, Madison

Andreas Jucker, University of Giessen, Germany

Juhani Klemola, University of Helsinki, Finland

Willem Koopman, University of Amsterdam, Holland

Barbara Kryk, University of Poznan, Poland

Merja Kytö, Uppsala University, Sweden

Bettelou Los, Vrije Universiteit, Amsterdam, Holland

Angelika Lutz, University of Erlangen, Germany

David Matthews, University of Newcastle, New South Wales, Australia

Chris McCully University of Manchester, England

Frances McSparran, University of Michigan, Ann Arbor

Michael Montgomery: South Carolina, Columbia, S.C.

Robert Murray, University of Calgary, Canada

Mieko Ogura, Tsurumi University, Japan

Päivi Pahta University of Helsinki, Finland

Theo Vennemann, University of Munich, Germany

Katie Wales, University of Leeds, England

Tony Warner, University of York, England

Gilbert Youmans, University of Missouri, Columbia, MO

Presenters and Session Chairs at SHEL-1 (UCLA, May 2000)

Noriko Akatsuka, UCLA

Henning Andersen, UCLA

Richard Bailey, University of Michigan, Ann Arbor

Lawrence Besserman, Hebrew University, Jerusalem

Mary Blockley, University of Texas, Austin

Laurel Brinton, University of British Columbia

Thomas Cable, University of Texas at Austin

Ruth Carroll, University of Turku, Finland

Don W Chapman, Brigham Young University

Anne Curzan, University of Washington, Seattle

Edwin Duncan, Towson University

Nancy Elliott, Southern Oregon University

Blaine Erickson, Kanazawa Institute of Technology, Japan

Susan Fitzmaurice, Northern Arizona University

Robert Fulk, Indiana University

Elly van Gelderen, Arizona State University

Michael Getty, University of Toronto

Gwang-Yoon Goh, Ohio State University

Chris Golston, Fresno State University

Maurizio Gotti, University of Bergamo, Italy

Kristin Hanson, UC Berkeley

Yukio Haraguchi, Kumamoto Gakuen University, Japan

Bruce Hayes, UCLA

Richard Hogg, University of Manchester, England

Richard Janda, Ohio State University

Dieter Kastovsky, University of Vienna, Austria

Edward L Keenan, UCLA

Henry Ansgar Kelly, UCLA

Ans van Kemenade, Katholieke Universiteit, Nijmegen, Holland

Willem Koopman, University of Amsterdam, Holland

18 Foreword

William A Kretzschmar, Jr., University of Georgia, Athens

Anthony Kroch, University of Pennsylvania

Barbara Kryk, University of Poznan, Poland

Ian Lancashire, University of Toronto

Eva Delgado Lavin, University of the Basque Country, Spain

Jeong-Hoon Lee, University of Texas, Austin

Anatoly Liberman, University of Minnesota, Minneapolis

Hans Lindquist, Växjö University, Sweden

Soon-Ai Low, University of Maryland

Joseph P McGowan, University of San Diego

Frances McSparran, University of Michigan

Anna Meskhi, Isik University, Turkey

Donka Minkova, UCLA

Colette Moore, University of Michigan

Betty Phillips, Indiana State University Terre Haute

Thomas Riad, Stockholm University, Sweden

Philip Rusche, University of Nevada, Las Vegas

Geoffrey Russom, Brown University

Herbert Schendl, University of Vienna, Austria

Catherine Smith, Northern Arizona University

William Spruiell, Central Michigan University

Robert Stockwell, UCLA

John Sundquist, Indiana University

Akinobu Tani, Hyogo University, Japan

Elizabeth Traugott, Stanford University

Geoffrey Russom, Brown University

Benji Wald, Los Angeles

David L White, University of Texas, Austin

Donka Minkova and Robert Stockwell

Los Angeles, March 2002

Note

1 For the bibliographical details on all short-form references embedded in the

sum-maries in our Foreword, please see the reference lists attached to the appropriate

paper.

Elizabeth Closs Traugott

0 Introduction1

One of the much-discussed themes of linguistic work in the last hundredyears has been how to think about the relationships between synchrony anddiachrony, between structure and use, between arbitrariness and motivated-ness, between what Roberts has called “a random ‘walk’ through the spacedefined by the set of possible parameter values” (Roberts 1993: 252) andobserved directionalities of change

Throughout the twentieth century diachronic theory lagged behind chronic While rapid advances have in recent years been made in the study

syn-of historical morphosyntax, no aspect syn-of language change has lagged hind synchronic theory more than pragmatic-semantic change Though theseeds of many ideas that dominated the end of the century are to be found inwork at its beginning, neither the semantics nor the pragmatics were suffi-ciently far advanced for those ideas to be developed in principled ways Re-cently, however, there has been the possibility of catch-up, and in thatcatch-up it has been the functionalist, pragmatic aspects of linguistics thathave carved out a path that in diverse ways link up with older concerns.What began as work on the semantic correlates of grammaticalization,often using texts in ways that privilege discourse and genre, has amongsome practitioners now come to be known as “historical pragmatics” (seeJucker 1995) There are many paths that one could follow with such a topic,but I will highlight work on certain aspects of the regularities and direc-tionalities in change that have been explored, with particular attention towork on the interplay between language and use (see most recently Croft2000), as it is conceptualized in the study of grammaticalization and es-pecially historical semantics and pragmatics

be-In recent years we have heard that the search for tendencies and tionalities is a hold-over from nineteenth century historicism, that change is

direc-an epiphenomenon, even that history itself is direc-an epiphenomenon (Lightfoot1999: 261) In other words, it falls out from things other than itself, from

20 Elizabeth Closs Traugott

what people do There is no doubt that language change occurs because offorces outside of language, most particularly processes of production andinterpretation If language acquisition is a prime reason for change, if com-peting motivations such as “be clear and processable” vs “be quick andeasy” (Slobin 1977; Langacker 1977) or maxims such as have been pro-posed by Keller (1994) are valid reasons for language change, then it is ob-vious that language change is an epiphenomenon We have a tradition ofcalling change within a relatively homogeneous community that is broughtabout by language acquisition “internal change”, as opposed to “externalchange” brought about by contact, but the first is actually no more “inter-nal” than the latter – it does not happen “in the language”, or “in the gram-mar”, only “in” transmission Expressions like “grammars change” areshort-hands for differences between grammars over time brought about bysystem-external factors such as acquisition, and not to be taken literally.Likewise, “directionalities” are not deterministic tendencies that requiresome change, and definitively not tendencies that live some reified exist-ence as cognitive paths, trajectories, or whatever other metaphor might beused, but they are nevertheless powerful tendencies that demand historicalthinking

To couch debates about such issues in language referring to “formal” and

“functional” theories is admittedly to reproduce old dichotomies many ofwhich were fortunately falling by the wayside in the nineteen nineties (seee.g Croft 1995; Newmeyer 1998) Nevertheless, it is useful in that it allows

us to acknowledge that the questions different researchers pose may be damentally different (see also Kemenade 1999)

fun-To say on the “formal” (generative) side of the debate that all one needs

in order to explain language change is: “(a) an account of how trigger periences have shifted and (b) a theory of language acquisition thatmatches PLD (primary linguistic data) with grammars in a deterministicway” (Lightfoot 1999: 225) simply puts the explanation off WHY does thetrigger change? Has the alleged “determinism” of the search for direc-tionality been replaced by a different determinism that predicts that if twochildren were to have the same trigger experience they would acquire thesame E-language? On this view, language change is the result of innovation

ex-in the ex-individual compared to some other, older, ex-individual Same ex-input(PLD) into the individual (LAD) will produce same output (G) In thisscenario, the individual is a processor of systems, largely passive, a logic

machine, a “language acquisition device”, presumably devoid of personaldifferences or preferences It is an implausible scenario not only because

we know children negotiate and experiment with language, but above allbecause it does not allow for abduction in language acquisition (see An-dersen 1973; Anttila [1972] 1989 on abduction as the key mechanism inlanguage change)

From a “functionalist” perspective, by contrast, change is the result ofstrategic interaction, specifically of choice-making on the part of speak-ers/writers in interactional negotiation with addressees/readers This in-cludes, but is not limited to, conveying of information On this view, lan-

guage change is the result of innovation in the individual and spread of

the innovation to the community, as suggested by Weinreich, Labov andHerzog (1968) This individual is active, an abducer who is a producer aswell as processor of language, and likely to have personal differences andpreferences Furthermore, innovation is not limited to early child lan-guage acquisition (see e.g Andersen 1973; Labov 1994; Ravid 1995) Onthis view too explanations for observed directionalities (and failures of di-rectionality) should be especially interesting and precisely what historicallinguists need to seek to provide Here I hope to suggest some ways inwhich to organize our thinking as we look for ways to account for direc-tionality

Unidirectionality is a thread common to the study of grammaticalizationand at least some branches of historical semantics and pragmatics Somearguments against unidirectionality are primarily philosophical – dependent

on what the researcher thinks a scientific argument is or should be – ments of this type have been put forward by Lightfoot (1999), and Lass(2000) Other arguments have been primarily of an empirical sort, forexample the work of Joseph and Janda (1988), Janda (1995), and Ramat(1992) Both kinds are presented in Newmeyer (1998) and Campbell(2001).2 I take these arguments seriously Nevertheless I do not find it use-

argu-ful to take a position such as “I take any example of upgrading as sufficient

to refute unidirectionality” (Newmeyer 1998: 263) If change is a socialproduct, the result of the interaction of language and use, how could there

be no counterexamples? Humans are not machines, and do not use guage mechanically They use it for purposes of strategic interaction Strat-egies must be flexible not rigid 100%s of anything! But likewise, sinceunidirectionality is not exceptionless, it is also not useful to adopt Leh-

lan-22 Elizabeth Closs Traugott

mann’s ([1982] 1995) and Haspelmath’s (1999) claims that there are no, or

at least no genuine, counterexamples to grammaticalization

Our road map is as follows Section 1 provides some background ongrammaticalization, with focus on English studies in the last hundredyears, and discusses aspects of the “deconstruction” of grammaticalization

in Newmeyer (1998: 5) Section 2 provides some background on semanticchange and the role of pragmatics in semantic change In section 3 I sug-gest a possible model of grammaticalization informed by historical prag-matics Section 4 puts forward some challenges for the future

The term grammaticalization, as is well known, seems to have originatedwith Meillet ([1912] 1958) Since then it has been closely associated withreanalysis that involves a change from lexical material in constructions tofunctional category status, e.g from nominal or verbal status to markers ofcase, tense, aspect, mood, conjunctions, etc

Although the term grammaticalization did not find its way into muchwork on English until the early nineteen seventies, Jespersen’s view ofhistorical syntax was in several ways germane to work on it We recoiltoday from Jespersen’s ideological claims that the modern languages are

“better” than older ones, that they have made “progress” or that cation is “beneficial” (Jespersen [1922] 1959: 363) We probably reject theidea that perfect languages would “express the same thing by the same,and similar things by similar means sound and sense would be in perfectharmony” (Jespersen 1959: 442) Yet his interest in “the transition fromfreedom in word position to greater strictness” (Jespersen 1959: 363), inshifts to more regular paradigms, and to shorter forms, coupled with hisidea that language originated in “half-musical unanalyzed expressions words and quasi-sentences” (Jespersen 1959: 441) resonates with some ofthe thinking behind Givón’s famous “cycle” (as he calls it), reproduced

simplifi-in (1):

(1) discourse > syntax > morphology > morphophonemics > zero

(Givón 1979: 209)

We should note in passing that Jespersen rightly saw that it is incoherent to

think of inflectional loss as preceding fixing of word order, i.e in a cycle:

“if this were true we should have to imagine an intervening period inwhich speech was unintelligible and consequently practically useless”(Jespersen 1959: 361) This is exactly what Lehman (1985) found himselfhaving to repeat again sixty years later

Despite Jespersen’s major work, Modern English Grammar on

Histori-cal Principles, that spanned forty years (1909–49), and Visser’s English Historical Syntax that spanned ten (1963–73), the book that probably did

most to jump-start historical English morphosyntax was Lightfoot’s

Prin-ciples of Diachronic Syntax (1979) It put the spot-light on the history of

modals, and of for-to constructions, both of which continue to be major areas of research From this work in (1979) to The Development of Lan-

guage twenty years later (1999), Lightfoot has focused on the question of

what a scientific approach to language change might be, and has found ananswer in language acquisition Follow-up work in the formal generativetradition (broadly construed) includes work by Kroch (e.g 1989), Fischer(1994, 2000), Kemenade (1987, 1999), Warner (1993), and several papers

in Kemenade and Vincent (1997), to mention only a few In all there is thebasic assumption that the interesting question is how grammars change (nothow people bring about changes in certain parts of grammars via strategicinteraction), and how child language acquisition may explain grammarchange

It is usually argued that because children bring about change, there not be unidirectionality In perhaps the clearest position statement from thisperspective, Lightfoot has said that if a child were to be thought to accessunidirectionality one would have to falsely postulate a “racial memory ofsome kind” (1999: 209) “Racial memory” is obviously an incoherent idea.But Lightfoot’s hypothesis that the trigger experience is simply the dataheard, and the frequency of it, is not a good substitute because it is unreal-istic The child is thought of as receiver and perceiver, apparently withoutany social purposes or awareness of the social purposes to which language

can-is put, of differences in language-use correlated with age, or of construction

of identity through language, and so forth, all of which can give the childaccess (or partial access) to layers of history

In his (1979) book Lightfoot discussed catastrophic change – innovation

of new syntactic categories or orders (what more recently has been

re-24 Elizabeth Closs Traugott

thought as parametric change), while acknowledging that the challenge toresearchers was to identify the local changes leading to them It is preciselythese local changes that have generated work in the more functional per-

spective The same year that Lightfoot’s Principles of Diachronic Syntax appeared, Givón’s On Understanding Grammar was also published This

highlighted the importance of cross-linguistic and typological studies (e.g.Lehman 1995 (though only remotely “functional”); Bybee 1985; Heine andReh 1984; Heine, Claudi, and Hünnemeyer 1991; Kemmer 1993; Svorou1993; Bybee, Perkins and Pagliuca 1994; Haspelmath 1997) Sometimeshistorical data from English was used to illustrate in depth the kind ofchange that is sketched or even reconstructed for other languages Just as

Latin cantare habeo ‘to sing have I’ to Fr chanterai has become a type change in discussions of Romance, so the development of be going to

proto-has dominated the pages of work on grammaticalization (see Hopper andTraugott 1993; Bybee, Perkins, and Pagliuca 1994, both strongly in-fluenced by Pérez 1990) But a great deal has also been devoted to English

as a subject in itself (e.g Brinton 1988, 1996) Central in this line of workhas been the research conducted on English in the context of the EnglishDepartment at the University of Helsinki (e.g Rissanen, Kytö, and Heik-konen 1997) The contributions to grammaticalization in English are fartoo many to mention here, but some (introduced in short form in Rissanen

et al., but developed into full-length monographs) deserve special mention:Kytö (1991) on modals, Nevalainen (1991) on the development of the fo-

cusing adverbials but, only, just, and Palander-Collin (1999) on the maticalization of I think and methinks as epistemic parentheticals In all

gram-these works, the pragmatic-semantic correlates, indeed putative precursors

of, grammaticalization have played a major role Change is conceptualized

as motivated by discursive and social practices Note that this is not thesame as saying discourse > syntax – Givón was hypothesizing change instructure from loose to tighter syntax, etc “Discursive practices” as I amusing the term assumes structure and syntax, and treats speakers and ad-dressees’ interaction in discourse as a motivator of change not a stage orphase in historical development On this view, the local changes investi-gated are often so particular, and so deeply entrenched in the idiomatic,constructional, semantic and pragmatic nature of language structure anduse that they cannot be usefully conceived of as the products of smallchildren setting parameters, but rather of older teens and adults engaged in

discourse practices for specific social purposes or of language-acquirers ingeneral.

Many of these issues are discussed in Newmeyer’s (1998) book on

Lan-guage Form and LanLan-guage Function A whole chapter is devoted to

“de-constructing grammaticalization” He characterizes grammaticalization asdescribing “the loss of grammatical independence of a grammatical struc-ture or element” (Newmeyer 1998: 227) and cites as “the standard defini-tion”: “grammaticalization where a lexical unit or structure assumes agrammatical function, or where a grammatical unit assumes a more gram-matical function (Heine, Claudi, and Hünnemeyer 1991: 2) In Newmeyer’sview, and indeed, that of most researchers on grammaticalization to date,the “or” is interpreted as inclusive – see also the following rather similardefinition: “the process whereby lexical items and constructions come incertain linguistic contexts to serve grammatical functions, and, once gram-maticalized, continue to develop a new grammatical function” (Hopper andTraugott 1993: xv) Newmeyer’s arguments are based on formal assump-tions: change comes about through child language acquisition, only phe-nomena that are universal in the absolute sense that they are predicted tooccur 100% of the time are considered to be explanatory I am not the first

to find that despite himself Newmeyer actually constructs ation to some extent (see Haspelmath 2000, Heine 2000), partly because hefinds that in the end he has to admit that “unidirectionality is almost true”(277) because it is so prevalent

grammaticaliz-Let us consider how Newmeyer approaches what I would call the vation and mechanism issue, and what he, Roberts, Lightfoot and otherscall the epiphenomenon issue The move that Roberts and Newmeyer try tomake is that grammaticalization is an epiphenomenon not only of themechanisms that bring change about (e.g reanalysis, analogy/rule exten-sion) but of other types of change, specifically downgrading analysis (de-categorialization), phonetic reduction (erosion), appropriate semanticchange (bleaching) Newmeyer models this as in Figure 1

moti-This model may be intended simply to illustrate and critique what meyer interprets some of the more extreme claims in the grammaticaliz-ation literature to be But it has some properties worth discussing In onesense this model appears to have no directionality However, it has an in-herently inbuilt directionality in two of its dimensions (downgrading andphonetic reduction) But what about semantic change? Is it not directional?

New-26 Elizabeth Closs Traugott

I will argue that it is – and in this directionality lies the key to the whole set

of changes we think of as grammaticalization This model shows that there

is phonetic reduction independent of grammaticalization, and semanticchange independent of grammaticalization This is indeed correct Butwhat downgrading is there outside of semantic change and phonological re-duction that is not a case of grammaticalization?

Figure 1. Grammaticalization as an epiphenomenon (Newmeyer 1998: 260).

Legend: GR = grammaticalization, ASC = appropriate semantic change,

DA = downgrading analysis, PR = phonological reduction.

Reprinted by kind permission of the MIT Press.

I propose that we think of grammaticalization differently We need to teaseapart two aspects of the theory of unidirectionality in grammaticalization.The possibility to do this is interestingly anticipated in the definitionsabove and the differences between them, provided we read “or” as exclu-sive in Heine, Claudi and Hünnemeyer’s definition above, and note that

“new grammatical function” is not the same as “more grammatical tion” in Hopper and Traugott’s, also cited above

func-One aspect of unidirectionality has to do with the grammatical function

of categories: this I will call “primary grammaticalization” Primary

gram-PR

ASC DA

GR

maticalization is the development in specific morphosyntactic contexts ofconstructions and lexical categories into functional categories Forexample, in English, lexical verbs become auxiliaries in certain of theiruses (or, in a different metalanguage, only a tiny subset of verbs that oncecould “raise to I” can still do so) Primary grammaticalization is compatiblewith recent work of Kemenade (1999), and Roberts and Roussou (1999).Indeed, that is how they define it; Kemenade specifically identifies the finalstage of grammaticalization as “base-generation as a functional head”(1999: 1001) There is almost no argument against this type of unidirec-tionality On this interpretation of unidirectionality, it makes no sense tosay that grammatical material becomes more grammatical, or that it necess-arily erodes, bonds, reduces syntactic scope, etc Indeed, it is feasible tothink of grammaticalization as allowing increased scope/freedom in exter-

nal syntax (for example, that as a complementizer has wider syntactic scope than that as a pronoun serving an argument function (Tabor and

Traugott 1998))

The other aspect of unidirectionality, and the one that is most sial, has to do with the form rather than the function of the categories underconsideration On this view grammaticalization is the development of mor-phophonemic “texture” associated with the categories in question; here theissue is the degree of morphological bonding/fusion, phonetic erosion,bleaching, etc (see specially Bybee, Pagliuca, and Perkins 1991; Bybee,Perkins, and Pagliuca 1994) With respect to such changes, it does makesense to talk about shifts to a more grammatical status, although it would bemore accurate to say that “expressions of functional categories become

controver-more bonded over time” Thus auxiliaries can undergo reduction (will > ’ll,

would > ’d, have > ’ve) This kind of change I call “secondary

grammati-calization” It is presumably what Kurylowicz has in mind when he wrotethe definition that lies behind those cited above: “the increase of the range

of a morpheme advancing from a lexical to a grammatical or from a lessgrammatical to more grammatical status, e.g from a derivative formant to

an inflectional one” (Kurylowicz [1965] 1976: 52) The crucial differencebetween this definition and the others is the “e.g.” This part is presumablyusually left off because of the difficulty of determining exactly what rolederivation has in grammaticalization

These two kinds of change are linked in ways still to be understood, but

in general we can say that changes of type B are later than, or at least start at

28 Elizabeth Closs Traugott

the same time as changes of type A, and crucially not before them I thesize that the shift from construction or from lexical to functional cat-egory is a subset of possible morphosyntactic changes (others, like wordorder are of a different kind), and licensed by the set of regular pragmatic-semantic changes (though not all pragmatic-semantic changes involvegrammaticalization, for example the development of performative mean-ings for speech acts verbs)

As traditionally understood, semantic change would seem to have verylittle to offer to grammaticalization beyond the well-known mechanisms,articulated by Bréal ([1900] 1964), Meillet ([1905–6] 1958) and othersabout a century ago, mostly in contrasting pairs, such as metaphor, meto-nymy, specialization and generalization, pejoration and amelioration

Studies almost exclusively in terms of nominal referents (e.g master,

mis-tress, sentiment, telephone, bead) have little to contribute to

grammatical-ization or to questions about unidirectionality But there were some hints ofwhat were to become predominant themes in the effort to conceptualize theinterface of meaning and grammaticalization Bréal speaks of subjectifi-cation, a concept I will return to, Meillet of the linguistic contexts for

change (for example, how homme ‘man’, chose ‘thing’ were affected by

ne-gation, interrogatives, conditional contexts (Bréal 1964: 239)), an idea solutely fundamental to work on recurrent patterns of semantic change.Though methodologically alien now, Stern’s account of meaning changewas particularly suggestive In his chapter on “permutation” he wrote of:

ab-“A shift in the apprehension of a complex referent, denoted by a phrase, [which] will, by repeated use, become associated to the word expressingthe earlier apprehension of it” (Stern [1931] 1968: 353) His example of the

transfer of meaning from beads ‘prayer’ to ‘beads of a rosary’ is

well-known Perhaps less well known is how he treated it: not as a transfer tonymy) in the world of non-linguistic action, but in the linguistic world ofthe phrase (language use) This example is one from the domain of nom-inals, with all their attendant problems attributable to changes in the realworld referents (are rosary beads made of wood, obsidian, plastic?) But healso discussed some other meaning changes that have no real-world refer-

(me-ents such as while with its development “from temporal to adversative and

concessive import” (Stern 1968: 379)

With hindsight one can see emerging in Stern’s attempt to characterize

“permutation” as what in Grice’s ([1975] 1989) and Levinson’s (1983)conceptions came to be known as the conventionalizing of conversationalimplicatures.3 I have developed the Grice/Levinson insight into the In-vited Inferencing Theory of Semantic Change or IITSC (Traugott 1999;Traugott and Dasher In press; for an earlier version see Traugott andKönig 1991) Here I will mention only two things about IITSC becausethey will be crucial later I have extended Geis and Zwicky’s (1971) no-tion of “invited inference”, seeing in this term (rather than “implicature”)

the possibility of alluding to both speaker’s strategic action (inviting) and

the hearer’s response (inferencing) Furthermore, following Levinson(1995, 2000), I distinguish between invited inferences that arise “on thefly” or are not salient in the community (these are Invited Inferences orIINs), and those that are well established (Generalized Invited Inferences,

or GIINs) An example of a GIIN is the causative inference from after in

certain contexts; this has been available since Old English It is not onlystable but well known and usable, but has not yet been semanticized as apolysemy

Recent work on semantic change motivated by pragmatic factors has itsroots in various places Four are of particular importance: lexical field the-ory, cognitive linguistics, grammaticalization, and historical pragmatics/semantics These in some sense span the century, with lexical field theoryemerging first, then grammaticalization and cognitive linguistics, and fin-ally historical pragmatics

Working at a time when Trier’s famous (1931) work on terms of intellect

in Middle High German around 1200–1300 was much discussed (it waspublished in the same year as Stern’s book), Stern attempted to get awayfrom the contemporary focus on discrete components of meaning divided

up as in a mosaic He took a psychological view of meaning and argued thatchange within a lexical field could be independent of culture His special

interest was in changes evidenced by the field RAPIDLY (e.g OE swifte,

the action as a unit” (Stern 1968: 185–191) He argued that this change fected all words that meant RAPIDLY before about 1300; words from othersources that came to mean RAPIDLY or were borrowed with that meaning

af-30 Elizabeth Closs Traugott

after 1400, e.g fleetly (1598) and rapidly itself (1727) did not undergo this

change

The theoretical interest of Stern’s hypothesis is several-fold He aimed toprovide evidence that related words could show parallel semantic changes,and that semantic changes can take place over a certain period, and thencease to be in effect He interpreted the data as an example in semantics ofregular change similar to that posited by the neogrammarians for phono-logical change More importantly, he suggested that the changes were de-

pendent on the lexical aspect of the verb, rapidly being associated with cess (imperfective) events, immediately with punctual (or perfective) ones.

pro-In other words, the change was associative, arising out of the syntactic

“phrasing”, in this case use in what we would now think of as cal contexts He also showed that it was implausible for the reverse change

non-canoni-to take place: “it is evident that if a person rides rapidly up non-canoni-to another, theaction is soon completed; but we cannot reverse the argument and say that

if a person soon rides up to another, then the action is also rapidly formed” (Stern 1968: 186) This is an argument for unidirectionality, andfor the conventionalizing of implicatures

per-While Stern’s work was often ignored because of problems with his tempt to equate semantic with phonological change, lexical field analysisgained momentum in the sixties and seventies, largely influenced by an-thropological taxonomies (for example Berlin and Kay’s 1969 work oncolor terms) English historical lexical field studies include such works asWilliams (1976) on synaesthesia, the metaphorical extension of terms ofthe five senses onto the domain of the other senses and onto social beha-

at-viors, language, and so forth (cf acid remark, bitter irony); Dahlgren

(1985) on the social construction of kingship; Lehrer (1985) on the

meta-phorical use on zoological terms (e.g baboon, vulture) for pejorative

pur-poses; Goossens (1985) on verbs of speaking, to name just a few The mostrecent line of research in lexical fields has been developed within historicalprototype theory (Geeraerts 1997), e.g Koivisto-Alanko (2000) on the field

of wit, Molina (2000) on the field of give sorrow.

Some of these studies began to suggest patterns of directionality,whether increase in the set of color terms along predictable paths, or pat-terns of change in synaesthesia Not, however, till the rise of cognitive lin-guistics and of grammaticalization as active fields of research did interest inunidirectionality in semantic change really come into its own, for in these

contexts the domains explored became increasingly conceptual and stract Nouns were paid less attention than verbs, most especially func-tional categories like tense, aspect, modality (e.g Bybee and Pagliuca1985), adverbials with temporal properties (e.g König and Traugott 1982),focus particles (e.g König 1991), etc.

ab-Especially significant for the study of semantic change was Sweetser’s(1990) work on modals and her suggestion (largely based on synchronicdata) that one way to conceptualize the relation between deontic and epis-temic modality was to think of it in terms of mapping of image-schemata

Sweetser hypothesized that in the case of the modal may sociophysical

ob-ligation is metaphorically mapped onto obob-ligation in the world of ing; and further mapped onto what she called speech act meaning in the

reason-world of speaking, e.g Kim may go (permission) onto Kim may be tired (epistemic), and onto Kim may be a nice guy, but I don’t trust him.

(“speech-act modal”) Sweetser argued that meaning change is from tent to reasoning and to “speech act” meanings, not vice versa (to avoidconfusion with speech acts and strictly performative meaning, the terms

con-“metatextual” (Dancygier 1992) or “procedural” (see Blakemore 1987) arepreferable) This kind of thinking was fundamental to work on grammati-calization in African languages associated with Heine and his colleaguesstarting with Claudi and Heine (1986)

Meanwhile I was working on an alternative approach, one that it turnsout was more in keeping with Stern’s ideas, since I proposed looking not somuch at the discontinuities of metaphorical mappings, at the sources andtargets of change, but at the inferential processes at work in the flow ofspeech, and the local continuities across times and texts Here the focus was

on associative meanings arising in the production and reproduction of course Much of the work on unidirectionality in semantic change was done

dis-in the context of grammaticalization, for example Traugott (1982), gott and König (1991), Bybee and Pagliuca (1985), Bybee, Perkins, andPagliuca (1994) But this was only an entry-point into the large territory ofsemantic change in general

Trau-Historical semantics and pragmatics is of course a very diverse field.Studies in semantics range from focus on the sociocultural macro-context

of changes in discursive practices to focus on the micro-contexts of ing development in communication, including locally constrained strategicinteractions in conversation, narrative, and other genres One thing that

mean-32 Elizabeth Closs Traugott

these lines of research have made crystal clear is that we can go beyond thewell-known contradictory-seeming (and therefore arbitrary-seeming) pairs

of semantic change that we have all learned: amelioration vs pejoration,generalization vs specialization, euphemism vs taboo, metaphor vs meto-nymy Likewise we can go beyond the apparent randomness of reference-

based meaning changes (such as changes in the meaning of chalk or horse,

cf Eckardt 1999) When we move from thinking about reference change tosense change, and from concepts expressed by nouns to concepts expressed

by other parts of speech, especially verbs or conjunctions, a new pictureemerges in which unidirectionalities predominate:

(2) a SPACE > TIME not vice versa

b DEONTIC > EPISTEMIC not vice versa

c CONTENT > PROCEDURAL not vice versa4

d SOCIOPHYSICAL > EPISTEMIC > METATEXTUAL not viceversa

An even larger picture emerges when we move from thinking about wordsout of context to constructions in context, and from the study of the OED orthe MED to study of corpora Now we find ourselves squarely in the do-main of discourse and genre analysis; of rhetorical purposes such as in-formation, explication, persuasion, strategic interaction, and rhetoricalforce (Fitzmaurice 2000) In this line of work, metaphor and metonymy,generalization and specialization, amelioration and pejoration still play animportant role However, though necessary, they are no longer sufficient.Rather, they are seen as outcomes (epiphenomena, yes) of two mechanismsthat are well-known to lead to morphosyntactic change: analogy and rea-nalysis Metaphor is closely allied to analogy because it is largely paradig-matic, and metonymy is closely allied to reanalysis because it is largelycontext-dependent and therefore associative (see Anttila 1989: 141–142).The hypothesis is that as speakers strategize discourse, these two mechan-isms guide and constrain the conventionalizing of conversational implica-tures A third important mechanism is subjectification, a mechanismwhereby meanings come over time to encode or externalize the speaker/writer’s perspectives and attitudes as constrained by the communicativeworld of the speech event, rather than by the so-called “real-world” char-

acteristics of the event or situation referred to (Traugott 1989; Stein andWright 1995; Traugott and Dasher In press5) Amelioration and pejorationare allied to subjectification at a relatively trivial level, and the development

of metatextual meanings at a more substantive level These three isms are not only cognitive but also communicative and arise out of com-municative practices in the (minimally dyadic) speech situation, and inwriting

mechan-The overarching directionalities we find are shifts from content ings (what I used to call “propositional meanings” and Sweetser calls “so-ciophysical meanings”), to non-truth-conditional, and procedural mean-ings Examples include:

mean-(3) a as long as ‘measure of spatial length’ > ‘measure of temporal

length’ > ‘conditional’ = ‘provided that’

b actually ‘in an affective manner’ > ‘epistemic adversative’ >

‘addi-tive discourse marker’

The history of actually illustrates an increase in semantic scope (the

manner adverbial is VP-internal, the discourse marker is nal).6 It, and many others like it, also illustrates a shift from meanings un-related to discourse to meaning entirely associated with discourse (the con-tent > procedural meaning change) Procedural meanings index metatex-tual relations between propositions or between propositions and the non-linguistic context; they cue addressees to speakers’ and writers’ attitudes tothe discourse and the participants in it Therefore many pragmatic/semanticchanges, including those illustrated in (2) and (3) can, I hypothesize, beseen as the outcome of subjectification (Traugott and Dasher In press)

sentence-exter-To return to grammaticalization, we can now better understand how tointerpret and probe the hypothesis that constructional or morphosyntacticlexical > functional category change will not occur without prior semanticchange Given the IITSC model of semantic change in which GIINs play amajor role, while semantic change often does precede, as in the case of themodals, it is not obvious that semantic as opposed to pragmatic change

must precede, in the sense that a new coded polysemy must have arisen

be-fore the category shift occurs What has to happen minimally bebe-fore maticalization can take place is that there is a widely understood, and oftenexploited GIIN (which is a pragmatic polysemy, see Horn 1985; Sweetser