Negation in the History of English [Reprint 2011 ed.] 9783110806052, 9783110161984

Citation preview

Negation in the History of English

m 1749

I

1999

I

Topics in English Linguistics 26 Editors

Bernd Kortmann Elizabeth Closs Traugott

Mouton de Gruyter Berlin · New York

Negation in the History of English edited by

Ingrid Tieken-Boon van Ostade Gunnel Tottie Wim van der Wurff

W DE Mouton de Gruyter G Berlin · New York 1999

Mouton de Gruyter (formerly Mouton, The Hague) is a Division of Walter de Gruyter & Co., Berlin.

© 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

Negation in the history of English / edited by Ingrid TiekenBoon van Ostade, Gunnel Tottie, Wim van der Wurff. p. cm. - (Topics in English linguistics ; 26) Papers originally presented at a conference held at the University of Leiden on Dec. 15-16, 1994. Includes bibliographical references and index. ISBN 3-11-016198-2 (cloth : alk. paper) I.English language-Negatives-Congresses. 2. English language—Grammar, Historical-Congresses, I. Tieken-Boon van Ostade, Ingrid. II. Tottie, Gunnel, 1937III. Wurff, Wim van der. IV. Series. PE1359.N44 N435 1998 425—dc21 98-42858 CIP

Die Deutsche Bibliothek — Cataloging-in-Publication-Data Negation in the history of English / ed. by Ingrid Tieken-Boon van Ostade .... - Berlin ; New York : Mouton de Gruyter, 1998 (Topics in English linguistics ; 26) ISBN 3-11-016198-2

© Copyright 1998 by Walter de Gruyter GmbH & Co., D-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. Printing: Werner Hildebrand, Berlin. Binding: Lüderitz & Bauer-GmbH, Berlin. Printed in Germany.

Contents Abbreviations

vii

Introduction Ingrid Tieken-Boon van Ostade - Gunnel Tottie - Wim van der Wurff Five ways of saying no: the development of sentential negation in English in a Government and Binding perspective Frits Beukema

1

9

English negation from an interactional perspective Jenny Cheshire

29

On Negative Raising in the history of English Olga Fischer

55

Negative Concord and Verb Projection Raising in Old English and West Flemish Eric Haeberli - Liliane Haegeman

101

Multiple negation in Middle English verse Yoko Iyeiri

121

Sentential negation and clause structure in Old English Ans van Kemenade

147

The facts end nothing but: the (non-)grammaticalisation of negative exclusives in English Terttu Nevalainen

167

Isn't it? or is it not? On the order of postverbal subject and negative particle in the history of English Matti Rissanen

189

vi

Contents

The origin and development of the "Neg ... neither·" construction: a case of grammaticalisation Ingrid Tieken-Boon van Ostade

207

Affixal and non-affixal negation - a case of stable variation over time? Gimnel Tottie

233

On the scope of Negative Concord Masatomo Ukaji

269

On expletive negation with adversative predicates in the history of English Wim van der Wurff

295

Index

329

Abbreviations Adj AgrOP AgrSP Aux BNC BUC COBUILD CI CP El E2 E3 EModE Fin FP GB HC ICE Infi IP IPP LAGB LALME LDOCE LF LHM LLC LME LModE LOB Ml M2 M3 M4 ME

adjective Agreement Phrase for the object Agreement Phrase for the subject auxiliary British National Corpus Brown University Corpus of written American English Collins COBUILD English Language Dictionary clitic complementiser phrase Helsinki Corpus Early Modern English subperiod 1, 1500-1570 Helsinki Corpus Early Modern English subperiod 2, 1570-1640 Helsinki Corpus Early Modern English subperiod 3, 1640,1710 Early Modern English finite functional projection Government and Binding Helsinki Corpus International Corpus of English Inflectional head Inflection Phrase infinitivus pro participio Linguistic Association of Great Britain A Linguistic Atlas of Late Mediaeval English Longman Dictionary of Contemporary English Logical Form Long Head Movement London-Lund Corpus of spoken British English Late Middle English Late Modern English Lancaster-Oslo/Bergen Corpus of written British English Helsinki Corpus Middle English subperiod 1, 1150-1250 Helsinki Corpus Middle English subperiod 2, 1250-1350 Helsinki Corpus Middle English subperiod 3, 1350-1420 Helsinki Corpus Middle English subperiod 4, 1420-1500 Middle English

viii MED NC Neg NegP NP NPI NR OE OED PDE PP S SOV Spec Subj TP V2 Vfin Vlex VP VPR VR WF

Abbreviations The Middle English Dictionary Negative Concord negation negation phrase noun phrase Negative Polarity Item Negative Raising Old English The Oxford English Dictionary Present-Day English prepositional phrase sentence subject-object-verb Specifier subject Tense Phrase Verb second finite verb lexical verb verb phrase Verb Projection Raising Verb Raising West Flemish

Introduction1 Ingrid Tieken-Boon van Ostade - Gunnel Tottie - Wim van der Wurjf

In spite of the attention that has been devoted to negation in the synchronic and diachronic study of English, there are still vast lacunae in our knowledge of the forms, meanings and functions of this phenomenon. This volume contains twelve papers on negation in the history of English which attempt to fill some of these gaps. Most of the papers were presented at the conference on Negation in the History of English held at the University of Leiden on December 15-16, 1994. All are empirically based, representing different theoretical and methodological approaches to corpus evidence of various kinds. Four of the papers are written within a generative framework, and also offer theory-internal argumentation based on historical negative material. Most of the other papers take a quantitative perspective and seek functional explanations of the problems under study. One of the phenomena that have been extensively studied is multiple negation or negative concord, i.e. the copying of the negative element on to more than one clause element without the cancelling of negative meaning, as in Chaucer's characterisation of the Knight: (1)

He nevere yet no vileynye ne sayde. In cd his lyf unto no maner wight (Chaucer General Prologue 70-71)

Five studies included in this volume deal with different aspects of multiple negation, i.e. those by Haeberli and Haegeman, Iyeiri, van Kemenade, Tieken-Boon van Ostade, and Ukaji. Although multiple negation was frequent in Old English, Middle English is the locus of both its peak and its decline — as is well known, negative concord disappeared in standard English (see Tieken-Boon van Ostade 1995 for a description of the early stages of this process) but it remains in non-standard dialects. Yoko Iyeiri studies itsriseand fall in a large corpus of Middle English poetry. She distinguishes three (sometimes intersecting) main types of multiple negation, (i) with ne... noght, (ii) with the negative conjunctions netnor, and (iii) the type where

2

Ingrid Tieken-Boon van Ostade - Gunnel Tottie - Wim van der Wvrjf

the negative element fuses with an indefinite, such as never and no. Iyeiri finds that more than half of all negative sentences in her material have multiple negation (somewhat surprisingly from a modern point of view, the highest proportions occur in formal texts), but that the vast majority of these contain only two negatives. The decline of multiple negation sets in first in type (i) constructions, whereas types (ii) and (iii) are more resistant to change and non-assertive forms such as any and ever only begin to appear very late in the period. Interestingly, one kind of multiple negation, involving a negative sentence followed by neither, only seems to have emerged in Middle English and also to have survived well into Early Modern English. This type is seen in (2)

It is not for your health ... Nor for yours neither

where the meaning is 'also not' and in (3)

I hope things are not so very bad with you neither

where the meaning is 'nevertheless'. Ingrid Tieken-Boon van Ostade shows that these two types seem to have appeared in the thirteenth century and survived even in the language of highly literate speakers and writers well into the Early Modern English period, when they were stigmatised by normative grammarians. They are still very much alive in non-standard Present-Day English. Usually, the scope of negation in sentences with negative concord is restricted to a single clause, but Masatomo Ukaji shows in his paper that it was also possible for the scope to extend beyond clause boundaries (wide negative scope), as in the following example from Bacon: (4)

Shall we not thinke, that God aboue, that knowes the Heart, doth not discerne, that fraile Men ... intend the same thing...

In a corpus of textsfromthe eleventh to the seventeenth centuiy, Ukaji found some sixty instances of wide negative scope (not including expletive negation, which is the subject of Wim van der WurfPs paper; cf. below). He found that negative scope could go deeper than the first level of embedding and also, surprisingly, extend into relative clauses. Interestingly, the set of governing predicates that may take

Introduction

3

wide-scope negation is almost identical with the predicates that allow negative raising (treated in this volume by Olga Fischer) and Ukaji discusses possible reasons for this. Ans van Kemenade deals with an early type of negative concord: she studies sentential negation conveyed by ne ... na, as in (5)

ponne ne mihi pu na pcet mot ut ateon of dees marines eagan 'Then you could not draw the speck out of the man's eye'

Although this is a minority pattern, van Kemenade collected well over 300 instances of it, two thirds of them with the negative in sentence-initial position. She shows that when the subject is a pronoun, na tends to follow it, while in sentences with nominal subjects, it precedes the subject. Van Kemenade uses this as evidence both for a particular view of Old English sentence structure and for the status of pronouns as clitics in Old English. Related problems of word order involving negatives and pronouns are dealt with by Rissanen (see below). Another aspect of negative concord in Old English is studied by Eric Haeberli and Liliane Haegeman. They show that although there are great similarities between Old English and West Flemish both as concerns word order and the use of negative concord, there are also interesting differences in the contexts often referred to as Verb Projection Raising (VPR) structures, and that superficial similarities may have to be explained in different ways. In Old English, the sentence in (6), (6)

pcet heora nan ne mehte nanes wcepnes gewealdan 'that none of them was able to wield a weapon'

has only the reading 'that none of them could wield a weapon', but in West Flemish example (7), (7)

da Valere nie en-durft niets zeggen 'that Valere doesn't dare to say nothing'

can only mean 'that Valere does not dare not to say anything' with the two negatives cancelling each other out. Thus VPR interferes with negative concord in West Flemish, while there are no such restrictions in Old English; this phenomenon

4

Ingrid Tieken-Boon van Ostade - Gunnel Tottie - Wim van der Wurff

may be related to a different categorial status of infinitival clauses in the two languages as well as to the existence of different rules for the overt movement of negative elements. Wim van der Wurff deals with another type of negation which is redundant from the point of view of Present-Day English, i.e. expletive negation by means of ne after adversative predicates, as in the Middle English sentence in (8), (8)

no man douteth that he ne is strong in whom he seeth strengthe

meaning 'no man doubts that he is strong ...'. Expletive negation could only be conveyed by ne, and when ne disappeared from the language after the Middle English period, so did expletive negation. From a theoretical point of view, this can be explained if we assume that the adversative predicate selects a NegP with a negative element as its head and a truth-value operator in its specifier. The disappearance of expletive negation in English can then be explained by the fact that not is not a negative head and could not take the place of ne after its demise. Negation is a highly mobile element, 'which can surface in different positions in the sentence. Matti Rissanen surveys the placement of not in relation to the postverbal sentence subject throughout the history of English. For the older periods of the language the typical position was after the subject if this was a personal pronoun, as shown in (9), (9)

ne dorste he nawuht hrcedlice ut of pcere ceastre faran 'he dared not go quickly out of the city'

but with the loss of ne and grammaticalisation of not, cliticisation of the negative adverb to auxiliaries preceding pronominal subjects became possible. Rissanen suggests that negative cliticisation with auxiliaries and the copula may have taken place in speech as early as the beginning of the sixteenth century, rather than around 1600 as proposed by Jespersen. This could then have contributed to the popularisation of the auxiliary do in negative sentences. Corpus evidence shows a neat progression from 40% not!-n't preceding personal pronoun subjects in the period 1650-1800 to 94% in Present-Day English, with contracted forms becoming totally dominant at the later stages of the language. Frits Beukema also discusses the placement of not. He focuses on the pattern /

Introduction

5

not say, which Ukaji (1993) has called a bridge phenomenon between I say not and I do not say. Beukema makes a comparison with a similar and similarly short-lived phenomenon in the acquisition of negation by French children and argues, within the framework of minimalist theory, that the bridge pattern in English is "bound to collapse [because] it is an inherently unstable pattern in the sense that it has different values for strength of Infi and Neg". Olga Fischer's paper deals with a particular instance of the mobility of the negative element. She treats negative raising, also known as negative transportation or negative transfer, as in (10)

I don't think he's coming at all

From the presence of the negative polarity item at all we can see that the negative element which surfaces in the matrix clause semantically belongs in the embedded clause. This type of construction continues to puzzle synchronic grammarians, and Fischer suggests that by taking a diachronic approach and studying its emergence in English (Old English does not show any examples), we may arrive at a better understanding of the phenomenon. Based on a survey of the Helsinki Corpus, Fischer finds a few doubtful examples with believe, think and ween in Middle English and more frequent occurrences in Early Modern English, mostly involving seem and think. She points to several factors that are likely to have promoted the rise of the neg-raising construction, among them the loss of multiple negation (negative concord), subject raising, the rise of epistemic modality, the change in the position of the adverb not, and the introduction of do in the Aux category. Another area that has to do with the mobility of the negative element and which has received little attention in a diachronic perspective so far is that of the variation between affixal and non-afFixal negation, as in (11)

a. b.

It was impossible It was not possible

In her paper, Gunnel Tottie looks at the use of affixal negation in late Middle English, focusing especially on actual and potential variability of forms. She compares the constraints operating on either type of negation in Middle English with those in twentieth-centuiy English and addresses the issue of stable variation

6

Ingrid Tieken-Boon van Ostade - Gwwel Tottie - Wim van der Wiaff

over time. The patterns are obscured by lexical idiosyncrasies, such as the non-existence of one variant or semantic non-equivalence of variants. In Middle English there are also additional problems such as the lack of lexicographical resources and native speaker intuitions. On the whole, however, it seems clear that most constraints on the use of affixal and non-affixal negation have remained constant through time. Terttu Nevalainen looks at the set of negative exclusive adverbs from Middle English onwards. After demonstrating the grammaticalisation of ne ... but to but between 1150 and 1500 and the subsequent rise and fall of but, she goes on to her central topic, the failure of nothing but and not/nought but to grammaticalise as exclusive adverbs. Nevalainen comes down in favour of a multicausal explanation: adverbialisation may have been impeded by modifiability and paradigm cohesion with other indefinite pronouns in the case of nothing ... but, and the possibility of splitting the first and second elements and the decrease in popularity of but may have played a part in the non-grammaticalisation of both expressions. Jenny Cheshire takes a corpus of Reading teenager English and other contemporary data as her point of departure for a discussion of the meaning of never. She argues that never is frequently used as a simple negator equivalent to not, as in (12)

I never went to school today

Linguists have failed to observe this use because of the influence of prescriptive grammar and work in theoretical linguistics based on isolated examples. Cheshire demonstrates how quantifiers (never as well as other lexical items) contribute to the involvement of participants in discourse, and that the use of never as a universal temporal quantifier or a negator referring to a single occasion is negotiated by discourse participants. She also adduces evidence showing that this has been the case throughout the history of English. In thus bringing together different approaches to the subject of negation in all periods in the history of English, ranging from Old English to the present day, the present volume aims at providing a reflex of the most recent research developments in the field of negation on the one hand and in that of diachronic syntax on the other. By also indicating where further research may still profitably be carried out in the areas that have been the object of analysis here, the authors hope to be able to contribute to fruitful further research on the history of negation in English.

Introduction

7

Note 1.

We are grateful to Leendert Plug for his assistance in producing the camera-ready copy of this volume, and also for his insightful comments on textual matters.

References Rissanen, Matti - Ossi Ihalainen - Terttu Nevalainen - Irma Taavitsainen (eds.) 1993 History of Englishes: New methods and interpretations in historical linguistics. Berlin: Mouton de Gruyter. Tieken-Boon van Ostade, Ingrid 1995 The two versions of Malory's Morte Darthur. Multiple negation and the editing of the text. Cambridge: D.S. Brewer. Ukaji, Masatomo 1993 " Ί not say': Bridge phenomenon in syntactic change", in: Matti Rissanen - Ossi Ihalainen - Terttu Nevalainen - Irma Taavitsainen (eds.), 453-462.

Five ways of saying no: the development of sentential negation in English in a Government and Binding perspective1 Frits Beukema

1. Introduction This article is concerned with the development of sentential negation in English, i.e. with the way in which not is used to deny the prepositional content of a sentence. It is not concerned with other expressions of negation, such as no, as in He bought no records as a variant of He did not buy any records, or not... either as a variant of neither, as m He did not buy any records either versus Neither did he buy my records. This sort of variation is undoubtedly relevant in a wider discussion of negative strategies in English (see Tottie 1991 for much useful discussion of the various negative strategies), but it cannot be dealt with in this article. The interaction between sentential negation and Negative Concord must likewise be left out of account. A great deal of interesting work has been carried out in this area (see for example Haegeman 1995; Rowlett 1994; van der Wouden 1997), but extending the discussion of the development of English sentential negation to the area of Negative Concord and negative polarity is beyond the scope of this article, and must be deferred to a later moment. This article is structured as follows. In section 2 the diachronic development of English sentential negation is surveyed, with particular focus on the I not say type, which is shown to have functioned as a bridge between Middle English sentential negation and the way it is now expressed in Modern English. Sections 3 and 4 discuss a parallel - albeit an inverse one - between the development of sentential negation in English and the acquisition of negation in French. Section 5 presents an analysis of the facts in the light of Government and Binding Theory and the Mimimalist Framework. The differences between English and French are argued to result from the interaction of two parameters: the strength of the V-feature of Infi and the potential blocking effect of the functional head in which negation is located.

10

Frits Beukema

2. The diachrony of English negation: facts and figures The way in which sentential negation is expressed has undergone a number of changes in the historical development of English. The history of English sentential negation neatly reflects Jespersen's (1917, 1924) generalisation that there is a developmental pattern in systems of sentential negation, and that the various stages of the developmental pattern are individually attested in languages. This pattern is sometimes referred to as the Negative Cycle. It can be summed up as follows: original negative markers weaken phonologically, drifting towards the left periphery of the clause in the Germanic and Romance languages. There they often become (pro)clitic to the finite verb, whereby they lose so much of their negative import that they need to be reinforced by an independent, usually postverbal, element, which eventually comes to be the sole negative marker, with the clitic marker becoming optional and eventually disappearing. The diachronic development of English sentential negation accurately mirrors Jespersen's Negative Cycle. The various stages are described in (1), with a first set of examples illustrating them in (2): (1)

a. b. c.

d.

"classical" Old English: ne, always preceding the finite verb (2a) Late Old English and throughout the Middle English period: ne strengthened by not, finite verbs placed between ne and not (2b) beginning in the Late Middle English period, ne in the ne ... not periphrastic negation is commonly left unexpressed; finite verbs placed in front of not (2c) beginning in the fifteenth century, two parallel alternatives to (2c) (which continues to be in use up to the late seventeenth century) arise in the domain of "lexical" (i.e. non-auxiliary) verbs: i. the inflected lexical verb follows not (2d) (this type, which was never very frequent, disappears in the second half of the eighteenth century); ii. the uninflected lexical verb follows not while a finite form of the dummy verb do precedes not (2e) (survives as the only way of marking clausal negation with lexical verbs; not develops an enclitic form -n't (2e') from around 1600)

Development of sentential negation in a GB perspective a. b. c. d. e. e'.

11

Ic ne secge I ne seye not I say not I not say [+Fin] I do not say [-Fin] I don't say [-Fin]2

A Romance language such as French seems to have undergone roughly the same development; compare (2) with (3): a. b. c. d. e.

Jeo ne di Je ne dis (pas) Je ne dis pas (standard written French) Je (ne) dis pas (standard spoken French) Je dis pas (colloquial French) Ί don't say'

Roughly speaking, French has traversed the same developmental stages in the Negative Cycle as English, but Modern Italian and Modern Spanish are only at the first stage of the Negative Cycle, where sentential negation is marked by a proclitic element. Some examples are given in (4a) for Italian and (4b) for Spanish: (4)

a.

b.

Gianni nan parla Gianni not speaks 'Gianni doesn't speak' No estamos contentos aqui not we-are happy-pl-masc here 'We are not happy here'

Having illustrated the Negative Cycle, let us now consider in more detail the diachrony of English. In surveys of the historical development of English sentential negation, the pattern exemplified by (2d) is rarely dealt with; recent discussion of this pattern, however, can be found in Denison (1993: 449ff), who calls it "moderately common", and especially in Ukaji (1993), who considers it a bridge between stages (2c) and (2e). The pattern in (2d), of which some more examples

12

Frits Beukema

are given in (5) (taken from Ukaji 1993: 454), has been characterised descriptively in a variety of ways in the earlier literature. The OED and Andrew (1940: 69) describe it as poetic; Söderlind (1951: 218) and Traugott (1972: 147) consider that it is used for emphasis, while Partridge (1953: 4), like Visser (1969), looks upon preverbal not in (2d) as a continuation of the Old English practice of placing ne before the verb (see (2a)). a. b. c. d. e. f. g·

I seyd I cowde not tellyn that I not herd (Paston Letters 705.51/2) I not holde agaynes luste al vttirly (ca. 1412. Hoccleve, The Regement of Princes) my young nouice (what euer thou be) not yet crept out of the shell (Nashe, Christs Teares; Works 11.86) ... it not belongs to you (Shakespeare, 2H4 IV.i.95/6) for who not needs shall never lack a friend (Shakespeare, Hamlet III.ii.207) God... staffers a wrong to be done, but not commands it (Dryden, Polybius 45) They... possessed the island, but not enjoyed it (1740. Johnson, Life of Drake; Works IV.419)

The "bridge" construction in (2d) was rare before 1500, but became moderately frequent in the work of Shakespeare and Jonson. From the incidence of the construction in the seventeenth century, Tieken (1987:46) concludes that the pattern "was at that time very likely ... characteristic of colloquial language, i.e 'the language spoken in everyday situations...'". Tieken found no more than three cases from the eighteenth century (from Richardson, Lady Mary Wortley Montagu and Defoe), which shows that the pattern declines rapidly and becomes obsolescent in the course of the eighteenth century (Ukaji 1993:455). Table 1 details the incidence of the I not say construction across the four centuries of its occurrence; it is based on Ukaji's data:

Development of sentential negation in a GB perspective

Table 1.

13

Distribution of the I not say construction

15th century

16th century

17th century

18th century

Hoccleve (2)

Ascham (1)

Shakespeare (33)

Defoe (1)

Pecock (1)

Marlowe (1)

Jonson (20)

Lady Mary (1)

Paston Letters (2)

Nashe (5)

D'Avenant (7)

Johnson (1)

Shakespeare (12)

Dryden (3)

Richardson (1)

Two properties of the bridge construction in (2d) and the construction in (2e), which is common now, merit special attention; they are summed up in (6): (6)

a. b.

(7)

a. b.

the patterns in (2d) and (2e) arise at roughly the same time if the head of the VP in the complement of Infi is an auxiliary, (2d) and (2e) are not attested - finite auxiliaries must end up to the left of the negation marker, as in (7) she not {denies/*has denied/*is denying} it she does not (deny/*have denied/*be denying} it

The development of English negation in the Middle English and Early Modern English periods is the central focus of this paper. What is so interesting about negation in this period is that it traverses three stages that can also be discerned in the acquisition of Present-Day French, as the next section will show.

3. The acquisitional development of French negation: facts and figures In the initial stages of acquisition, children learning French, just like children learning English, Dutch or German, produce many so-called "root infinitives", i.e. infinitival verbs forming the root clause predicate, alongside verb forms bearing finite inflection. An example of a root infinitive in early child French is given in (8):

14

Frits Beukema

(8)

'tacha ouvrir Natacha open [-Fin] (Nathalie, 2;02;02 (Ferdinand 1994))

Root infinitives are extensively documented in the extant corpora of child French, Dutch, English and German, but almost exclusively with lexical verbs. Auxiliaries (both modal and aspectual) are virtually exceptionlessly finite. But finite lexical verbs appear from the very earliest stages, too. It is worth noting, however, that even in the domain of finite inflection there is a difference between auxiliaries and lexical verbs. While auxiliaries alwaysfrontto a position to the left of negation (see e.g. (9)), for a short period of time (roughly a month) children acquiring French can place finite lexical verbs to the right of the negation marker. This period, which is illustrated by the examples in (10) (see Verrips - Weissenborn 1992, from which the data in (10) are taken) is very short-lived. Like the root infinitival pattern in (8), the pattern in (10) is characterised by the fact that auxiliaries never take part in it. (9)

(10)

est pas gros is [+Fin] not fat (Nathalie, 2;02;02 (Ferdinand 1994)) a. pas joue le chat not plays [+Fin] the cat (Fabienne, 2;00;13) b. pas compte not counts [+Fin] (Fabienne, 2;00;23) c. pas chante moi not sing [+Fin] I (Benjamin, 2;02;18) d. pas met not put [+Fin] (Benjamin, 2;02;18) e. pas saute not jump [+Fin] (Benjamin, 2;03;01)

Development of sentential negation in a GB perspective

15

Fabienne's utterances between the ages of 1 ;05; 11 and 2;00;23 feature precisely two examples of a finite main verb following the negation pas; Benjamin produces five utterances of this type between the ages of 1 ;09; 19 and 2;03 ;08. In the same period, instances of non-finite lexical verbs following pas, though rare in Fabienne's (not very extensive) corpus, are relatively frequent in Benjamin's speech. It is interesting to note, however, that in this age range Fabienne and Benjamin produce not a single instance of a finite lexical verb preceding the negation marker pas (even though Philippe's utterances around this time are overwhelmingly of this type: 41 out of a total of 72 verb-containing utterances, 2;01;19 - 2;02;10). Fabienne and Benjamin's data are reproduced in Table 2: Table 2.

Fabienne and Benjamin's utterances (N = total number of verbs, including all auxiliaries) /xas-Vlex [-Fin]

pas-V\ex [+Fin]

Fabienne (N= 59)

2

2

-

Benjamin (7^=155)

19

5

-

Vlex[+Fin]-/xs

Given this (very limited) survey of child French data, what we can say about the longitudinal development of the placement of lexical verbs in negative constructions in French is that the child starts out with root infinitives and (occasionally) finite verbs following the negator pas, and that the adult French pattern, in which pas follows the finite lexical verb, is attained later. Auxiliary verbs are a separate story; they are virtually consistently finite and precede pas.

4. Child French and historical English compared The constant factor in the development of English verb placement in main clauses appears to be that auxiliary verbs are always finite and always find themselves in a position in the "inflectional domain" of the sentence, i.e. in the functional projections associated with tense and agreement; thematic verbs, on the other hand, initially undergo overt movement as finite verbs to the inflectional domain, and subsequently cease to move overtly, either remaining finite or surfacing as

16

Frits Beukema

infinitives. In the acquisitional development of French a remarkably similar pattern was discovered: children consistently front auxiliaries, which are always finite, just as in English; thematic verbs start out to the right of negation, either as "root infinitives" or (for a short period of time) as finite forms; and subsequently, finite thematic verbs take on the properties of finite verbs in the adult language. The comparison is schematised in Table 3, where by the "inflectional domain" we understand the entire extended projection of the verb (i.e. the AgrS, Τ and AgrO projections): Table 3. Historical English and child French (P-I = position in inflectional domain; P-V = base position)

histoiy of English child French

A +Aux +Fin P-I -Aux +Fin P-I

Β +Aux +Fin P-I -Aux +Fin P-V

C +Aux +Fin P-I -Aux -Fin P-I

eats not cannot eat

not eats cannot eat

does not eat cannot eat

mange pas peut pas manger

pas mange peut pas manger

pas manger peut pas manger

The diachronic development of English and the French acquisitional sequence, though each other's inverses, are highly parallel: (11)

a. b.

Diachronic development of English: A —> A/B/C —> C Acquisitional development of French Vlex placement: C ->A/B/C —»A

Patterns A and C are at the extremes of the developmental sequence in both cases. Β and C co-exist for a certain period of time, and are fairly neatly set apart from the A pattern: in child French, Β and C are both gradually superseded by A (though all three patterns co-exist for some time); in English, Β and C arise in tandem in the fifteenth century, at a point at which A's demise starts (though, again, there is a transition period in which all three options are attested). Pattern Β is snort-lived both in the history of English and in the acquisition of French.

Development of sentential negation in a GB perspective

17

5. The diachrony of English negation: analysis We need the comparison with French child language to make it clear that pattern Β in the historical development of English negation is not just an idiosyncratic trait of some authors (mainly Shakespeare and Jonson) or an insignificant artifact of poetic or emphatic language, but that it is a real pattern in language development, both diachronically (as English shows) and acquisitionally (as the French data indicate). In the remainder of this paper I will no longer be concerned with the French acquisition data (although I will occasionally make reference to the French facts); my focus will be on the diachronic development of English negation.

5.1. The structure of the clause In order to provide an account of the historical development of negation in English a model of clause structure must be sketched first. This is given in (12) and is built on the idea that inflectional features are represented as functional heads which project full phrasal categories. Inflectional projections are always associated with lexical projections, which they dominate. Consider the order of the more central functional categories in (12): (12)

C - AgrS - Τ - AgrO - V

The order presented in (12) incorporates the "split Infi" idea. In the early stages of English, verbs move to get associated with the morphological features associated with the heads of functional projections. One of the diagnostics for movement of the verb is its position vis-a-vis negation, adverbs and clitics. Suppose that a verb moves from the V-position in (12) to the very left, to C, as in an interrogative structure. The Head Movement Constraint requires it to have moved through every intervening head on its way up. Such a movement would in itself not reveal anything about the functional positions which the verb has moved through, however; only the interaction of verb order and particular adverb positions or negation would reveal the functional position of the verb. It is therefore necessary to first identify the nature and position of negation and incorporate this in (12). As Rowlett (1994: 6) has observed, Jespersen's Negative Cycle can be provided

18

Frits Beukema

with a structural framework if we accept Pollock's (1989) proposal that there is a separate functional category NegP, which harbours the sentential negative. Various alternatives to a NegP analysis may be envisaged, one of which is to see the sentence negator as a VP adjunct, as illustrated in (13): (13)

[ w not [w ... ]]

However, Kayne's recent monograph on the "antisymmetry of syntax" (Kayne 1994) provides arguments to dismiss the VP-adjunct view. Kayne argues that specifiers, for example subject positions, are adjuncts, which entails that the X' level is superfluous. He also argues that more than two identical XPs over each other are excluded, as shown in (14): (14)

•[vp/KrftvpNPtvp...]]]

Thus, NegP is the only available option if we follow Kayne's theoretical argument. A further argument for the adoption of NegP as the locus of sentential negation can be derived from the fact that negation is a typical "thematic intervener", blocking head movement of the verb across it. Roberts (1993: 38), basing himself on Rizzi (1990), discusses this phenomenon in the context of Long Head Movement (LHM) of a non-finite verb to C across Agr in the Romance and Slavic languages. LHM occurs where a non-finite verb "topicalises" over a finite verb: the finite verb sum in (15) is in Agr, and the non-finite participial verb proöel is moved to C across Agr, in apparent violation of the Head Movement Constraint. Consider the Bulgarian sentence in (15): (15)

ProCel sum knigata Read (I)-have book-the Ί have read the book'

Such a long movement is blocked by negation, as shown by the ungrammatical sentence in (16): (16)

*Proöel ne statt knigata Ί have not read the book'

Development of sentential negation in a GB perspective

19

The ungrammaticality can be explained on the assumption that the non-finite verb pmCel cannot skip the head of NegP if Neg is treated as an A'-head, thus causing a case of "improper binding" in (16). A proposal treating sentential negation as an operator-like head of a functional category NegP would capture the LHM facts in a natural way. A more practical argument for the adoption of NegP would be that languages with periphrastic sentence negation can easily be accommodated in an approach featuring a NegP. Let us from now on assume that the presence of NegP in the functional inventory is well-established. With Belletti (1994) we propose the structure in (17): (17)

C- A g r S - N e g ]-T-AgiO-V

To illustrate the interaction of V-movement and negation Belletti cites the wellknown contrast between French and Italian reproduced in (18): (18)

a. b. c.

Gianni ha deciso [di non tornare mai\ *Jean a decidi [de ne rentrer jamais] Jean a decide [de ne jamais rentrer] 'John has decided never to return'

The French counterpart of Italian (18a) is ungrammatical: only (18c) is grammatical in French. Assume that the structure of the NegP in (17) is as in (19). The Italian verb picks up infinitival morphology in T, moves to the head of NegP where the clitic head non left-adjoins to it, and the V-complex moves to AgrS. (Another option is to say that there is a clitic chain and a V-chain with the same target, AgrS; then the V may have skipped Neg.) In French the infinitival verb moves no further than T, and only the clitic Neg head ne left-adjoins to AgrS. (19)

NegP Spec mai/jamais

Neg' \ Neg nonJne

20

Frits Beukema

Thus the difference in the position of the negative polarity item mai/jamais indicates the functional position taken up by the verb.

5.2. A Minimalist reinterpretation We will consider the approach to V-movement outlined in the previous section to be well-established. Let us now see how V-movement fares in the Minimalist Framework (Chomsky 1993), the most recent version of Government and Binding Theory. In the Minimalist Framework the analysis sketched above is maintained in outline, although the implementation differs. There is now a feature matching mechanism that matches the features associated with the inflectional morphology of lexical categories with the features represented in the functional heads. If we follow Chomsky's theory, inflectional morphology is present on the lexical head and it must be matched by the feature of the functional category. Checking matching features is subject to locality, and results in triggering movement of the lexical category to the functional domain. Moving the lexical element to a functional position and checking that the relevant features match causes the inflected element to be licensed. Matching features are eliminated as soon as they are checked. Tense and Agreement are the inflectional features operating in Vmovement while Case is the relevant feature in NP-movement. Obviously, Vfeatures on functional heads are relevant for checking features of heads while Nfeatures on functional heads are instrumental in moving XPs to specifiers. Features checked in overt syntax are strong while features checked covertly are weak. Ultimately, the strong/weak distinction is what is responsible for parametric variation among languages; this distinction squarely locates parametric variation in the domain of functional categories. Now that we have established the general framework, let us return to the analysis of the development of English negation.

5.3. Auxiliaries One robustly constant factor in the schema in Table 3 is the placement of auxiliary verbs: in all three patterns, they consistently find themselves in a position in the

Development of sentential negation in a GB perspective

21

inflectional domain. Viewed from a Minimalist perspective, this cannot be due to the fact that the V-feature of [+Fin] Τ is strong, for otherwise we would not expect thematic verbs to behave differently. Nor can the explanation lie in a putative difference in the strength of Agr-features in [+Aux] and [-Aux] verbs. Theoretically, the problem with such an approach would be that it localises a parametric difference outside the domain of functional heads (auxiliaries being V-heads, not F-heads). And while a "strength of Agr" approach might at first sight gain support from the fact that the verbs moving to the left of pas in child French in stage Β seem to be those verbs that (in Verrips - Weissenborn's 1992 terms) have a "suppletive agreement paradigm" (e.g. avoir, etre, aller), it is to be rejected as a general approach on account of the fact that it does not generalise to English. In English, there evidently is no correlation between V-placement and the morphological richness of the agreement paradigm, either synchronically or diachronically (cf. also Schäufele 1994: "the direction of V-AGR merger in English... has historically had no connection with the language's relativerichnessor poverty of subject agreement marking"). I would like to argue that the only way to arrive at a proper understanding of the obligatorily overt fronting of auxiliaries throughout all diachronic stages is to assume that since auxiliaries are invisible to LF processes, they must move prior to Spell-Out, i.e. they must always move in the overt syntax. By having this perspective eliminate other options, further evidence is provided for Chomsky's (1993) approach to auxiliaries.

5.4. Thematic verbs I argue that the A pattern in Table 3 is properly characterised by the postulation of a strong verb-attracting feature in Infi, which causes all verbs, including lexical verbs, to vacate their base positions prior to Spell-Out. Option B, on the other hand, instantiates a system with a weak V-feature in Infi: finite main verbs move covertly (while auxiliaries of course continue to undergo overt movement; cf. section 5.3). The diachronic development of English verb placement thus witnesses a weakening of Infl's V-feature; in the acquisitional sequence of French, by contrast, this feature starts out as weak (plausibly the initial setting of this feature) and strengthens in the course of time.

22

Frits Beukema

The most challenging pattern is option C, instantiated by c/o-support in English (and by root infinitives in child language; the account of English stage C offered here differs from that offered for stage C in child language). This pattern has been accommodated by postulating a head position Neg in the inflectional domain which harbours the Neg-operator. To this A'-head position we then attribute the property of blocking movement of θ-assigning verbs to or through it. Due to the postulated θ-opacity of this F-head at stage C, all thematic verbs are prevented from raising to Infi, not just in overt syntax, but at LF as well. As a result, thematic verbs at stage C surface without finite inflectional morphology. Stage C thus vindicates Pollock's (1989) original insight that the θ-opacity of a particular head in the functional domain is responsible for blocking the raising of θ-assigning verbs across negation. We diverge from Pollock's view, however, in identifying this blocking head as Neg (and thereby strengthen the case for a functional projection housing negation) and not as Agr: the blocking head certainly could not be AgrS, since it is not the case that lexical verbs are denied finite inflectional features in all finite sentence types: it is only negative sentences that block LF-raising of lexical verbs.

5.5. Why pattern Β is short-lived A final question to be addressed is why pattern B, both in child French and in the history of English, is so short-lived (to recapitulate: in French acquisition, pattern Β persists for only roughly one month; in the history of English, pattern Β was moderately common in the sixteenth and seventeenth centuries only). A possible approach to this problem would be as follows. Suppose we interpret the property "θ-opaque" as "weak", and "θ-transparent" as "strong", along Minimalist lines. We can then reinterpret the "θ-opacity" parameter in a way similar to the "strength of Infi" parameter: it is the strength or weakness of F-heads that matters in both cases. The historical development of English negation can then be represented as in Table 4.

Development of sentential negation in a GB perspective

23

Table 4. The historical development of English negation Pattern A

Pattern Β

Pattern C

Infi

strong

weak

weak

Neg

strong

strong

weak

It can now be understood why pattern Β is bound to collapse: it is an inherently unstable pattern in the sense that it has different values for strength of Infi and Neg.

6.

Summary

We have argued that two parameters are responsible for the variation in the wordorder patterns illustrated in Table 3: (20)

a. b.

the strength of Infl's verb-attracting feature the θ-opacity/transparency of the F-head harbouring negation

The strength of Infl's verb-attracting feature dwindles in the early fifteenth century, as a result of which LF-visible (i.e. lexical) verbs no longer move to Infi in overt syntax, and hence no longer end up to the left of the negation marker. This is confirmed by Kroch's (1989) observation that adverbs like always and probably also started to shift from postverbal to preverbal position at the beginning of the fifteenth century. Auxiliary verbs (which are not θ-assigners, and are therefore invisible to LF processes) are oblivious to the "strength of Infi" parameter, and continue to front indiscriminately (to the left both of adverbs like probably and of the negation marker) in overt syntax (since they are invisible to LF processes and hence cannot postpone movement to LF). Lexical verbs move overtly if and only if Infl's V-feature is strong, and stay in their base positions otherwise, carrying finite inflectional morphology (checked by LF movement to Infi) if the F-head harbouring negation is θ-transparent (so that LF raising to Infi is legitimate), and surfacing with non-finite morphology if the F-head is θ-opaque. The diachrony of English shows a development from strong Infi and Θtransparent F (cf. A), via a transitional stage in which Infl's strength dwindles and

24

Frits Beukema

F is variably θ-transparent (B) or θ-opaque (C), to a final state in which F's Θopacity wins out. With θ-opaque interpreted as weak and θ-transparent as strong, the fact that pattern Β is very short-lived can be understood, given the fact that this pattern is characterised by opposite strength values of the Infi and Neg parameters. Coming from a double-strong system (Old and Early Middle English), English simultaneously develops two new patterns, a double-weak one and a "mixed" system. Of these two alternatives, the former wins out.

Notes 1.

This paper is part of a research project being carried out by Hans Bennis, Marcel den Dikken and the present author on the development (both diachronic and acquisitional) of verb placement. The first results of this project are reported in Bennis - Beukema - den Dikken (1997). Thanks are due to Marcel den Dikken and Ton van der Wouden for their comments on an earlier version of this article.

2.

In Jespersen's accounts the stages illustrated by (2e) and (2e') are separate. They are evidently separate in that (2e') is later than (2e), but it is equally obvious that (2e') could not have developed if (2e) had not been there first. Here I attach greater weight to this causality than to the difference in temporal sequencing.

Development of sentential negation in a GB perspective

25

References Andrew, S.O. 1940

Syntax and style in Old English Cambridge: Cambridge University Press.

Belletti, Adriana 1994 "Verb positions: Evidence from Italian", in: David Lightfoot Norbert Hornstein (eds.), 19-40. Bennis, Hans - Frits Beukema - Marcel den Dikken 1997 "Getting verb movement", Linguistics 35: 1003-1028. Chomsky, Noam 1993 "Bare phrase structure", in: Gert Webelhuth (ed.), 383^439. Denison, David 1993 English historical syntax. London: Longman. Fasold, Ralph - Deborah Schiffrin (eds.) 1989 Language change and variation. Amsterdam: John Benjamins. Ferdinand, Astrid 1994 "Semantic verb types and the acquisition of verb movement in French", in: J. Fuller - H. Han - D. Parkinson (eds.), Cornell working papers in linguistics: ESCOL 1994 proceedings. Haegeman, Liliane 1995 The syntax of negation. Cambridge: Cambridge University Press. Jespersen, Otto 1917 "Negation in English and other languages", Det Kgl. Danske Videnskabernes Selskab. Historisk-filologiske Meddelelser 1: 1— 151. Copenhagen. [1962] [Reprinted in Selected writings of Otto Jespersen. London: George Allen and Unwin.] 1924 The philosophy of grammar. London: George Allen and Unwin. Kayne, Richard 1994 The antisymmetry of syntax. Cambridge, Mass.: MIT Press. Kroch, Anthony 1989 "Function and grammar in the history of English: Periphrastic dd\ in: Ralph Fasold - Deborah Schiffrin (eds.), 133-172.

26

Frits Beukema

Lightfoot, David - Norbert Hornstein (eds.) 1994 Verb movement. Cambridge: Cambridge University Press. Meisel, Jürgen (ed.) 1992 The acquisition of verb placement. Dordrecht: Kluwer. OED = The Oxford English Dictionary 1989 Prepared by J.A. Simpson - E.S.C. Weiner. (2nd edition.) 20 Volumes. Oxford: Clarendon Press. Partridge, A.C. 1953 Studies in the syntax of Ben Jonson 's plays. Cambridge: Bowes and Bowes. Pollock, Jean-Yves 1989 "Verb movement, UG and the structure of IP", Linguistic Inquiry 20: 365-424. Rissanen, Matti - Ossi Ihalainen - Terttu Nevalainen - Irma Tavitsainen (eds.) 1993 History of Englishes: New methods and interpretations in historical linguistics. Berlin - New York: Mouton de Gruyter. Rizzi, Luigi 1990 Relativized minimality. Cambridge, Mass.: MIT Press. Roberts, Ian 1993 Verbs and diachronic syntax. Dordrecht: Kluwer. Rowlett, Paul 1994 "The negative cycle, negative concord and the nature of spechead agreement", University of Salford Working Papers in Language and Linguistics 7. Schäufele, Steven 1994 "Do as I do, not as I say: The history of V-AGR merger, VPnegation and ί/o-support in English, 1350-1750". [Abstract, 3rd Diachronic Generative Syntax Conference, Amsterdam, Spring 1994.] Söderlind, Johannes 1951 Verb syntax in John Dryden's prose. I. Uppsala: A.-B. Lundequistska. Tieken-Boon van Ostade, Ingrid 1987 The auxiliary do in eighteenth-century English: A socio-historical linguistic approach. Dordrecht: Foris Publications.

Development of sentential negation in a GB perspective

27

Tottie, Gunnel 1991 Negation in speech and writing. San Diego: Academic Press. Traugott, Elizabeth Closs 1972 A history of English syntax. New York: Holt, Rinehart and Winston. Ukaji, Masatomo 1993 " Ί not say': Bridge phenomenon in syntactic change", in: Matti Rissanen - Ossi Ihalainen - Terttu Nevalainen - Irma Taavitsainen (eds.), 453-462. Verrips, Maaike - Jürgen Weissenborn 1992 "Routes to verb placement in early German and French: The independence of finiteness and agreement", in: Jürgen Meisel (ed.), 283-331. Visser, Fredericus Th. 1969-1973 An historical syntax of the English language. Leiden: E.J. Brill. Webelhuth, Gert (ed.) 1993 Government and binding theory and the minimalist program: Principles and parameters in syntactic theory. Oxford: Blackwell. Wouden, Ton van der 1994 Negative contexts: Collocation, polarity and multiple negation. London - New York: Routledge.

English negation from an interactional perspective1 Jenny Cheshire

1. Introduction This paper discusses the use of never as a negative marker in Present-Day English. I will argue that its importance has been overlooked because of the convention in linguistics of analysing forms in isolated sentences rather than in their conversational contexts. This practice forces us to rely on our intuitions about the role of a given form in the structure of a language, so that weriskunwittingly describing the internalised prescriptive norms concerning its use rather than the way in which it really functions. In section 2 I briefly set out the history of English never and in section 3 I describe the problems that linguists have encountered in its analysis. I then suggest the insights that can be gained if we consider never in the conversational contexts in which it is used.

2. A brief history of English never Jespersen (1917: 4) sets out a common pattern for the evolution of negative expressions, which is sometimes referred to as the Negative Cycle: The history of negative expressions in various languages makes us witness the following curious fluctuation. The original negative is first weakened, then found insufficient and therefore strengthened, generally through some additional word, and this in its turn may be felt as the negative proper and may then in due course of time be subject to the same development as the original word. The history of the expression of negation in English fits into this pattern. Old English ne was too weak to survive unaccompanied, and by the Middle English period had been strengthened by the addition of not. Thus the earlier ic ne secge Ί do not say', became in Middle English ic ne seye not, with "embracing" negation

30

Jenny Cheshire

(see Horn 1989: 454). Horn points out that amongst the world's languages negation is typically strengthened by minimisers (like the French pas 'step') or by indefinites. Speakers of English seem to prefer the latter way of reinforcing a negative: our present-day not derives from nowiht/nawiht 'nothing', itself from no/na + wiht 'not ever anything'. This "negative existential pronoun" (Horn 1989: 455) was reanalysed as a simple adverb, presumably facilitated by contexts allowing both interpretations (such as I sowed nought!I sowed not) and then spreading into contexts where there were no semantic indefinites (Bossuyt 1988: 311-312). From this point on most histories of English focus on not. The continuing development of the use of not conforms to Jespersen's Neg First principle, "the natural tendency ... for the sake of clearness, to place the negative first, or at any rate as soon as possible" (Jespersen 1917: 5). This was achieved by the introduction of ί/o-support, already attested in the fifteenth century and becoming standard by the seventeenth century, so that I seye not became I do not say. With modal verbs and auxiliaries not remains after the verb (/ may not, cannot and so on) but in ordinaiy main verb clauses the negative marker is before the main verb once again, as it was when the negative was expressed by the simple ne. The cycle has continued: in its turn not has now become phonetically weakened to the clitic -n % first in spoken English and now, in Present-Day English, in some written texts too. Not became semantically weakened also, no longer interpreted as referring to universal time, but used as a simple negator to refer to a specific moment or period of past time. We can see from early texts how writers - and presumably speakers too - turned once more to the universal temporal negator to reinforce the expression of negation: examples (1) and (2) contain naeure and neuer, both from ne cefre, via ncefre, literally 'not ever' or 'never', referring here not to all time but to past time: (1)

di moder was an hore, for nuste heo naeure dene your mother was a whore because knew-not she never the man öat de streonde hire on man that you begot her on 'Your mother was a whore because she didn't know the man who begot you' (La3amon's Brut, 1205)

English negationfroman interactional perspective (2)

31

He asked what that was and his wiff saide she wost neuer he asked what that was and his wife said she knew never 'He asked what that was and his wife said that she didn't know' {The Book of The Knight of La Tour-Landry, 1450)

In Present-Day English never continues to be used as a negative marker, sometimes with the possibility of the literal meaning of universal temporal negation, as in (3), where the utterance could be interpreted as 'on all occasions that Sally is offered meat, she will refuse it', and sometimes without universal temporal reference, as in (4) to (6), where the utterance refers to the past (unless otherwise specified, all examples in this paper are taken from my own recordings of spontaneous conversation: names given to speakers, such as "Benny" and "Nobby" in example (6) are pseudonyms): (3) (4) (5) (6)

Sally's a vegetarian ...she never eats meat You Ή never catch that train tonight (Quirk et al. 1985: 601) I never went to school today Benny:... we all went up there andjumped on him Nobby: you never ... you you hit him with a stick and then booted him

If the process of standardisation had not intervened we could expect the cycle to have continued, with never eventually replacing not as the conventional marker of negation, just as not eventually replaced ne. Never, in its turn, would have become phonetically reduced through rapid speech processes, and would have lost its expressive force through frequent use. The spelling in example (1) suggests that the intervocalic consonant in forms resulting from ne cefre may already have been lost by the thirteenth century; and the spelling ne 'er in the representation of speech in nineteenth century novels indicates how never must have been pronounced at that time. With standardisation, however, came the desire for the speech of the educated to resemble the form of the written language. During the nineteenth century short forms such as howe 'er, e 'er - and presumably also ne 'er - were branded as vulgar by schoolmasters (Jespersen (1905 [1982]: 219), presumably because they did not preserve the "v" of the written form and perhaps also because they were used by social groups from whom speakers who considered themselves to be educated

32

Jenny Cheshire

wished to disassociate themselves. Today the shortened form of never has disappeared from most varieties of English, surviving only in some rural dialects. Not surprisingly, prescriptivists objected not only to the reduced form of never but also to the restriction of its meaningfromuniversal temporal negation to simple negation. Although there is no record until the mid-eighteenth century of a prescription involving never (see Sundby - Bjorge - Haugland 1991), guides to good usage published during the twentieth century virtually always comment disapprovingly on the use of a single form, never, with two meanings that are apparently incompatible (referring to universal time, on the one hand, and to one specific occasion, on the other hand). The comments range from the severely prescriptive, as in Wood (1981), to the merely precautionary, as in Fowler (1965) or Collins (1981), and they tend to become more severe as time goes on: Wood (1981): "Never means 'not ever, on no occasion'. It is common to hear sentences such as I never saw you at the party. It is, however, incorrect to use never when referring to one occasion. Never can only be used in continuous context: Bob: I didn 't see you at the party, Jim Jim: I've never been to any of Sue's parties." Fowler (1965): "this use of never, however illogical, is idiomatic, at least colloquially". Collins (1981): "In good usage, never is not used with simple past tenses to mean not". The cycle has been interrupted, then, by the processes of prescription and codification that are part of the standardisation of language. The phonetic reduction of never has been reversed and its semantic restriction tends to be frowned on. Nevertheless, never does still occur with simple past tenses with the meaning 'not', both in non-standard English and in what is usually considered to be Standard English - in other words, in published written prose and in the speech of people who consider themselves to be educated (see Trudgill 1984). Examples (5) and (6) were uttered by working-class adolescents who used many features of non-standard

English negationfroman interactional perspective

33

English syntax, but examples (7) and (8) are from stretches of discourse that otherwise entirely conform to the norms of Standard English: (7)

(8)

He got ready to spring downfromon high right among the spears of the goblins..... But he never leaped (Tolkien, J. 1978. The Hobbit. 4th edition. Allen and Unwin, p. 95) Kay never went to Delft on Tuesday ... she stayed with our friends in Rotterdam (American university professor)

Prescriptivists tend not to comment on features that are in frequent use by the favoured sections of society, so the relative lack of comments in the eighteenth and nineteenth centuries presumably indicates that educated speakers used never more frequently in this way then than they do now. We do not know the extent to which educated speakers today use never to mean simply 'not', though this of course is an empirical question which could be answered by analysing a data base such as the Lund Corpus. Cheshire (1997) discusses the acceptability of never in a range of linguistic contexts: it seems that educated speakers of British English find examples such as (7) more acceptable than examples such as (8), perhaps because in (7) the time reference is to an indeterminate period when the protagonist could have jumped, but did not (whereas in (9) Tuesday clearly specifies a precise day to which never refers). Examples such as (6), where never stands alone, were found to be the least acceptable. This use of never, however, is veryfrequentin the nonstandardised varieties of English: it is reported in all the British urban centres surveyed by Cheshire, Edwards and Whittle (1989), and it is used in many Englishbased Creoles as a simple negator referring to past time (Holm 1988). For educated speakers, however, expressions such as I never or he never seem to have become a social marker of groups with whom they do not want to be associated. Perhaps the most accurate statement, then, concerning the use of never in PresentDay English is that its development as an emphatic negative marker has been slowed down in Standard English, so that this use is probably more widespread amongst speakers of the non-standard varieties. This way of expressing negation in English has not, however, been given proper recognition by linguists, as we will see in the following section.

34

Jenny Cheshire

3. Linguistic descriptions of never It is usual to distinguish between two types of non-affixal sentence negation in English: firstly, negation with not or -n 't; and secondly, negation with the negative words never, neither, nobody, no, none, nor, nothing and nowhere. Tottie (1991), for example, terms thefirsttype "iVctf-negation" and the second type "iVonegation". Quirk et al. (1985: 782) give a list of the negative words together with their corresponding non-assertive forms, pointing out that there are two negative equivalents for a positive sentence containing an assertive form: thus We've had some lunch has the two negative forms We haven't had any lunch and We've had no lunch (Quirk et al. 1985: 782). In the same way, these authors tell us, He sometimes visits us has the two negative forms He doesn't ever visit us and He never visits us. The inclusion of never in the TVo-negation category of negative words can be traced back to Klima's classic paper on negation in English (Klima 1964). Klima assumed that the same peculiarities of occurrence characterise ever as crry: namely, ever occurs in negative, interrogative and restrictive sentences whose corresponding simple sentences do not permit its occurrence (Klima 1964: 280). Thus *We've had any lunch and *He ever visits us are both unacceptable. If we move beyond our intuitions, however, and examine the way that speakers actually use these forms, it becomes clear that ever occurs veiy rarely in negative utterances. For example, in the 53,000 word corpus of adolescent speech that I recorded in adventure playgrounds in Reading during the late 1970s there are 73 occcurrences of never, but no occurrences whatsoever of not plus ever. The adolescents used a number of non-standard linguistic features (analysed in Cheshire 1982), so it might be reasonable to wonder whether their linguistic behaviour is typical of other speakers of English. However, my figures correspond almost exactly with the results of Tottie's analysis of50,000 words drawn from the London-Lund Corpus of educated adult speech. Tottie found only one token of not plus ever, but 78 tokens of never. She also found that there were more tokens of never in her corpus than of the other negative words with which never is conventionally classed: she writes that she had originally intended to analyse the class of negative words together, but that there were so many occurrences of never that she was obliged to remove them to a separate data sample in order to avoid skewing the analysis. She was led to pose

English negation from an interactional perspective

35

two questions that she hoped future research would address: why does never occur so frequently; and why does the not ever variant occur so rarely? It seems to me that part of the answer, at least, is that never can be a marker of strong negation in English, as I outlined in the previous section. This means that it should be seen as a negative item in its own right rather than as equivalent to the words in the class of Afo-negation expressions with which it is conventionally classed. In fact, Quirk et al. (1985: 783) report a stylistic difference between never and the other negative words, saying that in all cases "except possibly with never''' the combination of not (-« V) and the non-assertive word is more colloquial and idiomatic than the negative variant: this stylistic difference, presumably, is a reflection of the functional difference between never and the class of M>-words. Our failure to recognise the status of never as an independent marker of negation has left unresolved questions in other analyses, too. For example, Labov (1973) in his analysis of Hawaiian Creole English was puzzled by the occurrence of never in utterances such as (9) and (10): (9) (10)

He never like throw first 'he didn't like to throw first' And that thing was coming and something black on top the horse never have head '... something black on top of the horse didn't have a head'

Influenced by Klima's analysis, Labov assumed that never was "the same" as not ever and that it must therefore refer to "universal" time, with a gloss 'not on any occasion'. This assumption led him to discuss at some length an apparently insoluble semantic problem, that of identfying the kind of cognitive process that could connect never as an indefinite item, with universal reference, to never as it is used in (9) and (10), where it refers to a single event that took place in the past. As we saw in the previous section, however, the use of never with a time reference that is less than universal is not restricted to Hawaiian Creole English. We saw earlier that it occurs in "standard" and "non-standard" English alike. Nevertheless sociolinguists and dialectologists have failed to observe this widespread use, labelling the use of never to refer to a single occasion in the past as "non-standard" or "dialect" (see, for example, Cheshire 1982; Coupland 1988; Hughes - Trudgill 1979; Orton et al. 1963-1969). All of us, like Klima and Labov, seem to have

36

Jenny Cheshire

assumed that standard never corresponds to not ever and that it can be glossed as 'not on any occasion'. This assumption leads us then to classify other uses of never as non-standard or, in Labov's case, to conclude that although the limitation of never to a particular point in the past may be a possibility, albeit an "extraordinary" one, he would be reluctant to include it in "a general grammar of English" (Labov 1973: 59). This is obviously an unfortunate conclusion, given how widely this usage occurs. Since the use of never to refer to a past occasion is more than just a possibility, then, linguists need to decide how to analyse it. Should we assume that we are dealing here with polysemy? This would lead us to propose that there are two never forms in Present-Day English: one meaning 'not ever' and one simply meaning 'not', expressing emphatic negation. Or should we decide instead that Present-Day English has a single form never, whose meaning varies from referring to universal time to referring to a single point of time, depending on the context in which it is used? The question has important implications for variationist analyses, for the answer determines the forms that are considered to constitute the linguistic variable. For example, when I analysed the non-standard English of adolescents in Reading I considered only variation between never and didn 't, with reference to a single past occasion in both cases (Cheshire 1982). Tottie (1991), on the other hand, analysed only variation between never and not ever, disregarding, therefore, those tokens of never where it referred to a single occasion. Each of us performed a different analysis, and each of us failed to obtain a full picture of the way in which never functions in Present-Day English. I will attempt to provide a fuller picture in the following sections. My argument will be that it makes more sense to consider never as a single form, an emphatic negator, whose time reference is determined by the linguistic context in which it occurs.

4. Never in interaction It makes good sense to consider the function of never in conversational contexts rather than on the basis of our intuitions, for negation, in spoken language at least, often relates to a previous utterance. In order to understand how negation is expressed in a language, therefore, it seems obvious that it should be analysed within the context in which it occurs.

English negationfroman interactional perspective

37

In the examples that follow we see how speakers use negation to ensure that addressees have the same orientation to the topic as they have themselves, in terms either of shared background knowledge or of their personal stance. In (11), for example, Jacky corrects my presupposition that she has two parents, explaining in her negative clause why it is her mother who scolds her most. (11)

Jenny: Who is it who tells you off in your family... your mother or your father? Jacky: Well my mum 'cos I haven't got a dad now ...so it's my mum worse luck

Example (12) is from a conversation where Wendy and I were discussing indoor fireworks and, in particular, whether or not an indoor rocket was dangerous. Wendy's negative clause responds to the meaning that she infers from my oh yes, uttered with rising intonation on oh and a fall-rise on yes. The intonation suggests that I am not convinced by Wendy's previous account of the safety of the rocket, as is shown by her response, which aims to reassure me: (12)

—>

Wendy: but you know you just put it in the bo in the bottle and em... you know it kind of it... only it goes round the room Jenny: oh yes Wendy: it don't hit the ceiling it just goes round and round Jenny: goodjob Wendy: you have to mind your head mind you 'cos it goes up and round and round

These two examples show speakers using not to form a negative clause in order to ensure that their interlocutor's orientation to the topic is in tune with their own, by negating an implicit inference that they draw from their interlocutor's contribution, and that their negation then makes explicit (in (12), that Jacky has two parents; in (13), that an indoor rocket could shoot up into the air and hit the ceiling). Speakers use never in exactly the same way. Example (13) is from a conversation during which Nobby and his friend had been telling me about three other boys that they "went around" with sometimes, even though they disliked them. I was trying to obtain details of this "going around", and asked a question which presupposes that

38

Jenny Cheshire

a group of boys from another playground, at Shinfield, sometimes comes to Nobby's part of town. Nobby's negative clause negates this presupposition: (13) —>

Jenny: What about when there's a real bigfight like with the Shinfield lot or something like that... would you sort ofjoin in on their side? Nobby: Shinfield never come down here ... they're scared of us

In (14) we see my earlier example (5) in its wider discourse context. In the same way as Nobby does in (13), Marie corrects my presupposition - this time that she had been to school that day: (14)

—>

Marie: I had to do a lot of banging and my η my hands as you can see took quite a long time ... about three or four hours just to do it Jenny: Was that at school you made that? Marie: No I never went to school today

In (13) never has the meaning of universal temporal quantification, referring to all possible occasions when I might have expected the Shinfield lot to come down to the playground where we were talking. In (14), on the other hand, never has a more restricted time reference, to the day in question: this is the "problematic" use of never discussed earlier. I will discuss this use in more detail below, for the time being, I want to stress that negation generally, whether expressed by not or by never, has an interactional role in ensuring the coherence of the emerging discourse. In other words, it can link the current turn to the previous one, by negating a presupposition that has just been expressed, whilst simultaneously ensuring that the interlocutors have a shared orientation to the topic that they are pursuing, so that the subsequent turn is felicitous. This cohesive function is perhaps one reason why negation occurs more frequently in spoken discourse than in written discourse: Tottie (1991) found twice as much negation in her sample of spoken English as in the sample of written English. We can relate this discourse function of negation to the further function of creating interpersonal involvement in discourse (Chafe 1982, 1986), for Tottie also found that negation tended to occur in speaker turns that clearly testified to the co-operative effort that is necessary for conversation to be successful, such as in tag questions seeking corroborationfrominterlocutors (Tottie 1991:43). In her sample, negation also showed a correlation with mental state verbs

English negationfroman interactional perspective

39

such as know or think which, as Chafe (1982) has pointed out, indicate the involvement of speakers in what they are saying and occur more frequently in spoken English than in written English. The concept of involvement is very relevant to an understanding of why the universal temporal quantifier has been a favoured negative strategy throughout the history of English. Although different writers use the term "involvement" in somewhat different ways it stems, in all cases, from the assumption that spoken discourse is a collaborative production, with speakers and addressees working together to produce meaning as the discourse unfolds (see Chafe 1982; Gumperz 1982; Tannen 1989). Quantifiers have an important role in securing interpersonal involvement since they require the addressee to determine their scope and their precise interpretation. In (15), for example, the universal quantifier all is in construction with his hand in the first clause and head in the second; it is unlikely, however, that all refers to Nobby's brother's entire hand and still less likely that it refers to his entire head. Instead, addressees determine the extent of the quantification on the basis of their knowledge of the world: by using all Nobby alerts his addressees to the need to interpret bashed up as referring to the widest possible extent of his brother's hand, given the context: (15)

Nobby: My brother had all his hand bashed up ... all his head was bleeding.

All therefore functions as a very effective intensifying device, as has been pointed out by Labov (1984) and Cheshire (1989). We can think of quantifiers in terms of scalar implicatures - as members of an implicational set, such that an utterance containing one item from the scalar set entails the items lower down on the scale. Just as excellent entails 'good', so that this is an excellent meal entails 'this is a good meal', so all entails 'most', 'many', or 'some'. Never can similarly be seen as the high point on a scale containing never, often, sometimes and once. Thus when speakers use never, they invite the addressee to fix as wide a time reference as is possible in that context. Sometimes the time reference will be all possible occasions, as in (13) (Shinfieldnever come down here), where my previous turn had specified the time reference as indefinite, with what about ..., and where the immediate context in which Nobby uses never specifies all possible occasions through the use of the "timeless" present tense. In other utterances the tense of the

40

Jenny Cheshire

verb or the presence of a time adverbial specifies a time reference that cannot be to all possible occasions: thus in (16) the verb is in the present perfect tense, indicating that the period in question is past time up to and including the present: (16)

... you've never read it have you? (Svartvik - Quirk 1980: 626)

In exactly the same way, the tense of the verb and the adverbial today specify the time reference of never in example (5) (I never went to school today). In the case of example (6) (now (17)), when never occurs alone, we need to look beyond the current turn to the previous one, considering its function across speaker turns: (17)

Benny: we all went ψ there andjumped on him Nobby: you never ... you you hit him with a stick then booted him and then I had to do the rest

Nobby'syou never follows onfrom- and negates - Benny's we alljumped on him, with the time reference of never specified by the tense of the verbs in Benny's utterance. We see here, then, how syntax can be constructed jointly, across speaker turns (see Jeanneret 1992); we also see how important it is to consider a form in its conversational context in order to understand its syntactic function. In (17) the interlocutors also have to use their knowledge of the world in order to understand that the reference is to one specific past occasion (you "jump on" someone once only, of course, as the first stage in a fight). We can now recognise, then, that the different uses of never discussed in section 3 are essentially the same: in each case, interlocutors can fix the time to which the utterance refers by scanning the context in which never occurs, and by bringing into play their knowledge of the world. Sometimes the time reference can be to all possible occasions, in which case never can be said to be acting as the universal temporal negator; but sometimes the time reference is restricted by other linguistic forms in the utterance, or by the knowledge on which we draw in order to interpret the utterance. Variationist analyses that investigate the use of only one of the meanings of never, then, have given only a partial account of its functions. It is worth stressing that the uses that are commonly considered to be nonstandard are no different from the other uses of never. In each of the examples just

English negationfroman interactional perspective

41

discussed an alternative clause is possible using not, as can be seen by considering the pairs of sentences in (18). The difference is simply that using never invites the addressee to fix the time reference of the quantifier and it can be said, therefore, to actively involve them in constructing the meaning of the emerging discourse: (18)

a. b. c. d.

Shinfield never come down here / Shinfield don't come down here ... you've never read it have you? / you haven't read it have you? I never went to school today /1 didn't go to school today You never... / You didn't...

If, as I have argued, quantifiers can facilitate the creation of interpersonal involvement, it is hardly surprising if speakers turn to the universal temporal negator as a way of reinforcing a weakened negative. In time, as the sense of quantification becomes lost and as phonetic changes make the form semantical ly opaque (as has happened with English not) the one-time quantifier becomes a simple non-emphatic negative marker, no longer recognised as a quantifier and no longer, therefore, actively involving the addressee in fixing the time reference of the form.2 Never does not seem to have reached this stage yet, and it may well never become fully opaque: we saw in section 2 that the contracted form ne 'er is rarely used today, so that ever remains as part of its form, as a reminder of its etymology (see further Cheshire 1997). In my recordings of conversational English there are many stretches of discourse where never co-occurs with other features that express interpersonal involvement, suggesting that it can be one way, amongst others, of involving the interlocutor. In the extract below, for example, never co-occurs with a cluster of addressee-oriented forms: several occurrences of see, the deictic that accompanying a gesture, and the quantifier all (the addressee-oriented forms are in roman type): (19)

Jacky: The other day ... when we was up Ridgeway ... when we was at primary school... she always used to get me in trouble ... and I used to hate that ... I didn't mind getting in trouble ... but her ... she kept on getting me in trouble... and one day I was sitting in class... and a student was reading us a story ...I wasn t listening anyway ... bid she kept on fiddling with my bracelet... and trying to pull it off me... and I went like that... see ... and she sent me outside the door ...but it was her see ... and

42

Jenny Cheshire when the teacher come ... Mr. Mayhews ...he come in and told me off... see ... and he blamed everything onto me... so I told him what happened and when he asked Wendy... Wendy said that she never ... and they all agreed with her see.

Note that Jacky' s use of never requires the addressee to refer back to the previous clause in order to fix its time reference, like Nobby's you never in (6) (later (17)). If never does function as a way of securing the involvement of the addressee, we would expect it to occur in conversational contexts where addressee-involvement is particularly important. It is no surprise, therefore, to find that never is frequently used in friendly arguments, as one of a cluster of addressee-oriented features which together can be interpreted as positive politeness devices, allowing the interlocutors to attend to each other's positive face as they disagree with each other. As an example of this consider (6) once more, this time in a still wider conversational context: the other addressee-oriented items are all, the second person pronouns, the address form mate and the intensifying overstatement half killed. The slang terms boot and bollocks can also be seen as positive politeness forms, as can the nonstandard verb form done: (20) —>

Benny: I went and grabbed him ...he went and told him and Mike and all our other mates ... and we all went up there andjumped on him Nobby: you never ... you you hit him with a stick then booted him and then I had to do the rest Colin: I kicked him in the bollocks Benny: I lacked him Nobby: I done the most to him mate I half killed him

Similarly, never frequently occurs in a cluster of addressee-oriented forms at the beginning of a narrative, when the speaker needs to secure the interest of the interlocutor in order to keep the floor for an extended turn. In (21) we see never used in the formulaic introduction of a narrative (I'll neverforget the time when...) as well as in the orientation section, together with multiple negation and the intensifier shit:

English negationfroman interactional perspective (21)

43

Jacky: I... I'll never forget the time when I went up to bed ...I heard a creaking sound... I was the only one in the house ...my sister was with my mum and my brother was out ...I went to bed early 'cos I never had nothing to do and I had no supper... and I heard a creaking upstairs and I was shit scared ...I wouldn 't stay ... and I had all the lights on I was shit scared.

Its use in the formulaic introduction of a narrative suggests that never can have a further role in the turn-taking system, engaging the involvement of the interlocutor and therefore allowing the current speaker to take an extended turn. It frequently occurs at locations for speaker change, when the current speaker is eliciting talk from others. In (22), for example, I was the fieldworker trying to elicit talk from a group of 13-year-old girls. My rather uninspired questions were not succeeding in eliciting more than veiy short utterances, but my utterance containing never, together with the quantifier ever in whatever, was at last followed by a longer sequence of talk, jointly constructed by two speakers: (22)

—>

Jenny: What's your favourite food? Wendy: Favourite food? Marie: That's easy... chips Jenny: Chips Wendy: Roast Jenny: Roast dinner Wendy: Yes andfor pudding gypsy tart Jenny: Whatever's that? I've never heard of that Marie: Ohl can't explain 'cos ... it's hard to explain isn't it? Wendy: Gypsy well it's pastry on the bottom it's sort of Jenny: yes Marie: coffee Wendy: coffee on the top Marie: yeah coffee on the top Wendy: ugh

A similar example occurs in (23), where the monosyllabic Debbie finally launches

44

Jenny Cheshire

into a longer turn after my reformulation of a negative clause with not into a negative clause with never. (23)

—>

Jenny: Are you havingfireworksthis year? Debbie: yes Sharon: we have indoor fireworks as well Jenny: yes ...I like them... sparklers Debbie: yes Sharon: not only sparklers... rockets Jenny: indoor rockets? Sharon: indoor rockets Jenny: oh I didn 7 know you could get those Debbie: yes Jenny: I've never seen those Debbie: indoor rockets you know all thefireworksyou can have outside you can have inside as well

Thus utterances containing never often have an interactional role that appears to reflect the function of never as an involvement strategy: speakers use never when they wish to take an extended turn, when they wish to show their interest in the contributions made by their interlocutor, or when they wish to attend to their interlocutor's positive face in potential face-threatening events such as arguments.

5. The serial effect In all these examples speakers could equally well have chosen the alternative way of expressing negation, with not. Bolinger (1977) argues that variation in language always has a function: if we accept this possibility, we can explore a further dimension of English negation, that of the variable use of not and never. In particular, we can investigate what Scherte andNaro (1991,1992) have termed "the serial effect": a preference for similar clausal patterns to occur within a section of discourse. For example, Weiner and Labov (1983) discovered in their data a tendency for one passive form to lead to another; Schiffrin (1981), studying the use of tense in narratives, observed a general tendency for particular grammatical forms

English negation from an interactional perspective

45

to cluster together; and several writers have identified a parallel effect for noun phrase concord, such that one plural marker tends to lead to another or, conversely, that the absence of a plural marker tends to lead to further absences within the same stretch of discourse (see Poplack 1980, on Spanish; Scherre - Naro 1991, on Brazilian Portuguese). In the conversations that I have analysed it is sometimes possible to observe a serial effect in the use of quantifiers. Although previous research on the serial effect has been quantitative, attempting to determine its statistical significance, it is inappropriate to do this for never, since it is not always clear whether it is used as a straightforward negative or whether its temporal meaning comes into play. Problems of this kind are common in the analysis of syntactic variation (see Cheshire 1987 for discussion). Instead, it is revealing to analyse stretches of conversation where never or ever co-occur, and to observe the conversational outcome of this syntactic parallelism. In (24), for example, I introduce a quantifier with the question Do girls ever have fights with each other? It would of course have been equally possible to have constructed an interrogative without ever, phrasing the question as Do girls havefightswith each other? The point of interest here, though, is not that ever is used, but that once it has been used, it is repeated in the series of clauses that follow, on every occasion when it is possible to use the quantifier. Marie and Wendy co-operate in answering my question, with Wendy emphatically affirming that girls do indeed have fights; my next question again includes a quantifier, and the two girls continue to co-operate, with Marie choosing never for her negative clause, followed by Wendy's ever in her response to my small joke: (24)

—>

—»

—>

Jenny: Do girls ever havefightswith each other? Marie: Yes Wendy: yes ... you're telling me Jenny: Have you ever been in a fight with a girl? Marie: yes I have I have Wendy: I have as well Marie: but we ve neverfightedtogether Jenny: not yet (laughter) Wendy: not yet you know I don't think we ever will

46

Jenny Cheshire

The sequence is unremarkable and the conversation proceeds harmoniously. In just the same way, the not strategy, once chosen as an expression of negation, is often continued. This is illustrated in (25) where the conversation develops from the previous discussion about "gypsy tart": (25)

—> —» —» —>

Jenny: Do you have dinner with her then is that where you've had it? Wendy: no we has it at school Jenny: Oh do you? Marie: I has it at school but she don't Wendy: No I go to home to dinner I used to stay to school dinners but I don't now Jenny: Why don't you stay then? W: well they hadpig's heart once and I didn't like that and they made me - eat it and I was sick so my mum said I didn't have to stay anymore Jenny: yes

Syntactic harmony of this kind typically occurs when the conversation is proceeding harmoniously, with speakers co-operating to produce felicitous discourse as in the two extracts above. When the conversation takes a less harmonious turn, this can be marked by a disruption of the pattern of syntactic parallelism. Consider (26), where Nobby and Benny are teasing Ronny, first in a friendly fashion and then in a more hostile way. All three speakers use slang and much swearing, expressing positive politeness (Brown - Levinson 1987). Nobby's first teasing question rests on his professed presupposition that Ronny had stolen the carpet which was in his bag. Ronny negates the presupposition crossly, swearing and addressing Benny with the insult you puff. The other boys laugh, and the teasing then becomes more intense, with Nobby repeating the presupposition, despite Ronny's previous denial. Benny then insults Ronny, using pretty, an adjective that is normally used to refer to girls (see Kuiper 1991 for discussion of male insults using terms that usually refer to females), and presupposing that they have succeeded in making Ronny angry. In this second part of the teasing Benny changes the syntax, choosing never rather than not, which was the negative marker that Ronny had introduced; Nobby continues the pattern of linguistic changes, using the slang expression pinch instead of their previous nick (also slang).

English negationfroman interactional perspective (26)

—> —>

47

Jenny: What's he got in there? Benny: A fucking carpet what else Nobby: Where'd you nick it? Ronny: Ifucking didn 't nick it you puff (laughter) Benny: you sure you never nicked it Nobby: Where 'd you pinch it mate Benny: You don't half look pretty when you 're angry (laughter)

In this case, then, the absence of syntactic parallelism mirrors the absence of conversational harmony, with the heightened teasing of the unfortunate Ronny. In (27) there is a similar absence of syntactic parallelism. The extract is part of a conversation between a married couple, A and B, and two of their friends. Speaker Β had been chastising her husband earlier for not reading enough, and she suggests that he reads the novel Cold Comfort Farm. A's reply is incongruous in this context, as is shown by the laughter with which it is received: not only is A male, but there had been no mention of babies in the previous discourse, and there is no obvious reason to anyone other than Β why this remark is relevant. The absence of harmony in the content of A's contribution is matched by the absence of harmony in the syntax: Β uses never whereas A responds with not (he could have said I've never had a baby either). The incongruous remark allows A and Β to collaborate (after an intervening lateral sequence) in telling the story of A going to visit Β in the evenings whilst Β was in hospital after the birth of their baby, and A finding that Β had laughed so much at the book that she had burst her stitches. (27)

B: Why darling why don't you bribe Jo to lend you her Cold Comfort Farm ... you've never read it have you? A: No but I haven't had a baby either {(laughter) B: {Oh honestly it saved my life in hospital it really did A: actually every evening I used to d: it's all right Arthur (approximately 6 seconds of intervening talk) A: anyway I used to go into the hospital in the evenings andfind her..sort

48

Jenny Cheshire

of in real great pain because she'd laughed so much . she'd burnt a couple . burst a couple of stitches (Svartvik - Quirk 1980: 626) Thus the existence of two methods of negation allows speakers to generate discourse meanings over and above the local meaning of simple negation, by giving them the choice of following their interlocutor's selection of either not or of ever and never. The choice is available to all speakers of English, both "standard" and "non-standard" varieties: (24), (25) and (26) are from my recordings of workingclass adolescents, but (27), from the London-Lund Corpus, is from the speech of educated adults.

6.

Conclusion

As educated people, linguists are in a double bind. Try as we may, our intuitions will be influenced by norms based on standardised, written language. Analytical approaches based on intuitions, such as Klima (1964), are likely to be influenced by these norms. Even those approaches based on the empirical observation of language use are susceptible to the influence of educated norms, if they involve setting up pre-determined categories within which to carry out an analysis (as in variationist analysis or in dialectology, for example). The case of never with reference to a single past event shows how a feature that is seen as "incorrect" by guides to good usage has been incorrectly labelled "non-standard" by sociolinguists, and still more incorrectly labelled as "non-existent" by linguists working within a framework based on isolated sentences. I have argued that this has led us to overlook an important way of expressing negation in English, one that conforms to Jespersen's Neg First principle and that illustrates the continuation in Present-Day English of Jespersen's Negative Cycle. Yet in order to perform a linguistic analysis we obviously have to be educated people, trained in linguistics. A way forward is to avoid pre-determined categories and to see instead how forms are used in discourse. This includes observing how speakers orient to the syntax that is used by their interlocutors, borrowing the methods used in conversation analysis. I hope to have shown that in the case of never this allows us to recognise a hitherto neglected yet important method of expressing negation in

English negationfroman interactional perspective

49

English, which involves the interlocutor in relating the negative form to the linguistic and nonlinguistic context in which it occurs, in order to determine the extent of its temporal reference. It is therefore especially well suited to the demands of face-to-face interaction, and this must surely account for the fact that throughout the history of English speakers have used it as a way of reinforcing a negative marker. It is of course only by observing never in its conversational context that it is possible to observe its interactional functions and to resolve some of the questions that have perplexed linguists until now. Thus our understanding of the present-day syntax of a language can benefit from taking a more comprehensive approach than is usual, considering a syntactic form from the point of view of its history, its normative status, its syntactic relationships, its discourse functions and its occurrence in conversation.

Notes 1.

An earlier version of this paper was published in Mondada (1995) and reprinted in Trudgill - Cheshire (in press).

2.

This can be seen clearly in different versions of some early texts. For example, Tieken-Boon van Ostade (1995: 78) points to the use of never with a quantifier function as an alternant to no in the first pair of phrases and as an alternant to none in the second pair: that there is no knyght wollfyghtfor hym (Winchester edition 139. 1-2) that there nys neuer a hryghte wyllefyghte for hym (Caxton edition 99. 3-4) that never man drew hit but he were dede or maymed (Winchester edition 987. 17-18) that none ne drewe it but he were dede or maymed (Caxton edition 484. 32-33)

50

Jenny Cheshire

References Bolinger, Dwight 1977 Meaning andform. London: Longman Bossuyt, A. 1983 "Historical functional grammar: An outline of an integrated theory of language change", in: Simon C. Dik (ed.), 301-325. Brown, Penelope - Stephen C. Levinson 1987 Politeness: Some universals in language usage. Cambridge: Cambridge University Press. Chafe, Wallace 1982 "Integration and involvement in speaking, writing, and oral literature", in: Deborah Tannen (ed.), 35-53. 1986 "Writing in the perspective of speaking", in: C.R. Cooper Sidney Greenbaum (eds.), 12-39. Cheshire, Jenny 1982 Variation in an English dialect: A sociolinguistic study. Cambridge: Cambridge University Press. 1987 "Syntactic variation, the linguistic variable and sociolinguistic theory", Linguistics 25 (2): 257-282. 1989 "Addressee-oriented features in spoken discourse", York Papers in Linguistics 13: 49-64. 1997 "Involvement in 'standard' and 'nonstandard' English", in: Jenny Cheshire - Dieter Stein (eds.), 68-82. Cheshire, Jenny (ed.) 1991 English around the world: Sociolinguistic perspectives. Cambridge: Cambridge University Press. Cheshire, Jenny - Viv Edwards - P. Whittle 1989 "Urban British dialect grammar: The question of dialect levelling", English Worldwide 10: 185-225. Cheshire, Jenny - Dieter Stein (eds.) 1997 Taming the vernacular: From dialect to written standard language. Harlow: Longman. Collins - Collins Dictionary of the English Language 1981 London - Glasgow: Collins.

English negationfroman interactional perspective

51

Cooper, C.R. - Sidney Greenbaum (eds.) 1986 Studying writing: Linguistic approaches. Beverley Hills: Sage. Coupland, Nikolas 1988 Dialect in use. Cardiff: University of Wales Press. Dik, Simon C. (ed.) 1983 Advances in functioned grammar. Dordrecht: Foris Publications. Fodor, Jerry A. - Jerrold J. Katz (eds.) 1964 The structure of language. Englewood Cliffs, NJ: Prentice Hall. Fowler, H.W. 1965 Modern English usage. (2nd edition.) London: Oxford University Press. Gumperz, John J. 1982 Discourse strategies. Cambridge: Cambridge University Press. Holm, John A. 1988 Pidgins and Creoles. Vol. 1. Cambridge: Cambridge University Press. Holmes, Janet 1990 "Hedges and boosters in men's and women's speech", Language and Communication 10(3): 185-205. Horn, Laurence R. 1989 A natural history of negation. Chicago: University of Chicago Press. Hughes, G.A. - Peter Trudgill 1979 English dialects and accents: An introduction to social and regional varitieties of English. London: Arnold. Jeanneret, T. 1992 "Modes de structuration en conversation", in: Georges Lüdi (ed.), 59-69. Jespersen, Otto 1905 Growth and structure of the English language. London: Edward Arnold. [1982] [Reprinted.] 1917 "Negation in English and other languages", Det Kgl. Danske Videnskabernes Selskab. Historisk-filologiske Meddelelser 1: 1-151. Copenhagen.

52

Jenny Cheshire

[1962]

[Reprinted in Selected writings of Otto Jespersen. London: George Allen and Unwin.] Klima, Edward S. 1964 "Negation in English", in J. A. Fodor - J.J. Katz (eds.), 246-323. Kuiper, Koenraad 1991 "Sporting formulae in New Zealand English: Two models of male solidarity", in: Jenny Cheshire (ed.), 200-211. Labov, William 1973 "Where do grammars stop?", in: Roger W. Shuy (ed.), 43-88. 1984 "Intensity", in: Deborah Schiffrin (ed.), 43-70. Labov, William (ed.) 1980 Locating language in space and time. New York: Academic. Lüdi, Georges (ed.) 1992 Approches linguistiques de 1'interaction; Bulletin CILA (Commission interuniversitaire suisse de linguistique appliquee; Neuchätel, Switzerland). Mondada, L. (ed.) 1995 Formes linguistiques et dynamiques interactionnelles (Cahiers de l'lnstitut de Linguistique et des Sciences du Langage de 1'Universite de Lausanne 7). Orton, Harold - MV. Barry - W.J. Halliday - P.M. Tilling - MF. Wakeling 1963-1969 Survey of English Dialects. Vols. 1-4. Leeds: Arnold. Quirk, Randolph - Sidney Greenbaum - Geoffrey Leech - Jan Svartvik 1985 A comprehensive grammar of the English language. London: Longman. Poplack, Shana 1980 "The notion of the plural in Puerto Rican Spanish: Competing constraints on /s/ deletion", in: William Labov (ed.), 55-67. Scherre, Maria Marta Pereira - Anthony J. Naro 1991 "Marking in discourse: 'Birds of a feather'", Language Variation and Change 3 (1): 23-32. 1992 "The serial effect on internal and external variables", Language Variation and Change 4 (1): 1-14. Schiffrin, Deborah 1981 "Tense variation in narrative", Language 57 (1): 45-62.

English negationfroman interactional perspective

53

Schiflrin, Deborah (ed.) 1984 Georgetown University Bound Table on Languages and Linguistics 1984. Washington, D.C.: Georgetown University Press. Shuy, Roger W. (ed.) 1973 Georgetown University Round Table on Languages and Linguistics 1972. Washington, D.C.: Georgetown University Press. Sundby, Bertil - Anne Kari Bjorge - Kari E. Haugland 1991 A dictionary of English normative grammar 1700-1800. Amsterdam: John Benjamins. Svartvik, Jan - Randolph Quirk 1980 A corpus of English conversation. Lund: Gleerup. Tannen, Deborah 1989 Talking voices: Repetition, dialogue and imagery in conversational discourse. Cambridge: Cambridge University Press. Tannen, Deborah (ed.) 1982 Spoken and written language: Exploring orality and literacy. Norwood, NJ: Ablex. Tieken-Boon van Ostade, Ingrid 1995 The two versions of Malory's Morte Darthur. Multiple negation and the editing of the text. Cambridge: D.S. Brewer. Trudgill, Peter 1984 "Standard English in England", in: Peter Trudgill (ed.), 32-44. Trudgill, Peter (ed.) 1984 Language in the British Isles. Cambridge: Cambridge University Press. Trudgill, Peter - Jenny Cheshire (eds.) in press The sociolinguistics reader. Volume 1. London: Arnold. Weiner, J. - William Labov 1983 "Constraints on the agentless passive", Journal ofLinguistics 19: 29-58. Wood, Frederick T. 1981 Current English usage. London: Macmillan.

On Negative Raising in the history of English1 Olga Fischer

1. Introduction Negative Raising as in I don't think I can handle these facts is a phenomenon whereby the negative element, which logically belongs to the subclause, has been "raised" into the matrix clause. In spite of the fact that it has been studied extensively, Negative Raising (NR henceforth) has remained somewhat intangible. It would be of interest to look at the subject historically (especially because the earliest stages of English do not show it) in the hope that the diachronic facts may throw some new light on the nature of the construction. However, before I embark on the subject of NR, I should warn the reader that this paper will not constitute a neat and straightforward exposition of how and why NR (or negative attraction / negative transfer / negative transportation, as it is also called) developed in English; rather it will contain a list of questions and suggestions that slowly emerged while I was looking at a large collection of historical data that I had extracted from the diachronic part of the Helsinki Corpus.2 My aim in this paper will be mainly to show what data I have found, to order these as far as possible, and to point out possible connections between the development of NR and other developments that were taking place in English at more or less the same time. This, then, will be essentially a surface approach, which, I believe, may be a good way into a rather diffuse topic. It is the nitty gritty of the data which, initially, will give us most insight into what factors may have been responsible for the rise of NR. The problem with NR is that it is not a simple matter, as Horn (1978), among others, quite clearly found when he was looking for syntactic evidence for the rule of NR. He shows convincingly that NR is not just a syntactic matter but "betrays a fundamental syntactic, semantic and pragmatic process" (1978: 130, 216). Since there is so much uncertainty as to the nature of NR on the synchronic level, it seems wise to take all possible details into account when looking at it diachronically. After this initial stage, it will be more profitable to look at the wider (theoretical) issues involved. The data investigated appear to indicate that the rise

56

Olga Fischer

of NR is intimately connected with a whole range of developments, and in my subsequent discussion I will touch on such factors as the rise of epistemic modals (and in general the expression of modality in English, which has moved from a morphological system (the subjunctive), via, or in combination with, a more or less open lexical system (the use of the premodals in Old English and a range of adverbial expressions, cf. Goossens 1982) to a more grammaticalised modal auxiliary system, cf. Plank 1984); the development of verbs like think, expect, suppose etc. (i.e. the verbs that typically occur with NR) into a kind of "hedges", elements used in discourse to tone down assertions; the loss of multiple negation in the standard language and the grammaticalisation of implicit negatives like any etc. in negative contexts (i.e. the replacement of constructions such a s i n e have noon children by I haven't any children);3 the changes taking place in the placement of certain adverbials (including the negative not, cf. Ellegard 1953: 200ff); and the rise of Subject Raising and the loss of impersonals. It must be clear that I cannot hope to discuss all this in the scope of a single article. Still, I hope that the data and the questions presented here will lead to further explorations.

2. Collecting the data 2.1. The phenomenon of Negative Raising The collection of data was in itself a problem because it was not so clear from the start how much should be included. Linguists writing on the (synchronic) topic of NR have often made remarks about the fuzziness of the phenomenon and its unclear boundaries. Bublitz (1992: 551), for instance, begins his contribution on NR as follows: "This paper is about a particular type of negation in complex utterances widely known as 'transferred negation' ... and also still widely regarded as 'an unclear phenomenon' ..., notorious for a considerable number of attempts, frequently unsuccessful, to explain it adequately"; compare also Klooster (1984: 86): "The process of NEG Raising,... is in many respects a mysterious affair" (my translation). Let us first look at some facts of NR in Present-Day English. When we look up the phenomenon in the handbooks (e.g. Quirk et al. 1985: §14.36), we are usually given sample sentences like the following:

On Negative Raising in the history of English (1)

a. b. c.

57

He didn 't think that Mary was pretty I don't suppose he has paid I don't expect he cares

These sentences can be instances of NR but are not necessarily so because they can be read in two possible ways: either the negative has the matrix clause as its scope, or the scope of the negative is restricted to the subclause. This would then give the following readings respectively for e.g. (la): 'it is not the case that he thought that Maiy was pretty' (often called the "strong reading") and 'he thought that Mary was not pretty' (the "weak reading", cf. Bublitz 1992: 558, note 13; Trask 1993: 180). Only the context (and in spoken language the differences in intonation) can decide what is what. The usual analysis given for the second type of interpretation, NR proper, is that the negative element is base generated in the subclause and subsequently transferred, "raised", to the higher level clause (cf. Fillmore 1963; R Lakoff 1969; G. Lakoff 1970; Carden 1971; Bartsch 1973; Bhat 1975).4 That Raising has taken place can be argued from the way the clause must be "tagged" in context (cf. Lakoff 1969; but Cattell 1973 and Horn 1978: 154ff show there are problems with this too): (lb)

i. ii.

You don't think he has paid, do you? You don't think he has paid, has he?

In the first (strong) reading, the main clause must be negative for the positive tag to occur; in the second (weak) reading it is the subclause that is tagged, implying that the subclause is inherently negative (i.e. NR has taken place). It must be clear that when the main clause subject is I, only the weak reading is likely from a pragmatic point of view (cf. Bublitz 1992: 558-559, notes 13, 14). In the case of reported speech, as in (la), both readings are presumably equally likely to occur. However, tagging does not help as a diagnostic device in cases with infinitival complements, where NR may also occur. These cases are often ignored in the literature. The use of NR here depends to some extent on the type of lexical verb used. (2)

I don't need to go

Example (2) means only 'it is not necessary that I go' and cannot convey 'it is

58

Olga Fischer

necessary that I do not go'. In contrast, (3)-(5) usually mean that the speaker expects or advises (someone) not to go, i.e. they are then instances of NR. (3) (4) (5)

I don't expect to go there I don't expect him to go I don't advise you to go there

Klooster (1984: 89) discusses how it can be shown that a sentence like (5) may indeed involve NR. In normal usage, he writes, a verb like (Dutch) aanraden 'advise' can be combined with an adverbial adjunct of manner such as in (6): (6)

Ik raadje dringend aan niet naar China te gaan Ί strongly advise you not to go to China'

This adjunct can also be used with the negative lexical counterpart of aanraden, i.e. afraden, 'to advise against it, to dissuade someone' as in (7): (7)

Ik raadje dringend af naar China te gaan Ί strongly "dis-advise" you to go to China'

However, the modifier cannot be used in a NR construction: (8)

Ik raadje (*dringend) niet aan naar China te gaan Ί (*strongly) don't advise to go to China'

Klooster explains this as follows (Klooster 1984: 89-90, my translation): The reason is that dringend in [(8)] distorts the special relation that exists between niet ['not'] and aanraden ['advise'] (i.e. it cancels the N[egative]T[ransportation]-character of the negation) and turns Ik raadje niet aan [Ί do not advise you'] etc. into the 'normal' negation of Ik raadje aan [Ί advise you'] with the consequence that the sentence becomes ill-formed because adjuncts like dringend can only be combined with predicates that express an activity [i.e. you can combine such adjuncts with activity verbs like listen ('he carefully listened to her story') but not with the negation of this activity ('he

On Negative Raising in the history of English

59

carefully didn't listen to her story')]. Dringend upsets the litotes-effect of the NT-sentence, and through it also its well-formedness. Another reason that the non-finite complements are of interest is the fact that some verbs allow NR. only with non-finite clauses, while with finite complements they behave like "normal" verbs. For instance: (9)

a. b.

I don't wish to see you again Ί wish not to see you again' I don't wish that I will see you again *'I wish that I will not see you again'

(These facts are discussed in Horn 1978: 151.) Similarly hope allows NR. only with an infinitival complement in English (unlike e.g. German hoffen and Dutch hopen, where it is also usual with finite complements): (10)

a. b.

I never hope to see you again Ί hope that I will never see you again' I don't hope that I will see you again *'I hope that I will not see you again'

Another way in which NR may be recognised is by looking at the behaviour of negative polarity items, NPIs for short (i.e. items like any, ever, yet, at all, in weeks, until etc.), which can only occur in a negative context (so in negative clauses, questions, conditional clauses; cf. Klima 1964).5 Thus, these items would occur with a clause like (3) above as follows (NPIs in roman type): (11)

a. b.

*I expect to go there at all / yet / until Sunday I don't expect to go there at all / yet / until Sunday

That is, the NPI can only occur in the subclause in (1 lb) because the negative is underlyingly part of the subclause, not the main clause; it cannot occur in the subclause in (1 la) because there is no negative to attach to. In contrast, with a nonNR verb, like claim, NPIs cannot be used in clauses similar to (1 la) and (lib), because in both cases the subclause is not negative:

60

Olga Fischer

(12)

a. b.

*I clcdm to have seen her at all / in weeks */ don't claim to have seen her at all / in weeks6

Cases of NR. involving non-finite complements as discussed above may for our purpose be more interesting than the usual ones involving finite complements, because it looks as if, diachronically, NR took place there first. We will see below that in the historical material examined the majority of the unequivocal cases of NR involve non-finite complements: all examples of NR with the verb seem are nonfinite (see section 2.3), and with the verb think the majority (eight out of eleven, see section 2.4.1) are.

2.2. Periods and data considered As far as my collection of data is concerned, I have looked only at the Middle and Early Modern English periods contained in the Helsinki Corpus because there is no evidence that NR occurred in Old English.7 Traugott (1992: 271) writes in this respect that "[t]here do not appear to be examples of 'negative raising'", but she points to two constructions that are rather "similar-looking" to NR constructions: (13)

a.

b.

ac he ne com na to demenne mancynn... ac to but he not came not to judge mankind but to gehcelenne. save 'but he didn't come to judge mankind but to save them' OECHom I, 22 320.5, Traugott 1992: 270) Ne cwced he deah no dcet dcet he cwced fordcemde he not said he however not that that he said because he gesinscipe tcelde, ac fordcemde he wolde da sorga marriage censured, but because he wanted those sorrows awegdrifan disse middangeardes away-drive of-this world 'He said what he said, however, not because he disapproved of marriage but because he wanted to drive away the sorrows of this world' (CP 51 401.11, Traugott 1992: 271)

On Negative Raising in the history of English

61

In both cases, it looks as if the negative is raised to the main clause, which should be positive because the actions of "coming" and "saying" do indeed take place. However, the difference with true NR. cases is that the subclause is also marked negative and must be interpreted as negative so that these might be cases of "forward-looking" Negative Concord (Traugott 1992:272) rather than NR.8 Mitchell (1985: §31810) also comments on the use of the negative in (13b), and mentions that two types regularly occur here in Old English, both (13b) and the "normal" one with the negative in front of the conjunction and not "illogically attached to the verb of the main clause" (Mitchell 1985: §3181). It is noteworthy, though, that the cases of Old English negative attraction (or, better, forward-looking concord) only occur in sentences such as (13), expressing a contrast. It rather looks as if its use here is related to the fact that Old English sentence structure is still more paratactic than hypotactic.9 When one analyses complex clauses like the ones in (13) into separate, independent ones, we get 'he didn't come to judge mankind; he came to save them', which makes it more understandable that the first main clause is negated, because indeed 'he did not come in order to judge them'. I have found an example in Middle English which shows that the first clause is indeed underlyingly negative in these cases: (14)

We prayon not here pat we be not tempted...; but we praye pat God lede vs not in-to temptacion (M3/4 IR SERM ROYAL 11)10

The type of contrastive clause represented in (13) still occurs in Middle English, as (15) shows, and may be of possible interest when Negative Concord is lost (which gradually happens in this period) because in that case (i.e. in (13)) either the matrix verb negative or the non-matrix verb negative would be lost. If the second is lost, it would automatically result in a NR. structure. (15)

He ne askep non outrage, noper wyn ne flesch ne fische, ne more ne biddep he not for pe 3er ne for al pe woke, but onliche to dryue forp pe day (M3/4 IR RELT VICES4 108)

Another type of negative construction that occurs in Old English (and in later periods too) is the type discussed by Ukaji (this volume) and van der Wurff (this volume), where the scope of the negative extends across the clause boundary (see

62

Olga Fischer

also Mitchell (1985: §§2039-2040), who refers to this usage as "expletive ne"): (16)

Fordcem ne pearf rumne wisne rrton tweogan, p[oet\ da yflan ncebben Therefore not need no wise man doubt that the evil not-have eac ecu edlean hiora yfles also eternal rewards of-their evil 'Therefore no wise man need doubt that also the evil ones will suffer eternal punishment for their evil deeds' (Boethius 113.21; Mitchell 1985: §2039)

Again, this is not a case of NR. because it is the main clause that is logically negative, while the subclause becomes syntactically negative by a type of Negative Concord.11 Of a different kind are Mitchell's examples in §2040, where the matrix verb is not explicitly negative and attracts a negative in the subclause. The difference with (16) is that the subclause is indeed negative from the point of view of the speaker, and would be in the negative if it was used independently (compare the remarks made for (13) above). We find a similar instance in Early Modern English (18). (17)

(18)

and us forbyt pcet we ne sceolon hogian ymbe ure frcetewunge and us forbids that we not should think about our ornaments 'and forbids us to think about our ornaments' OECHom ii 464.16; Mitchell §2040) Nulle ich pet nan iseo ow bute he habbe... spetiale leaue. Not-will I that none see you but he have special leave Ί want that no one sees you / 1 don't want anyone to see you, unless ...' (Ml IR RELT ANCR 33)

Example (18) is more interesting than (17) because here both matrix clause and complement are negated. Used as independent clauses both are indeed negative. (18) is of special interest to the development of NR, because, with the loss of Negative Concord (see also below), this clause could indeed become a NR clause (as the second and more usual translation given for (18) shows). Concluding that there are no true cases of NR in Old English, I have restricted myself as far as the data search is concerned to the Middle and Early Modern

On Negative Raising in the history of English

63

periods. For these periods, I have looked at the most frequent of those verbs that are known to be involved in NR. in Present-Day English.12 In Table 1 there is an overview of the overall frequency with which each relevant verb was found in each separate period (the first column given for each period), followed (in square brackets) by the number of constructions in which they were involved with negatives (i.e. all the negative types discussed in (19)-(22) below). I have looked at all the instances in which these verbs were combined with a negative. On account of the fuzziness of the NR phenomenon (mentioned in 2.1 above), it seemed better to include too much rather than too little, and this procedure also seemed useful in order to get the widest possible overview of all the factors that might be involved. Thus, I have included cases where the matrix verb itself is negated (19), or where an explicit negative is present elsewhere in the matrix clause (20); also cases where there is an explicit negative in the complement clause, whether finite or non-finite (21), and finally cases where a negative is incorporated into another element in the sentence, main or subclause (22a-b). (19)

(20)

(21)

(22)

But when they haue begönne to be cd one thynge then they be goodes. Do it not happen that these thynges be good by adoption or optaynyng of vnitie? (El XX PHILO BOETHCO 782) even they that are wisest amongst us living, compared with the Prophets, seem no otherwise to talke of God, then as if the children which are caried in armes should speake of the greatest matters of state. (E2 IR SERM HOOKER 5) how hapneth it he is not brought Face to Face to justifie this matter, neither hathe bin of all this time? (El XX TRI THROCKM I,66.C2) a. then at the last the wyckedfolke be turmented with more greuous ponyshments, when they seme to be vnponyshed. (El XX PHILO BOETHCO 101) b. For it semeth perchaunce incredible to some folke, that it behoueth that wyckedfolke be more vnhappye when they haue accomplyshedther desyres, then... (El XX PHILO BOETHCO 98)

For only a veiy small number of verbs did I find examples that can be interpreted as NR constructions. These are indicated separately in Table 2.

64

Olga Fischer

Table 1. NR verbs in the Helsinki Corpus, indicating the overall number of NR verbs, the use of the verb as a "hedge" (*), and the number of constructions containing a negative (square brackets). ME1

ME2

ME3

ME4

El

E2

E3

believe

16 [4]

26 [1]

76 [9]

27 [2]

88* [9]

131 * [9]

82 * [17]

desire

1

6 [1]

60 [10]

50 [4]

91 [9]

89 [9]

116 [10]

22 [2]

45 [7]

44 [3]

25 [2]

28 [2]

7*

73 * [8]

68 [10]

12 [1]

19 [4]

21 * [6]

expect

—

—

happen

—

—

hope

7 [1]

11

36 [1]

1

imagine

34* [2] —

—

4

30 [6]

50 [4]

48 * [3]

67 [13]

148 [23]

suppose

_

_

_

40 * [8]

17* [1]

32 * [3]

32 * [2]

think

46 [9]

55 * [6]

122 * [9]

125 * [9]

139* [27]

179* [39]

206* [38]

trow

16 [1]

2*

54 * [10]

38 * [6]

•ween

73 * [10]

[1]

58 * [6]

35 * [4]

—

_

2

wish

_

*

4 [1]

OS m

seem

—

—

—

—

—

—

15 [2]

46 [6]

29 [3]

I have also marked in Table 1 (with an asterisk) in how far the verbs concerned could be used as inteijections or conversational hedges, in which they to some extent lose their lexical or referential meaning, like Present-Day English suppose as

On Negative Raising in the history of English

65

in You don't go there very often, I suppose. Going through the data, it became clear to me that the development from full lexical verbs to such inteijections - a kind of grammaticalisation - may be relevant to the emergence of NR with these verbs. In the following sections, I will discuss those verbs for which I have found some evidence in the corpus for NR, or which are otherwise of interest, i.e. the verbs believe, expect, happen, hope, seem, think, trow and ween. Table 2. Occurrence of Negative Raising in the Helsinki Corpus ME1

ME2

ME3

ME4

El

E2

E3

believe

-

-

1?

1?

-

-

desire

-

-

-

-

-

-

expect

-

-

-

-

-

happen

-

-

-

-

-

-

1?

hope

-

-

-

-

-

-

-

imagine

-

-

-

-

-

-

-

seem

-

-

-

-

-

-

8

suppose

-

-

-

-

-

-

-

think

-

-

-

trow

-

-

-

-

-

-

-

ween

1?

-

-

-

-

-

-

•wish

-

-

-

-

-

1?

1?

2?

1 + 1?

2?

1+5?

1 -

4?

9+1?

2.3. Negative Raising with impersonal verbs: seem and happen To begin with seem: the following are all the instances I have found where the verb seem is itself negated All examples are late, i.e. from period E3 (1640-1710).

66

Olga Fischer

(23)

so that it seems this Creature has no very goodforesight: It does not seem to have any eye-lids, ... (E3 EX SCIO HOOKE 13.5,211) letting one creep on my hand, that it immediately fell to sucking, and did neither seem to thrust its nose very deep into the skin, ... (E3 EX SCIO HOOKE 13.5,212) It did not seem at all, though I viewed it a good while as it was sucking to thrust more of its nose into ... (E3 EX SCIO HOOKE 13.5,212) and that their Parents do engage they shall keep good order, and be cleanly and neat in their apparel; that they may not seem to disgrace their fellowes, or to be disdained by them for their poverty. (E3 IS EDUC HOOLE 228) she therefore said a thousand things to appease the raging of his flame, and to prepare him to hear who it was with calmness: but before she spoke, he imagin'd who she meant, but wou'd not seem to do so, but... (E3 Μ F1CT BEHN 159) If every man mayjudge for himself there will be nothing but confusion in Religion, there will be no end of Controversies: so that an infallible Judge is necessary, and without this God had not made sirficient provisionfor the assurance of men's Faith, andfor the Peace and Unity of his Church: Or as it is expressed in the Canon Law, (aliter Dominus non videretur flösse discretus), ('otherwise our Lord had not seem'd to be discreet^). (E3 IR SERM TILLOTS II:ii445) He never seemed to understand foreign affairs well: and yet he meddled too much in them (E3 NN HIST BURNETCHA 1,1,169) but if we look upon the Judgments of Men, who is there to whom these things seem not only not to be believed, but also not to be heard? (E3 XX PHILO BOETHPR 182) That it may not seem impossible, that Electrical (Effluvia) should be able to insinuate themselves into the pores of many other bodies, I shall adde, that I found them subtile enough to attract not onely Spirit of Wine, but that ['also'] the fluid aggregate of Corpuscles we call Smoak (E3 EX SCIO BOYLE 13)

(24)

(25)

(26)

(27)

(28)

(29)

(30)

(31)

On Negative Raising in the history of English (32)

67

for this which I am now about to say will not seem less wonderful, but it necessarily follows from what hath been before proposed. (E3 XX PHILO BOETHPR 179)

Most of these examples are clear instances of NR, except for (26) and (30). In (26) seem seems to have more lexical meaning (it means something like 'to have the appearance of) 1 3 so that it is possible (and likely) that seem itself is negated here rather than the subclause (so there is no NR). In (30) not should be taken with only, in other words it does not negate seem. Examples (31) and (32) are different from the others because they are monoclausal (they contain a small clause). It is not clear whether these too should be regarded as involving NR. Semantically, they seem to be cases of NR because quite clearly it is only the small clause that is in the scope of the negation. Syntactically, however, one could maintain that Raising is only Raising when a boundary is crossed, i.e. an S or NP node (or in more up-to-date terms a CP, IP or NP node), which is not the case with small clauses.14 These monoclausal cases (i.e. (31)—(32)) are, however, of great interest because it is possible that here the loss of multiple negation led to the NR character of seem. Because the usual development after loss of multiple negation was for only the matrix verb to become negated, the negator in the small clause, where the negative "logically" belonged, got lost, automatically as it were resulting in a NR construction. Thus, when multiple negation is lost in structures like (33), it is to be expected that the negative that is preserved after the loss of multiple negation goes with the main verb because that is the natural development also with other verbs (verbs of the non-NR kind) leading to the Present-Day English situation, where / didn't claim anything is unmarked in comparison to I claimed nothing}5 (33)

For cd be it so that this ne seme nat credible thing peraventure to some folk, yit moot it nedes be that... (M3 XX PHILO BOETHCH 446.C2)

Cases of NR with impersonal verbs like seem are clearer than with nonimpersonal verbs like think, suppose etc. because a non-NR reading is not possible.16 We have seen above (and compare also the discussion of examples (6)-(8)) that a strong reading of clauses such as He didn't think Mary is pretty is possible when the verb think is interpreted as an activity verb, i.e. in the meaning

68

Olga Fischer

of 'to entertain a thought'. With seem such a strong reading is not possible (except when it is used in a different sense, as in (26)) because seem cannot assign a thematic role to its subject, or, to put it differently, seem does not have an agentive subject. It is therefore interesting that all the cases of NR with seem in the corpus are also cases that show Subject Raising. I have not come across any examples such as It doesn 't seem to be true: we always get It seems not to be true or It seems to be untrue, as we can seefromthe examples below in (35)—(39). The only exception I have found is (34), but here semeth has the differed more fully lexical meaning 'befit', and can therefore itself be negated. (34)

(35)

(36)

(37)

(38)

(39)

'Sir, hit nedith natyou to put me to no more payne, for hit semyth nat you to spede thereas all thes othir knyghtes have fayled.' (M4 Μ ROM MALORY 47) For it would almost seem incredible to any one, and it is yet true, that evil Aden must necessarily be more unhappy when they have compassed what they desire, than when they cannot do so ... (E3 XX PHILO BOETHPR 177) Next, it seem'd nothing more difficult to give an intelligible reason, why Cork is a body so very unapt to suck and drink in Water, (E3 EX SCIO HOOKE 13.5,113) and yet the blood seem 'd to run through its head very quick andfreely,so that it seems there is no part of the skin but the blood is dispers 'd... (E3 EX SCIO HOOKE 13.5,212) And yif I turne ayein to the studies of men, who is he to whom it sholde seme that he ne scholde nat oonly leven thise thinges, but ek gladly herkne hem? " (M3 XX PHILO BOETHCH 448.C1) Wherfore it seemef) jrat Jiere is no more dignyte ne worthies pat a man mai come to, pan for to haue tribulacioun for Cristis loue. (M3/4 IR RELT HILTON 10)

Likewise note the first occurrence of seem in (23) above (so that it seems this Creature has no very good foresight). Similarly, we do not get NR with other (older) syntactic uses of the verb seem, which have either a dative experiencer and no subject (41), or which have the thematic role of cause as subject (40):

On Negative Raising in the history of English (40)

(41)

69

"Now confesse I wel, " quod I, "that Y see wel now certeynly withouten doutes the thinges that whilom semeden uncerteyn to me. " (M3 XX PHILO BOETHCH 436.C1) And jitte, prouj Cornell of ham, him semede J)at he hade nou3t ynou3,... (M3 NN HIST BRUT 225)

It is only in Negative Concord constructions with it that seem is negated, but these can be left out of consideration because it is usual in multiple negation for all elements that can be negated to be negated within the same clause. This is shown in examples (42)-(43): (42)

(43)

lauerd loke to me ant haue merci of me. softe me mi sar swa & salue mine wunden pt hit ne seme nohwer ne suteli ο mi samblant pt ich derf drehe. (Ml NN BIL MARGME 62) For cd be it so that this ne seme nat credible thing peraventure to some folk, yit moot it nedes be that... (M3 XX PHILO BOETHCH 446.C2)

The question arises: why should Subject Raising and NR be connected? The answer to this might be that both Subject and Negative Raising are to be expected the moment the verb seem comes to be treated like a non-impersonal (following the trend of other impersonal verbs which were also becoming personal; cf. Fischer van der Leek 1983), and as a result comes to look veiy much like a modal auxiliary: like the modals, seem expresses a type of epistemic modality, "likelihood". Like modal auxiliaries the verb is syntactically "transparent". The moment Subject Raising occurs with a verb like seem, it becomes syntactically similar to modal verbs, in that both have a surface subject deriving its theta-role from the embedded predicate.17 Horn (1978: 145), in his synchronic study of NR, notes another link between modals and NR verbs. He shows that both can appear with NPIs in the subclause if the main clause is negative. Thus, we can have both He isn't supposed to leave until 5 and He isn't able to leave until 5, while this is not possible with non-NR verbs: *I don't claim that I have seen her in weeks. With modals, too, the negative is placed next to them, whether or not it logically belongs there. Thus, Seuren (1985:169)18 shows that in (44), meaning 'it is possible that she hasn't arrived yet', the negative is raised, while in (45), meaning 'it is not

70

Olga Fischer

possible that she has arrived yet', this is not so: (44) (45)

She may not have arrived yet She can't have arrived yet

This can be seen from the impossibility of (44') compared to the grammatical (45'): (44') (45')

*I don't believe she may have arrived yet I don't believe she can have arrived yet

The reason for this is that the negative cannot have moved up all the wayfromthe lowest clause in (44') to the highest level, skipping one level. In (45'), however, the negative was already one level higher to begin with. It is perhaps not surprising that I have not found many instances of NR with the verb happen (a verb similar to seem in that they are both impersonals, i.e. occurring in non-argument /^-constructions), because happen as a verb is introduced into English much later (see Table 1). In addition, happen, unlike seem, is not used as a discourse hedge, nor does it express epistemic modality in the same way (it expresses "coincidence" rather than "likelihood"). The only example I have found is (46): (46)

if such Person who was Goaler or Keeper of such Goal or Prison on the said Five and twentieth Day ofDecember One thousand six hundred ninety and five shall not happen to bee the Goaler or Keeper of such Goal or Prison att the time of the making such Su[m]mon ... (E3 STA LAW STAT7 VII,75)

Note, however, that the verb is accompanied by a modal, which may well be responsible for the fact that NR has taken place here; compare the influence of the modal pattern that I discussed with seem. We will see that the modal plays a special role in the discussion of other verbs too.19 Notice that again Subject Raising is present in this example. In cases in which happen is used without Subject Raising (i.e. with non-argument it, or when the subject has the thematic role of "theme", as in (49), rather than "experiencer"), the negative is never raised:

On Negative Raising in the history of English (47)

(48) (49)

71

how hapneth it he is not brought Face to Face to justifie this matter, neither hathe bin of all this time? (El XX TRI THROCKM I,66.C2) Whereby it happeth that hatred hath no place emongeste wise me[n], (El XX PHILO BOETHCO 103) You shall further note, that if at any time your Causticke doe happen not to worke so well and sufficiently to your minde, as happily you would with it should do, then apply the same Cautery againe (E2 EX SCIM CLOWES 31)

Since the earliest examples of Subject Raising with happen are found in the corpus in period E2, we would not really expect NR to occur before that period. As with seem, it seems likely that Subject Raising must precede NR.

2.4. Negative Raising and non-Subject Raising verbs 2.4.1.

Think

The other verbs in the corpus all occurred with an agentive subject, which means they did not have to develop Subject Raising first for NR to become possible (if that hypothesis is indeed correct). It may be for this reason that NR occurs earlier here. Think is the verb found most frequently with NR, and here also the clearest instances are encountered. With the other verbs in Table 1, NR is much less certain, and most often it is found there only in connection with a modal, where the modal may be partly responsible for NR. Let us have a look at NR with think first, in order to try and find out what subtle (or unsubtle) shifts may have led to its emergence. The earliest examples I have come across in the corpus are in period M4: (50)

A servant of our hath made a complainte of him I cannot thinke that he hath informed us all truely, yet not for that we will not suffer him to disobey our writinge (M4 XX CORP CPASTON 201)

72

Olga Fischer

(51)

he js wel avysyd pat sehe seyd sehe wuld neuer have to done wyth all, nere he kan not fienk Jjat sehe hath non ryth to have jt. (M4 XX CORP MPASTON 223)

There may well be even earlier ones, like the one I found in the Roman de la Rose (Fischer 1992: 285): (52)

He thenkith nought that evere he shall / Into ony syknesse fall {Roman de la Rose, Benson 1988: 11. 5621-5622)

It has to be remembered that the Helsinki Corpus is relatively small and that it is therefore not unlikely that even earlier instances may be found. Notice that both the examples from the Paston Letters have a modal, which may have influenced the presence of NR.20 Moreover, to establish whether NR has taken place with think in the third person is in any case difficult because in an example like (52) it is possible that the meaning is 'the thought never entered his head that he might fall ill', which is a less likely reading for (50), where the subject is I. When we look at the context in which (52) occurs, the non-NR reading might even be preferable, since the subject he stands for any poor man (thepovre, 1.5601); in other words, the situation is not an actual one, since the writer is not reporting the thoughts of the subject. Likewise (51) with he as subject is not entirely clear; moreover, it contains multiple negation. In period El, we find one example used without a modal, and one with: (53)

(54)

(R Roysterj I am sorie God made me so comely doubtless. For that maketh me eche where so highly fauoured, and all women on me so enamoured (M Mery) Enamoured quod you; haue ye spied out that; Ah sir, mary nowe I see you know what is what. Enamoured ka; mary sir say that againe, But I thought not ye had marked it so plaine. (El XX COME UDALL L. 165) And ouer this I can never thincke, my lordes, that so many worthye Bishoppes, so many honorable parsonages, and so many other worshippfidl, vertuous, wise and well learned men as att the makinge of that lawe were in the parliament assembled, ever ment to haue any man pvnished by death in whom ... (El NN BIO ROPER 89)

On Negative Raising in the history of English

73

Example (53), however, is not a clear case of NR. Klooster (1984) describes a number of conditions on NR. The first one is: "a negated NR-verb expresses a 'structural fact' and not a 'phenomenal fact'; this condition is related to the epistemic status of NR-verbs" (Klooster 1984: 92, my translation).2' He explains these terms, which he takes from Woisetschläger (1976), as follows: structural facts are objects of knowledge such as theories, hypotheses, conventions, laws etc., whereas phenomenal facts are objects of observation. For this reason, NR cannot occur with matrix verbs in the progressive or in the perfect, because these aspectual forms always refer to phenomenal facts, (55)

a. b.

I have not advised you to go to China I am not advising you to go to China

In both these cases the giving of advice is factive, something you do or do not, i.e. it can be denied. The negative, in other words, is part of the matrix clause, not of the complement. In (53) above, the context suggests that the addressee has indeed "marked it", but that the thought did not occur to the speaker that he had. The addition of so plaine characterises the whole clause as an incidental fact, a thought or observation that had not entered the speaker's head; it does not seem to convey an attitude on the speaker's part. (54) with subject / seems a relatively clear example of NR: the speaker expressing that he "thinks" that it is "impossible" that anyone would etc. In period E2 I have come across six potential examples of NR. The following instances seem similar to (53): (56)

(57)

and asked him, How now, Perrott, (quoth the Kinge) what is the Matter that you make this great Moane? To whom Sir John Perrott answered, And it fyke your Myestie, I did not thinck that your Highness had byn there: Yes, sayd the King, we heard you well inough:... (E2 NN BIO PERROTT 33) What! Sir John Perrott, Who sent for you? He answered, Why, my Lord, I did not think you would have ask'd me that Question; and I had imagined soemuch, you should have sentfor me twise before I would have come once, (E2 NN BIO PERROTT 38)

74

Olga Fischer

(58)

Doe you thinke there is truth in them? (Pag.) Hang 'em slaues: I doe not thinke the Knight would offer it But these that accuse him in his intent towards our wines, are ayoake of his discarded men: very rogues, now they be out ofseruice. (E2 XX COME SHAKESP 44.C2)

Example (56) seems to refer to an observation: the speaker did not think of the fact that the king had been there. In (57) the clause and ['if] I had imagined so much suggests that the thought did not enter his head, i.e. again the main clause itself is negative. (58) is similar to (56) in that the speaker has heard that "the knight did offer if' (even if it is not true, he has heard the knight's men say so), but that he had not thought of it, so the matrix verb is negative. Example (59) contains a small clause. We have seen (cf. note 14) that in those cases it is not clear whether one can speak of NR. at all; moreover, (59) is interrogative so that we also do not know if the speaker is asked whether he has this thought or not, or whether he agrees with the idea that "blissfulness is not good". The latter seems less likely. (59)

"Dost thou not think blisfidnes good? " "Yea the greatest, " quoth I. (E2 XX PHILO BOETHEL 64)

The following instance may be an example of NR.: (60)

Bewitched doe you thinke I am bewitched? Ifeele no harme in my body, you make me more afraide. (ADan.A) Nay I doe not thinke that the olde woman hath bewitched you, or that your body is bewitched, but the diuell hath bewitched your minde, ... (E2 IS HANDO GIFFORD B1V)

The old woman has not "factually" bewitched the addressee and therefore the subclause could be negative ("I think that the woman has not..."). (61)

Afterward she fel out with that N. She sent her Cat, who told her, that she had giuen him that, which hee should neuer recouer: and indeed the man died Now doe you not thinke the woman spake the trueth in all this?

On Negative Raising in the history of English

75

Would the woman accuse her selfe falsely at her death? Did not the Cat become her seruant? Did not she send her? Did shee not plague and kill both man and beaste? What shoulde a man thinke of this? (E2 IS HANDO GIFFORD E1R) On the other hand, in (61) from the same text, it is unlikely that NR is involved because the speaker asks the addressee what he is actually thinking; notice the phrase What shoulde a man thinke of this. In other words, he wants to know whether the addressee entertains that thought or not (note that this is again in a question, like (59)). A similar interpretation (so without NR) could be given for (60), where the addressee is also asked doe you thinke. Example (62) again contains a modal and has a small clause, which obscures the presence of NR: (62)

Therefore we ought not to thinke any study, paines, or trauaile too great, wherein we hope of much benefit to our patients, and also profit and commodity to our selues:... (E2 EX SCIM CLOWES 18)

Finally, there are ten potential examples in period E3. The following look like good candidates for NR since they describe an attitude or hypothesis rather than an observational fact: (63)

(64)

(65)

For if a depraved Temper be, as it were, the Sickness of the Soul, since we do not think those whose Bodies are distempered to be worthy of our Hate, but rather of our Compassion, ... (E3 XX PHILO BOETHPR 185) He had a very ill opinion both of men and women; and did not think there was either sincerity or chastity in the world out ofprinciple, but that some had either the one or the other out of humour or vanity. (E3 NN HIST BURNETCHA 1,1,167) when his Spirits were so low and spent, that he could not move nor stir, and he did not think to live an hour; (E3 NN BIO BURNETROC 21)22

76

Olga Fischer

Example (66), like (59) and (61), is again in a question (and contains a modal), where the addressee is asked whether he has the thought or not. In other words, it is more likely that the main verb is negated, and that there is no NR. (i.e. it is similar to example (61)): (66)

shou'd you not think 'twere possible I might go one step farther, and enquire his Name? (E3 XX COME VANBR 1,34)

Examples (67)-(69) are good instances of NR but they are again cases with a small clause: (67)

(68)

(69)

since he don't belong to our Fraternity, we may betray him with a safe Conscience; I don't think it lawful to harbour any Rogues but my own. (E3 XX COME FARQUHAR 8) Then they enacted, that Edwi\ri\ Brother of Edmund, a Prince of great hope, should be banish't the Realm. But Canute not thinking himself secure while Edwi[ri\ liv'd, consulted with Edric how to make him away;... (E3 NN HIST MILTON X,275) And did you come up, Sir Thomas? (Sir Thomas Whitegrave) No, I did not; I had no Subpoena; and being a Justice of the Peace, I did not think fit to leave the Country at that time without a Subpoena (E3 XX TRI OATES IV,85.C2)23

It should perhaps also be noted that there are many examples in the corpus which do not show NR, but which easily could show it in Present-Day English under similar pragmatic conditions (so I am suggesting in fact that there may have been lexical and/or syntactic obstacles which prevented the earlier use of NR; I will come back to this in section 3). Out of all the negatives used with think {137 in all; see Table 1), 40 could have shown NR but do not. Some examples: (70)

PHI: Thynkest thou that there be anye thynge in these worldly and transitorye thynges yet may bring in or shew any such state. BOE: I say I thinke not. '... I don't think so' (El XX PHILO BOETHCO 71)

On Negative Raising in the history of English (71)

(72)

(73)

(74)

77

Beside, a Prince may not pardon a wilfidl murderer, yet I think that no man wil say in hast, that he which hath committed murder may not take a pardon. '... I don't think that any man ...' (E2 IR SERM SMITH E6R) a Portugal Gentleman, who ask 'd him if the King ofEngland was crown'd yet? To whom he answered, I think not yet but he shall be shortly. Nay, saith the Portugal, that shall never be, for his Throat will be cut by Don Raleigh and Don Cobham before he be crowned. '... I don't think yet...' (E2 XX TRI RALEIGH I,215.C2) we lodged at an Inne, where Master lames Acmooty paid all charges: but at Barwhicke there was a grieuous chance hapned, which I thinke not fit the relation to be omitted. '... of which I don't think it fit to omit the relation' (E2 NN TRAV JOTAYLOR 139.C2) It may be doubted concerning whiping ...at what time and by whom it should be donne, whether presently upon the commiting of the fault whilst it is yet fresh and hot, and whether the parents themselves should beat their children. As to the first I thinke it should not be donne presently, least passion mingle with it, ... '... I don't think it should be done ...' (E3 IS EDUC LOCKE 58)

Examples like I think not (70), and I think not yet (72) are interesting because of the position of not. It is in fact ambiguous whether not should be connected with think or with the implicit complement (not (yet) would then be a small clause). This suggests that NR may have arisen in monoclausal cases through the ambiguity in the position of not. In Middle English, not comes to be used as a sentence negator (replacing ne and ne... not), negating the verb, always in the position after the verb. This happens after the loss of ne, which used to be the sentence negator and was positioned before the verb. When not is connected with yet, as in (72), its scope is only the small clause, and the usual position of not would indeed be before the clause. In the later Middle English, Early Modern periods, not begins to shift to a position before the verb (cf. Ellegärd 1953: 180ff). As soon as not occurs before think, NR is a fact, and the NR phenomenon can spread. Considering these facts, it is perhaps not surprising that it is after verbs like think, know, say etc. that not lingers much longer in postverbal position as has often been noted, but not really

78

Olga Fischer

satisfactorily accounted for (cf. Ellegärd 1953: 199fif; Ukaji 1992, note 2, which refers to Tieken-Boon van Ostade 1987).24 This may be due to a certain unwillingness to accept NR with these verbs. (75) is an interesting example in this respect: (75)

Sir George hearing this, thought it was not best to come there, fearing Ioane wouldfather a child vpon him, ... (E2 NI FICT DELONEY 84)

Not has clearly been raised from the infinitive ('he thought it was best not to come there'), but not raised all the way up to think, in spite of the fact that not best sounds very odd. In fact, not has kept the position that the negative element normally had with think in small clauses of this type, as we can see in (76), (76)

a.

b.

Syr, I may have CCC. marcs in joyntur, and I to take pe lesse when I may have pe more, my fjrendes wold Jjenke me not wyse &c. (M4 XX CORP TMULL 1,126) the old man thought himselfe no earthly man, they honoured him so much (E2 NI FICT ARMIN 44)

Not best sounds odd because a predicate like best does not really represent a prototypical NR verb. Horn (1978) and others have shown that it is possible to predict what type of verbs may become NR verbs on the basis of what Horn has called the Uncertainty Principle. He shows that there is a link between the place of the negative and "uncertainty", noting that "the negative force weakens with the distance of the negative element from the constituent with which it is logically associated" (Horn 1978: 132-133). This distancing is frequently employed with predicates that are involved in matters of (un)certainty, i.e. predicates clustering around "the notion of opinion" (Horn 1978: 207) in order to tone down the speaker's opinion for reasons of politeness (cf. also Bublitz 1992).25 One more thing should be noted in connection with think and NR. In Middle English, the Old English verbs pencan and pyncan (the first personal, the second impersonal) coalesce. However, whenever think is used as an impersonal verb with a dative experiencer, NR never occurs. We can link this up with the seem-case: i.e.

On Negative Raising in the history of English

79

Subject Raising seems to be a necessary step again before NR. Some examples: (77)

(78)

(79)

(80)

andfor dare euele jewune ne öincö hit hem no misdade, ac bied ihecdden for wise menn andfor jeape. (Ml IR RELT VICES 1 79) And 3if J^ee J)ink }jat Jdis maner of worching be not acordyng to pi disposicion in body & in soule, pou maist leue it, & take anoper... (M3 IR RELT CLOUD 129) Me thinke, Throckmorton, you neede not to haue the Statutes, for you haue them meetely perfectly. (El XX TRI THROCKM I,72.C2) Truely no, I thanke God I haue had my health pretily well, but yet me thinke my meate doth me no good of late. (E2 IS HANDO GIFFORD A4R)

Note that (77) with ne before the finite verb is not a case of NR but of multiple negation.

2.4.2.

Other verbs

The other NR verbs behave more or less like think, but believe, expect and wish all seem to develop NR later. Of the six potential examples with expect, four seem to be factive - cf. (81) (see the discussion of example (53) of think). In (81a), it seems that Mr. Bacon indeed had no such expectation; if he had, there might have been no reason for him to allter [his] speeche. (81b) is not so clear: was there no such expectation on the part of the speaker that they would meet, or was there an expectation that they would not meet? The former seems to me more likely. In (82) we find again the main verb in combination with a modal and in the perfect (cf. the discussion of (55) above), so no NR here; (83) occurs with a small clause and has a modal.

80

Olga Fischer

(81)

a.

(82)

(83)

Then Mr. Bacon spake to this effect: I expected not (quoth he) that the matter of defence shold have bene alledged for excuse, therefore I must allter my speeche ... (E2 XX TRI ESSEX 14) b. Dear Amanda, I did not expect to meet with you in Town. (Aman.) Sweet Cousin, I'm over-joy'd to see you (E3 XX COME VANBR I,34)26 yet wee command you, to tell him plainely, thct wee could never haue expected (at this tyme when wee are so fidl of trouble and expectations of danger to our state) wee should haue ben by him denyed any such matter. (E2 XX CORO ELIZ 401) I shall not expect them home untell to morrow. (E3 XX CORP EOXINDEN 322)

Example (83) is interesting: untell to morrow is a NPI, so it looks as if the negative must be part of the subclause, i.e. NR has taken place. Of the three examples with believe, the most recent one seems a clear case of NR, because it presents an attitude or hypothesis: (84)

He tells me he doth not believe that the Duke of York will go to sea again, though there are a great many about the King that would be glad of any occasion to take him out of the world... (E3 NN DIARY PEPYS VII,411)

However, the two from the Middle English period are both doubtful: (85)

a.

b.

For jee may wel knowe pat tombes & sepultures ne ben not made of such gretness ne of suche highness, Wherfore it is not to beleue pat pei ben tombes or sepultures. (M3 NI TRAV MAND 34) and more I haue nought hadde to do with pe seyd John; and I can nought beleue jjat in pis cas pe same John myght by yowr lawe any swich sute haue ageyn me as yowr lettre specifieth (M4 XX CORO WPASTON1 5)

On Negative Raising in the history of English

81

In (85a) not goes clearly with the verb be used as a modal, i.e. 'it cannot be the case that we believe...'. The same is true for (85b); not is again connected with the modal: Ί am not able to believe'. The two instances that have been found with the verb wish are both doubtful: (86)

a.

b.

therefore I would not wish any Horse-man of vertue at any time to be without it (E2 IS HANDO MARKHAM 78) But (says he) I would not wish you to approach; for I am sure you will be in love as soon as you behold her (E3 Μ FICT BEHN 189)

Both contain a modal verb. Note too, that wish is not used as a hedge verb in the periods examined (see Table 1), and also that in Present-Day English NR is allowed with wish only in infinitival constructions (so similar to hope·, see examples (9)-(10)). One would have thought that the verbs wenen and trowen would have been ideal verbs to develop NR. Semantically, they fit the class of NR verbs (they are "predicates of opinion", cf. 2.4.1 above), and they also developed as hedges early on in their history (see Table 1). There are, however, next to no examples of NR with these verbs. The reason for that must be that these two verbs became obsolete more or less at the time that NR began to be more frequent, i.e. in the Early Modern period, as the evidence given above has shown. None of the negative constructions that I have encountered with trowe (seventeen in all) do in any way represent possible cases of NR, neither when they contain a small clause, nor when they have finite or non-finite complements. In (87a) it is clearly the matrix verb itself which is negated; the subclause is not inherently negative. In the instance with a small clause (87b) there is no ambiguity in the position of not, which is clearly part of the small clause, it nought: (87)

a.

"Whan I considere thi resouns, " quod I, Ί ne trowe nat that men seyn any thing more verrayly. And yif I turne ayein to the studies of men, who is he to whom it sholde seme that he ne scholde nat oonly leven thise thinges, but ekgladly herkne hem? " (M3 XX PHILO BOETHCH 448.C1)

82

Olga Fischer

b.

"Certes, " quod I, "Y trowe it nought," and thow hast schewyd me wel that over thilke good ther nys no thyng more to ben desired " (M3 XX PHILO BOETHCH 430.C2)

Wenen is a little more interesting. There is one potential example of NR, (88), but it seems to me that what the speaker (King Lear) expresses is that he never even entertained the thought that his daughter might put him to shame. The fact that she does put him to shame (in his eyes at least) and that he had never thought of her doing it, makes the shock all the greater, as is clear from his outcry against her. (88)

for nauere ich ne wende. J)at })u me woldes J)us scanden; par-fore pu scalt beon deed ich wene;flij ut of min eceh-sene. (Ml NN HIST LAYBR 1,80)

There are quite a few examples of wenen with a negative in the finite complement, which in Present-Day English would be translated by means of NR (so with the negative in the matrix clause); quite clearly this potential is not being used yet: (89)

pe haued michel sineged and nele lete. ne bete, ne milce bidde. for pat he weneö Jiat god ne wile swo michel sinne forgiue; '... because he doesn't think that God will forgive such a great sin' (MX/1 IR HOM TRIN12 75)

There is one very interesting example, however, with a multiple negative: (90)

Nv mine leoue sustren monie temptatiuns ich habbe ow inempnet under pe seoue sunnen. nawt pah pe pusentfoldpt me is wiö itenptet. Ne mähte ich wene ham namon nomeliche nempnin. (Ml IR RELT ANCR 116)

The following is the translation given by Salu (1955: 100): "I have now, my dear sisters, mentioned to you many temptations connected with the seven sins, but not the thousandfold by which people are tempted. And I do not think that anyone could enumerate them all". Here, one can see quite clearly how the loss of Negative

On Negative Raising in the history of English

83

Concord may lead to the emergence of NR. There are also some other examples, with small clauses, where loss of double negation may be the beginning of a (semantically) raised negative: (91)

a.

b.

j e mine leoue sustren jef ei is anewil to seon ow; ne wene 3e jaer neauer god; ah leued him pe leasse. "Ah, my dear sisters, if anyone insists on seeing you, believe no good of it / don't believe any good of it and trust him the less for it" (Salu 1955: 24) (Ml IR RELT ANCR 3) adelest aire kinge. Longe bid ceuere; Jjat no wene ich nauere. pat ceuere Moddred mi mcei; wolde me biswiken. (Ml NN HIST LAYBR 11,736)

Finally, a verb that we should briefly look at, even though no cases of NR have been found in the data, is the verb hope. This is an interesting verb, because it allows NR in other languages, closely related to English, and it does allow NR in non-finite complements in Present-Day English, as observed in (10). The data related to hope show the following. Although it is a well-established verb already in Middle English (see Table 1), it develops much later as a "hedge" verb compared to think, ween and trow. The reason for this, I think, is that hope retains a strong lexical meaning and has not weakened enough in the Middle English period to be used as a hedge. This is clear from the way it is used in the Middle English examples: (92)

Y shed drede pe from pe hejt ofpe daye; y for-sope shed hope in pe. (M2 XX OLDT MPPSALT 66)

It is often used with a prepositional phrase such as in (92), which expresses in whom the hope is placed. The following examples, which show hope in combination with a negative, also make clear that the strong lexical meaning of hope cannot allow NR:

84

Olga Fischer

(93)

Why what haste, what haste, mistris Frank (quoth he) I pray you stay and drinke ere you goe, I hope a cup of newe Sacke will doe your old belly no hurt. (E2 NI FICT DELONEY 78) (AMis. Ford,.Λ) You are the happier woman. (AMis. Page.A) Let's consult together against this greasie Knight: Come hither. (AFord.A) Well: I hope, it be not so. (ΛPist:A) Hope is a curtall-dog in some affaires: Sir (AIohnA) affects thy wife. (AFordA) Why sir, my wife is not young. (E2 XX COME SHAKESP 44.C1)

(94)

It is clear in both these cases that the use of NR would clash with the hope or good intention that the speakers express, in other words it would be rather impolite. The strong lexical sense of hope is made clear in (94) by the repetition of the noun hope in the next line.27 With non-finite complements we come across examples where the negative is in an ambiguous position, i.e. it could go with the matrix verb as well as the non-finite one. These are presumably the cases that lead to NR in non-finite complements later on when it became usual for the negative adverb to be placed before the matrix verb: (95)

a.

b.

So with a friendlyfarewell, I left him as well, as I hope neuer to see in a worse estate: for he is amongst Noblemen and Gentlemen; that knowe his true worth — (E2 NN TRAV JOTAYLOR 138.C1) whose haf) eny god, hopef) he nout to holde, bote euer pe leuest we leosep alast. (M2 XX XX HUSB 7)

In the Early Modern period hope develops the weaker sense of 'expect', but in spite of this NR does not become possible. A reason may be that hope is not veiy frequent with small clause or infinitival constructions, which may have furthered the rise of NR through the ambiguous position of the negative, as observed above. It is also important that the weakened sense of hope was soon lost again. According

On Negative Raising in the history of English

85

to the OED, the senses of 'suppose', 'expect', 'think' became obsolete after 1632. In other words, I believe that the lexical meaning of hope has precluded its use as a Negative Raiser in English. The verb is still strongly connected with the noun hope. It is still a type of activity verb, 'to entertain hope', unlike in Dutch and German, where the verb has been drawn into the epistemic range expressing "likelihood" or "probability" rather than "hope". For a view that the stronger lexical meaning of English hope is not responsible for the absence of NR, see Horn (1978: 184-187).

3. Concluding remarks and suggestions for further research In this investigation, I have tried to lay bare some of the factors that may have contributed to the rise of NR in the history of English. As we have seen, there is no clear evidence for the existence of NR in the Old English period. The reasons for its absence seem to be several. I give a summary of factors obstructing and promoting NR in (96). (96)

Syntactic obstacles to NR -presence of multiple negation -presence of subjunctive (fulfilling a similar pragmatic role as NR) Factors promoting NR -loss of multiple negation -loss of subjunctive -rise of epistemic modals -emergence of Subject Raising (impersonal verbs following the path of epistemic modals) -change in the position of the adverb not -the introduction of do in the Aux category, and its similarity to the modals -bleaching of verbs expressing "opinion" to hedges, thus entering the domain of modality

First of all, NR would not be visible in monoclausal constructions because multiple

86

Olga Fischer

negation was the rule there (at least in the prose; cf. Mitchell 1985: §1628). Because of this, any element that could be negated was usually negated, including the finite verb, even if logically no negative was required there. In constructions consisting of more than one clause, the rule was that only the clause to which the negative logically belonged would be in the scope of the negation. Cases such as (13) and (16)-( 17), where we found the negative "illogically attached" to the matrix verb or the subclause, were found to be cases of Negative Concord (either forward or backward-looking) and not of NR. However, these cases, and the small-clause and infinitival ones, are of interest because upon the loss of multiple negation some of these may develop into NR constructions. We have also seen that there seems to be a link between NR verbs and modals. Semantically, the group of NR verbs resembles the modals in that they, too, are concerned with the expression of modality.28 NR verbs typically express an opinion or attitude, and are used to convey the likelihood, possibility or probability that a certain event will take place. Syntactically, theriseof epistemic modal constructions in the course of the Old and Middle English periods, in which the subject no longer received a thematic role from the modal verb (so in contrast to root modals), may well have contributed to the rise of NR with verbs like seem and happen, which also developed a non-thematic subject during this period. In my data a clear link could be established between the occurrence of Subject Raising and NR with the impersonal verbs seem, think and happen. In addition, the data also seem to show that the presence of a modal in the matrix clause may promote the use of NR (see e.g. examples (46), (50) and (62) and note 19). In view of the link between the expression of modality and NR, it is also possible that the rise of NR is connected with the loss of the morphological subjunctive. The subjunctive in Old English could be used after "predicates of opinion" - i.e. the same semantic group to which NR verbs belong - to express uncertainty. Like NR the subjunctive could be used but did not have to be used (cf. Mitchell 1985: §2034). In other words, subjunctive and NR are pragmatically linked, in that both serve to tone down the certainty of a statement. (Horn 1978: 140ff refers to the possibility of a link too, but notes that there are quite a few problems with the French and Spanish data adduced to prove this.) It could be that we have a case of grammaticalisation here. Due to the need to find new and fresh ways of expressing politeness, we find an ongoing process - in which new means are called upon to express the same thing - starting with the morphological subjunctive, moving on

On Negative Raising in the history of English

87

to epistemic modals, and next to metaphorical expressions (for Halliday's phrase, see note 28) such as I think or it probably is so. We could compare this process to the one in which for ever new items have to be invented to express intensity, i.e. very, awfully, terribly, terrifically, enormously etc., although in the "negative" process, other more purely linguistic factors have obviously played a role as well (such as the phonetic attrition of the subjunctive endings in Late Old English, Early Middle English). It should also be noted that in the negative grammaticalisation process the position of the negative does not really change because it remains all through close to the core of these expressions, i.e. the finite verb. When we look at the development of NR in the later period, I think it can be shown that there is a connection between NR and the development of NR verbs as hedges. (This link has also been extensively discussed synchronically; cf. Horn 1978: 188ff) Similarly, a link has been established between the occurrence of NR and the development of a weakened lexical sense of the proto-NR verb. This was especially clear in the case of seem. In cases where seem was used as a full verb (i.e. (26) and (34)), NR did not occur. The non-occurrence of NR with the English verb hope in the corpus shows the other side of this coin. This "weakening" would tie up nicely with the idea of grammaticalisation suggested above, because semantic bleaching is a natural ingredient in any grammaticalisation process. Finally, on the basis of the evidence presented here, it would be worthwhile to study in more detail the possible route that NR has taken, i.e. from a monoclausal (small clause) construction via infinitival complements to finite clauses. A very influential factor in this route could have been the loss of multiple negation, because after this loss there arose a tendency to attach the remaining negative to the matrix verb.29 It can hardly be an accident that with the impersonal verbs seem and happen all instances of NR involve infinitival clauses. As the historical data show, the finite clauses, which lack Subject Raising, are all without NR Similarly, the clear cases of NR with personal verbs (fourteen in all) mostly involve monoclausal (seven) or non-finite complements (three); only four having a finite clause. If this is indeed the route that NR has followed, it would tie up with the idea that natural syntactic change tends to sneak through a language, appearing first where it is least noticeable or salient (cf. Romaine 1982: 212).

88

Olga Fischer

Notes 1.

I would like to thank Willem Koopman, Bettelou Los, Wim van der Wurff, and especially Wim Klooster, for their willingness to read an earlier version of this paper, for their many helpful and constructive comments, and for preventing me from making some painiiil blunders.

2.

For more information on the Helsinki Corpus, see Kytö (1991). This corpus was found to be very useful for this investigation since it covers an extensive period, from Old English to Late Modern English, and includes most genres, from highly formal to colloquial.

3.

This change in the history of English is discussed in Tottie (1991), who notes that Neg-incorporation or 'Wonegation", as she calls it (/ have no children), is the rule until Late Middle English, and that this begins to be replaced by "Mrf-negation" (/ haven't any children) in Early Modern English.

4.

In contrast, Klima (1964) and Jackendoff (1968) propose a deep structure with a negative element in both main and subclause, followed by a rule of Neg-absorption. There seems to be little evidence for this, so not surprisingly this analysis is not followed by most linguists writing on the subject. Also, when one looks at the matter diachronically, this analysis does not seem empirically warranted. Although multiple negation was the rule in Old English prose (cf. Mitchell 1985: §§1607-1609, 1628-1629), and to some extent also still in Middle English, there was no multiple negation normally (as will be shown below) in those cases which later show negative attraction. We do find occasional instances where both the main and the subclause are negated, but as a rule only the finite verb in the subclause is negated. Although semantically the NR analysis seems very plausible, Klooster (1994) points out that there are a number of syntactic problems with this analysis within the present generative framework.

5.

Horn (1978: 137-138) makes a distinction between more liberal NPIs such as any, ever, and also yet, which "can appear freely in embedded clauses under any negated higher-clause predicates and thus cannot be used to argue for any lexically governed extraction rule", and "narrow"

On Negative Raising in the history of English

89

NPIs. Thus, it might be safer to use only narrow NPIs such as until, can help, sleep a wink, in years/weeks etc. in order to diagnose NR. 6.

Of course the sentence does work if we take at all as a modifier of claim, which is negative; but it quite clearly cannot modify have seen.

7.

The period division in the Helsinki Corpus is as follows: Ml, M2, M3 and M4 cover the Middle English period, i.e. the years 1150-1250, 1250-1350,1350-1420,1420-1550respectively.El, E2, E3 (1500-1570, 1570-1640, 1640-1710) refer to the Early Modern periods. These tags (Ml etc.) will also be used in the references to indicate the time of occurrence.

8.

For the time being I assume that the second negative elements in (13a) and (13b) are part of the subclause, i.e. that na and no are part of the infinitival phrase and the complement respectively, and do not form a socalled "bipartite negation" as with French ne ... pas (but cf. van Kemenade, this volume). So I believe that in Old English these are cases of Negative Concord (as indeed Traugott 1992: 170 describes them). The reasons for this assumption are that one does not find, as far as I know, many examples in Old English such as Ic ne cet na, where na stands unsupported by another phrase (which can be a NP, a PP, a to-infinitive or an adverbial phrase; Mitchell (1985: §1605) gives a few examples but adds that "when na negates some other element in the sentence it usually immediately precedes that element"); another reason is that cases of Negative Concord involving more than two negative elements are very frequent, and the use of ne by itself is also quite common. The situation is different in Middle English, where ne comes to be supported by not/not, where the second negative element acquires a fixed position (immediately after the finite verb; this was not the case yet in Old English), and where ne no longer occurs by itself (from the later period onwards) except in situations where the context is already negative (cf. Fischer 1992: 280 ff). The question of the nature of Old English negation clearly needs further investigation.

9.

For the problems one encounters in the analysis of Old English clause structure, see Donoghue - Mitchell (1992), and also Foster (1975), who writes on the ambiguous status of Old English pa.

90

Olga Fischer

10.

The references are exactly as given in the Helsinki Corpus. The first element identifies the period (cf. note 7), the second the "prototypical text category" (i.e. whether it represents instruction, narrative text etc.), the third the genre involved, the last is an abbreviation of the text itself followed by the page number. In Texts examined there is a list of these abbreviated text titles, which gives further information about the editions used.

11.

Klooster (forthcoming) suggests that adversative predicates like doubt select CPs which contain an abstract Neg operator. Evidence for this he finds in the fact that NPIs occur freely in such complements. For instance, I doubt that we need ask anything more, is perfectly acceptable whereas with predicates that do not govern a negative CP such sentences are out, cf. *I think that we need ask anything more. Example (16) is therefore of interest because it shows the presence of the negative explicitly.

12.

Want, although frequent in current English, was not included because in my corpus it only appears in the sense of 'lack', not yet in its modern sense.

13.

This is a fairly common meaning of seem also in Middle English; cf. this example from Chaucer (Benson 1988, Bk of the Duch 6566): Hir yen seemed anoon she wolde /Have mercy... 'Her eyes gave the appearance at once that she would have mercy...'

14.

For the notion that NP and S function as bounding nodes or bounding categories, see Chomsky (1981: 79ff). Small clauses are not maximal projections, i.e. they are clausal structures lacking Infi (cf. Chomsky 1981: 107), so in a way they are in between a full clause and a direct constituent. Because of their half-clausal status, they are of interest to the phenomenon of Negative Raising because they may have eased the negative across the clause boundary.

15.

Tottie (1991) discusses the general development in English from Nonegation (i.e. the incorporation of the negative into indefinites), which was the rule in Old English and Early Middle English, to Not-negation

On Negative Raising in the history of English

91

(the negative adverb which has the clause as its scope) from Late Middle English onwards, remarking that JVoi-negation is now the rule. 16.

Except in contrastive cases with heavy stress, as in John doesn 't seem ill, he is ill. I owe this example to Wim Klooster (p.c.).

17.

It seems to me that it is possible that for this reason, too, the auxiliaries be and have in Present-Day English are morefrequentwith Mtf-negation than the main verbs be and have, which usually have TVo-negation. These auxiliaries are also syntactically transparent, and also do not assign a thematic role to the subject. Tottie (1991:448ff) has noted this difference between main verb and auxiliaiy be and have, but has ascribed it to different factors, the main one being frequency. She believes that the high frequency of the copulas be and have would cause them to take part in the changefromNo to Mrt-negation later. (Note, however, that Aitchison (1991:78-79) writes that "[t]here is a growing body of evidence that frequently used words quite often get affected early".) Also the factors which according to Tottie (1991: 458ff) support the idea that the high frequency of copula be and have works to resist the change do not seem very convincing.

18.

I owe this reference to Wim Klooster, who discusses these cases in his lecture notes on Negative Raising; see Klooster (1994).

19.

Quirk et al. (1985: §14.36) also note the peculiar role played by the modal. They compare I don't believe that they are married with I can't believe that they are married and note that the modal prevents NR. In other words, they see the negative in the second example as belonging to the matrix clause, not the subclause. It is also possible, however, that the sentence does contain NR, but that the modal has been raised along with the negative, meaning Ί believe that they cannot be married' rather than 'it is not possible for me to believe that they are married'. It would be of interest to find out whether diachronically the NR construction with the modal was earlier, and to investigate the hypothesis that NR with nonmodal constructions became morefrequentonly after the introduction of the auxiliary do. Because do belongs to the same syntactic category as the modals, it might also take over the possibility of having the negative next to it even when logically it does not belong there (cf. the use of negatives

92

Olga Fischer

and modals in examples (44)-(45) above). 20.

The presence of a modal does not automatically imply that NR. has taken place. The following is an instance with a modal but without NR: I saye I cannot certaynelye tell or thynkefromwhence any sorow maye happen to any such thynge. (El XX PHILO BOETHCO 69) Notice the presence of the adjunct of manner certaynelye, which, as we have seen (examples (6)-(8) above) precludes the possiblity of NR

21.

The context in Klooster makes clear that this does not mean that NR verbs are themselves epistemic. It refers to the fact that the interpretative status of NR verbs is intimately connected with whether they refer to incidental or structural facts.

22.

The instance in E3 XX COME VANBR 1,64, should possibly also be included among these examples.

23.

See also E3 IR SERM TILLOTS II:ii427 and E3 NN BIO BURNETROC 136.

24.

Ellegärd writes that a number of verbs (among them many that semantically belong to the NR group) were reluctant to adopt the do-form when used in the negative and instead continued to use the order I think not rather than I don't think. The reason for this, Ellegärd (1953: 200) believes, is that "their frequent use in negative expressions had turned these into fixed phrases". Tieken-Boon van Ostade (1987) looks at Jo-less constructions in eighteenth-century prose. Again it is noteworthy that the I think not construction is especially prevalent with later NR verbs. It seems to me worthwhile to investigate whether an avoidance of NR which was a new construction, also has something to do with the late arrival of do here.

25.

Horn (1978: 131) refers here to Poutsma (1928: 105), who already noted that "the shifting of not often has the effect of softening down the negativing of a sentence". See also Bublitz (1992) for this pragmatic use

On Negative Raising in the history of English

93

of NR. It should therefore not come as a surprise that NR. seems to be associated with modality, as noted for some of the examples discussed; it is quite possible that the loss of the subjunctive, which could have a similar "softening down" function may also have led to the rise of NR constructions, but this needs further investigation. 26.

The other two instances are E2 XX CORO ELIZ 401 and E3 NN DIARY PEPYS VIII, 317.

27.

Since hope expresses "strong hope", the non-use of NR with this verb is comparable with the impossibility of NR in (8) above: just as to advise strongly clashes with the "litotes-effect" of NR, so does hope, because it lexically expresses "strong hope".

28.

Halliday (1994: 354) indeed refers to NR verbs as "metaphors of modality". He writes that in this type of modality "the speaker's opinion regarding the probability that his observation is valid is coded not as modal element within the clause, which could be its congruent realisation, but as a separate, projecting clause in a hypotactic clause complex. To the congruent form it probably is so corresponds the metaphorical variant / think it is so, with I think as the primary or 'alpha' clause". I owe this reference to Wim van der Wurff.

29.

That there has always been a strong tendency to attach the negative to the finite verb even when it does not fall under its scope was already noted by Jespersen (1917: 53).

94

Olga Fischer

Texts examined Ml IR RELT ANCR Ml NN HIST LAYBR Ml NN BIL MARGME MX/1 IR HOM TRIN Ml IR RELT VICES M2 XX XX HUSB M2 XX OLDT MPPSALT M3 XX PHILO BOETHCH M3 NN HIST BRUT M3 IR RELT CLOUD M3NITRAVMAND M3/4 IR RELT HILTON M3/4 IR SERM ROYAL M3/4 IR RELT VICES4 M4 NI ROM MALORY M4 XX CORP CPASTON M4 XX CORP TMULL El XX PHILO BOETHCO El NN BIO ROPER El XX TRI THROCKM El XX COME UDALL

Ancrene wisse, Tolkien (1962) Layamon's Brut, Brook and Lesley (1963, 1978) Margarete, D'Ardenne (1977) Old English homilies of the twelfth century, Morris (1873) Vices and virtues, Holthausen (1888) "Song of the husbandman", Historical poems of the XlVth and XVth centuries, Robbins (1959) The earliest complete English prose psalter, Bülbring (1891) Chaucer's Boethius, The Riverside Chaucer, Benson (1988) The Brut or the chronicles of England, Brie (1960 (1906)) The cloud of unknowing, Hodgson (1958 (1944)) Mandeville's travels, Hamelius (1919) Walter Hilton's eight chapters on perfection, Kuriyagawa (1967) Middle English sermons, Ross (1940) The book of vices and virtues, Francis (1942) The works ofSir Thomas Malory, Vinaver (1954) Paston letters andpapers of thefifteenthcentury, Davis (1971) The letters of Thomas Mull, in The Stonor letters and papers, 1290-1483, Kingsford (1919) George Colville, Boethius 'consolation of philosophy, Bax (1897) William Roper, The life of Sir Thomas Moore, Hitchcock (1958 (1935)) The trial ofSir Nicholas Throckmorton, Hargrave (1730) Nicholas Udall, Roister Doister. The Malone Society Reprints, Facsimile (1934)

On Negative Raising in the history of English

E2 NI FICTION ARMIN E2 XX PHILO BOETHEL E2 EX SCIM CLOWES E2 NI FICT DELONEY E2 XX CORO ELIZ E2 XX TRI ESSEX E2 IS HANDO GIFFORD E2 IR SERM HOOKER E2 IS HANDO MARKHAM E2 NN BIO PERROTT E2 XX TRI RALEIGH E2 XX COME SHAKESP E2 IR SERM SMITH E2 NN TRAV JOTAYLOR E3 NI HCT BEHN E3 XX PHILO BOETHPR E3 EX SCIO BOYLE E3 NN HIST BURNETCHA E3 NN BIO BURNETROC

E3 XX CORP EOXINDEN

95

Robert Armin, A nest of ninnies, The Shakespeare Society (1842) Queen Elisabeth's Englishings ofBoethius, Pemberton (1899) William Clowes, Treatise for the artificialI cure of struma, Da Capo Press, Facsimile (1970) The novels of Thomas Deloney, Lawlis (1961) The Edmondes papers, Butler (1913) "The trial of the Earl of Essex", The Dr. Farmer Chetham Ms, Grosart (1873) George Gifford, A handbook on witches and witchcraft, Oxford Univ. Press, Facsimile (1931) Richard Hooker, Two sermons upon part of S. Judes epistle, Da Capo Press, Facsimile (1969) Gervase Markham, Countrey contentments 1615. Da Capo Press, Facsimile (1973). The history ofthat most eminent statesman, Sir John Perrott, Rawlinson (1728) The trial of Sir Walter Raleigh, Hargrave (1730) William Shakespeare, The merry wives of Windsor, Facsimile ed. Kökeritz (1955) Henry Smith, Of usurie, Amsterdam, Facsimile (1975) All the works ofJohn Taylor the water poet, The Scolar Press, Facsimile (1977) Aphra Behn, Oroonoko, Henderson (1960) Richard Lord Viscount Preston, Boethius, (1695) Robert Boyle, Electricity & magnetism, Univ. of Oxford, Facsimile (1927) Burnet's history of my own time, Airy (1897, 1900) Gilbert Burnet, Some passages of the life and death of the Right Honourable John, Earl of Rochester, The Scolar Press, Facsimile (1972) The Oxinden and Peyton letters, 1642-1670,

96

Olga Fischer

E3 XX COME FARQUHAR E3 EX SCIO HOOKE E3 IS EDUC HOOLE

E3 IS EDUC LOCKE E3 NN HIST MILTON Χ E3 XX TOI OATES E3 NN DIARY PEPYS E3 STA LAW STAT7 E3 IR SERM TILLOTS II E3 XX COME VANBR E3 IS HANDO WALTON

Gardiner (1937) George Farquhar, The beaux stratagem, The Scolar Press, Facsimile (1972) Robert Hooke, Mcrographia, Facsimile, ed. Gunther (1968 (1938)) Charles Hoole, A new discovery of the old art of teaching schoole, The Scolar Press, Facsimile (1969) John Locke, Directions concerning education, Kenyon (1933) The works ofJohn Mlton, Vol. 10, Krapp (1932) The trial of Titus Oates, Hargrave (1730) The diary of Samuel Pepys, Latham and Matthews (1972) The statutes of the realm, Vol. VII, Dawsons of Pall Mall (1963 (1820)) Three Restoration divines: Barrow, South, Tillotson: Selected sermons, Vol. II, Simon (1976) The complete works of Sir John Vanbrugh, Vol. I, Dobree and Webb (1927) Izaak Walton, The compleat angler, Bevan (1983) (for fuller details, see Kytö 1991)

References Aitchison, Jean 1991

Language change: Progress or decay? (2nd edition.) Cambridge: Cambridge University Press.

Ahlqvist, A. (ed.) 1982 Papers from the 5th ICHL. Amsterdam: Benjamins. Bartsch, Renate 1973 '"Negative transportation' gibt es nicht", Linguistische Berichte 27: 1-7.

On Negative Raising in the history of English

97

Benson, Larry D. (gen. ed.) 1988 The Riverside Chaucer. Oxford: Oxford University Press. Blake, N.F. (ed.) 1992 The Cambridge history of the English language. Vol. II: 1066-1476. Cambridge: Cambridge University Press. Bhat, D.N.S. 1975 "A semantic constraint underlying the NEG-raising rule", Papers in Linguistics 8: 125-131. Bublitz, Wolfram 1992 "Transferred negation and modality", Journal of Pragmatics 18: 551-577. Carden, Guy 1971 "A dialect argument for not-transportation", Linguistic Inquiry 2: 423-426. Cattell, Ray 1973 "Negative transportation and tag questions", Language 49: 612-639. Chomsky, Noam 1981 Lectures on government and binding. Dordrecht: Foris Publications. Cole, Peter (ed.) 1978 Syntax and semantics. Vol. 9: Pragmatics. New York: Academic Press. Donoghue, Daniel - Bruce Mitchell 1992 "Parataxis and hypotaxis: A review of some terms used for Old English syntax", Neuphilologische Mitteilungen 93: 163-183. Ellegärd, Alvar 1953 The auxiliary do, the establishment and regulation of its use in English Stockholm: Almqvist & Wiksell. Fillmore, Charles J. 1963 "The position of embedding transformations in a grammar", Word 19: 208-231. Fischer, Olga C.M 1992 "Syntax", in: N.F. Blake (ed.), 207-^08.

98

Olga Fischer

Fischer, Olga C.M. - Frederike C. van der Leek 1983 "The demise of the Old English impersonal construction", Journal of Linguistics 19: 337-368. Fodor, Jerry A. - Jerrold J. Katz (eds.) 1964 The structure of language. Englewood Cliffs, NJ: Prentice-Hall. Foster, R. 1975 "The use of pa in Old English and Middle English narratives", Neuphilologische Mitteilungen 76: 404—414. Goossens, Louis 1982 "On the development of the modals and of the epistemic function in English", in: A. Ahlqvist (ed.), 74-84. Halliday, Michael A.K. 1994 An introduction to functional grammar. (2nd edition.) London: Edward Arnold. Hogg , Richard Μ (ed.) 1992 The Cambridge history of the English language. Vol. I: The beginnings to 1066. Cambridge: Cambridge University Press. Horn, Laurence R. 1978 "Remarks on Neg-Raising", in: Peter Cole (ed.), 129-220. Jacobs, Roderick A. - Peter S. Rosenbaum (eds.) 1970 Readings in English transformational grammar. Waltham, Mass.: Gin and Company. Jackendoff, Ray 1968 "On some incorrect notions about quantifiers and negation", Studies in transformational grammar and other topics. [Air Force Document AFCRL 68-0032, Bedford, Mass.] Jespersen, Otto 1917 "Negation in English and other languages", Det Kgl. Danske Videnskabemes Selskab. Historisk-filologiske Meddelelser 1: 1-151. Copenhagen. [1962] [Reprinted in Selected writings of Otto Jespersen. London: George Allen and Unwin.] Kastovsky, Dieter (ed.) 1991 Historical English syntax. Berlin: Mouton de Gruyter.

On Negative Raising in the history of English

99

Klima, Edward S. 1964 "Negation in English", in: Jerry A. Fodor - Jerrold J. Katz (eds.), 246-323. Klooster, Wim G. 1984 "Ontkenning en noodzakelijkheid. Observaties met betrekking tot negatie en 'moeten'", Glot 7: 63-120. 1994 Negative transportation [Unpublished MS, Instituut voor Neerlandistiek, University of Amsterdam.] forthcoming "Monotonicity and scope of negation". Kytö, Merja 1991 Manual to the diachronic part of the Helsinki Corpus ofEnglish Texts: Coding conventions and lists of source texts. Helsinki: University of Helsinki. Lakoff, George 1970 "Pronouns, negation and the analysis of adverbs", in: Roderick A. Jacobs - Peter S. Rosenbaum (eds.), 145-165. Lakoff, Robin 1969 "A syntactic argument for negative transportation", Chicago Linguistic Society 5: 140-147. Mitchell, Bruce 1985 Old English syntax. Vols. I-II. Oxford: Clarendon Press. OED = The Oxford English Dictionary 1989 Prepared by A.J. Simpson - E.S.C. Weiner. (2nd edition.) 20 Volumes. Oxford: Clarendon Press. Plank, Frans 1984 "The modals story retold", Studies in Language 8: 305-364. Poutsma, H. 1928 A grammar of Late Modern English. Part I. Groningen: Noordhoff. Quirk, Randolph - Sidney Greenbaum - Geoffrey Leech - Jan Svartvik 1985 A comprehensive grammar of the English language. London: Longman. Rissanen, Matti — Ossi Ihalainen - Terttu Nevalainen - Irma Taavitsainen (eds.) 1992 History of Englishes: New methods and interpretations in historical linguistics. Berlin - New York: Mouton de Gruyter.

100

Olga Fischer

Romaine, Suzanne 1982 Socio-historiccd linguistics, its status and methodology. Cambridge: Cambridge University Press. Salu, MB. 1955 The ancrene riwle. (The Corpus MS, translated by M.B. Salu.) London: Burns and Oates. Seuren, Pieter A.M. 1985 Discourse semantics. Oxford: Blackwells. Tieken-Boon van Ostade, Ingrid 1987 The auxiliary do in eighteenth century English: A socio-historical linguistic approach. Dordrecht: Foris Publications. Tottie, Gunnel 1991 "Lexical diffusion in syntactic change: Frequency as a determinant of linguistic conservatism in the development of negation in English", in: Dieter Kastovsky (ed.), 439^167. Trask, RL. 1993 A dictionary of grammatical terms in linguistics. London: Routledge. Traugott, Elizabeth Closs 1992 "Syntax", in: Richard Μ Hogg (ed.), 168-289. Ukaji, Masatomo 1992 " Ί not say': Bridge phenomenon in syntactic change", in: Matti Rissanen - Ossi Ihalainen - Terttu Nevalainen - Irma Taavitsainen (eds.), 453—462. Warner, Anthony R 1982 Complementation in Middle English and the methodology of historical syntax. London: Croom Helm. Woisetschläger, Erich F. 1976 A semantic theory of the English auxiliary system. Bloomington, Indiana: Indiana University Linguistics Club.

Negative Concord and Verb Projection Raising in Old English and West Flemish1 Eric Haeberli — Liliane Haegeman

1.

Introduction

In this paper we will consider certain aspects of Negative Concord in Old English. Negative Concord (NC) is the phenomenon whereby two or more negative constituents in a clause do not cancel each other out but together express a single negation. Our discussion will be based on a comparative study of Old English and West Flemish, a dialect of Dutch. The motivation for this comparative approach is that Old English and the West Germanic languages with SOV word order, such as Dutch and German and their dialects, show striking similarities with respect to word order. As for the West Flemish dialect of Dutch, it also shares basic properties with Old English with respect to the syntax of negation. However, we will see that there are interesting contrasts between Old English and West Flemish as far as NC readings are concerned, particularly in contexts which have been referred to as Verb Projection Raising structures. Apart from describing this contrast between Old English and West Flemish, our aim will be to consider briefly its theoretical implications and to propose a possible explanation for it. The paper is organised as follows. In section 2, we will consider the word order properties of Old English which are parallel to the ones found in the modern West Germanic languages. Section 3 deals with some basic aspects of the syntax of negation. We will discuss the similarities between Old English and West Flemish, and we will show that in Verb Projection Raising contexts Old English and West Flemish differ with respect to NC. In section 4 we turn to theoretical issues raised by the contrast observed in the preceding section. Adopting certain recent theoretical approaches, in particular Kayne's (1994) "antisymmetry" framework, we will propose a way of explaining the contrast between Old English and West Flemish.

102

2.

Eric Haeberli — Liliane Haegeman

Old English word order

2.1. Verb Final and Verb Second It has often been observed (see for example van Kemenade 1987) that Old English has certain word order properties which are reminiscent of languages like Dutch or German. First, with respect to subordinate clauses, the finite verb often occupies the final position. This is illustrated in (1) (finite verb in roman type). (1)

a.

b.

gif hie him pees rices ujx>n if they him the kingdom granted 'if they granted him the kingdom' (van Kemenade 1987: 16; Parker, 755) pcet Darius hie mid gefeohte secan wolde that Darius them for battle visit wanted 'that Darius wanted to seek them out in order to battle with them' (van Kemenade 1987: 19; Orosius, 45, 31)

The same pattern can be found in the modern West Germanic languages. As an illustration, consider the following data from West Flemish (see Haegeman 1992 for a description): (2)

a.

b.

da Valere in Gent een us eet (West Flemish) that Valere in Gent a house has 'that Valere has a house in Gent' oa Valere Marie morgen ziet if Valere Marie tomorrow sees 'if Valere sees Marie tomorrow'

Another similarity between Old English and the modern West Germanic languages concerns the position of finite verbs in main clauses. As shown in (3), a finite verb often occupies the second position in an Old English main clause, following any type of constituent such as a subject NP (3a), a PP (3b) or an object NP (3c). This phenomenon has been referred to as Verb Second (V2).

Negative Concord and VPR in Old English and West Flemish

(3)

a

b.

c.

103

Se swicola Herodes cwaeö to dam tungel-witegum the treacherous Herod spoke to the star-wise-men 'the treacherous Herod spoke to the astrologers' (van Kemenade 1987: 17; AHTh, I, 82, 15) On pare tide waes sum oder witega on Iudea lande on that time was some other prophet in Jews' land 'In these days there was another prophet in the land of Judah' (van Kemenade: 1987: 18; AHTh, I, 570) Maran cydde habbaö englas to Gode pome men more affinity have angels to God than men 'angels have more affinity to God than men' (van Kemenade 1987: 17; AHTh, I, 10, 3)

V2 can also be found in the West Germanic languages, as exemplified again by West Flemish: (4)

a. b.

c.

Valere eet een us in Gent (West Flemish) Valere has a house in Gent In Gent eet Valere een us in Gent has Valere a house 'Valere has a house in Gent' Marie ziet Valere morgen Marie-Obj sees Valere-Subj tomorrow 'Valere sees Marie tomorrow'

2.2. Verb Projection Raising The data in (1) and (3) illustrate the fact that finite verbs in Old English often occur in final position in subordinate clauses and in second position in main clauses and examples (2) and (4) show that these word order properties seem to be shared by the modem West Germanic languages. However, in Old English we can find several exceptions to the word order patterns in (1) and (3) (see van Kemenade 1987 for discussion). Here, we will focus on one of these exceptions, namely subordinate clauses containing a finite modal

104

Eric Haeberli — Liliane Haegeman

and a non-finite main verb. In such constructions it is possible for the finite modal to occur clause-finally (as in (lc)) but also in a position preceding the infinitive and one or more of the infinitive's complements and adjuncts. The latter option is shown in (5) (infinitival clause in square brackets). (5)

a.

b.

poet he mehte [his feorh generian] that he might his life save 'so that he might save his life' (van Kemenade 1987: 20; Orosius, 48,18) jxet hie ne mehton [pa gefarerum to eorpan bringan] that they not could the dead to earth bring 'so that they could not bury the dead' (van Kemenade 1987: 20; Orosius, 49,23)

Yet, the examples in (5) are not incompatible with the assumption made in section 2.1. that Old English shares basic word order properties with the modern West Germanic languages. For there are indeed certain West Germanic dialects, like West Flemish or Swiss German, which also allow a finite modal to occur to the left of an infinitival main verb and its complements in subordinate clauses as in (6): (6)

a.

b.

da Valere koste [zen leven redden] (West Flemish) that Valere could his life save 'that Valere could save his life' da Valere wildige [tnoaste weke no Gent goan] that Valere wanted next week to Gent go 'that Valere wanted to go to Gent next week'

The structures in (5) and (6), where the finite modal precedes the infinitive and its complements, have generally been referred to as Verb Projection Raising (VPR) structures (see e.g. Haegeman - van Riemsdijk 1986). This is the kind of construction we will focus on here. We will see that although (5) and (6) look superficially identical and thus seem to be another instance of the parallelism discussed in section 2.1. between Old English and the modern West Germanic languages, they cannot be analysed in an entirely identical way. Evidence for this claim is provided by phenomena related to the syntax of negation.

Negative Concord and VPR in Old English and West Flemish

105

3. The interaction of negation and VPR in Old English and West Flemish 3.1. Negation in Old English Before looking at the interaction of negation and VPR, let us consider some basic properties of Old English negation and the syntax of negation in general. The main aspects of Old English negation which will be relevant for our discussion are illustrated in (7). (7)

a.

b.

c.

Ic ne toweorpe da burg I not destroy the city Ί will not destroy the city' (Haeberli 1991: 54; CP, 399.31) ne mceg nan man sodre secgan not can no man more-truly speak 'Nobody can speak more truly' (Haeberli 1991: 58; Bo, 94.8) ic wyrce pa tacnu pe ncefre nan man ne geseah I do the miracles that never no man not saw cer on nanum lande before in no land Ί will do miracles that no man has ever seen before in any land' (Haeberli 1991: 58; Exodus, 34.10)

Example (7a) shows that the preverbal negative element ne is sufficient to express sentential negation. However, one or more negative elements can be added and we get what has been referred to as a Negative Concord (NC) reading. Thus, in (7b) the two negative elements ne and nan man do not cancel each other out (in contrast to, for example, a Standard Modern English sentence like Nobody didn't speak truly) but they jointly express a single negation. Similarly, in (7c) the four negative elements ncefre, nan man, ne, on nanum lande enter into a NC relation and therefore express a single negation.

106

Eric Haeberli - Liliane Haegeman

3.2. The expression of sentential negation: heads and maximal projections The way of expressing sentential negation illustrated in (7) can also be found in other languages. In Italian, for example, the preverbal element non is also sufficient to express sentential negation and it can also enter into NC readings with additional negative elements (for a discussion of negation in Italian see Zanuttini 1991): (8)

a.

b.

Gianni non ha visto Maria (Italian) Gianni not has seen Maria 'Gianni has not seen Maria' Gianni non ha visto nessuno Gianni not has seen no one 'Gianni has not seen anybody'

Similarly, West Flemish has a preverbal negative element en. Contrary to Old English however, the presence of en depends on the occurrence of at least one other negative element, as in (9a). As in Old English, additional negative elements enter into a NC relation; see (9b). (9)

a.

b.

da Valere dienen boek nie en-kent (West Flemish) that Valere that book not not-knows 'that Valere does not know that book' da Valere an niemand niets en-zegt that Valere to no one nothing not-says 'that Valere does not say anything to anybody'

Preverbal negative elements like ne, non or en have generally been analysed as heads which incorporate into the verb (cf. e.g. Pollock 1989). This hypothesis explains the fact that these elements immediately precede the verb whatever position it occupies. As for the other negative elements in the examples above, they have been treated as maximal projections which, at some level of representation, have to be in a certain configuration with the negative head to satisfy what has been referred to as the Neg Criterion (cf. Haegeman - Zanuttini 1991; Haegeman 1995a). In terms of these hypotheses, it is the configuration established due to the Neg Criterion which makes NC readings possible (cf. Haegeman 1995a: 138ff).

Negative Concord and VPR in Old English and West Flemish

107

33. Negation and Verb Projection Raising Given the general aspects of sentential negation and NC discussed in the previous subsections, let us turn here to VPR and its interaction with negation. As discussed in 2.2 above, VPR structures are embedded clauses in which the modal does not occur clause-finally but to the left of the infinitival main verb and one or more of the main verb's complements or adjuncts. What we will focus on here is the effect of VPR with respect to NC readings. Starting with Old English, we can observe that NC relations can be established between a negative element to the left of the finite modal and a negative element to the right of the modal (negative elements in roman type): (10)

a.

poet heora nan ne mehte [nanes waepnes gewealdan] that of-them none not was-able no weapon wield 'that no one of them was able to wield a weapon' (Mitchell 1985: 660, § 1596; Or. 194.18)

b.

poet him nan man ne pearf [to feormfidtume na Jsingc syllari] that him no man not need for help-in-food no thing give 'that no person has to give him help in food' (Haeberli 1991: 125; Law II, Cn, 69.1)

c.

for dam de pa Iudeiscan noldort [naefre

brucan]

because the Jews not-wanted never use nanes Ringes mid pam hoepenum no thing with the heathens 'because the Jews never wanted to use anything with the heathens' (Mitchell 1985: 663, § 1604; AEHom 5.124) In all these examples a negative constituent contained within the infinitival clause enters into a NC relation with the negative head ne and another negative constituent.2 What is interesting now is that West Flemish shows fundamentally different properties as far as the interaction of VPR and negation is concerned. Consider (11):

108

Eric Haeberli - Liliane Haegeman

(11)

a.

b. c.

da Valere niets nie en-durf [zeggen] that Valere nothing not not-dares say 'that Valere does not dare to say anything' (NC) *da Valere en-dwjt [niets zeggen] that Valere not-dares nothing say da Valere nie en-durfi [niets zeggen] that Valere not not-dares nothing say 'that Valere does not dare not to say anything' (Double Negation) *'that Valere does not dare to say anything' (NC)

In (1 la) just the order main verb - modal is inverted. This phenomenon has been referred to as Verb Raising (VR). In such a configuration, Negative Concord between the complement of the infinitival main verb (i.e. niets) and the negative elements nie and en is possible. If we turn to VPR now, (lib) already gives a first indication that VPR structures interfere with negation in West Flemish. The negative head en which depends on the presence of another negative constituent (cf. section 3.2) apparently cannot be licensed by the negative object to the right of the modal. But what is crucial now for our discussion of Old English is that VPR also interferes with NC relations in West Flemish. In (11c) the negative head en is licensed due to the presence of the negative constituent nie to the left of en. The sentence is therefore grammatical, contrary to (lib); but interestingly, (11c) can only be interpreted with a Double Negation reading. Double Negation contrasts with NC: in the former, two negative constituents in a domain cancel each other out, whereas in the latter, two or more negative elements in a domain express a single negation. In (1 lc) the negative element to the left of the modal (nie 'not') and that to the right of the modal (niets 'nothing') cancel each other out. As suggested by the second translation under (11c), a NC reading is impossible in this case. (11c) therefore contrasts clearly with the Old English examples in (10) where VPR does not seem to interfere with NC readings.

3.4. Summary To summarise, we have seen in sections 2 and 3 that Old English and West Flemish

Negative Concord and VPR in Old English and West Flemish

109

show considerable similarities with respect to word order and the syntax of negation. However, one type of word order, namely VPR, only superficially seems to be identical in the two languages. This conclusion is based on a contrast with respect to NC readings in VPR structures. We have indeed observed that in Old English embedded clauses a negative constituent occurring between a finite modal and the non-finite main verb can enter into a NC relation with a negative constituent preceding the modal. In West Flemish however, NC is not possible with such a word order pattern. The question we have to raise then is how the surprising contrast observed in section 3.3 between Old English and West Flemish can be accounted for. Given that the superficially identical word order patterns in the two languages show different properties with respect to the interpretation of negative elements, there must be something which is crucially different in Old English as compared to West Flemish. In the following section we address this issue on the basis of some recent theoretical proposals within the generative framework.

4. The contrast between Old English and West Flemish VPR structures 4.1. Different types of movement One way of accounting for the contrast between Old English and West Flemish is discussed in detail by Haeberli and Haegeman (1995). Based on standard assumptions with respect to the syntax of the West Germanic languages - in particular the assumption that projections can be head-final in these languages - the analysis proposed in Haeberli and Haegeman (1995) suggests that VPR-type patterns are the result of different kinds of movement in Old English and West Flemish. In Old English a finite modal precedes the infinitival clause due to leftward movement of the modal whereas in West Flemish this order is obtained through rightward movement of the infinitival clause. This analysis of the contrast between Old English and West Flemish is represented schematically in (12). (12)

a. b.

C C

XP XP

AuXi [YP V] t, ^ Aux [YP V]k

(Old English) (West Flemish)

110

Eric Haeberli - Liliane Haegeman

As for the contrast with respect to NC readings, the proposal made by Haeberli and Haegeman (1995) is roughly that rightward movement of the infinitival clause creates a scope island.3 A negative element contained in an infinitival clause which has been moved to the right therefore cannot enter into a NC relation with a negative constituent to the left of the modal. Hence the impossibility of the NC reading in the West Flemish example (1 lc). As for Old English, the infinitival clause has not been moved in (12a) but stays in its base position. The infinitival clause therefore does not constitute a scope island and NC readings are possible in the way illustrated in the Old English examples in (10). For more detailed discussion of this approach, the reader is referred to Haeberli and Haegeman (1995). Here we would like to pursue a diiferent approach which is in line with recent theoretical proposals made in particular by Kayne (1994).

4.2. Antisymmetric Minimalism Kayne (1994) proposes a system which is more constrained than the standard Principles and Parameters system and which therefore is potentially more promising for achieving the goal of explanatory adequacy in syntactic theory. The aspect of Kayne's so-called "antisymmetric" system which is crucial for our discussion here is that only leftward movements are allowed and that rightward movements are not and that all projections in the clause structure are head-initial. This obviously means that the explanation summarised under (12) cannot be maintained because the movement illustrated in (12b) is not part of the options provided by Universal Grammar in Kayne's system. As discussed for example by den Dikken (1994), Kayne's antisymmetric system implies that VPR patterns basically can always be treated as structures in which the infinitival clause occupies its base position because complements are always generated to the right of their head in Kayne's system. A West Flemish sentence like (13a) couid therefore simply be analysed as (13b) rather than in the standard way illustrated in (12b) above. In (13b) FP is a functional projection, i.e. a projection headed by a functional element such as a possibly abstract inflectional morpheme (see the discussion below).

Negative Concord and VPR in Old English and West Flemish

(13)

a.

b.

111

da Valere wilt dienen boek kuopen that Valere wants that book buy 'that Valere wants to buy that book' da Valere wilt [pp dienen boek kuopen]

But obviously the same analysis would also hold for Old English VPR structures. The contrast observed in section 3.3. between Old English and West Flemish NC could therefore not be explained in an entirely straightforward way in terms of an antisymmetric account of VPR. In the remainder of this paper, we will explore a way of dealing with the NC data within an antisymmetric framework. Our account will be based on two main hypotheses discussed in the following two subsections.

4.3. Truncation and VPR The first aspect of our analysis concerns the structure of the infinitival clause. In (13b) we did not determine the precise categorial status of the infinitival clause. What one might suggest now is that the contrast between Old English and West Flemish with respect to NC readings in VPR structures could actually be related to the status of the FP in (13b), i.e. that the FP is not identical in Old English and West Flemish. Evidence for the assumption that the structure of infinitival clauses may vary can be found in West Flemish. In West Flemish the distribution of elements in finite subordinate clauses is roughly as given in (14) (cf. Haegeman 1994 for more details): (14)

C - Subj - Cls - Objs - Advs - Objs - Neg - PP/AP - V

As illustrated in (15), VPR infinitival clauses contain, apart from the complementiser and the overt subject, the same elements in the same order. (15)

a.

da Valere doarom wildige [an niemand niets nie zeggen] that Valere therefore wanted to no one nothing not say 'that for that reason Valere wanted to say nothing to anyone'

112

Eric Haeberli - Liliane Haegeman b.

da Valere doarom wildige [/ morgen an Marie togen] that Valere therefore wanted it-Cl tomorrow to Marie show 'that for that reason Valere wanted to show it to Marie tomorrow'

What is interesting here is that the clitic domain (i.e. the domain following the subject, cf. 14) is not always present in infinitival structures. In the so-called IPP (Infinitivus pro Participio) construction, the infinitival clause following the modal can contain a clitic only if it follows the finite auxiliary (16a) but not when it precedes the finite auxiliary (16b): (16)

a.

b.

da Valere oa [willen [die boeken/ze an Marie togen]] that Valere had want the books/them-Cl to Marie show 'that Valere had wanted to show them to Marie' da Valere [willen [die boeken/*ze an Marie togen ]]i oat t.

The contrast in (16) suggests that when the infinitival clause in an IPP context precedes the finite auxiliary, it has a truncated structure and it therefore does not contain the structural domain where clitics are licensed.4 For reasons of space we cannot pursue this issue here (but see Haegeman 1996). What is crucial for us here is simply that, given (16), it seems to be plausible to assume that there can be variation with respect to the structure of infinitival clauses. With this in mind, let us return to the contrast between Old English and West Flemish. Two of the relevant examples, (11c) and (10a), are repeated in (17): (17)

a. b.

da Valere nie en-dürft [FP niets zeggen ] (WF: Double Negation; *NC) pcet heora nan ne mehte [pp nanes wsepnes gewealdan ] (OE: NC)

What we could argue now is that the FPs in (17a) and (17b) are different. In (17a) FP contains the locus of negation, i.e. NegP according to most recent work on negation (cf. e.g. Pollock 1989; Haegeman 1995a). (17a) would thus contain two NegPs, one in the matrix domain and one in the embedded domain, hence the Double Negation reading. As for (17b), we could assume now that the FP does not contain the projection which encodes sentential negation (NegP), i.e. that the infinitival FP is truncated in a similar fashion as the infinitival clause in

Negative Concord and VPR in Old English and West Flemish

113

(16b). Hence, there would only be one NegP in (17b), namely the matrix NegP. We assume (following Haegeman - Zanuttini 1991; Haegeman 1995a) that negative constituents have to satisfy a well-formedness condition (Neg Criterion) according to which negative constituents have to be in a specifier-head relation with a negative head at a certain level of representation. We also assume that Negative Concord is the by-product of the Neg Criterion (cf. Haegeman 1995a: 139). Based on these assumptions we can explain the NC reading in (17b) straightforwardly. The negative constituent to the right of the modal has to establish a specifier-head relation with a negative head in the matrix domain since there is no negative head in the infinitival clause.5 Hence, due to its meeting the Neg Criterion in the matrix domain, the negative object can enter into a NC relation with the negative elements of the matrix domain. In West Flemish (cf. (17a)), the presence of NegP in the infinitival clause allows the negative constituent to the right of the modal to satisfy the Neg Criterion within the infinitival clause, and a NC reading with negative elements in the matrix domain is therefore ruled out.

4.4. Movement and the Neg Criterion Since the negative object in (17b) occupies a position within the infinitival clause, the analysis outlined in section 4.3 can only hold if negative constituents in Old English do not have to move in the overt syntax to satisfy the Neg Criterion in order to enter into NC relations. This conclusion is also supported by extraposition phenomena as illustrated in (10c), repeated here under (18): (18)

for dam de pa Iudeiscan noldon [nasfre brucan] nanes Ringes midpam hcepenum (NC)

Apart from the negative constituent ncefre contained within the infinitival clause, we have another negative constituent in (18) (nanes pinges) which generally has been referred to as an extraposed constituent as it occurs to the right of the verb.6 A NC reading seems to be possible with this extraposed element. This fact from Old English contrasts with West Flemish where extraposed constituents such as the PP in (19a) cannot enter into NC relations. NC is only possible if the PP is moved to a position where it can satisfy the Neg Criterion in the overt syntax as in (19b) (see

114

Eric Haeberli - Liliane Haegeman

Haegeman 1995a: 134ff for discussion): (19)

a.

b.

da Valere nie ketent en-is van niemand (West Flemish) that Valere not pleased not-is of no one 'that Valere is not pleased with no one' (Double Negation) da Valere van niemand nie ketent en-is 'that Valere is not pleased with anyone' (NC)

The contrast between (18) and (19) thus suggests that apart from the hypothesis discussed in section 4.3 (i.e. truncation of the infinitival clause in Old English) another factor comes into play for the analysis of the VPR contrast in (17), namely the movement requirements based on the Neg Criterion. Whereas negative constituents in West Flemish have to move in the overt syntax to meet the Neg Criterion in order to enter into NC relations, negative constituents in Old English do not seem to be forced to move in the overt syntax for NC readings.7 This further assumption would account then for the fact that the negative constituent in (17b) does not have to occur in the matrix domain but can stay in the infinitival clause.8

5.

Conclusion

In this paper we have seen that although Old English and the modern West Germanic languages manifest interesting parallelisms with respect to word order, a closer investigation shows that superficially identical word order patterns can have distinct properties. The distinct properties we have observed concern the syntax of negation in VPR contexts. Whereas VPR structures interfere with NC in West Flemish, NC readings seem to be possible without restrictions in Old English VPR contexts. In the second part of this paper, we briefly sketched a possible theoretical approach to account for the surprising contrast between Old English and West Flemish. Basing our analysis on recent theoretical proposals, we suggested that the difference between Old English and West Flemish might be related to the distinct categorial status of infinitival clauses in Old English and West Flemish and to a contrast with respect to overt movement of negative elements in the two languages.

Negative Concord and VPR in Old English and West Flemish

115

Notes 1.

This paper is part of research project 11-33542.92 subsidised by the Fonds National Suisse de la Recherche Scientifique. We would like to thank the following for their help and suggestions: Anthony Kroch, Susan Pintzuk, Beatrice Santorini and the participants of the LAGB Spring Meeting 1992 (Brighton, April 1992) and the Second Diachronic Generative Syntax Workshop (University of Pennsylvania, November 1992).

2.

Some additional remarks might be necessary here with respect to examples (10b) and (10c). Contrary to the VPR structures we have seen so far, two elements precede the modal in (10b), namely a pronominal object and the subject. As we will see below (cf. 11c), the occurrence of two or more constituents to the left of the modal is not an unexpected phenomenon given the properties of VPR in the modern West Germanic dialects. We can also find other clear examples of this type in Old English: öcet se reccere da deawas & da undeawas cunne wel toscadan that the teacher the virtues & the vices can well discern 'that the teacher can distinguish virtues and vices well.' (Haeberli 1992: 42; CP, 20, 149.16) In the particular case of (10b) however, one might also argue, following van Kemenade (1987) and Pintzuk (1991), that the object is a clitic and thus has been clitcised to the position to the immediate right of the complementiser. As for (10c), we can observe that the negative complement of the verb brucan occurs to the right of the infinitive rather than to the left as one would expect. This phenomenon has been referred to as extraposition. We will return to this issue briefly below in connection with the analysis of NC readings in Old English (cf. section 4.2).

3.

This claim is supported by data involving interrogative elements (cf. Haeberli - Haegeman 1995: 96). Consider the following sentences:

116

Eric Haeberli - Liliane Haegeman (i)

(ii)

(iii)

(iv)

Wien goat-er wa moaken? who goes there 'what' make 'Who will make what?' / 'Who will make something?' 'k een wa vu jun. I have 'what' for you Ί have something for you.' Wien zont er wa willen een? who would there 'what' want have 'Who would want to have what?' / 'Who would want to have something?' Wien zout-er willen wat een? who would there want what have 'Who would want to have something /*what?' *'Who would want to have what?'

We can see from West Flemish (i) and (ii) that wa ('what') has two interpretations: the interrogative reading ('what') and the existential quantificational reading ('something'). In (i) wa has the interrogative reading by virtue of its being related to the sentence initial w/z-word, wien ('who'). In (iii) wa is outside the verb cluster and it may either be given the interrogative reading by virtue of being related to sentence initial wien or it may be given the existential reading. In (iv) wa is within the verb cluster (willen wa een) and its scope is contained by the verb cluster: wa cannot be related to the interrogative wien. Hence the interrogative reading available in (iii) is absent and only the existential reading is available. In (iv) the verb cluster is a scope island: the scope of wa is delimited to the verb cluster. 4.

Cf. Haegeman (1995b) and Rizzi (1994) for the notion of "truncated" structures.

5.

Either through LF movement or through a non-overt expletive operator (for discussion of this point, see Haegeman 1995a).

6.

In an antisymmetric system, "extraposition" would obviously not be an adequate term here as nanes fringes would not be a constituent which has been extraposed (i.e. moved to the right), but simply a constituent which has not undergone leftward movement. For simplicity's sake we will

Negative Concord and VPR in Old English and West Flemish

117

nevertheless continue using the term "extraposition". 7.

Whether negative elements have to move in the overt syntax or not seems to be subject to parametric variation. The parametric choice found in Old English would correspond, for example, to the one of Modern English or the Romance languages (for discussion of this issue, cf. Haegeman 1995a: Ch. 4).

8.

As suggested by Marcel den Dikken (p.c.), one could even argue that the VPR contrast between Old English and West Flemish is just related to this difference with respect to the effect of the Neg Criterion and that the assumption made in section 4.3 might be superfluous. One could thus propose that NegP is always optional in infinitival clauses (both in Old English and West Flemish) and that the NC reading in West Flemish (17a) is simply ruled out because the negative constituent has not moved to a position where it could meet the Neg Criterion. (17a) would then have the same status as (19a). Such an analysis would have the advantage that we could avoid a problem raised by the hypothesis put forward in section 4.3, namely the problem why the infinitival FP should have a different status in the two languages, or more precisely what the overt evidence for the language acquirer would be to make this kind of distinction. However, this alternative proposal which would not depend on the assumption made in section 4.3 would raise another problem, namely the problem of the status of NegP. The account in the text is based on the simple assumption that, if the clause structure allows it, NegP is projected whenever there is a negative constituent in a clause. Thus, an infinitival clause contains NegP if there is a constituent with negative force. To avoid projection of NegP in Old English infinitival clauses, the truncation hypothesis proposed in section 4.3 would therefore be necessary. However, if we adopted the approach discussed in this footnote, the presence or absence of NegP could not be accounted for in an entirely straightforward way.

118

Eric Haeberli — Liliane Haegeman

References Dikken, Marcel den 1994 "Minimalist Verb (Projection) Raising", Groninger Arbeiten zur Germanistischen Linguistik 37: 71-88. Haeberli, Eric 1991 The Neg Criterion and Negative Concord. [Memoire de Licence, Universite de Geneve.] 1992 Some aspects of Old English word order. [MA. dissertation, University College London.] Haeberli, Eric - Liliane Haegeman 1995 "Clause structure in Old English: Evidence from Negative Concord", Journal of Linguistics 31: 81-108. Haegeman, Liliane 1992 Theory and description in generative syntax: A case study in West Flemish Cambridge: Cambride University Press. 1994 "The typology of syntactic positions: I^Relatedness and the A/A' distinction", Groninger Arbeiten zur Germanistischen Linguistik 37: 115-157. 1995a The syntax of negation. Cambridge: Cambridge University Press. 1995b "Dutch child root infinitives and truncated structures", Language Acquisition 4: 205-255. 1996 "IPP constructions and V-movement in West Flemish", Geneva Generative Papers 3.1: 50-76. Haegeman, Liliane - Henk van Riemsdijk 1986 "Verb Projection Raising, scope, and the typology of rules affecting verbs", Linguistic Inquiry 17: 417-466. Haegeman, Liliane - Raffaella Zanuttini 1991 "Negative heads and the Neg criterion", Linguistic Review 8: 233-252. Kayne, Richard 1994 The antisymmetry of syntax. Cambridge, Mass.: MIT Press. Kemenade, Ans van 1987 Syntactic Case and morphological case in the history of English Dordrecht: Foris Publications.

Negative Concord and VPR in Old English and West Flemish

119

Mitchell, Bruce 1985 Old English syntax, Oxford: Clarendon Press. Pintzuk, Susan 1991 Phrase structures in competition: Variation and change in Old English word order. [Ph.D. diss., University of Pennsylvania.] Pollock, Jean-Yves 1989 "Verb movement, UG and the structure of IP", Linguistic Inquiry 20: 365—424. Rizzi, Luigi 1994 "Some notes on linguistic theory and language development: The case of root infinitives", Language Acquisition 3: 371-393. Zanuttini, Raffaella 1991 Syntactic properties of sentential negation: A comparative study of Romance languages. [Ph.D. diss., University of Pennsylvania.]

Multiple negation in Middle English verse Yoko Iyeiri

1. Introduction One of the most striking features of the development of English negative sentences is that multiple negation, which commonly occurs in Old and Middle English (Mazzon 1994), came to be replaced by single negation in Present-Day English.1 In Middle English it is usual for a clause to employ several negative elements which do not cancel out each other's negative sense. Examples include the following (negative elements in roman type): (1) (2) (3)

(4)

Herto ne uindestu neuer andsware (The Owl and the Nightingale 657) Hit nere [= ne were] no3t for loren (King Horn 479) pat he ne moucte no more line For gol[d\ ne silver ne for no gyue (Havelok 356-357) Wheper ['yet'] hade he no helme ne hawbergh naujser, Ne no pysan ne no plate pat pented to armes, Ne no Schafte ne no scheide to schwue ne to smyte (Sir Gawain and the Green Knight 203-205)

As examples (l)-(4) illustrate, multiple negation consists of various types. In the case of (2), for instance, the contracted adverb ne is supported by nojt 'not'. In example (4), the negative connectives nauper 'neither' and ne 'nor' are involved in multiple negation. Three major types of multiple negation are discussed later in section 3.2. Here I would define multiple negation as clauses with two or more negatives which do not cancel each other out. This definition of multiple negation does not exclude the neither... nor type, since neither and nor do not cancel each other out. Knowing that the issue is disputable, I will nevertheless include this type.

122

Yoko Iyeiri

I do not find any reasons to regard neither... nor as single negation simply because the construction is authorised in Present-Day "standard" English. As discussed in 3.2.1, examples of neither ... or are available in Middle English. It may be a historical accident that neither... nor rather than neither ...or has been established in Standard English. Fortunately, the proportion of the clauses with neither ... ne 'nor' or nor alone to the whole sample of multiple negation is as small as 1.5%, so that the overall view provided in the present paper is useful even for scholars who insist upon a different definition of multiple negation. Although multiple negation is found since the Old English period onwards, its relative frequency has to my knowledge never been investigated in any detail. It seems at least likely that the phenomenon reaches its peak only in Middle English. Miyabe (1968: 92) argues that it is only in the early period of Middle English that multiple negation overtakes single negation, and that negative clauses marked only by the adverb ne are much more common in Old English. Middle English eventually undergoes a decline in multiple negation, especially in its later period. A recent study by Mazzon (1994: 161) shows that the phenomenon had receded to a great extent already in the fifteenth century. The present paper first of all discusses the overall development of multiple negation in Middle English verse2 and then investigates in detail three different patterns of multiple negation.

2. The development of multiple negation in Middle English 2.1. Chronological and dialectal factors Multiple negation is fairly frequent from Early Middle English onwards, as Table 1 shows. The texts that I have investigated are as much as possible presented chronologically.3 As the table shows, the phenomenon reaches its peak around the time of the texts of King Horn and The South English Legendary, where more than 70% of negative clauses display multiple negation.4 After this peak, the phenomenon gradually decreases.5 Dialectal factors also play a role. English Metrical Homilies and Cursor Mundi, both from the North (LALME, I: 88, 108), clearly show an earlier reduction of multiple negation than contemporary texts from other areas. In general, northern texts are progressive in their use of negation, whereas southern texts are relatively

Multiple negation in Middle English verse

Table 1. The proportions of multiple negation Texts

Multiple negation

Single negation

Poema Morale

69.1%

30.9%

The CM and the Nightingale

59.5%

40.5%

King Horn

71.6%

28.4%

Haveiok

49.8%

50.2%

The South English Legendary

73.5%

26.5%

English Metrical Homilies

11.6%

88.4%

Genesis and Exodus

38.2%

61.8%

William of Shoreham

58.6%

41.4%

Cursor Mundi

18.2%

81.8%

Sir Ferumbras

52.3%

47.7%

Confessio Amantis

8.6%

91.4%

Handlyng Synne

21.6%

78.4%

Kyng Alisaunder

56.7%

43.3%

Sir Gawain

18.4%

81.6%

The Canterbury Tales

25.9%

74.1%

Alliterative Morte Arthure

19.9%

80.1%

Alexander and Dindimus

45.9%

54.1%

The Destruction of Troy

30.7%

69.3%

Stanzaic Morte Arthur

21.7%

78.3%

123

124

Yoko Iyeiri

conservative. Havelok, though contemporary with The South English Legendary, presents a much smaller proportion of multiple negation, which correlates with the fact that the location of Havelok is further north than that of The South English Legendary. Mcintosh (1976: 36) argues that the language of the manuscript of Havelok is attributable to a scribe from west Norfolk, whereas the manuscript of The South English Legendary is localised in Berkshire by LALME (I: 62).6 Furthermore, the relatively conservative figures of Sir Ferumbras and Alexander and Dindimus can also be explained on these grounds. The manuscript of Sir Ferumbras is localised in Devonshire, while Alexander and Dindimus was composed somewhere around Gloucester (Skeat 1878: xxix; Lumiansky 1967:107). Iyeiri (1992: 81-85) classifies the texts under discussion into the following three groups according to the extent to which the adverb ne declines in them: Group 1 includes texts where the adverb ne is preserved well; Group 2 includes texts where the use of the adverb ne has been reduced to a notable extent; and Group 3 includes texts where the adverb ne remains only sporadically. The classification is largely valid in respect to the development of multiple negation (the figures following the texts indicate the ratio of multiple negation to the whole sample of negative sentences in them): Group 1: Group 2:

Group 3:

Poema Morale (69.1%), The Owl and the Nightingale (59.5%), King Horn (71.6%), The South English Legendary (73.5%). Havelok (49.8%), Genesis and Exodus (38.2%), The Poems of William of Shoreham (58.6%), Sir Ferumbras (52.3%), Kyng Alisaunder (56.7%), Alexander and Dindimus (45.9%). English Metrical Hormlies (11.6%), Cursor Mundi (18.2%), Confessio Amantis (8.6%), HandlyngSynne (21.6%), Sir Gawain and the Green Knight (18.4%), The Canterbury Tales (25.9%), The Alliterative Morte Arthure (19.9%), The Destruction of Troy (30.7%), The Stanzaic Morte Arthur (21.7%).

The texts of Group 1 show frequent occurrence of multiple negation. They are all relatively early texts. Supposing that King Horn and The South English Legendary represent the peak of multiple negation, the texts of Group 2 show that multiple negation is in the process of reduction. They are by and large slightly later than the Group 1 texts, but progressive texts like English Metrical Homilies and Cursor

Multiple negation in Middle English verse

125

Mimdi are not included here. Finally, the Group 3 texts show a significant reduction of multiple negation. They tend to be Late Middle English texts and/or texts from northern areas of England. English Metrical Homilies and Cursor Mundi belong to this last group.

2.2. Multiple negation and style Apart from the chronological and dialectal factors, style also appears to affect the frequency of multiple negation. Among the texts that I have investigated, The Canterbury Tales illustrates the point most clearly. The proportions of multiple negation are significantly larger in formal texts such as The Knight's Tale and The Clerk's Tale than in informal texts such as The Reeve's Tale, The Canon's Yeoman's Tale and The Wife of Bath's Tale, as shown in Table 2.7 On the whole, texts in an informal style display a significant reduction of multiple negation. In the case of The Reeve's Tale, as many as 90% of negative clauses contain one negative only. Table 2. Multiple negation vs. single negation in some of The Canterbury Tales Tales

Multiple negation

Single negation

The Knight's Tale

30.0%

70.0%

The Clerk's Tale

30.7%

69.3%

The Reeve's Tale

10.7%

89.3%

The Canon's Yeoman's Tale

22.6%

77.4%

The Wife of Bath's Tale

21.6%

78.4%

For the sake of comparison, I have also investigated The Tale of Melibee and The Parson's Tale, which are both largely prose texts (Table 3). These works are in formal style and indeed multiple negation is relatively common in them. Around 40% of the negative clauses include more than one negative. The Modern English concept of incorrectness associated with multiple negation was not present around this time, since the phenomenon was welcomed more in formal style.

126

Yoko Iyeiri

Table 3. Multiple negation vs. single negation in The Tale of Melibee and The Parson's Tale Multiple negation

Single negation

The Tale of Melibee

39.6%

60.4%

The Parson's Tale

43.3%

56.7%

Tales

2.3. Multiple negation with more than two negatives

The final point to be made about the overview of Middle English multiple negation is that most examples provide only two negative elements, and those including more than two are significantly less frequent. This is the case all through the Middle English period. I have collected 1,437 examples of multiple negation in Group 1, of which as many as 1,274 (88.7%) include only two negatives. Likewise, Group 2 provides 1,174 examples of multiple negation, of which 938 (79.9%) have only two negatives. Group 3 shows the same tendency. The texts of this group present 1,092 (75.1 %) examples with only two negatives among the total of 1,454 instances of multiple negation. Especially in Group 1, the proportion of multiple negation with two negatives approaches 90%.

3. Various forms of multiple negation 3.1. Multiple negation with ne and multiple negation of two other types

Multiple negation includes various types of clauses with more than one negative, among which at least three types are worthy of note in view of the decline of the phenomenon. The first is multiple negation with the adverb ne 'not',8 as in (5) (Type 1). The second type is multiple negation with the negative conjunction ne 'norΊηοτ, as in (6) (Type 2). The third type is multiple negation where never, no, nowhere, etc.9 are repeated or occur with not (Type 3), as exemplified by (7).

Multiple negation in Middle English verse

(5) (6)

(7)

127

Wydur to wende ne wat he noght (Cursor Mundi 64) pat per ne valle in pulke hous . no qualm in al pe sere Ne gret siknesse ne honger strong. pat per of ne be[o] no fere (The South English Legendary 158/91-92) I will noghte wonde for no werre to wende whare me likes {The Alliterative Morte Arthure 3494)

The three types of multiple negation are not always exclusive. For example, multiple negation with the adverb ne can also include the conjunction ne. Indeed, it is to some extent unavoidable in Early Middle English for a negative clause to illustrate Type 1 as well as another type, since the adverb ne is almost always included in Early Middle English negation. For instance: (8)

Naue [= ne haue] no man none sikerhede (The Owl and the Nightingale 1265)

The discussion in 2.3 shows, however, that clauses with two negatives rather than three or four are dominant in Middle English, so that it should not be so frequent for two or three of the above types to occur simultaneously in one negative clause. Table 4 shows how often each type occurs in every one hundred negative clauses. The table indicates that multiple negation of Type 1 is the most frequent at least up to Group 2, while in Group 3, multiple negation of this type becomes much less frequent. By contrast, multiple negation of Types 2 and 3 is not asfrequentto begin with, and remains more or less stable during Middle English. Thus, the rise and fall of Type 1, but not that of Types 2 and 3, directly affects the rise and fall of multiple negation. Type 1 shows a gradual increase up to King Horn and The South English Legendary partly due to the increase of the type ne ... not (the adverb ne reinforced by not), after which it starts to decline.10 Clearly, the sharp decline of Type 1 and the sharp decline of multiple negation as a result arise from the decline of the adverb ne in later Middle English. That negative clauses with the adverb ne are the key to the development of Middle English multiple negation would also explain the change of the type of single negation. Table 5 shows that the negative element in single negation differs significantly in Groups 1,2, and 3. In Group 1, the negative element is nearly always

128

Yoko Iyeiri

Table 4. Multiple negation of Types 1, 2, and 3 (per one hundred negative clauses) Texts Group 1

Group 2

Group 3

Type 1

Type 2

Type 3

Poema Morale

67.0

36.2

7.4

The (M and the Nightingale

59.5

11.7

3.2

King Horn

71.6

8.8

4.9

S. E. Legendary

72.1

6.8

4.0

Havelok

40.1

22.1

0.7

Gen and Exodus

33.3

10.1

2.2

William of Shoreham

55.3

13.9

4.2

Sir Ferumbras

48.7

6.6

6.8

Kyng Alisaunder

53.1

17.8

6.0

Alexander and Dindimus

29.7

24.2

7.0

English Metrical Homilies

6.9

5.6

0.4

Cursor Mundi

6.5

10.7

3.7

Con/. Amanlis

3.8

3.6

2.5

Handfyng Syime

6.3

14.8

5.8

Sir Gawain

3.2

12.1

8.9

The Canterbury Tales

17.5

13.9

6.8

All. Morte Arthure

2.6

15.0

7.8

The [destruction of Troy

1.3

26.1

8.6

Stanzaic Morte Arthur

14.4

7.6

3.7

Multiple negation in Middle English verse

129

Table 5. Elements used in single negation ne 'not'

not

Poema Morale

28

0

0

1

The Owl and the Nightingale

98

0

1

1

King Horn

28

0

0

1

S.E. Legendary

374

10

26

5

Havelok

50

24

53

7

Gen and Exodus

51

33

53

4

Wm. ofShoreham

96

47

40

5

Sir Ferumbras

109

81

91

1

Kyng Alisawider

149

20

55

5

Alexander arid Dindimus

6

21

60

6

English Metrical Homilies

13

96

95

1

Cursor Mundi

71

262

442

12

Conf. Amantis

130

362

492

6

HandfyngSynne

103

453

518

45

Sir Gawain

18

40

92

5

The Canterbury Tales

147

543

688

19

All. Morte Arthure

17

39

188

2

The Destruction of Troy

9

128

240

40

Stamme Morte Arthur

24

81

193

2

Texts

never, no etc.

neither, ne, nor

'

130

Yoko Iyeiri

the adverb ne. Clauses with the adverb ne alone are potential candidates for multiple negation in that they can later develop the form ne ... not. Multiple negation is still in the process of increasing in these texts. That never, no, nowhere, etc. and not hardly stand alone in Group 1 suggests that the decline of the adverb ne has not made much progress by this stage, although in The South English Legendary there are a handful of instances of single negation with not alone or never, no, nowhere, etc. alone, which is a product of the disappearance of the adverb ne.n In Group 2, never, no, nowhere, etc. and not are more and more frequently employed as a marker of single negation. In general, single negation of this new type has surpassed in number the single negation marked by the adverb ne in this group. One notable feature of Table 5 is that single negation with the adverb ne alone has receded significantly in Alexander and Dindimus. It is the latest text of this group and shows the linking feature to Group 3. In Group 3, the decline of the adverb ne is reaching the final stage, although single negation of the early type marked by the adverb ne is still available.

3.2. Multiple negation with the negative conjunction ne 'nor'/nor 3.2.1.

The negative conjunction ne 'nor,trior and the conjunction or

In contrast to Type 1 discussed above, multiple negation of Types 2 and 3 is relatively stable throughout Middle English and does not reveal any obvious process of decline (see Table 4). Multiple negation of Type 2 merits attention, however. It has been noted that the conjunction netnor is freely employed in negative clauses in Middle English where Present-Day English would employ and or or (Jack 1978b: 29). That the occurrence of Type 2 is relatively stable throughout the Middle English period suggests that the replacement of ne 'nor'trior by and or or did not make much progress in Middle English. Late Middle English texts, however, reveal a few instances of or where netnor could have been employed. The following is an example from Cursor Mündt (9)

Bot moght i neuer gete hider til, Μ child a-gain, for god or il (MS Cotton Vespasian A iii of Cursor Mundi 8677-8678)

Multiple negation in Middle English verse

131

cf. and mijt I neuer gete bidder tille. my childe agayne for gode ne ille (MS Fairfax 14 of Cursor Mundi) cf. Bot miht i neuer hider tille, Mi child gett for godd nor ille (MS Göttingen theol. 107 of Cursor Mundi) cf. But myjt I neuer hidur tille No childe gete for good nor ille (MS Trinity College Cambridge R.3.8. of Cursor Mundi) Since ne/nor alternates with or depending upon the manuscript, netnor and or are free alternatives in this context, and the above example shows that the scribe of MS Cotton Vespasian, at least, was progressive enough to employ or. In some cases, or occurs in a parallel combination with netnor. (10)

And wyle nat jeue hym for to lere, Clergye ne craft or ouper mystere {Handlyng Synne 1201-1202)

Or was used instead of nelnor even in a context where a modern reader would expect the use of ne/nor. (11)

For pe soules are no pyng Wrsheped wyp pat offiyng, NoJjer vs to cunseyle or to rede (Handlyng Synne 7965-7967)

I came across just under 70 examples of or in negative clauses that could have been written as ne/nor, around half of which arefromChaucer. In view of the total number of instances of Type 2 (1,381 examples), this figure is very small indeed, making up only about 5%. What is striking is that most of the instances in question are found in texts which belong to Group 3, i.e. Cursor Mundi, Confessio Amantis, Handlyng Synne, The Canterbury Tales, The Destruction of Troy, and The Stanzaic

132

Yoko lyeiri

Morte Arthur. Thus the development of or in negative sentences which leads to the decline of Type 2 is attested mainly in Late Middle English, and even here examples of or are veiy much limited.

3.2.2.

The clause conjunction ne 'nor,/nor and and

In contrast to the case of or, the conjunction and does not seem to be a free variant to ne 'norΊηοτ. The context where and appears is in many ways different from the context of ne/nor. For example, Jack (1978b: 30-33; 1978c: 67-69), having examined some Middle English prose texts, points out that and is preferred to ne/nor when a negative clause is conjoined to a preceding positive clause, while conversely ne/nor is preferred to and when two negative clauses are combined. In example (12), for instance, a negative clause is attached to a preceding positive one by and, while in example (13), two negative clauses are combined by ne: (12)

(13)

Fyrst he watzfitndenfautlez in his jyue wyttez, And efte fayled neuer pe jreke in his jyue fyngres (Sir Gawain and the Green Knight 640-641) Pe ponder ne had noujth ben herd, Ne pe sonne ne had ben yseye, For pe dust and pe poudreye (Kyng Alisaunder 2174-2176)

This applies to verse texts. So long as the functional distinction between and and ne/nor is preserved, the replacement of ne/nor by and (hence the reduction of Type 2) does not occur. My investigation of Middle English verse texts reveals that on the whole the function of ne/nor and that of and are not interchangeable. In the interest of space, Table 6 shows the situation of only the last three texts in Group 3.12

Multiple negation in Middle English verse

133

Table 6. And and minor which connect two clauses Positive+Negative

Negative+Negati ve

Alliterative Morte Arthure

and miner

29 0

0 9

The Destruction of Troy

and minor

62 3

4 58

Stanzaic Morte Arthur

and nelnor

9 0

0 4

The functional distinction between nelnor and and is preserved even in these rather late texts. As for earlier texts, the tendency is, needless to say, very much the same. The tendency to employ nelnor in connecting two negative clauses and to employ and in conjoining a negative clause to a preceding positive one is generally attested in Middle English, which is consistent with the fact that Type 2 does not decline even in Late Middle English. Besides this major point, Jack (1978b: 30) proposes that interrogative clauses tend to be introduced by and instead of nelnor in Early Middle English prose. Since nelnor comes to be gradually replaced by and, later Middle English can be expected to present an even more pronounced tendency to employ and in this syntactic context. However, Jack (1978c), who discusses later Middle English prose, no longer mentions this syntactic context. Apparently, either the tendency is rather weak or no relevant examples are attested. At least the data of the present study does not contradict the inference that and is common before an interrogative clause all through the Middle English period. The sample provides fifteen examples of interrogative clauses introduced by a conjunction, which is always and. An illustrative example is provided below: (14)

Hyt may weyl be hymselfhap myght, Syn je drede hys croys by syght. And ys nat jyt pe same tre Hap pat tokene more myght pan je? (.Handlyng Synne 8237-8240)

134

Yoko lyeiri

The examples are all preceded by a positive clause, however, and therefore do not strongly support the conjecture. The employment of and in these examples may simply be ascribable to the nature of the preceding clause, which is positive. However, the tendency to employ and before an interrogative clause is expected in any case, since negative interrogative clauses often provide a positive presupposition, which perhaps induces the employment of and rather than minor.

3.2.3.

The negative conjunction ne 'norΊηοτ in clause-internal positions and and

The conjunction nelnor is also employed to connect elements or phrases in a negative clause, and in this case there is a semantic distinction between the use of nelnor and and in Middle English. As Jack (1978b: 33-38; 1978c: 69-70) argues, there is a single scope of negation when and is employed whereas the use of nelnor indicates that two elements connected by the conjunction are negated one by one. Hence the double scope of negation. This is the case not only in prose but also in the verse texts that I have investigated. And and nelnor are not free alternatives. In the following case, for example, hole and fere are combined under a single scope of negation and therefore hole andfere is treated as a set: (15)

Alias! that he nere [= ne were] hole and fere ['well and strong']/ (The Stanzaic Morte Arthur 411)

As a result of the semantic feature of and which connects elements tightly under a single scope of negation, and isfrequentlyfound in a context where the syntactic connection of elements is fairly tight. For instance, and often occurs where two elements are dominated by a single shared preposition, as illustrated by (16) and (17): (16) (17)

Of erf and wimmen leue ic nogt (Genesis and Exodus 3079) Sua that thar was na herberie To Josep and his spouse Marie (English Metrical Homilies 63/5-6)

Multiple negation in Middle English verse

135

The constituents dominated by a single shared preposition are by nature combined together under the scope of the preposition. In the above examples, both erfand wimmen and Josep and his spouse Marie are a set of closely linked elements. By contrast, when ne 'norΊηοτ is employed, prepositions are often repeated: (18)

(19)

By-lefpou in no wychecrqfi, Ne ine none teliinge, Ne forpe inne none ymage self {The Poems of William ofShoreham 92/177-179) Certes, that were grete pyte, So As man nod neityr yit more Off biaute ne of bounte {The Stanzaic Morte Arthur 1737-1739)

The tendency is also attested with infinitives. An infinitive marker can be shared by two elements combined by and, as the following example illustrates: (20)

Vnto the fyfty day at the morne They lefie not for to synge And Rede {The Stanzaic Morte Arthur 3896-3897) cf. Jt ne was non so litel knaue For to leyken ne for to plawe, Pat he ne wode with him pleye {Havelok 950-952)

This is by no means absolute, however, but simply a tendency which arises from the semantic distinctions found between the usages of ne/nor and and. In fact, the following example provides both a repetitive preposition and a shared preposition along with the conjunction ne/nor. (21)

Pat wyl I welde wyth guod wylle, not for pe wynne golde, Ne pe saynt, ne pe sylk, ne pe syde pendaundes, For wele ne for worchyp, ne for pe wlonk werkkez {Sir Gawain and the Green Knight 2430-2432)

136

Yoko Iyeiri

As discussed so far, Middle English occasionally uses the conjunction ne/nor where Present-Day English would employ and or or instead. The development of and and or in place of ne/nor is one of the aspects of the decline of multiple negation, which did not however take place to any significant extent in Middle English, although the slightly extended use of or in Late Middle English suggests that the process has certainly started by this time. Multiple negation of Type 2 remains in the language well into the Early Modern English period. In fact, the fairly frequent use of nor as one of the negative elements in a clause in Early Modern English has been well documented (Sugden 1936: 439; Franz 1939: 410).

3.3. Multiple negation where never, no, etc. are repeated or occur with not To turn to multiple negation of Type 3, relevant examples are by no means common in Middle English, as mentioned above. Linguistically, the development of non-assertive forms such as ever and any instead of redundant never, no, nowhere, etc. causes the decline of multiple negation of this type (Jack 1978c: 70; Burnley 1983: 60; Blake 1988: 106; Fischer 1992: 283-284; Tieken 1995: 125-127).13 The development of non-assertive forms is attested in the texts which I investigated, especially in Group 3, but never to a great extent. This is consistent with the fact that the occurrence of Type 3 does not decrease much even in Late Middle English. Some illustrative examples of non-assertive forms in negative clauses follow: (22)

(23) (24)

(25)

pat was par neuer par-of mad oght ['anything'], Til pat pe croice ['cross'] par-of was wroght (iCursor Mundi 8847-8848) Sua hei na-thing was euer wroght (Cursor Mundi 9465) That I unto mi lyves ende Ne wol hire serven everemo (Confessio Amantis 273/1734-1735) She shal noujt to any be sette Wyp outyn leue of my maumette (Handlyng Synne 189-190)

Multiple negation in Middle English verse

(26)

(27)

137

Late vs neure wyp vnryght Coueyte oght averts hys wyll, No wyp auaryce to holde for yll (Handlyng Syrme 6512-6514) Hit was neuer herd, as I hope, sith heuyn was ο loft, In any coste where ye come but ye were clene victorius (The Destruction of Troy 1100-1101)

They tend to be found in relatively late texts. The sample provides 48 examples of non-assertive forms in negative clauses,14 of which as many as 39 occur in texts of Group 3. The instances are found in: Cursor Mundi (5x), Confessio Amantis (lOx), Handlyng Syrme (18x), and The Destruction of Troy (6x). By contrast, non-assertive forms occur in negative clauses only sporadically in texts of Groups 1 and 2. For example: (28) (29)

(30) (31)

(32)

For hi ne ssolde horn temie nojt. eny ping to lede {The South English Legendary 333/170) Ne acombre naujt py wyt any mo, To meche to penche hou hyt was po (The Poems of William ofShoreham 138/253-254) Ac noujth ysen hym in any place (Kyng Alisaunder 1312) We maken no medisine . no no man prayen Wip ony hapelene help . to helyn owe bodius (Alexander and Dindimus 319-320) & jit may per no man . in any maner wise Wip solepne sacrifice . serue hem at onus (Alexander and Dindimus 734-735)

Examples are found in The South English Legendary (lx), William of Shoreham (1 x), Sir Ferumbras (1 x), Kyng Alisaunder (1 x) and Alexander and Dindimus (5x).15 The South English Legendary is the only text in Group 1 that yields an example of non-assertive forms (see example (28)). As a matter of fact, examples are sparse even in the texts of Group 2, except in Alexander and Dindimus, where five instances are found. Alexander and Dindimus is the latest in date in this group.

138

Yoko Iyeiri

4. Conclusions The points discussed above can be summarised as follows: from the beginning of the Middle English period, well over 50% of negative clauses include more than one negative, and most examples of multiple negation include only two negatives rather than three or four. Middle English observes both the rise and fall of multiple negation, although the former process does not stand out since it occurs concurrently with the decrease of multiple negation. As far as the texts investigated in the present study are concerned, the peak of the phenomenon was reached around King Horn and The South English Legendary. Multiple negation then starts to decline, which was caused mainly by the decline of the adverb ne. By the stage of Group 3, multiple negation shifts to single negation to a large extent. From a dialectal point of view, the decline of multiple negation occurs much earlier in northern parts of England than in southern areas. Two Northern texts, English Metrical Homilies and Cursor Mmdi, display a particularly early decline of the adverb ne and hence the early decline of multiple negation. Moreover, Chaucer's Canterbury Tales shows that formal writings tend to include a larger proportion of multiple negation than informal writings. In relation to the decline of multiple negation, this paper discussed the following three types separately: (1) multiple negation with the adverb ne (Type 1); (2) multiple negation with the negative conjunction ne 'nor'/nor (Type 2); and (3) multiple negation where never, no, nowhere, etc. are repeated or occur with the adverb not (Type 3). On the whole, multiple negation of Type 1 is the commonest, and at the same time this is the type that undergoes the sharpest decline in Late Middle English. Multiple negation of Types 2 and 3 is not common, but is preserved even in Late Middle English. The use of or (instead of ne/nor) and of non-assertive forms such as ever and any (instead of never and no), which leads to the decline of multiple negation of Types 2 and 3, is encountered particularly in Late Middle English, although never to a large extent.

Multiple negation in Middle English verse

139

Notes 1.

As Barber (1976:283) and Austin (1984: 142) note, multiple negation becomes non-standard during the seventeenth century. Even today, the phenomenon has not entirely been eliminated. See also Tieken (1982: 284).

2.

Since Jack's (1978a, 1978b, and 1978c) substantial study deals with Middle English prose, the present paper analyses Middle English verse. I have investigated: Poema Morale (Digby A 4), ed. J. Zupitza (1878); The Owl and the Nightingale (Cotton Caligula A. ix), ed. E.G. Stanley (1960); King Horn (Cambridge University Library, Gg 4.27(2)), ed. J. Hall (1901); Havelok (Laud Misc. 108), ed. G.V. Smithers (1987); The South English Legendary (Cambridge, Corpus Christi College 145), ed. C. D'Evelyn and A.J. Mill (1956-1959), vol. I only; English Metrical Homilies (Edinburgh, Royal College of Physicians), ed. J. Small (1862); Genesis and Exodus (Cambridge, Corpus Christi College 444), ed. O. Arngart (1968); The Poems of William of Shoreham (British Library, Additional 17376), ed. M. Konrath (1902); Cursor Mmdi (Cotton Vespasian A iii), ed. R Morris (1874-1875), 11.1-10,122; SirFerumbras (Ashmole 33), ed. S.J. Herrtage (1879); Confessio Amantis (Fairfax 3), ed. G.C. Macauley (1900), pp. 1-300; Handlyng Synne (Bodley 415), ed. I. Sullens 1983); KyngAlisaunder (Laud 622), ed. G.V. Smithers (1952); Sir Gawain and the Green Knight (British Library, Cotton Nero A. x), ed. J.RR Tolkien, E.V. Gordon, and revised by N. Davis (1967); The Canterbury Tales (Ellesmere), ed. L.D. Benson, et al. (1987), excluding The Tale of Melibee and The Parson's Tale, both written wholly or mostly in prose; The Alliterative Morte Arthure (Lincoln, Cathedral Chapter Library 91), ed. M. Hamel (1984); Alexander and Dindimus (Bodley 264), ed. W.W. Skeat(1878); The Destruction of Troy (Hunterian 388), ed. G.A. Panton and D. Donaldson (1869-1874), pp. 1-330; The Stanzaic Morte Arthur (Harley 2252), ed. J.D. Bruce (1903).

3.

For the ordering of the texts, I have primarily relied on the manuscript dates of each text.

4.

MS Gg 4.27(2) (Cambridge University Library) of King Horn has been dated variously in the range between the mid-thirteenth and the midfourteenth centuries (Allen 1984: 3), whereas MS Corpus Christi College

140

Yoko Iyeiri

145 (Cambridge) of The South English Legendary perhaps dates back to the early fourteenth century (D'Evelyn 1956-1959,111:5; D'Evelyn 1970: 414; and Görlach 1974: 78), although the original legendaries may have started in the late thirteenth century (Görlach 1976: 8). 5.

Multiple negation has often been regarded as a means of emphasis (Brook 1958: 146; Miyabe 1968: 92; Elliott 1974: 401; Brook 1976: 65; Burnley 1983: 66). The frequent occurrence of multiple negation in Middle English as shown in Table 1 suggests, however, that it was not employed only to provide the negation with emphasis. See also Tieken (1995: 27-28).

6.

As mentioned above, I have investigated only the first volume of The South English Legendary, ed. D'Evelyn and Mill (1956-1959). The language of this part of the text is localised in Berkshire, although another part of the manuscript is characterised by linguistic features from Hampshire (LALME, I: 62).

7.

For stylistic differences within The Canterbury Tales, see Muscatine (1957: 66-243).

8.

Contracted forms of negation such as nis (< ne is), nam (< ne am), and nadde (< ne hadde) are included in this group.

9.

Nevermore, nothing etc. are all included under this category. Also orthographic variants are included.

10.

The form ne ... not is most frequently attested in The South English Legendary. 1,152 examples of multiple negation in the text include 435 examples (37.8%) of ne ... not. Compare these figures with Poema Morale, an earlier text, where 65 examples of multiple negation include only nine instances of ne ... not (13.8%).

11.

Since most examples of multiple negation include only two negatives, the disappearance of the adverb ne would lead to the rise of single negation.

12.

Examples in which ne 'norΊηοτ is the sole negative element are not included, since nelnor in this case cannot be replaced by and.

Multiple negation in Middle English verse

141

13.

This process may be accelerated by prescriptivism, once the use of nonassertive forms such as ever and any in negative clauses is reasonably established and the redundancy of negative elements comes to be recognised by English speakers. The idea that multiple negation is ungrammatical can occur when a large number of examples of multiple negation with the adverb ne plus another negative element undergoes the decline of ne, resulting in a large number of newly produced single negation. As often pointed out in previous studies, the influence of Latin grammar may also be one of the factors to encourage the use of single negation (Sweet 1892-1898,1: 1520; Curme 1931: 139-140; Jespersen 1909-1949, V: 451; Kisbye 1971-1972,1: 240; Leith 1983: 111; Tieken 1994: 360).

14.

As in the example below, non-assertive forms occasionally occur in a clause with an element which has a negative connotation. These examples are not included in the figure. Vnne|)e lastep oght pat men bye Wyp pt jtf wunne wyp marchaimdye. Vnnejse any pat hap gode grace To lyue weyl wyp swych purchace CHandlyng Synne 9477-82).

15.

All examples illustrate any and its orthographic variants.

142

Yoko Iyeiri

References Allen, Rosamund (ed.) 1984 King Horn: An edition based on Cambridge University Library MS Gg.4.27(2). Garland Medieval Texts 7. New York: Garland. Arngart, Olof (ed.) 1968 The Middle English Genesis and Exodus. Lund Studies in English 36. Lund: Gleerup. Austin, Frances 1984 "Double negatives and the eighteenth century", in: N.F. Blake Charles Jones (eds.), 138-148. Barber, Charles 1976 Early Modern English. London: Andre Deutsch. Benson, Larry D. et al. (eds.) 1987 The Riverside Chaucer. (2nd edition.) Boston: Houghton Mifflin. Blake, N.F. 1988 "Negation in Shakespeare", in: Graham Nixon - John Honey (eds.), 89-111. Blake, N.F. (ed.) 1992 The Cambridge history of the English language. Vol. Π: 1066-1476. Cambridge: Cambridge University Press. Blake, N.F. - Charles Jones (eds.) 1984 English historical linguistics: Studies in development. [CECTAL (The Centre for English Cultural Tradition and Language) Conference Papers Series 3. University of Sheffield.] Brook, G.L. 1958 A history of the English language. London: Andre Deutsch. 1976 The language of Shakespeare. London: Andre Deutsch. Bruce, J. Douglas (ed.) 1903 Le Morte Arthur: A romance in stanzas of eight lines. EETS e.s. 88. London: Oxford University Press. Burnley, David 1983 A guide to Chaucer's language. London: Macmillan. Curme, George O. 1931 Syntax. Boston: D.C. Heath.

Multiple negation in Middle English verse

143

D'Evelyn, Charlotte 1970 "Collections of Saints' Legends", in: J. Burke Severs (ed.), 413-429, 556-559. D'Evelyn, Charlotte - Anna J. Mill (eds.) 1956-1959 The South English Legendary. 3 Vols. EETS o.s. 235, 236, and 244. London: Oxford University Press. Elliott, Ralph W.V. 1974 Chaucer's English. London: Andre Deutsch. Fernandez, Francisco - Miguel Fuster - Juan Jose Calvo (eds.) 1994 English historical linguistics 1992. Amsterdam: John Benjamins. Fischer, Olga 1992 "Syntax", in: N.F. Blake (ed.), 207-408. Franz, Wilhelm 1939 Die Sprache Shakespeares in Vers und Prosa. (4th edition.) Halle: Niemeyer. Görlach, Manfred 1974 The textual tradition of the South English Legendary. Leeds Texts and Monographs n.s. 6. Leeds: The University of Leeds, School of English. 1976 An East Midland revision of the South English Legendary. Middle English Texts 4. Heidelberg: Carl Winter. Hall, Joseph (ed.) 1901 King Horn: A Middle English romance. Oxford: Clarendon Press. Hamel, Maiy (ed.) 1984 Morte Arthure: A critical edition. Garland Medieval Texts 9. New York: Garland. Heritage, Sidney J. (ed.) 1879 The English Charlemagne romances. I: Sir Ferumbras. EETS e.s. 34. London: Oxford University Press. Iyeiri, Yoko 1992 Negative constructions in selected Middle English verse texts. [Unpublished Ph.D. thesis, University of St. Andrews, Scotland.] Jack, George B. 1978a "Negative adverbs in Early Middle English", English Studies 59: 295-309.

144 1978b 1978c

Yoko Iyeiri "Negative concord in Early Middle English", Stadia Neophilologica 50: 29-39. "Negation in Later Middle English prose", Archivum Linguisticum n.s. 9: 58-72.

Jespersen, Otto 1909-1949 A Modem English grammar on historical principles. 7 Vols. Heidelberg: Carl Winter. Kisbye, Torben 1971-1972 An historical outline of English syntax. 2 vols. Aarhus: Akademisk Boghandel. Konrath, Μ (ed.) 1902 The poems of William of Shoreham EETS e.s. 86. London: Kegan Paul, Trench, Trübner. LALME = A Linguistic Atlas of Late Mediaeval English 1986 Angus Mcintosh - M.L. Samuels - Michael Benskin. 4 Vols. Aberdeen: Aberdeen University Press. Leith, Dick 1983 A social history of English. London: Routledge. Lumiansky, R.M. 1967 "Legends of Alexander the Great", in: J. Burke Severs (ed.), 104-113,268-273. Macaulay, G.C. (ed.) 1900 The English works of John Gower. EETS e.s. 81. London: Oxford University Press. Mazzon, Gabriella 1994 "OE and ME multiple negation: Some syntactic and stylistic remarks", in: Francisco Fernandez - Miguel Fuster - Juan Jose Calvo (eds.), 157-169. Mcintosh, Angus 1976 "The language of the extant version of Havelok the Dane", Medium /Evum 45: 36-49. Miyabe, Kikuo 1968 "On some features of negative sentences of Early Middle English". Studies in English Literature, English Number 1968: 83-103.

Multiple negation in Middle English verse

145

Morris, Richard (ed.) 1874-1875 Cursor Mmdi: A Northumbrian poem of the XFVth century. Parts Ι-Π. EETS o.s. 57 and 59. London: Oxford University Press. Muscatine, Charles 1957 Chaucer and the French tradition: A study in style and meaning. Berkeley: University of California Press. Nixon, Graham - John Honey (eds.) 1988 An historic tongue: Studies in English linguistics in memory of Barbara Strang. London: Routledge. Panton, G.A. - D. Donaldson (eds.) 1869-1874 The "Gest Hystoriale " of the Destruction of Troy: An alliterative romance translatedfrom Guido de Colonna's "Historia Troiana". 2 vols. EETS o.s. 39 and 56. London: Oxford University Press. Severs, J. Burke (ed.) 1967 A manual of the writings in Middle English 1050-1500. Vol. 1. New Haven: The Connecticut Academy of Arts and Sciences. 1970 A manual of the writings in Middle English 1050-1500. Vol. 2. New Haven: The Connecticut Academy of Arts and Sciences. Skeat, Walter W. (ed.) 1878 The alliterative romance of Alexander and Dindimus. EETS e.s. 31. London: Oxford University Press. Small, John (ed.) 1862 English metrical homilies. Edinburgh: Paterson. Smithers, G.V. (ed.) 1952-1957 Kyng Alisaunder. 2 Vols. EETS o.s. 227 and 237. London: Oxford University Press. 1987 Havelok. Oxford: Clarendon Press. Stanley, Eric Gerald (ed.) 1960 The Owl and the Nightingale. London: Nelson. Sugden, Herbert W. 1936 The grammar of Spenser's "Faerie Queene ". [Language Dissertations 32.] Philadelphia: University of Pennsylvania Press for Linguistic Society of America.

146

Yoko fyeiri

Sullens, Idelle (ed.) 1983 Robert Manrtyng of Brunne: Handlyng Synne. [Medieval and Renaissance Texts and Studies 14.] Binghamton, NY: Center for Medieval and Early Renaissance Studies. Sweet, Henry 1892-1898 A new English grammar: Logical and historical. 2 vols. Oxford: Clarendon Press. Tieken-Boon van Ostade, Ingrid 1982 "Double negation in the eighteenth century", Neophilologus 61: 278-285. 1994 '"After a copye unto me delyverd': Multiple negation in Malory's Morte Darthur", in: Francisco Fernandez - Miguel Fuster - Juan Jose Calvo (eds.), 353-364. 1995 The two versions of Malory's Morte Darthur. Multiple negation and the editing of the text. Cambridge: D.S. Brewer. Tolkien, J.R.R. - E.V. Gordon (eds.) 1967 Sir Gawain and the Green Knight. (2nd edition.) Revised by Norman Davis. Oxford: Clarendon Press. Zupitza, Julius 1878 "Zum Poema Morale", Anglia 1:5-38.

Sentential negation and clause structure in Old English Ans van Kemenade

1. Introduction In this article I will consider a minor pattern of sentential negation in Old English that foreshadows the later Middle English pattern with ne ... not. Although it is an infrequent one, the pattern shows that multiple sentential negation must have been an option in the grammar of English rather earlier than has been assumed so far. The facts I give for this pattern are largely new, and have been culled from a corpus of major Old English prose texts. I will show that the make-up of the pattern gives us interesting evidence about the architecture of the Old English clause, and illuminates in particular the position of personal pronouns, and that of the finite verb. In section 2, I will define what I view as sentential negation, and outline the assumptions made for the analysis of negation. I will also lay out some of the issues with respect to Old English clause structure for which my new facts will turn out to be illuminating. Sections 3 and 4 will then present the Old English negation evidence and discuss it in the context of Old English clausal structure. Section 5 is the concluding section.

2. Sentential negation The historical development of negation in Old and Middle English is in many ways a good illustration of what has come to be known as "Jespersen's cycle" (Jespersen 1917, 1924), as in (1):

148

Ans van Kemenade

(1)

Stage 1: Stage 2: Stage 3:

Stage 4:

negation is expressed by one negative marker negation is expressed by a negative marker in combination with a negative adverb or noun phrase the second element in Stage 2 takes on the function of expressing negation by itself; the original negative marker becomes optional the original negative marker becomes extinct

It is important to note that this cycle concerns primarily the negation of whole sentences, and this is what makes the scenario sketched by Jespersen an attractive and plausible one for the early history of English negation. Sentential negation is predominantly marked by preverbal ne immediately preceding the finite verb. Here are two examples of this: (2)

a.

b.

ne sende se deofol da fyr of heofenum, peah pe hit not sent the devil the fire from heaven though it iffan come from above came 'The devil did not send fire from heaven, though it came from above' (^Hom.i.6.13) He ne andwyrde dam wife cet fruman he not answered the woman at first 'He did not answer the woman at first' (TEHom.ii.l 10.33)

It may be useful to add a brief word about how to distinguish Old English sentential negation from other instances of negation, such as constituent negation. In practice, this is no easy matter.1 When there is, morphologically, a constituent negation, we nearly always find a multiple negation with ne on the left of the finite verb (Mitchell 1985: 663 ff). This is illustrated in (3): (3)

a.

... pcet he na sippan geboren ne wurde that he never afterward born not would-be 'that he would not be bom afterward' (Oros. 139.11)

Sentential negation and clause structure in Old English b.

149

... pcet heora nan ne mehte names wcepnes gewealdan that of-them none not could no weapon wield 'that none of them could wield any weapon' (0ros.103.25)

Presumably this indicates that what is morphologically a constituent negation has scope over the whole clause. In (3a), na is morphologically the negator for the adverb sippcur, in (3b), two constituents are negated: heora nan and nanes wcepnes. In both cases, we also find the negator ne on the immediate left of the finite verb. Although this results in a semantically negative sentence, we seem to have here a morphological constituent negation where the negation has sentential scope. This is not what we will term sentential negation here. We will restrict that term to sentences which can (but need not) have more than one negative element, but where the second negative does not negate a constituent. See (4): (4)

a.

b.

ponne ne miht pu na pcet mot ut ateon of dees then not could you not the speck out draw of the mannes eagan man's eye 'then you could not draw the speck out of the man's eye' 0EHomP.Xin.153) Ne bid na se leomingeniht furdor ponne his lareow not is not the apprentice further than his master 'The apprentice is not ahead of his master' (/EHomP.XIII. 134)

In both of these cases, the element na cannot be interpreted as modifying the constituent it precedes; in (4a), na does not modify pcet mot, in that the sentence does not mean: 'it is not the speck that you can draw ...'; similarly in (4b) it does not modify se leomingeniht, so that the sentence does not mean: 'it is not the apprentice who ...'. Rather, it is the whole sentence that is being negated by both negation elements. Examples such as those in (4) have not received much attention in the literature. Mourek (1903) makes a distinction between "qualitative" (sentential) and "quantitative" (constituent) negation. Mitchell (1985) follows Einenkel (1912) in wondering whether the distinction reflects a reality in Old English grammar and scraps it altogether, thereby glossing over the distinction as made

150

Ans van Kemenade

here between cases as in (3) and those in (4). While I agree with Mitchell and Einenkel that the distinction is not always easily made, I nevertheless think that it is possible to isolate the instances as in (4) by considering for each case whether the negator can reasonably modify the constituent or not, and this is precisely what leads to the distinction between (3) and (4): in (4) an interpretation as constituent negation does not seem to be reasonable. If this is correct, we have a pattern of multiple sentential negation in Old English that is several centuries earlier than it is customary to assume: for instance, Jack (1978a,b,c) traces the development of Middle English multiple sentential negation on the assumption that it is a novel pattern in Middle English. The pattern of multiple sentential negation as in (4) is a minority pattern in Old English. I have therefore found it necessary to collect data from a large corpus of major Old English prose texts. I chose prose texts for the usual reasons: it is fairly uncontroversial to assume that the prose more closely reflects the grammatical options in Old English. The texts chosen that are generally taken to represent Early Old English (ninth century) are: The Old English Orosius, ed. Bately (1980); King Alfred's Version of St. Augustine's Soliloquies, ed. Carnicelli (1969), King Alfred's Old English version of Boethius' 'De Consolatione Philosophiae', ed. Sedgefield (1899); King Alfred's West-Saxon version of Gregory's Pastoral Care, ed. Sweet (1885); the early part of the Parker MS of the Anglo-Saxon Chronicle, from Two of the Saxon Chronicles Parallel, ed. Plummer (1892-1899). From the tenth century were chosen: volume I of the Homilies of the Anglo-Saxon Church, ed. Thorpe (1846); both volumes of Mfric's Lives of Saints, ed. Skeat (1881-1900); both volumes of the Homilies of Mfric, ed. Pope (1967); the Homilies of Wulf stem, ed. Bethurum (1957); the later part of the Parker MS of the Anglo-Saxon Chronicle, ed. Plummer (1892-1899). All the texts have been exhaustively searched via the machine-readable version of the textcorpus composed for the Old English dictionary project in Toronto, available through the Oxford Text Archive. This has yielded some 330 examples of the required pattern. Perhaps the number is marginally lower because of the difficulty of separating constituent negation with sentential scope from sentential negation as discussed above. I have tried as much as possible to eliminate ambiguous examples, but nevertheless some examples of the former may have crept in. I now outline the assumptions that I make for the analysis of negation. Follow-

Sentential negation and clause structure in Old English

151

ing the line of argument started by Pollock (1989) in the Principles and Parameters framework, I assume that like other lexical and morphological information, negation is represented in a separate projection which is defined according to the general X'-schema (5). (5)

The evidence that the negation element ne is a head comes primarily from the fact that it is always on the immediate left of the finite verb (Vfin). It is accordingly assumed that it is a head that incorporates Vfin, as in example (6).2 (6)

& seo pruh ne mceg na itnc begen ymbfon and the tomb not can not us both surround 'and the tomb cannot hold us both' (GD.226.4) CP

mceg

unc begen ymbfon

152

Ans van Kemenade

The second sentential negator na is assumed to be a specifier element: the primary evidence for this is its fixed position, which will be considered below. This is a tensed clause, which has as its main verb mceg. VP is dominated by NegP. NegP has ne as its head. NegP is dominated by IP, which defines the tense and agreement characteristics of the sentence. The Specifier of IP hosts the subject, seo pruh, as in any tensed clause. CP defines the conjunction positions. The base position of the verb is signalled by that of the non-finite verb. Finite mceg moves to the I position in order get its finiteness characteristics tense and agreement. On its way, it moves through the head position of NegP, picking up ne, which moves along with it.3 This is why ne always appears on the immediate left of the finite verb; it also serves to account for the fact that in clauses with V-movement, we always find the order ne + Vfin ... + na. In clauses without such movement, the order na + ne + Vfin is frequently attested, as in (7). (7)

pcet hie it na gebetan noldan that they it not atone-for not-wanted 'that they did not want to atone for it' (Oros.38.17)

3. Sentential negation with two negation elements We will now introduce the evidence provided by the patterns involving sentential negation with two negation elements. We will focus this discussion on the pattern with na and its primary variant no as the second sentential negator. The most important reason for this is that naJno seems to function unambiguously as a negative constituent or sentential modifier. This is where it differs from other negatives that might potentially function as the second sentential negator, like nahtlnoht/nowihtl nawiht,4 These often function as modifiers, but their use as such is often difficult to separate from their use as a negated noun. It is therefore often unclear whether they occupy a nominal position or a (negative) adverbial position, and I leave more detailed consideration of these for further research. We will further restrict the discussion to those cases where na/no is clearly in a high position close to Vfin in main clauses, since this is where its main interest lies from the point of view of word order. There is some evidence in Old English that there is a second relatively fixed position for the (second) negative

Sentential negation and clause structure in Old English

153

adverb, close to the non-finite verb. We will have to leave the study of this pattern, too, for further research. In (8), I give a sample of illustrations of the pattern with a second, "high" sentential na: (8)

a.

b. c.

d.

e.

f.

g.

Ne het he us na leornian heofonas to wyrcenne not ordered he us not learn heavens to make 'He did not bid us learn to make the heavens' (7ELS.XVL127) ponne ne miht pu na pcet mot ut ateon of dees mannes eagan (yEHomP.XIII.153) ... ne meahtest pu hi na forleosan not could you them not lose 'you could not loose them' (Boeth.3.7.17.20) Nis na se halga gast wuniende on his gecynde swa swa he not is not the holy ghost existing in his nature as he gesewen wees seen was 'The Holy Ghost is not in his nature existing as he was seen' (7EHom.i.322.17) Ne bid na se leorningeniht furdor ponne his lareow Not is not the apprentice further than his master 0EHomP.XIII.134) Ne scede na we Drihten pcet he mid cynehelme odde mid not said not our Lord that he with diadem or with purpuran gescryd, cuman wolde to us purple clothed come wanted to us 'Our Lord said not that He would come to us with a diadem or clothed with purple' 0ELS.XXXI.762) Ne wende na Ezechias Israhela kyning dcet he not thought not Ezechias Israel's king that he gesyngade, da... sinned, when... 'Ezechias, king of Israel, did not think he was sinning when ...' (CP.39.2)

154

Ans van Kemenade

One very striking observation stands out when we consider these examples, which are representative of the corpus: when the subject is a pronoun, as in (8a-c), it appears on the left of na, when the subject is a noun, as in (8d-g), it appears on the right of na. Furthermore, object pronouns optionally occur on the left of na, as in (8a-b). This observation provides a new perspective on the discussion of the status of personal pronouns in Old English. Let us consider the background of these issues first.

3.1. Issues of clause structure and the status of Old English pronouns Old English clause structure cannot be considered without regard to the verb second (V2) constraint. While Old English can be reasonably analysed as an SOV language, it is also clear that the sentence has some satellite positions in main clauses that are reserved for some first constituent and Vfin. Some initial illustration of this is given by the following sentences: (9)

a.

b.

hwi wolde God swa lytles pinges him forwyrnan why would God so small thing him deny 'why should God deny him such a small thing?' OEHom.i.14.2) On twam pingum hcefde God pees marines sawle gegodod in two things had God the man's soul endowed 'With two things God had endowed man's soul' OEHom.i.20.1)

Observe that these examples illustrate the phenomenon of subject-auxiliaryinversion that we still find in Present-Day English. In Old English, however, it is not restricted to interrogative and negative-initial contexts, as (9b) illustrates. Such constructions have been analysed by van Kemenade (1987) as involving preposing of the first constituent and Vfin to the satellite sentential positions in CP as in the following adapted structure:5

Sentential negation and clause structure in Old English

155

(10) Spec on twam pingun\ IP God pees marines sawle tj gegodod tj But not all Old English main clauses conform to this pattern. In sentences introduced by an interrogative constituent, a negative element, or the short adverbial pa, subject-verb inversion is indeed canonical, as (9a) and (11) illustrate: (11)

a.

b.

c.

For hwam noldest pu de sylfe me gecydan poet ... for what not-wanted you yourself me make-known that... 'wherefore would you not want to reveal to me yourself that...' OELS.XXXm.307) pa foron hie mid prim scipum ut then sailed they with three ships out 'then they sailed out with three ships' (ChronA.AD 897) Ne sceal he noht unalyfedes don not shall he nothing unlawful do 'he shall not do anything unlawful' (CP.60.15)

The same is not true for main clauses introduced by a non-subject other than an interrogative or negative, or pa. While in these cases, inversion is near-canonical when the subject is a noun, as in (9b) above, pronominal subjects occur on the left of the preposed Vfin, as in (12): (12)

a.

Foröon we sceolan mid ealle mod & mcegen to therefore we must with all mind and power to Gode gecyrran God turn 'therefore we must tum to God with all our mind and power' (Blickling.97)

156

Ans van Kemenade

b.

Be deem we magon suide swutide oncnawan dcet... by that we may very clearly perceive that... 'by that, we may perceive very clearly that...' (CP.181.16)

On the basis of this and a wider range of evidence, it was argued by van Kemenade (1987) that Old English personal pronouns are clitic-like elements that are procliticised to Vfin in topic-initial constructions. Such procliticisation is blocked when the first constituent position is occupied by an operator-like element as in the examples in (11). The essence of this analysis is that movement of Vfin is always to the C-position in the structure (10), essentially because the phenomenon of topicalisation plus verb-fronting is restricted to main clauses. There is some topicalisation in embedded clauses in Old English, but it is shown by van Kemenade (1997) that this is restricted to a well-defined set of constructions, including passives, impersonals and the like. An alternative analysis of V2 and pronouns in Old English has been presented by Pintzuk (1991, 1993). The essence of the proposal is that structures as in (10) are restricted to examples where the pronominal subject is postverbal, i.e. interrogative, negative-initial and initial constructions. In Pintzuk's view, topic-initial constructions represent a lower sentential level IP, with the topic in Spec,IP and the pronouns as a clitic adjoined to the topic. This would mean that example (12b) would have a structure like (13): (13)

Topic Be deem

Topic CI we

magon{ suide swutide t; oncnawan

Pintzuk's clitic analysis is very problematic for reasons that go beyond the scope of this article (but see van Kemenade, forthcoming). Furthermore, there are several arguments against regarding Spec, IP as the relevant topic position. The main argument is that genuine topicalisation is restricted to main clauses in Old English, whereas Pintzuk's analysis predicts that it should also occur in

Sentential negation and clause structure in Old English

157

embedded clauses. This and other arguments are elaborated in van Kemenade (1997). On the other hand, the attractive part of the analysis is that it assigns different positions to Vfin between topic-initial and w/z/Neg//?a-initial clauses, and the position I is also available in embedded clauses. It is, however, hard to assign a principled place to clitic pronouns in it. The combined insights of van Kemenade and Pintzuk can be preserved if we have an analysis where topicalisation is restricted to main clauses, as in (10), and where we can accommodate the position of pronouns in a principled way. This is precisely where the facts concerning double sentential negation as introduced above prove illuminating. We will turn to them in the next section.

4. Multiple sentential negation and clause structure in Old English In this section, we turn to discussing the negation evidence in more detail. It will lead us to a more articulate sentential structure for Old English than has been assumed so far. But let us reconsider the data first. Of particular relevance with respect to word order issues are those examples that are main clauses where ne + Vfin occurs at the beginning of the sentence. These make up about 220 of the total number of examples. Two illustrative examples are repeated here: (14)

a. b.

... ne meahtest pu hi naforleosan (Boeth.3.7.17.20) Nis na se halga gast wuniende on his gecynde swa swa he gesewen woes (^Hom.i.322.17)

As noted above, one very striking fact stands out when we consider these examples: when the subject is a pronoun, it appears on the left of na; when it is a noun, it appears on the right of na.6 Furthermore, object pronouns optionally occur on the left of na, as observed in connection with (8). This shows that the position for the nominal subject is really different from that of the pronominal subject, and that these two positions are separated by NegP. The regular position for the subject is to the right of na, and the position for pronouns on the left of

158

Ans van Kemenade

na is indeed restricted to personal pronouns rather than being a subject position. This observation provides a new perspective on the discussion of the status of personal pronouns in Old English, and on the character of the position of pronouns in topic-initial constructions. Before we go on to discuss this, let us consider a more articulate sentential structure than has been assumed so far for Old English, which is clearly suggested by the negation facts. This structure is in line with recent generative syntactic theorising, such as Pollock (1989) and Belletti (1990), in that each contentful morphological element, e.g. tense inflection, negation, agreement inflection, heads its own projection according to the general phrase structure format. This results in the structure in (15): (15)

CP Spec

Vfin Spec

F

Pron F Vfin

NegP Spec na

Neg' / ^ ^ X

Neg

TP

we-Vfin Spec

T'

Subj Τ

... VP

Vfin Spec

V' I ... ν

Sentential negation and clause structure in Old English

159

Let us assume that pronominal subjects (and, optionally, objects) inhabit a separate functional projection FP in the structure (15).7 Let us suppose furthermore that na as a second sentential negation element is in Spec,NegP. This is suggested first of all by its fixed position. The fact that this is a fixed position is suggested furthermore by the fact that na, even when it is morphologically a constituent negation, sometimes moves away from its constituent to the characteristic position for na as observed above. This is illustrated by the following examples: (16)

a.

b.

c.

Ne leofad se man na be Hafe anion, ac lyfad be not lives the man not by bread alone but lives by eallum dam wordum pe gad of godes rnude all the words that go from God's mouth 'Man does not live by bread alone, but by all the words that come from God's mouth' (/EHom.I.166.111) ne lifad na se mann be Hafe anum, ac lifadh be dam wordum de gad of godes mude (jEHom.1.168.26) Nis na godes wunung on dam groegum stanum, ne on not-is not God's dwelling in the grey stones nor in ceremmt wecgum, ac he wunad on heofonum brazen lumps but he dwells in heaven 'God's dwelling is not in the grey stones, nor in brazen lumps; but he dwells in heaven' (JELS.VII.135)

The contrast between (16a) and (16b) is interesting. The meaning of (16a) shows the essential constituent negation of na: 'man lives [not by bread alone]'. On the next page, we find the same sentence with the same meaning, but with na in the position characteristic in (14): on the left of the nominal subject. In (16c), the co-ordinations show that the intended negation is: na on dam grcegum stanum, giving the reading: 'God's dwelling is [not in the grey stones], nor in ... '. But na occupies the position on the left of the nominal subject, as characteristic of the sentence negation with two negators.8 We can account for this fixed position of na by assuming that the second sentential negator occupies Spec,NegP. When we have a constituent negation with sentential scope, as in (16), the constituent

160

Ans van Kemenade

negator na optionally moves to that position. In the structure (15), we can account for the basic V2 patterns as discussed above. Recall that the patterns are as follows: (17)

a. b. c.

^-element I ne I pa- Vfin - Subj ... Topic - Vfin - Noun Subj ... Topic - Pron Subj - Vfin ...

(9a) (9b) (12)

If the first constituent is an interrogative constituent, as in (17a), that constituent is in Spec,CP and Vfin moves to C (through intermediate head positions as indicated in (15)). If the subject is a noun, it is in Spec,TP in (15); if it is a pronoun, it is in Spec,FP.9 This difference in position is not visible on the surface in w/?-questions. It emerges very clearly, however, in (17b) and (17c). Topics move to Spec,CP like w/z-constituents (recall that this is shown by the fact that they occur only in main clauses). But they do not draw Vfin to C, as seems to be true for all historical stages of English.10 Rather, Vf moves to the head position of FP." There, it is followed by the nominal subject (in Spec,TP) as in (17b), or preceded by the pronominal subject (in Spec,FP), as in (17c). The advantage of this analysis over previous ones is that there is a separate position for clitic pronouns. Their position is not contingent on proclisis to Vfin or enclisis to the topic. Rather, they have a position of their own, and their position relative to Vfin is determined by the movement requirements on Vfin.

5.

Conclusion

I have shown in this article that in the Old English period, rather earlier than has so far been assumed, there was already a pattern with double sentential negation. I have looked at this phenomenon primarily from a syntactic point of view, since it tells us a good deal about certain vexed puzzles concerning Old English word order. The position of the standard sentential negation marker ne co-varies with the position of the finite verb. The second sentential negator nalno, as far as attested in the prose texts considered in this article, occupies a fixed position, with pronouns appearing on the left of it, and nominal subject on the right. I have argued that this distribution supports a "split Infi" hypothesis for Old

Sentential negation and clause structure in Old English

161

English clause structure and throws new light on the positional characteristics of personal pronouns and nominal subjects. Other important evidence that may be derived from this is that it yields a more precise characterisation of the position of the finite verb, especially in root clauses: although in topic-initial root clauses, the position of the finite verb is lower than in interrogative, negative-initial and pa- initial clauses, it is nevertheless higher than the second negation element. One interesting observation may be made in conclusion: careful scrutiny of a small pattern in a large corpus of texts may give us the kind of data that are easily glossed over without such detailed research, and in the case in point yields crucial evidence for larger questions of clause structure.

Notes 1.

See, for instance, the discussion in Mourek (1903), Einenkel (1912), and Mitchell (1985: §§ 1596 ft).

2.

The outcome of the issues discussed later on in this article will show that Old English clausal structure, as well as the position of subject and Vfin, is more complex than this. However, for the purposes of illustrating the points at issue now, i.e. verb movement through the head of NegP, the present simplified structure suffices.

3.

In the case of a main clause, ne moves along with the finite verb further, in accordance with the V2 rule.

4.

Haeberli (1991) states erroneously that all these elements behave in the same way.

5.

The coindexed traces tj and tj indicate the position from which movement has taken place.

6.

This observation was made also by Einenkel (1912), and followed up in a spirit similar to ours by Haeberli (1991). I am grateful to Eric Haeberli for pointing this out to me. I differ from Einenkel in separating cases of sentential negation from others. My analysis here differs from that in Haeberli (1991) in that I argue here that there is a fixed

162

Ans van Kemenade "high" position for na when it is a sentence negator and that this position is Spec,NegP. Haeberli assumes a variable position for na, adjoined to different functional projections.

7.

I use the term FP as short for a neutral Functional Projection, to avoid prejudging its precise status. Perhaps the fact that we find personal pronouns in that position can be brought in line generally with second position facts as observed by Wackernagel (1892) as some sort of agreement phenomenon. Issues of this sort take us far beyond the scope of this article.

8.

An anonymous reviewer objects that (16a) illustrates ambiguity between sentential and constituent negation reading. I would therefore like to emphasise the point that this is one case where this ambiguity would be unexpected because (16a-b) are quite clearly cases of constituent negation semantically.

9.

This presupposes an analysis of Old English personal pronouns as Germanic-style weak pronouns as in Cardinaletti and Starke (1994).

10.

This is in accordance with the observations in Kiparsky (1995). This analysis is an improvement over that of Kiparsky's in that it does justice to all the V-movement facts.

11.

This leaves open the question what it is that forces V-movement to Cl° in topic-initial constructions. I assume that C1P is part of the C-system and propose to derive this movement by assuming that topics are a weak trigger for V-movement, in the sense that they require movement of V to some head position in the C-system. This is further motivated in van Kemenade (forthcoming).

Sentential negation and clause structure in Old English

163

Texts examined Mfric's Lives of Saints. Edited by W. Skeat. EETS. o.s. 94 and 114. London 1881-1900 Homilies of Mjric. Edited by J.C. Pope. EETS, 259. Oxford 1967. King Alfred's Old English version of Boethius' 'De Consolatione Philosophiae'. Edited by W.J. Sedgefield. Oxford 1899. King Alfred's Version of St. Augustine's Soliloquies. Edited by Th. Carnicelli. Harvard University Press 1969. King Alfred's West-Saxon Version of Gregory's Pastoral Care. Edited by H. Sweet. EETS, 45 and 50. London 1885. The Homilies of the Anglo-Saxon Church. Edited by B. Thorpe. London: The Aelfric Society 1846. The Homilies ofWidfstan. Edited by D. Bethurum. Oxford 1957. The Old English Orosius. EETS s.s. 6. Edited by J. Bately. Oxford 1980. Two of the Saxon Chronicles Parallel. Edited by C. Plummer. Oxford 1892-1899.

References Battye, Adrian - Ian Roberts (eds.) 1995 Clause structure and language change. Oxford: Oxford University Press. Beletti, Adriana 1990 Generalized Verb Movement. Turin: Rosenberg & Sellier. Cardinaletti, Anna - Michal Starke 1996 "Deficient pronouns: A view from Germanic", in: Höskuldur Thrainsson - Samuel David Epstein - Steve Peter (eds.), 21-65. Einenkel, Eugen 1912 "Die Englische Verbalnegation", Anglia 35: 187-248. Haeberli, Eric 1991 The Neg Criterion and Negative Concord. [Memoire de Licence, Universite de Geneve.]

164

Ans van Kemenade

Jack, George 1978a 1978b 1978c

"Negation in later Middle English prose", Archivum Linguisticum n.s. 9: 58-72. "Negative adverbs in early Middle English", English Studies 59: 295-309. "Negative Concord in early Middle English", Studia Neophilologica 50: 29-39

Jespersen, Otto 1917 "Negation in English and other languages", Det Kgl. Danske Videnskabernes Selskab. Historisk-filologiske Meddelelser 1: 1-151. Copenhagen. [1962] [Reprinted in Selected writings of Otto Jespersen. London: George Allen and Unwin.] 1924 The philosophy of grammar. New York: Norton and Company. [1965] [Reprinted.] Kemenade, Ans van 1987 Syntactic case and morphological case in the history of English. Dordrecht: Foris Publications. 1997 "V2 and embedded topicalisation in Old and Middle English", in: Ans van Kemenade - Nigel Vincent (eds.), 326-352. forthcoming Verbal syntax in the history of English Kemenade, Ans van - Nigel Vincent (eds.) 1997 Parameters of morphosyntactic change. Cambridge: Cambridge University Press. Kiparsky, Paul 1995 "Indo-European origins of Germanic syntax", in: Adrian Battye - Ian Roberts (eds.), 140-171. Mitchell, Bruce 1985 Old English syntax. Oxford: Oxford University Press. Mourek, V.E. 1903 Zur Negation im Altgermanischen. Prague: Verlag der kgl. Böhmischen Gesellschaft der Wissenschaften. Pintzuk, Susan 1991 Phrase structures in competition: Variation and change in Old English word order. [Ph.D. diss., University of Pennsylvania.]

Sentential negation and clause structure in Old English 1993

165

"Verb seconding in Old English: Verb movement to Infi", The Linguistic Review 10: 5-35. Pollock, Jean-Yves 1989 "Verb movement, UG and the structure of IP", Linguistic Inquiry 20: 365-424. Thrainsson, Höskuldur - Samuel David Epstein - Steve Peter (eds.) 1996 Studies in comparative Germanic syntax. Vol II. Dordrecht: Kluwer. Wackemagel, Jakob 1892 "Über ein Gesetz der Indogermanischen Wortstellung", Indogermanische Forschungen 1: 333—435.

The facts and nothing but: the (non-)grammaticalisation of negative exclusives in English1 Terttu Nevcdainen

1. Introduction This paper discusses the extent to which negative expressions tend to grammaticalise as exclusive adverbs in English. The Present-Day English set of exclusive adverbs consists of such items as only, merely, just, simply, solely and exclusively. It also includes items that dictionaries label as literary or archaic, such as but, and regional variants such as nobbut. Semantically all English exclusive adverbs may be analysed as negative quantifiers. They add a negative implication to the propositions in which they occur. When we say John saw only Mary, we are committed to the truth of the proposition 'John saw nobody other than Maiy'.2 The main source domains for exclusive adverbs in English and other languages include the numeral 'one', which is the etymological source for e.g. only, alone and uniquely on the one hand, and simply and singly on the other (reflecting the two PIE stems *oi(no)- and *s(e)m-, respectively). Another common source domain consists of a set of related notions to do with purity, unmixed quality, lack of other qualities, and lightness and exactness (in merely, purely andjust). The third domain that is frequently attested are expressions of exclusion and restriction, either by means of negation and a preposition of exception or lexically (e.g. ne ... but, nobbut, exclusively) (see Nevalainen 1991: 123-157). In this article I shall use the term "negative exclusive" as shorthand for overtly negative expressions that have grammaticalised or shown signs of becoming grammaticalised as exclusive adverb(ial)s. A case in point is the Middle English construction ne ... but, which is synonymous with only in cases like Mary nis but a child 'Maiy is only a child'. A parallel can be found in French, where ne ... que is an equivalent of 'only'. Like Middle English, Modern French is also gradually losing its overt negative element ne in this construction. The adverbialisation of negative exclusives may involve one of two things. As would be predicted by Jespersen's cycle (1917: 9), the negative element may be

168

Terttu Nevalainen

suppressed, as in ne ... but, which gave rise to but in the course of the thirteenth and fourteenth centuries. The negative element may also be incorporated into the exclusive adverb by means of univerbation, no/na ...but yielding nobbut in some northern varieties of British English. What is of particular interest in this process of grammaticalisation are the intermediate stages; more or less fixed collocations such as nothing but. In this paper I shall largely concentrate on these would-be exclusive adverbs. Section 2 outlines the lexico-grammatical properties that justify the category of exclusive adverbs in English, and sections 3 and 4 examine the adverbialisation of negative exclusives in Middle and Modern English respectively. The study focuses on the long-term diachrony of the most obvious potential cases for adverbialisation, not/nought but and nothing but. While the process runs its course with ne ...but in Middle English, I will show that it is arrested in the case of both not/nought but and nothing but. Some possible reasons for this failure to grammaticalise are discussed in section 5.

2. The syntax and semantics of exclusive adverbs Exclusive adverbs form part of a larger adverbial category called focusing adverbs. Together with particularisers (e.g. especially, particularly) and additives {also, even), exclusives have the property of being both syntactically and semantically associated with another element in the sentence.3 Adverbs like only and also can thus focus on a part of a sentence as wide as the predication or as narrow as a single constituent of an element, such as an auxiliary within a verb phrase (see Quirk et al. 1985: 604). Focusing adverbs come towards the grammatical end of the lexical vs. grammatical continuum of adverbs. They constitute a limited set that differs from most other, less grammaticalised members of the English adverbial category in that they display a number of "negative" syntactic properties. A focusing adverb itself cannot usually be modified by other adverbials. It does not have a pro-form nor can it be pronominalised as an isolated expression. Hence we cannot say, for instance: (1)

*John will fly only to New York and so will Mary to London

Despite the fact that focusing adverbials can be associated with the verb phrase and

The (non-)grammaticalisation of negative exclusives in English

169

even an entire predication, they do not participate in clausal processes in the same way as free sentence and verb phrase modifiers do. Thus they cannot appear independently as the focus of cleft constructions, alternative negation, or interrogation. Consider the unacceptable cases in (2) and (3), where roman type is used to indicate focused elements. (2) (3)

*John didn't phone Mary only but also *Does Mary write only or also?

I shall return to the positional syntax of negative exclusives in section 4. At this point suffice it to note that such adverbialised items as but tend to assume a position immediately before the clause element that they are synsemantically associated with in the sentence. It has already been pointed out that exclusive focusing adverbs are semantically negative. Just like other negative operators, they also display the contrast between ordinary sentential scope and narrow scope. The difference between the two is illustrated by examples (4) and (5), both drawn from Taglicht (1984). The angular brackets mark the focus element of the exclusive, while Taglicht's notation is used to indicate scope. (4) (5)

Only for did he acquire the drawing ("They drove a hard bargain!'; ordinary scope) (Taglicht 1984: 153) °~For only ~° he acquired the drawing. ('They didn't know its real value!'; narrow scope) (Taglicht 1984: 153)

If exclusive focusing adverbs are negative in meaning, they should behave grammatically like other negative polarity items (cf. e.g. Quirk et al. 1985: 780-782). While it is not possible to discuss the matter in detail here for reasons of space, it should be pointed out that the status of exclusives as negative polarity items is not very clear in all respects (Nevalainen 1991: 65-68). On the one hand, the presence of an exclusive such as only may cause subject-operator inversion, as in example (4) above. It may also license non-assertive polarity items, as in (6). On the other hand, as Klima (1964) has shown, ow/y-clauses often fail other negative tests, such as association with negative question tags; compare (7a) and (7b).

170

Terttu Nevalainen

(6)

Only ever accept suggestions with any sincerity (Klima 1964: 311) a. John only takes , doesn't he? b. John does not take milk, does he?

(7)

Interestingly, the class of exclusives is divided concerning co-ordinating constructions with not. While we have the institutionalized co-ordinating correlative not only χ - but alsoy, meaning 'both χ and y', no negative correlative is permitted with the exclusive but. We thus get cases like (8a-b) and (9a), but not cases like (9b). (8) (9)

a. b. a. b.

I saw him only I saw him but I did not see him only but also *I did not see him but but also

These collocational limitations of but finally bring us to the topic of this study, negative exclusives. I would argue that exclusives like but, which themselves go back to overtly negative expressions, betray their origins in that they do not allow constructions which, with ordinary scope, would amount to a "double negative". In other words, but is a good illustration of what Hopper (1991: 159) calls the principle of persistence: details of its lexical history are reflected in constraints on its grammatical distribution. Incidentally, etymological conditioning can also be observed with other exclusives, notably only, merely and alone, albeit usually in a less radical form than with but (Nevalainen 1991: 256-259).

3.

The adverbialisation of negative exclusives in Middle English

Because the lexeme but is also involved in other negative exclusives, a few words should be said about its origins before I discuss the way in which it became adverbialised. But goes back to the Old English locative adverb and preposition beutan, butan, meaning 'on the outside'. Very early on, the preposition developed the figurative sense 'except, save', as in the phrase ealle buton anum, 'all except one' in example (10). This preposition also grammaticalised into a conjunction, first into a subordinating function, 'except that, unless', and later into co-ordinating functions,

The (non-)grammaticalisation of negative exclusives in English

171

equivalent to both German sondern and aber. (10)

Sceotend swcefon/ pa pcet homreced healdan scoldon/ soldiers slept who the gabled-building guard should ealle buton anion all except one "The soldiers were sleeping, those who were guarding the gabled building, all except one' {Beowulf 703-705)

The exclusive adverb but goes back to the Old English construction ne ... bid, meaning 'only, no more than', but here being either a preposition or a conjunction. As the negative particle ne disappears in Middle English, a category change takes place and the original preposition is left to carry the exclusive sense of the new focusing adverb. The gradual process of grammaticalisation of the negative exclusive but is presented in Table 1, which derives from my earlier research on the subject (Nevalainen 1990: 344). Table 1. Frequency of ne... but and but per 100,000 words in Middle English

ne... but

but

Ml 1150-1250

12.6

0.0

M2 1250-1350

9.9

5.4

M3 1350-1420

4.4

13.3

M4 1420-1500

1.0

43.6

Middle English period

The figures show thefrequenciesof ne... but and but per 100,000 words in the four Middle English periods of the Helsinki Corpus of English Texts (HC; pilot version). The gradual decline of ne ... but and the simultaneous rise of but emerge clearly from the data, with the simple form dominating from Chaucer's days onwards. Examples (11) and (12)fromthe third Middle English subperiod (1350-1420) clearly illustrate the development.

172

Terttu Nevalainen

(11)

But certes by nature ther nys but ο God; but by the participacioun of dyvinite ther ne let ne distourbeth nothyng that ther ne ben many goddis. 'But certainly by nature there is only one God; but by participation of divinity nothing prevents there being many gods' (HC, M3, Chaucer's Boethius 433) Pat cytee of Alizandre is wel .xxx. furlonges in lengthe but it is but .x on largeness And it is a full noble cytee & a fayr. (HC, M3, Mandeville's Travels 37)

(12)

What about the other possible negative exclusives, no but, not but, and nothing buf! We know that nobbut and nought but occur regionally as synonyms of only in British English, as can be seen from the map for only in Upton et al. (1987: 132), for instance. Nothing but is also commonly paraphrased as 'only' (LDOCE) or 'just' (iCOBUILD) in dictionaries of Present-Day Standard English. Any adverbialisation of these negative phrases is presumably correlated with the grammaticalisation of the initial elements of the collocations, not (< nought < nalno wiht) and nothing (< no thing). Judging from the evidence in the OED, both of them grammaticalised early, and appear to have been available at least from the Early Middle English period onwards. On the basis of the Helsinki Corpus, however, it appears that none of the three negative exclusives mentioned became fully adverbialised in Middle English. First of all, their exclusive readings are quite rare in the corpus. Not but has a frequency of eight (including naught/nought but), nothing but occurs three times and no but only once in the Middle English section of the Corpus. The low incidence of their 'only' readings means that they do not fulfil the requirement of frequency set by Traugott and Heine (1991: 9) for items in process of grammaticalisation. As we can see from examples (13)—(15), even the cases that are paraphrasable as 'only' or 'merely' may be interpreted compositionally as a negative element followed by a prepositional phrase of exception. Apart from an exclusive adverbial reading, no but in (13) could be interpreted as a prepositional compound meaning 'except' (OED s.v. nobbut 2) or, less plausibly, as a case of noun phrase ellipsis in which the determiner no is followed by an exceptive £«/-phrase. Similar ambiguity occurs with not but in (14), where the editorial punctuation suggests an independent reading of but as a preposition of exception. The case in (15), finally, comes closest to being an adverbialised instance of not but in the data.

The (non-)grammatica]isation of negative exclusives in English (13)

(14)

(15)

173

No man sai euer God, no but the oon bigetun sorte, that is in the bosum of the fadir, he hath teld out. (HC, M3, Wycliffe's New Testament 1,1) Therfor as Jhesu knew, that the Farisees herden, that Jhesu makith and baptisith mo disciplis than Joon, thouj Jhesus baptiside not but hise disciplis, he lefte Judee, and wente ajen in to Galilee. (HC, M3, Wycliffe's New Testament IV, 1) for all the comouns pere eten withouten cloth vpon here knees & pei eten all maner offlessch & lityll of bred, And after mete pei wypen here hondes vpon here skyrtes & pei eten not but ones a day. (HC, M3, Mandeville's Travels 144)

The material also contains one environment where not but persists until the present day as a paraphrase equivalent of 'only', namely the modal can. As in examples (13) and (14), however, the case in (16) may be interpreted as a negative instance of can followed by an exceptive clause, an exceptive conditional consisting of but and if (see further Traugott 1997). (16)

Pe mercy of pis fadur kan we not telle fully, for he ys pe moste worchere pat may ben in pis world; and he can not worche but jifhe medle mercy, for he wrohte by mercy wan he made pis world... (HC, M3, English Wycliffite Sermons I, 236)

Early instances of the construction nothing but are illustrated in (17) and (18). As (17) shows, nothing can be the complement of a preposition, and further modified by else, a clear indication of its status as a pronoun at the time. The clause is multiply negated by means of ne, which would further complicate any adverbial interpretation of the postverbal negative phrase. (17)

it happith diuersely som tyme in on hole & som tyme in an other / for lettere .D. ne serueth of nothyng ellis butfor to shewe the wher thow shalt bygynne thy reknyng in thy litel cercle ... (HC, M3, Chaucer, The Equatorie of Planetis 34)

174

Terttu Nevalainen

Example (18) is syntactically vague. The nothing but construction is open to the adverbial 'only' reading, but as there is little evidence supporting it, the compositional reading ('nothing except') remains the more likely one. (18)

theryn dwelleth he / and hunteth nomore / ne desyreth no wynnynge but he lyueth by almesse and taketh nothyng but suche as men gyue hym for charyte and doth grete penance for his synnes (HC, M4, Caxton, The History of Reynard the Fox 10)

4. Not but and nothing but after Middle English 4.1. Early and Late Modern English Having suggested that, apart from but, none of the negative exclusives under scrutiny had unequivocally adverbialised in Middle English, I shall move on to the Modern English period. The distributions of the different negative exclusive items are shown in Table 2. Table 2.

Frequency per 100,000 words of but, nothing (...) but and not (...) but in Modem English

ModE period

but

nothing (...) but

not/nought(...)but

EModE 1500-1560 1570-1630 1640-1700

59.4 104.2 68.3

8.4 13.7 21.6

1.5 4.2 2.3

LModE 1840-1900

19.4

7.0

3.0

The Early Modern English and Late Modem English frequencies of but are approximations based on the non-computerised 2.3 million word corpus I compiled for my research on Modern English exclusive adverbs (EModE: 1500-1700, LModE: 1840-1900; Nevalainen 1991: 102-119). The Early Modem English

The (non-)grammaticalisation of negative exclusives in English

175

figures for nothing but and not/nought but are from the Helsinki Corpus (1500-1710), and the corresponding Late Modern English datafromthe Corpus of Late Modern English Prose (1860-1919).4 There are thus small differences in the periodisation of the materials. The periods shown in Table 2 are those used in my own corpus (Nevalainen 1991). Table 2 clearly shows that the adverb but is much morefrequentthan the negative exclusive phrases in Early Modern English. In fact but is one of the quantitative prototypes of the exclusive adverbial paradigm at the time. The other one is, then as now, only. I have analysed their functional differentiation and overlap in detail using the VARBRUL program as my statistical tool in Nevalainen (1991). Table 2 also indicates that not but (including the two instances of naught but found in the Helsinki Corpus) remains relatively rare. As can be seen from cases like (19), nought is still treated as a pronoun. Neither does the phrase not but show any signs of univerbation in these data sources. It is indicative of the increasing confinement of not but to the modal context of can that these contexts take up more than half of the instances in the Helsinki Corpus data. A typical case is given in (20). In the Late Modern English prose corpus all three instances of not but appear in the context cannot but. (19) (20)

Which were indede nought els but Gyb our cats two eyes. (HC, El, Gammer Gvrtons Nedle 13) I would not mention any such thing as any piece of Evidence to influence this case, but I could not but tremble to think, after what I knew, that any one should dare so much to prevaricate with God and Mm, as to tell such horrid Lyes in the Face of a Court. (HC, E3, The Trial of Lady Alice Lisle IV, 121 C2)

The modal context is peculiar in that it also preserves the collocation in present-day standard language. The paraphrase options of cannot but vary from 'can only' to the relatively rare 'can but', and from 'must' (or 'has to') to 'all χ can do is', i.e., from constructions involving a negative existential quantifier to those containing an affirmative universal quantifier. The range is illustrated in (21) (cf. LDOCE s.v. cannot).

176

Terttu Nevalainen

(21) —>

One cannot but admire her. a. One can only admire her. b. One can but admire her. c. One must admire her. d. All one can do is admire her.

Table 2 suggests that the frequency of nothing but steadily increases in Early Modern English. The obvious question to ask then is, does it also show signs of grammaticalisation? As far as univerbation is concerned, the answer is an unequivocal "no" at the graphemic level. Most of my syntactic evidence also points in the same direction. There is, however, some evidence from outside the Helsinki Corpus suggesting that nothing hut could be construed as a unit that could assume the function of a modifier in the sentence. The examples in (22) and (23) come from the Oxford Shakespeare Corpus. (22)

(23)

Conceal them, or thou diest. Why, sir, they were nothing but about Mistress Anne Page, to know if it were my master's fortune to have her or no. CMerry Wives of Windsor 1597-8: 1, 42-45 ) Out, hyperbolical fiend, how vexest thou this man! Talkest thou nothing but of ladies? {Twelfth Night 1601: 1, 26-27)

In both cases, it is possible to interpret nothing but as a modifier of the following prepositional phrase rather than the complement of the preceding verb. However, cases like this are exceptional. In the entire Shakespeare Corpus they are the only ones of their kind that my systematic search has yielded, although there are about 150 relevant instances of nothing but. Evidence in favour of compositionality is also not difficult to come by. Cases like those in (24)-(26), all three of which unambiguously illustrate the status of nothing but j as a noun phrase, occur commonly in Shakespeare:

The (non-)grammaticalisation of negative exclusives in English (24)

(25)

(26)

177

What's you will? Nothing good monsieur, but to help Cavaliery Peaceblossom to scratch (Midsummer Night 1595, 21-23) nothing lets to make us happy both But this my masculine usurped attire Do not embrace me till each circumstance Ofplace, time, fortune do cohere and jump {Twelfth Night 1601: 1, 247-250) 'Mad' call I it, for to define true madness, What is't but to be nothing else but mad? (Hamlet 1600-1, 94-95)

Nothing and but are separated in (24) and (25). No such discontinuity would be allowed if the phrase were adverbialised. In (26) nothing is postmodified by else. This shows that nothing has not lost its pronominal semantics. Else would be incompatible with true exclusive adverbs, which in the rare cases in which they are modified only accept intensifiers like almost, as in almost exclusively. Examples (27) and (28) from the Oxford Shakespeare Corpus further support the compositional interpretation of nothing but as an indefinite pronoun followed by a preposition. In (27), nothing is postmodified by a /o+infinitive construction, and in (28) the complement of but is in the objective case, i.e. thee rather than thou. Similarly in (29), nothing occurs as the complement of the preposition in, and the Aw/-phrase repeats the same preposition. (27)

(28) (29)

Alas, the seas hath cast me on the rocks, Washed me from shore to shore, and left my breath Nothing to think on but ensuing death (Pericles 1607, 44^7) When there is nothing living but thee, thou shalt be welcome. (Timon of Athens 1605, 357-358) This is thy eld'st son's son, Infortunate in nothing but in thee. (Life and Death of King John 1596, 177-178)

178

Terttu Nevalainen

The Shakespeare Corpus also offers no evidence for nothing but modifying a finite verb. Example (30) illustrates the typical case, where nothing must be interpreted as a direct object of the main verb does. Cases of nothing but premodifying a tensed verb - *he nothing but smiles - do not appear to be allowed. (30)

Why, what's the matter? Does he rave? No, madam, he does nothing but smile. (Twelfth Night 1601: 1, 10-11)

Returning to the quantitative diachrony of the construction, the Helsinki Corpus suggests that the syntactic discontinuity of nothing ...but was more or less the rule in the first half of the sixteenth century. About two thirds (62.5%) of the cases have some material separating the two parts of the construction. In the second and third Early Modern English subperiods, discontinuous cases cover a little over 40% of the total. I must therefore conclude that there is little evidence in favour of the grammaticalisation hypothesis in the Early Modern English data, either in the Helsinki Corpus or in the Shakespeare Canon. The corpus of Late Modern English prose provides further evidence for the argument in favour of compositionality. As many as five of the seven instances of nothing but are discontinuous; see example (31). There are a further four cases of nothing followed by a preposition of exception, namely three instances of nothing ... except and one of nothing... save. The latter is illustrated in (32). (31)

(32)

But no sooner was I donned than down came the rain again, through the mud roof of my room too and there was nothing for it but to change sadly into ordinary clothes - and write to you (BELL (vol. I) 397:12) Otherwise nothing - save a glimpse every now & then of Smith or Berridge. (DOWSON 147:21)

The (ncm-)grammaticalisation of negative exclusives in English

179

4.2. Present-Day English What is the status of nothing but in the twentieth century? One answer can easily be obtained from the ICAME CD-ROM, which contains four Present-Day English corpora, i.e. the Brown Corpus of written American English, the Lancaster-Oslo/Bergen Corpus of written British English, the Kolhapur Corpus of written Indian English (one million words each), and the half a million word London-Lund Corpus of spoken British English.5 My results are shown in Table 3. For the sake of comparison, I also give the overall frequency of the nothing (...) but construction in the corpora discussed above. Table 3 tells us that nothing but has lost ground considerably since the Early Modem English period. It is most common in the American English corpus, and even there its frequency has dropped to one third of its frequency in the two Early Modem English corpora. The present-day British English data show the lowest incidence of the expression, lower still in speech than in writing. Table 3.

Frequency of nothing (...) but per 100,000 words in Modern and Present-Day English

Corpus

nothing (...)

Early Modem English

HC (1500-1710) Shakespeare

14.3 15.5

Late Modern English

LModE Prose

7.0

Present-Day English

Brown (BUC) Kolhapur Lanc.-Oslo/Bergen (LOB) London-Lund (LLC)

4.8 4.3 3.7 2.2

but

Despite the quantitative change, there does not appear to be any qualitative change in the use of nothing but between Early Modem English and the present-day data examined. Examples drawn from the present-day corpora are given in (33)—(35): (33)

You may ask what else there is, and the answer is nothing - nothing but empty space. And since you are made of atoms, you are nothing much but empty space, too. (BUC Religion D13:54)

180

Terttu Nevcdainen

(34)

He had saved nothing and left the children with nothing but a bare room and debts. (Kolhapur Catg.F Fl8:62) and they yound they were !so bVored# - cos ΛαΙΙ the 'other :\English peopled did Nothing but 'play g/olß and [@jn]. *AdrVink#* (LLC 1 13:744-747)

(35)

The frequency of discontinuous instances of nothing... but also remains noticeably high in the present-day data. It is about one third of the cases in both the American English and spoken British English corpora, and about one fifth in the written British English corpus.6

5. Discussion It may be easier to think of reasons why the collocation nothing but has failed to grammaticalise than to try to find out why it has steadily been losing ground since Early Modern English. As the research on language variation and change shows, actual processes of language change can often be best understood in terms of multiple motivation, both linguistic and extralinguistic (for a recent statement, see Smith 1996). There is no reason why multiple motivation should not be relevant in language maintenance and stable variation as well. In the case of nothing but two related factors suggest themselves. Although the phrase was gathering momentum in the seventeenth century, the adverbialisation of nothing but may have been impeded by the sheer phonological weight and syntactic flexibility of the phrase. Throughout the Modern English period, these factors would combine with the paradigm support nothing gets from the other indefinite pronouns, both affirmative and negative. As argued by Lehmann (1985: 306-309) and others, phonological attrition and loss of semantic integrity are typical prerequisites for a process of grammaticalisation to take place. We have little evidence of any "weight loss" of either kind in the case of nothing but. In a situation like this, paradigm cohesion with the other indefinites would be an obvious factor retarding grammaticalisation. As long as we have phrases like no-one but, nowhere but; anyone but, anywhere but; everyone but and everywhere but (both in combination with the negative particle not and without it), the presence of nothing but, anything but, and everything but is well motivated.

The (non-)grammaticalisation of negative exclusives in English

181

The non-grammaticalisation of nothing but in Modern English could be partly related to the fact that many of these compound pronouns are themselves relative newcomers to the language, dating from the Middle English period and gaining ground in Early Modern English (see Raumolin-Brunberg 1994). A second point concerns the internal variability of the phrase; that is, its prepositional part. But is, of course, not the only preposition used with nothing and other indefinites to express exception. Save and except both occur from Early Modern English onwards. The presence of choice here would naturally further strengthen the autonomy of the pronominal part of the phrase. The advantages gained from a phrasal expression in discourse are also considerable. As we have seen, modifiability is one of them. The possibility of splitting nothing ... but in the clause has its thematic value. It gives the speaker/writer the opportunity to organise his or her message flexibly. Sentenceinitially, nothing serves as the theme of the sentence, while the postponed butphrase can convey new information in the rheme. The two examples in (36) and (37) illustrate the type of textual motivation that may have contributed to preserving the compositional status of nothing but. In both cases nothing appears as the grammatical subject of the sentence in which it occurs. (36)

(37)

Lastly, it maketh his will to be no will, as though his goods were not his owne: for nothing is ours but that which wee haue rightlie got (HC E2 Smith, Two Sermons, D7V) Thus the unstressed of gets its significance from its use with : nothing can snow snow but "it". (BUC Learned J33:27)

If we take a paradigmatic view of exclusive focusing adverbs, we may find reasons why nothing but is not so much in demand today as it seems to have been in Early Modern English. Around 1600, but was one of the two quantitative prototypes of the exclusive adverbial paradigm, even morefrequentthan only in the materials that I have studied (Nevalainen 1991: 158). Thinking in terms of Jespersen's cycle, it would perhaps not be too far-fetched to argue that at that time nothing but was on its way to becoming a more emphatic variant of the monosyllabic but. The information given in Table 2 in section 4.1 will lend quantitative support to this argument. There were, however, also other developments that influenced the exclusive

182

Terttu Nevalainen

paradigm in Modern English. As shown by Table 4 (based on Nevalainen 1991), the most important of these developments must have been the growing frequency of only throughout the period. Others include the increase in popularity of simply and merely in Late Modern English and the rise ofjust as an exclusive adverb. In colloquial text types, such as personal letters and comedies, just was more populaf in the latter half of the nineteenth century than but, the use of which was dramatically declining. Table 4. Frequency per 100,000 words of some of the most common exclusive adverbs in Modem English ModE

only

simply

merely

just

EModE 1500-1560 1570-1630 1640-1700

56.7 74.9 101.5

0.8 0.7 0.4

0.0 3.1 1.5

0.0 0.0 0.6

LModE 1840-1900

145.4

14.6

13.8

14.0

The three incoming elements, simply, merely and just, are illustrated by examples (38) to (40) from Nevalainen (1991). (38)

(39)

(40)

I assure you, my dear Major, my dear Madam, the gracious young lady simply saved my life, nothing else. (G.B. Shaw, Arms and the Man, p. 85) The simplest operation, and therefore the first, is merely to place these planes on a piece of paper, and pass the pencil round them. (Herbert Spencer, Education, p. 94) If I can whisper into his soul that I am just a shepherd, I shall recover his youth. (Joseph Parker, The City Temple Pulpit, p. 274)

But had, moreover, undergone a register shift and was mostly found at the more literate end of the oral-literate continuum. Under the circumstances there was little support for nothing but from the simple exclusive. What is more, the new contender

The (non-)grammaticalisaticHi of negative exclusives in English

183

just was on its way to becoming one of the two prototypes of the exclusive adverb paradigm - a position which it has fully attained in present-day spoken English (Nevalainen 1991: 151-154; Tottie 1986: 99).

6. Conclusion Negative exclusives nicely illustrate the diachronic balancing act between stable syntactic variability and grammaticalisation, or discourse autonomy and the loss of it. While Jespersen's cycle ran its full course in the case of the exclusive adverb but, with not but it was only extended to the modal context of can. With nothing but the process seems, partly for discourse reasons, to have come to a halt in Standard English. Further motivation for this can be found in the paradigms that nothing but is a member of, in its uninterrupted cohesion with other indefinite pronouns on the one hand, and the diminishing support from the exclusive but, on the other. The rise of exclusive adverbs like just in Late Modern English is another paradigmatic factor that may have curbed the adverbialisation of nothing but and other negative exclusives in the twentieth century. The extent to which their course of development might have altered towards the end of this century will no doubt emerge from such new data sources as the British National Corpus (BNC) and the International Corpus of English (ICE).

Notes 1.

The first part of the title appeared as the heading of an editorial in The Independent on Sunday, on the 27th of March, 1994.

2.

Some of these items are polyfunctional, or homomorphs, as Quirk et al. (1985: 70-71) call functionally different but formally identical and etymologically related lexemes. Just is a good example of this kind of layering and specialisation (see Nevalainen 1991: 49-53, Hopper 1991).

3.

Although the terminology varies, most writers on the topic agree on a distinction between exclusives or restrictives and additives or inclusives (Nevalainen 1991: 55; see also König 1991).

184

Terttu Nevalainen

4.

The Corpus of Late Modem English Prose is based on my selection of Late Modern English correspondence in Nevalainen (1991); the material, about 100,600 running words, was computerised and provided with textlevel coding at the Department of English, University of Manchester (see Denison 1994).

5.

The ICAME CD-ROM also contains the diachronic part of the Helsinki Corpus of English Texts. It is available from the Norwegian Computing Centre for the Humanities, University of Bergen, Norway.

6.

Interestingly, the Kolhapur Corpus has one and the LOB as many as three instances of the phrase nothing but the truth. The instances may allude to the familiar oath that witnesses have to take in courts of law swearing to tell "the truth, the whole truth, and nothing but the truth".

The (non-)grammaticalisation of negative exclusives in English

185

References Adamson, Sylvia - Vivien Law - Nigel Vincent - Susan Wright (eds.) 1990 Papers from the Fifth International Conference on English Historical Linguistics. (Current Issues in Linguistic Theory 65.) Amsterdam - Philadelphia: Benjamins. Athanasiadou, Angeliki - Rene Dirven (eds.) 1997 On conditionals again. (Current Issues in Linguistic Theory 143.) Amsterdam - Philadelphia: Benjamins. COBUILD = Collins COBUILD English Language Dictionary 1987 London - Glasgow: Collins. Denison, David 1994 "A corpus of Late Modern English prose", in: Merja Kytö - Matti Rissanen - Susan Wright (eds.), 7-16. Fodor, Jerry Α.- Jerrold J. Katz (eds.) 1964 The structure of language. Englewood Cliffs, NJ: Prentice-Hall. Hopper, Paul 1991 "On some principles of grammaticalization", in: Elizabeth Closs Traugott - Bernd Heine (eds.), 17-35. Jespersen, Otto 1917 "Negation in English and other languages", Det Kgl. Danske Videnskabernes Selskab. Historisk-filologiske Meddelelser 1: 1-151. Copenhagen. [1962] [Reprinted in Selected writings of Otto Jespersen. London: George Allen and Unwin.] Kastovsky, Dieter (ed.) 1994 Studies in Early Modern English Berlin - New York: Mouton de Gruyter. Klima, Edward S. 1964 "Negation in English", in: Jeny A. Fodor - Jerrold J. Katz (eds.), 246-323. König, Ekkehard 1991 The meaning of focus particles. A comparative perspective. London: Routledge.

186

Terttu Nevalainen

Kytö, Meija - Matti Rissanen - Susan Wright (eds.) 1994 Corpora across the centuries: Proceedings of the First International Colloquium on English Diachronic Corpora, St Catharine's College Cambridge, 25-27March 1993. Amsterdam - Atlanta, GA: Rodopi. LDOCE = Longman Dictionary of Contemporary English 1987 London: Longman. Lehmann, Christian 1985 "Grammaticalization: Synchronic variation and diachronic change", Lingua e Stile 20: 303-318. Nevalainen, Terttu 1990 "Modelling functional differentiation and function loss: The case of buf\ in: Sylvia Adamson - Vivien Law - Nigel Vincent Susan Wright (eds.), 337-355. 1991 But only, just· Focusing adverbial change in Modern English 1500-1900. (Memoires de la Societe Neophilologique de Helsinki LI.) Helsinki: Societe Neophilologique. OED = The Oxford English Dictionary 1989 Prepared by J.A. Simpson - E.S.C. Weiner. (2nd edition.) 20 Volumes. Oxford: Clarendon Press. Quirk, Randolph - Sidney Greenbaum - Geoffrey Leech - Jan Svairtvik 1985 A comprehensive grammar of the English language. London: Longman. Raumolin-Brunberg, Helena 1994 "The development of the compound pronouns in -body and -one in Early Modern English", in: Dieter Kastovsky (ed.), 301-324. Smith, Jeremy 1996 An historical study of English. Function, form and change. London - New York: Routledge. Taglicht, Josef 1984 Message and emphasis: On focus and scope in English London: Longman.

The (non-)grammaticalisation of negative exclusives in English Tottie, Gunnel 1986

187

"The importance of being adverbial. Adverbials of focusing and contingency in spoken and written English", in: Gunnel Tottie Ingegerd Bäcklund (eds.), 93-118. Tottie, Gunnel - Ingegerd Bäcklund (eds.) 1986 English in speech and writing. (Studia Anglistica Upsaliensia 60.) Stockholm: Almqvist & Wiksell International. Traugott, Elizabeth Closs 1997 "Unless and but conditionals: A historical perspective", in: Angeliki Athanasiadou - Rene Dirven (eds.), 145-167. Traugott, Elizabeth Closs - Bernd Heine 1991 "Introduction", in: Elizabeth Closs Traugott- Bernd Heine (eds.), 1-14. Traugott, Elizabeth Closs - Bernd Heine (eds.) 1991 Approaches togrammaticalization. Vol. 1. Amsterdam-Philadelphia: Benjamins. Upton, Clive - Stewart Sanderson - John Widdowson 1987 Word maps: A dialect atlas of England. London - New York Sydney: Croom Helm.

Isn't it? or is it not? On the order of postverbal subject and negative particle in the history of English Matti Rissanen

1. Introduction The unmarked position of the English negative adverb or particle not (< OE nauht/nouht Aand he writes !cl\early# A> Aand !p\urposefidly a'bout them# A> this is Atr\ue (is it ^n\ot#}# Ο yes# Ο yes# (London-Lund 2 6 946)

6. Concluding remarks In the present paper I have attempted to show how a survey of the ordering of the negative adverb or particle and the postverbal subject can give us valuable information on the gradual grammaticalisation of naught > not, and of its weakening and cliticisation even before this phenomenon can be seen in spelling. Not was completely grammaticalised by the late fifteenth century, by the time the first instances of the order V+woi+Pers Pron Subj occur in unmarked contexts. The next step in grammaticalisation can be said to take place by the nineteenth century

200

Matti Rissanen

when the interrogative constructions with enclitic, contracted -n 't supersede those in which the non-contracted not immediately follows the auxiliary or be/have, or in which the personal pronoun subject precedes the negative particle. The rapid disappearance of the construction with uncontracted »οί+Pers Pron Subj from the late seventeenth century on might even imply that this order was never quite established in the language but that the Modern English change in the unmarked order from Pers Pron Subj +not to noi+Pers Pron Subj was caused by the sixteenth-century tendency towards cliticisation and contraction of the negative particle. This tendency could also be taken into account in the discussion of the early establishment of i/o-periphrasis in negative questions in English.15

Notes 1.

For the order «οί+fmite verb, which mainly occurs in Early Modern English, see Ukaji (1992) and Beukema (this volume).

2.

In this article the symbol V refers to both the verb and (pre)auxiliaiy categories, unless the contrast between the two classes is explicitly indicated. The capitalised abbreviated titles in square brackets refer to the text list of the Manud to the Helsinki Corpus (Kytö 1996: 169-230).

3.

The preverbal position of the negative adverb na, nawihtlnauht was not uncommon either.

4.

iVancmht (with variant spellings) was originally an indefinite pronoun 'nothing', 'naught'. Its adverbialisation, which marks the first stage in its grammaticalisation and Middle English development into the negative particle proper, began in Old English (cf. OED s.v. naught C.I.).

5.

In his unpublished licentiate thesis Mönkkönen (1984) gives a number of instances of both orders. He does not comment on the correspondence of the type of the subject with word order, but his examples confirm the pattern suggested here. As shown by Rissanen (1994), the placement of the demonstrative pronoun subject in negative questions is the same as that of the noun subject.

The order of subject and negative particle

201

6.

I am indebted to Anthony Kroch and Ann Taylor for providing me with a list of the instances of postverbal subject in negative sentences in the Penn-Helsinki Corpus. I am also grateful to Merja Kytö for valuable technical assistance in corpus handling.

7.

The grammaticalisation process of the negative adverb is of course closely connected with the gradual loss of the preverbal ne. When not had become a more or less compulsory part of the negation, the weakly stressed preverbal ne became optional and finally disappeared. The close link that had earlier existed between ne and the finite verb now developed between the finite verb and not. The earliest sporadic instances of the spellings natlnot, which distinguish the negative particle from the adverb naught/nought, occur in the early fourteenth century (OED s.v. not adv. and sb. 7; MED s.v. not adv. la). This spelling becomes common in the second half of the fourteenth century.

8.

From Early Modern English on, only those instances in which inversion is caused by interrogation are included in the figures. Inversion in conditional clauses with zero link, or caused by sentence-initial negative adverbials in Modern English follows somewhat different word order patterns. The number of such instances is negligible.

9.

Spelling is of course a highly unreliable indicator of pronunciation. It is worth pointing out, however, that the printers of Tyndale and Respublica are not inconsistent in their conventions of combining two words into one. In such combinations, one of the words seems to be weakly stressed.

10.

Cf. Ellegärd (1953: 162). As has been pointed out in most standard studies of the emergence of c/o-periphrasis in negations, the use of this auxiliary also solved the problem of placing the negative particle before the semantical ly meaningful part of the verbal group.

11.

Salmon (1966: 128-129) observes the relevance of the type of the subject in the placement of not in negative questions in Shakespeare, but she gives no distribution figures.

202

Matti Rissanen

12.

Including four instances of Aux+the enclitic contracted form -n't.

13.

The version used here is experimental and not yet quite complete. I am grateful to Douglas Biber and Edward Finegan for permission to use this Corpus.

14.

The only exception in the LOB Corpus occurs in a learned text: if Orpheus could call up his wife's shade in Erebus, could not he, Aeneas, also a descendcnt of the gods, mcke the same journey (J62: 37)

15.

The interesting question of the avoidance of the contracted forms of the negator in many dialects, particularly in northern and western England, Scotland and Ulster, falls outside the scope of this paper. The Helsinki Dialect Corpus, now in preparation, does not, at its present stage, include samples from most of these dialects. Reference can be made e.g. to Orton - Sanderson - Widdowson (1978: maps M9-15, M27, M31-33b, M37-38, M42-46, M49-50, S9); Viereck - Ramisch, eds. (1991: maps M46, SI7); Edwards - Weltens (1985: 107); Upton et al. (1994: 494-502). It is also worth pointing out that Anneli Meurman-Solin's Helsinki Corpus of Older Scots (1450-1700), which was completed in 1995, does not contain any instances of the contracted form -n't. The contrast to the Early Modern English section of the Helsinki Corpus, representing mainly the Southern Standard, is striking.

Hie crder of subject and negative particle

203

References Biber, Douglas - Edward Finegan - Dwight Atkinson - Ann Beck - Dennis Burges - Jena Burges 1994 "The design and analysis of the ARCHER Corpus: A progress report", in: Meija Kytö - Matti Rissanen - Susan Wright (eds.), 3-6. Brainerd, Barron 1989 "The contractions of not: A historical note", Journal of English Linguistics 22: 176-196. [1993] [Reprinted.] Edwards, Viv - Bert Weltens 1985 "Research on non-standard dialects of British English: Progress and prospects", in: Viereck (ed.), 97-123. Ellegärd, Alvar 1953 The auxiliary do, the establishment and regulation of its use in English. Stockholm: Almqvist & Wiksell. Jespersen, Otto 1917 "Negation in English and other languages", Det Kgl. Danske Videnskabernes Selskab. Historisk-filologiske Meddelelser 1: 1-151. Copenhagen. [1962] [Reprinted in Selected writings of Otto Jespersen. London: George Allen and Unwin.] Kastovsky, Dieter (ed.) 1994 Studies in Early Modern English. Berlin - New York: Mouton de Gruyter. Kytö, Meija 1996 Manual to the diachronic part of the Helsinki Corpus of English Texts: Coding conventions and lists ofsource texts. (3rd edition.) Helsinki: Department of English, University of Helsinki.

204

Matti Rissanen

Kytö, Meija - Matti Rissanen - Susan Wright (eds.) 1994 Corpora across the centuries: Proceedings of the First International Colloquium on English Diachronic Corpora, St Catharine's College Cambridge, 25-27March 1993. Amsterdam - Atlanta, Ga.: Rodopi. MED = The Middle English Dictionary 1956Ann Arbor, Michigan: Michigan University Press. Mönkkönen, Ilkka 1984 Negation in Old English prose. [Unpublished licentiate thesis, University of Helsinki.] OED = The Oxford English Dictonary 1989 Prepared by J.A. Simpson - E.S.C. Weiner. (2nd edition.) 20 Volumes. Oxford: Clarendon Press. Orton, Harold - Stewart Sanderson - John Widdowson (eds.) 1978 The linguistic atlas of England. London: Croom Helm. Quirk, Randolph - Sidney Greenbaum - Geoffrey Leech - Jan Svartvik 1985 A comprehensive grammar of the English language. London: Longman. Rissanen, Matti 1994 "The position of not in Early Modern English questions", in: Kastovsky (ed.), 339-348. Rissanen, Matti - Ossi Ihalainen - Terttu Nevalainen - Irma Taavitsainen (eds.) 1992 History of Englishes: New methods and interpretations in historical linguistics. Berlin - New York: Mouton de Gruyter. Salmon, Vivian 1966 "Sentence structures in colloquial Shakespearean English", Transactions of the Philological Society 1965. Oxford: Basil Blackwell, 105-140. Ukaji, Masatomo 1992 " Ί not say': Bridge phenomenon in syntactic change", in: Matti Rissanen - Ossi Ihalainen - Terttu Nevalainen - Irma Taavitsainen (eds.), 453-462.

The crder of subject and negative particle

205

Upton, Clive - David Parry - J.D.A. Widdowson 1994 Survey ofEnglish Dialects: The dictionary and grammar. London - New York: Routledge. Viereck, Wolfgang (ed.) 1985 Focus on: England and Wales. Amsterdam - Philadelphia: Benjamins. Viereck, Wolfgang - Heinrich Ramisch (eds.) 1991 The computer developed linguistic atlas of England. Tübingen: Niemeyer.

The origin and development of the "Neg ... neither" construction: a case of grammaticalisation Ingrid Tieken-Boon van Ostade

1. Introduction While in the only two extant versions of Malory's Morte Darthur (1469-70, 1485) a large variety of constructions with multiple negation is found (Tieken 1995), one type is conspicuous by its absence: the "Neg ... neither" construction, i.e. a type of negative sentence with an additional neither, as in (1): (1)

Nay, not so neither (ca 1608 Shakespeare Coriolanus IV.5.171)

Its absence is striking because in the eighteenth century it is by far the most frequently encountered double negative construction (Tieken 1982: 281-282). On the face of it, the "Neg ... neither" construction therefore appears to have arisen since the late fifteenth century, gaining great currency in the centuries intervening. In the eighteenth centuiy the construction was in fairly general use; I have come across it even in the relatively more formal styles of some educated writers, albeit not very frequently. This, however, is no longer the case, and Quirk et al. (1985: 784) describe only its counterpart with single negation, as: (2)

You 're no kid either, you know (1994 Michael Crichton Disclosure. London: Random House, 294)

At first sight it might therefore be surmised that in the standard language the construction disappeared along with multiple negation generally; in reality, however, the situation must have been more complicated. My evidence from the Morte Darthur suggests that the disappearing process of multiple negation was already well on its way at the end of the fifteenth century,1 when the new construction would not even have been in existence yet. So apart from looking for the origin and subsequent development of the "Neg ... neither" construction and discovering

208

Ingrid Tieken-Boon van Ostade

whether the construction with simple negation as in (2) did indeed arise out of the double negative - as it would seem to have done (see Jespersen 1917: 113) - I want to try and find out why and how the construction gained greater currency despite the fact that multiple negation had already been in the process of disappearing for some time. Because part of the source material for my analysis was obtained from the Oxford English Dictionary, the present article is at the same time an exploration into the possibilities of making use of the OED for the purpose of doing corpus analysis (cf. Jucker 1994: 149; and Fischer 1997).

2. A classification Jespersen (1917: 112-113) distinguishes two subcategories of the construction, one in which the first negative is nor, while in the other any negative other than nor occurs as first negative. The first type of construction, which is illustrated by (3), I shall refer to as Type 1 and the second, illustrated by (4), as Type 2: (3) (4)

Type 1: It is not for your health... Nor for yours neither (1599 Shakespeare Julius Caesar II.i.327) Type 2: I hope, sister, things are not so very bad with you neither (1814 Jane Austen Mansfield Park. Penguin 1966, 63)

Jespersen argues that the function of neither in Type 2 sentences is that of an "afterthought". He even distinguishes a third variant of the construction, which will henceforth be referred to as Type 3, in which the "Neg ... neither" construction follows a positive statement and serves to deny this statement emphatically. Some examples from my own corpus are given in (5) and (6). In Type 3 sentences, according to Jespersen, "neither merely 'clinches the argument' by making the negative veiy emphatic" (1917: 113, see also Jespersen 1940: 453). (5)

Type 3: EdKno'well You speake very well, cousse. M. Stephen. Nay, not so neither. (1616 Jonson Every man in his humour I.iii.104-105, ed. Herford - Simpson. Vol. III. 1927: 314)

The origin and development of the "Neg ... neither" construction

(6)

209

Type 3: "I'll walk "... "You won't neither. " (1931 Dashiell Hammett The glass key. London: Pan Books. [1975]: 145)

The "Neg ... neither" construction is also discussed by Ukaji (1979) and Austin (1984), though neither distinguishes between different subcategories. Despite the fact that Ukaji regards neither in this construction as "empty", he still assigns a meaning to it, i.e. "for all that, nevertheless" (1979: 113-115). To my mind, a distinction such as the one proposed by Jespersen, according to which only Type 3 sentences cany any emphasis should, indeed, be made. I do, however, believe that the meaning 'for all that, nevertheless' suggested by Ukaji should be assigned to Type 2 sentences only: neither in Type 1 has a different meaning, i.e. 'no more, also not' (cf. LDOCE s.v. either A). Present-Day English would require either here, as in example (2), and Quirk et al. provide this use of either with the label "additive adverbial" (1985: 782-783). The sense of implied contrast, which is typically characteristic of sentences belonging to Type 2, is thus absent from instances of the Type 1 construction. In Type 2 sentences, the substitution of either is therefore not possible, and this serves as an important distinction between the two categories (cf. Ukaji 1979: 115). Type 3 is in my opinion an emphatic variant of the Type 2 construction.

3. The data Over fifteen years ago, when I began my study of multiple negation in the Morte Darthur, I rather unsystematically collected each and every sentence with multiple negation I happened to set my eyes on. This resulted in a fairly substantial corpus of instances, ranging from around 1550 to 1970. Despite its size, the corpus is nevertheless not representative of usage at any given period due to the haphazard means of compiling it. In order to undertake a more systematic search for instances of the "Neg ... neither" construction I decided to make use of the text provided by the OED as illustrative material. Now that the OED is available in electronic form, it is possible to treat the quotation areas of all the dictionary entries as corpus material which can be made the subject of analysis (cf. Fischer 1997).2 As a corpus, the amount of material is enormous, but it is probably equally unrepresentative as

210

Ingrid Tieken-Boon van Ostade

my own corpus of usage due to the method by which source texts were originally selected for extraction. According to Schäfer (1980: 13), the material read for the OED was primarily literaiy in nature, and certain authors moreover received preferential treatment. Shakespeare is one such author. At the same time, certain periods, such as the late fifteenth century, are underrepresented as compared to others (Schäfer 1980: 4). A more representative body of material is formed by the Helsinki Corpus, which was set up to cover as wide a variety of genres, styles and authors as was feasible within the format chosen (Nevalainen - Raumolin-Brunberg 1989), though this corpus has its own shortcomings, mainly in terms of size (Fischer 1997: 161) and coverage: it does not go beyond the first decade of the eighteenth century (see also Rissanen 1994: 77). I therefore checked the Helsinki Corpus too,3 as well as the other text corpora supplied with it in its CD-ROM version (LOB, London/Lund, Kolhapur, Brown)4 for any additional instances. The results are the following. My own modern English corpus contains nearly eighty instances of the construction I am discussing here, spanning a period of over 400 years. The Helsinki Corpus produced nineteen instances.5 None of the other corpora yielded any instances, except for the Brown Corpus (7). Searching the OED on CD-ROM meant an interesting if time-consuming experience (cf. Jucker 1994: 154). As it proved impossible to combine searches for neither and not (the word not has apparently not been marked up to avoid causing a memory overflow), I had to check all 4479 instances of neither, distributed over 3901 entries, to identify the construction in question. Thus, all other uses of the word, as in the sense of 'neither of the two' and 'nor', had to be excluded, as well as all those instances in which neither did not appear in end position. For practical purposes I limited my search to just over half the number of entries (all instances under A-O, i.e. 2042/3901 lemmas). This search yielded 43 instances of the construction in question. Searching under alternative spellings of the word produced an additional 29 instances.6 In addition, I encountered two more instances under the entry nother which had not been marked up. My OED corpus thus consists of 74 instances. There is only one case of overlap between the various corpora collected, i.e. (7): (7)

agef me for nawiht ne jeoue ich for inc nowöer (a. 1225 Julicna 19)

The origin and development of the " N e g ... neither" construction

211

This sentence occurs both in the Helsinki Corpus and in my data from the OED. Altogether, there are 177 instances for analysis. For the sake of convenience, instances from each of the separate corpora will in the subsequent discussion be marked to indicate their origin (T = Tieken Corpus, HC = Helsinki Corpus, BUC = Brown Corpus, OED = OED quotation material). The oldest instance dates from around 1200: (8)

Forr

hellepitt

(OED:

Ne gredi3nesse 10216).

niss ncefre full,

c. 1200

Ormulum

nowwjjerr

The entire corpus thus spans a period of eight centuries. Chronologically, the instances may be arranged as in Table 1. Table I.

A

chronological

distribution

of

the

instances

encountered

in

the

various

subcorpora

Tieken

Helsinki

Corpus

Corpus

Brown Corpus

OED Corpus

Total

13thc

4

4

8

14thc

-

6

6

15thc

2

8

10

16thc

9

5

10

24

17thc

28

8

20

56

18thc

19

Β

13

32

19thc

10

13

23

20thc

11

7

(-)

18

Total

77

7

74

177

19

If the figures in this table are anything to go by, it would seem that the construction dates back to the Early Middle English period, that it gained greater currency after

212

Ingrid Tieken-Boon van Ostade

the Middle English period and that, upon reaching a peak in the seventeenth century, it began to disappear in the course of the eighteenth century. Moreover, it would apparently still be fairly current today. A closer look at the data, however, reveals a more complex picture. Thus it is striking that five of the earliest recorded instances are from texts belonging to the so-called Katherine group: three from St Juliana, one from St Margaret and one from Hali meidenhad; one more instance is from the closely related Ancrene riwle. This might merely point to another literary bias in the collection of the OED material; however, the two remaining ones of the earliest instances are from the Ormulum and from the South English legendary (Life of St Edmund), which suggests instead that the construction appears to have originated in the Midlands, being possibly more common in the West than in the East. Furthermore, in the majority of the early instances, neither occurs in the context of only two negated items, as in: (9)

pt tu nabbe pi wil wid him. ne weole nowöer (HC: ca. 1225 Hali meidenhad 150),

the two negated elements being pi wil and weole. In (10), the negated elements are wind and weder and in (11) rug and side: (10) (11)

... pat ne dreded na wind ne na weder nowöer {OED: c. 1225 Juliana 72) he ne sparede rug ne side nojjur (HC: 1285-1295 Life of St. Edmund 434)

Thus, the original meaning of the word neither, i.e. 'not each of the two' (cf. Hoad 1986, s.v. either, neither), seems to lie at the basis of the "Neg ... neither" construction. Some further evidence for this possibility may be found in the following Old English sentence from the Helsinki Corpus: (12)

ac hi[ne] ne went rwefre na|x>r. ne weder, ne unweder 'neither of the two, namely good or bad weather, ever changed it' (HC: AD 993 /Elfric De temporibus anni 64)

The origin and development of the "Neg ... neither" construction

213

In the first part of this sentence, which is structurally identical to the construction under investigation, nador is the subject of went, and it comprises the phrase ne weder ne unweder. This use of neither subsequently seems to have weakened into 'and not either' or 'no more, also not', one of its later meanings (cf. Hoad 1986, s.v. neither), as appears from the fact that later instances of the construction occur with more than two negated items. An example is the following sentence: (13)

Of kynge, ne conceyle ne of pe comunes nojjer (OED: 1399 Rieh. Redeies iv.60)

Neither thus acquired the function of a coordinating conjunction, though it follows rather than precedes the clause or phrase it links to the preceding one. Jespersen likewise regards neither as a co-ordinator, calling it a "supplementary connective" (1917: 112). The evidence from the corpus therefore suggests that the "Neg ... neither" construction originated during the Early Middle English period in the Midlands. It seems to have arisen out of a structurally identical construction which is considerably older, in which neither had the basic meaning of denying two alternatives (cf. OED, s.v. neither A.1). The subsequent change into a co-ordinator was accompanied by a process of semantic bleaching, the original sense 'not each of the two' having weakened into 'and not either, no more'. Furthermore, in all Middle English instances neither co-occurs with ne (or a spelling variant) meaning 'nor', which indicates that Type 1 was the oldest construction and that Type 2, with negatives other than nor, was a later development. In the course of time, constructions belonging to Type 1 would appear with negatives other than nor as well, as example (1) demonstrates. As for the geographical spread of the construction, it appears that with the appearance of Piers Ploughman the construction had moved to the London area, as the corpus contains two instances from this text: (14) (15)

Was no pride on his apparail ne no pouert nojjer (OED: 1362 Langland Piers Ploughman A.ix.l 11) Euery man helpe other ... And be we noujte vnkynde of owre catel ne of owre kunnynge neyther (OED: 1377 Langland Piers Ploughman B.xi.206)7

214

Ingrid Tieken-Boon van Ostade

It is not unlikely that the author, who is believed to have originated from Shropshire in the West Midlands, took the construction with him when he moved to London; nor is it even unlikely that Langland's text was instrumental in its spread, for the large number of manuscripts that have come down to us - Schmidt (1995: x-xii) records 69 manuscripts of all versions together - testifies to its popularity.8 Whatever may have been the case, by the beginning of the next century the construction had made its way into the Chancery records, as the following instance from the Helsinki Corpus indicates: (16)

I myghte not haue ben remedied ne myne neyghebores nother so sone at that tyrrn... (HC: 1414 XX Doc Pet3 198)

It is striking that the words of the petition from which (16) derives are not those of a Londoner but of someone from Cambridgeshire (Fisher et al. 1984: 198). It thus appears as if around the turn of the fourteenth century the construction entered the London area from a variety of directions at the same time. Almost a century later the construction is encountered in the language of Caxton: (17)

"I shall never sette foote there ". ' IMor / nother'sayd Richarde {OED·, ca. 1489 Caxton Sonnes of Aymon ix. 222)

We should nevertheless be careful in attributing the construction to Caxton's own usage, as the concordance to his own prose, i.e. the prologues and epilogues to the texts he published, contains not a single instance of the "Neg ... neither" construction (Mizobata 1990, s.v. neither, nother and variant spellings). Caxton claims to originate from Kent (Blake 1969: 19), and it is not unlikely that the construction had not spread as far as Kent yet at the time of his youth. That he still does not use it so many years later may be explained by the fact that the construction occurs in direct speech in (17): even if he had adopted it since his arrival in London, he may have considered the construction to be too colloquial to make use of it in the more formal styles of his prologues or epilogues. That the construction is not yet encountered in the language of Malory's Morte Darthur must therefore be due to the fact that it had not yet spread to the author's dialect.9

The origin and development of the "Neg ... neither''' construction

215

Classification of the instances in the corpus proved no easy matter. A simple substitution test, of "either" for Type 1 instances and "nevertheless" for Type 2, was not sufficient, as the (a) and (b) readings of example (18), a sentence from the Clift family correspondence (1792-1846), suggest (Austin 1984:146): (18)

the Housekeeper was not in neither a. 'the Housekeeper was not in either' b. 'the Housekeeper was nevertheless not in'

Nor is it the case, as Jespersen suggests, that the first Negative of a Type 1 sentence is always nor. (19)

M-s Cotton:... an 'e says "You're no lady to talk like that". An' I says "I know I'm not - but you're no greengrocer neither..." (T: 1947 J.B. Priestley The linden tree. In: Time and the Conways and other plays, Penguin 1969 [1973]: 288)

On the other hand it does seem to be the case that all instances with nor as their first negative belong to Type 1. One example of a Type 2 instance is the following quotation from the OED (s.v. carrot sb. 4): (20)

If I had said your head was Red I had not seen such a Liar neither; it was direct Carrot (OED·. 1671 Glanvill A further discovery of Μ Stubbe 28)

Another example, though with an implied negative in but, is (21): (21)

lacke Leiden ... had ...a peece of a rustic sword ... by his side ... it was but a foyle neither, and he wore it, to shewe that he should haue the foyle of his Enemies (OED·. 1594 Nashe The unfortunate traveller 21)

The combination of but and neither does not serve to Type 2 instances either, as neither in (22), also from the OED (s.v. baste c.) and likewise with but as an implied negative, has the sense of 'also', which classifies the instance as Type 1:

216

Ingrid Tieken-Boon van Ostade

(22)

The body of your discourse is sometime guarded with fragments, and the guardes are but slightly basted on neither {OED: 1599 Shakespeare Much ado about nothing, I.i.289)

Difficult though a classification may be, it is clear that Type 2 is a later development of the construction. The first unequivocal instances in the corpus, examples (20) and (21), date from the late sixteenth century. Type 3 is first attested around the same time.

4. A grammaticalisation process The above discussion of the data suggests two separate developments, one in which neither underwent a process of weakening and developed into a grammatical marker having the function of linking one negative clause to another (Type 1), and another, later, development in which neither acquired a different function along with the potential of carrying emphasis (Type 2). In the course of the former change, neither shifted category, from that of an adverb (cf. Hoad 1986, s.v. neither, OED, s.v. neither) to a co-ordinating conjunction, while the latter development did not affect the status of neither. The question is whether we have indeed to do with two independent developments or whether the two developments can somehow be reconciled into a single process. For obvious reasons the latter possibility would offer the most attractive solution, and I will therefore try to pursue it here. A particularly fruitful approach to the problem seems to be offered by regarding the various developments of neither as part of a grammaticalisation process, since in the course of the discussion so far two aspects have been identified in the rise of the Type 1 construction, i.e. semantic bleaching and category change, which are part of this process (see McMahon 1994: 168-170). Hopper and Traugott (1993: xv) define grammaticalisation as "the process whereby lexical items and constructions come in certain linguistic contexts to serve grammatical functions". "Once grammaticalized," the authors continue, the lexical items and constructions "continue to develop new grammatical meanings". In the case of the "Neg ... neither" construction, this is exactly what appears to have happened. For an item to become grammaticalised, it must first undergo a process of generalisation of function (Hopper - Traugott 1993: 97). Such a development does

The origin and development of the "Neg ... neither" construction

217

indeed characterise the early stages of the development of the "Neg ... neither" construction: while neither first collocated with two elements and with the negatives nelnor only, as in (7)-(ll), it gradually appears in collocation with more than two elements and with negatives other than nor. This points to a generalisation of function on two counts. At the same time, grammaticalised elements usually undergo a change of grammatical category. Hopper and Traugott (1993: 104) discuss a "cline of categoriality", along which items change from a major grammatical category (nouns or verbs) to a minor one (prepositions, conjunctions, auxiliaries, pronouns, demonstratives). The categories of adjectives and adverbs form an intermediate class in this respect. In the course of its history, neither in Type 1 instances changed from being an adverb to a conjunction. In its earliest forms of occurrence, it is usually described as an adverb (cf. OED and Hoad 1986, s.v. either, neither), while in its attributive uses neither functions as an adjective, as in neither approach was very fruitful. The latter type of construction is not relevant to the discussion here. Despite the development of neither sketched above, Quirk et al. (1985) do not distinguish between different uses in Present-Day English of neither as far as grammatical category is concerned, always treating neither - and either - as an adverb (e.g. 1985: 782-784, 883, 937). They do, however, observe that neither belongs to the "class of linking adverbs which most resemble coordinators" (1985: 938). In Type 2 constructions, neither still has the function of an adverb. While neither was undergoing grammaticalisation, its original function survived as well. Such a situation is characteristic of grammaticalisation generally, and is known as "divergence" (Hopper - Traugott 1993: 116-117). One example is the development of an into the indefinite article a, while the full form regularly developed into the numeral one. The origin of neither in Type 2 constructions, first as a marker of contrast as in (20) and (21) and subsequently, it seems to me, of emphatic contrast as in (5) and (6), may be explained as having arisen from the desire to use multiple negation as a marker of emphasis. Despite the disappearing process which multiple negation had already been undergoing for some time, the emphatic function of multiple negation was still very much in evidence in the Early Modem English period (see Tieken 1995: 123), which is when the new construction is first attested. According to Hopper and Traugott (1993 : 94), "once grammatical ization has set in, there are certain likely paths along which it proceeds". One such path they summarise as: "lexical item used in specific linguistic contexts>syntax>morphology" (1993: 95). Eventually, the grammaticalised word or construction may

218

Ingrid Tieken-Boon van Ostade

even disappear (McMahon 1994: 168). As noted in section 2 above, Ukaji labelled the use of neither in the constructions under discussion as "empty". While the semantic content of neither is indeed no longer very large, its grammatical content, even in Present-Day English, is still veiy much in evidence, as the following examples show: (23)

(24)

I don't think you've had a very good one, neither (T: 1956 J.B. Priestley Mr Kettle and Mrs Moon. In: When we are married and other plays, Penguin 1969 [1973] : 309) "Don't no-one get out the door neither." (T: 1969 Michael Caine in the film The Itdicn job)

Nor is it the case that neither is about to be cliticised, which would be a clear step towards the form turning into an affix (Hopper - Traugott 1993: 6). Clitics are defined as "unaccented forms that tend to be found attached to a more heavily accented form" (Hopper - Traugott 1993: 132), and it is clear from examples (23) and (24) that neither carries full stress. These are relatively modern examples, which, together with the evidence from the Brown corpus (cf. Table 1), indicates that the "Neg ... neither" construction has not disappeared from Present-Day English. The Brown corpus evidence also suggests that the construction has been subject to a demoting process, though this process has little to do with the grammaticalisation of neither in Type 1 constructions. An analysis of the British National Corpus would throw further light on both these developments.

5. The construction demoted The sharp increase in currency of both types of the "Neg ... neither" construction during the seventeenth century indicated by the figures in Table 1 is followed by a decrease a century later. The increase may well have been brought about by the usefulness of the "Neg ... neither" construction as a potential discourse marker.10 Neither in this construction, according to Jespersen (1940: 453), "clinches the argument", and thus typically "helps to signal the direction in which ... a conversation is going" (Chalker - Weiner 1994, s.v. "discourse marker"). It is striking that of the eight seventeenth-century instances from the Helsinki Corpus, six are from plays,

The origin and development of the "Neg ... neither" construction

219

notably Middleton's Chaste maid in Cheapside (1) and Vanbrugh's Relapse (5). Similarly, two-thirds of my own Early Modern instances are from the language of drama, while the Brown corpus is a corpus of modern American spoken English. I have already observed with reference to example (17) that even as early as the late fifteenth century the "Neg ... neither" construction may have been characteristic of the more colloquial registers. The decrease of the construction, in formal written English in any case, seems to be the result of the adverse comment from normative grammarians regarding the use of multiple negation generally. That Dryden, in his revision of Shakespeare's Troilus and Cressida (1679) leaves one instance of the "Neg ... neither" construction unchanged and even adds another one (Tieken 1990a: 162-163) suggests that at the time the construction was still regarded as acceptable. This situation remains to a large extent unchanged for some time as is clear from Fuami's (1991: 629) distinction of eighteenth-century instances of multiple negation into standard and non-standard usage: all but one of her Standard English examples belong to the "Neg ... neither" construction. That the construction was still acceptable at least during the early half of the century is confirmed by a comment made by the grammarian Daniel Duncan in his New English Grammar (1731): "Neither is sometimes correspondent to also", as if you will not go out, nor I neither, answers to if you will go out, and I also" (1731: 282). Whether Duncan's application of the rule "two Negatives make a positive" is relevant here or not, the point he makes is clear: the "Neg ... neither" construction was still acceptable, while most other types of double negation no longer were. Multiple negation generally was beginning to receive adverse comment, as by Gough (1754: 114) who wrote that "two Negative Words ... must not be written in the same simple sentence", and Ussher (1785: 77) who observes that "/ can not eat none, is very improperly said for ..." (see Tieken 1982). In time, such comment must have also have affected the "Neg ... neither" construction. The only overt reference proscribing the "Neg ... neither" construction I have come across is found not in a grammar but in a dictionary, viz. the one by Dr Johnson (1755, s.v. neither): 3.

Sometimes at the end of a sentence it follows as a Negative; and often, though not very grammatically, yet emphatically, after another Negative.

220

Ingrid Tieken-Boon van Ostade

If it be thought that it is the greatness of distance, whereby the sound cannot be heard; we see that lightnings and coruscations, near at hand, yield no sound neither. Bacon Men come not to the knowledge of which are thought innate, 'till they come to the use of reason, nor then neither. Locke [italics added] Johnson considers the construction to be emphatic, though neither of his instances actually are. According to my classification they belong to Type 1. As a result of the grammarians' proscriptions, the "Neg ... neither" construction came to be relegated to the realm of the non-standard language, the only acceptable construction being its positive counterpart illustrated by (2) above. Precisely when the "Neg ... either'''' construction came into being is unclear. The OED does not record a negative sentence with either until 1828 (s.v. either)·. (25)

Thy sex cannot help that either (1828 Scott F.MPerth xxxii)

The Helsinki Corpus provides a much earlier instance, though not a negative one: (26)

the best Earledome in England or Scotland either (1630 John Taylor The pennyles pilgrimage 136)

It does therefore indeed seem likely that the "Neg ... either" construction arose through the process of back formation as a result of the normative comments referred to above, but that the rise of the new construction was facilitated by the already existing pattern of sentences like (26). Even if the construction with either is the only acceptable construction in PresentDay Standard English, its use is still considered to be informal and colloquial (Quirk et al. 1985: 937, 784n). More formal or "literary" variants of the construction are, according to Quirk et al. (1985: 784n): (27) (28)

he couldn't speak, (and) neither could he walk he couldn't speak, nor could he wcdk

The origin and development of the "Neg ... neither''' construction

221

Already by the mid-nineteenth century the "Neg ... neither" construction was characteristic of non-standard usage, as the following instances from George Eliot's Adam Bede illustrate: (29)

(30)

And to think ο' your knowing no better, Molly, and been here a-going i' nine months, and not for want ο' talking to, neither (T: 1859 George Eliot Adam Bede, Signet classics [1961]: 82) That 'ud be unreasonable. It isn 't right for old nor young nayther to make a bargain all ο' their own side (T: 1859 George Eliot Adam Bede, Signet classics [1961]: 147-148)

Nearly a hundred years later, Zandvoort (1945: 210), the only modern grammarian I have come across who discusses the "Neg ... neither" construction, does indeed observe that "In colloquial (some might say uneducated) English it is sometimes used in the same meaning as either". The example he gives is an instance of the Type 2 construction: (31)

there was once a time, and not so long ago neither

The Brown corpus contains two instances which are non-standard as well, as is suggested by the presence of other non-standard markers in the sentence such as he don't and ain't: (32) (33)

He don't care at all, and I don't care to get my head busted neither (BUC: General fiction. K24: 121) there ain't nobody askin for your ticket stub, neither (BUC: Adventure. N29: 5)

My own corpus contains as many as eight instances from various plays by J.B. Priestley written during the period 1938 to 1955. Priestley likewise appears to have used the construction to mark the language of the characters in question as colloquial, and possibly non-standard as well:

222

Ingrid Tieken-Boon van Ostade

(34)

Maria (angrily) I wouldn 't give her a penny. Clara (quickly) Nor me, neither (Τ: 1938 J.B. Priestley When we are married. In: When we are married and other plays. Penguin 1969: 76)

6. Conclusion Present-Day British English speakers have a choice of three constructions when they wish to express agreement with a previous negative statement: (35) (36) (37)

/ can7 swim - Neither can I I can't swim - I can't, either I can't swim - Me neither

The choice, however, is not equally open to all speakers or equally characteristic of all registers of language. Thus, (35) is marked as formal and even literary, and is therefore most characteristic of the written language, while (36) is part of the more colloquial, though still standard, registers of speech and writing. (37) is generally regarded as non-standard, as a result of which it will be avoided by careful speakers. In all three cases, neither!either functions as a co-ordinating conjunction, though the word in question occurs in different positions in the sentence. Historically, however, only the last two types have something in common. As for (35) and (36), at first sight they appear to be positional variants of each other, neither occurring at the beginning of a positive clause in (35) and either at the end of a negative one in (36). Though I did not come across any instances of the "Neg ... neither" construction in Malory's Morte Darthur, I did find examples of sentences in which neither, or one of its variant spellings, occurred at the beginning rather than at the end of the clause, as in: (38)

that the fadir loved nat the sonne, nother the sonne loved nat the fadir (ed. Vinaver 1990: 882)

The origin and development of the "Neg ... neither'" construction

223

In a different dialect, this sentence might have read: "that the father loved not the son, and the son loved not the father neither". That the "Neg ... neither''' construction did not occur in Malory's dialect alongside sentences like (38) suggests that the two types of construction were not interchangeable after all. This gives strong support to the view that by the end of the fifteenth century the grammaticalisation of the "Neg ... neither" construction was already well under way. Grammaticalisation, according to Hopper and Traugott (1993: xvi) is "a process whereby a form may become fixed and constrained in distribution". The evidence from the Morte Darthur suggests that this was already the case with neither. With respect to the use of the OED quotations as source material for the purpose of corpus analysis, there is of course no way of saying on the basis of the above account whether as a corpus the material accurately represents usage over the centuries. It is, however, striking, that the OED corpus does not provide any evidence which is in conflict with that of the other corpora I have used. According to Hoad (1987: 51), "Angus Macintosh estimated that in order to undertake an adequate study of the geographical distribution of Middle English lexical items it would be necessary to take a corpus of 4.5 million words". According to the introduction to the second edition (1989: xxiii), the OED contains nearly 2.5 million quotations. At a minimum of ten words per quotation and given a small amount of overlap caused by quotations being used more than once,11 the OED corpus would consist of ca. 25 million words of English from all periods in history, five times as much as Macintosh believed to be possibly adequate for Middle English alone. As noted above, the material extracted for the OED shows a strong literary bias; as more research is done on the OED, other biases become apparent, such as a sexist one (Tieken 1990b: 414; Willinsky 1994: 11), a British one (Algeo 1990: 146-148; Willinsky 1994: 11), and a social class one (Willinsky 1994: 11). Osselton (1993: 129-130), moreover, discovered a clustering of quotations from the period in which the OED editors lived. As a corpus, the OED material thus appears to be far from random (cf. Fischer 1997: 163). As far as its representativeness is concerned, the figures in Table 2 below show that only the fifteenth century shows a similar proportion of instances of the "Neg ... neither" construction from the total number of instances recorded.

224

Ingrid Tieken-Boon van Ostade

Table 2. Proportion of instances from the Helsinki Corpus and the OED in view of the total number of instances in these subcorpora Helsinki Corpus

Proportion of total

OED

Proportion of total

13thc

4

21%

4

5.4%

14thc

-

6

8.1%

15thc

2

10.5%

8

10.8%

16thc

5

26.3%

10

13.5%

17thc

8

42%

20

27%

Total

19

-

74

If it is indeed the case that the Helsinki Corpus is representative of usage of the period covered by it, it must be concluded that the OED material is not, with the exception of the fifteenth century, which may possibly have been better studied, i.e. more randomly, by the readers for the OED than any other period. Fischer (1997: 170) argues that the OED can be used as a historical copus to supplement the Helsinki Corpus. On the basis of my findings presented here I am inclined to agree, though when the question of respresentativeness comes up the OED material must be treated with extreme care, primarily due to its various biases that have been identified. Rather than being a "splendid" corpus in its own right, as Fischer claims (1997: 170), I would argue that it is only potentially so. If for the third edition, which is planned to appear around the beginning of the next century, the OED quotation material is reviewed against biases such as the ones outlined above, the New OED project will offer extremely valuable possibilities from the point of view of diachronic corpus analysis.

Notes 1.

See also Mazzon (1994), who dates the beginnings of the disappearing process in the Early Middle English period, and Iyeiri (this volume).

The origin and development of the "Neg ... neither" construction

225

2.

For the purpose of my analysis I have made use of the CD-ROM version of the first edition of the OED as this was the only version available at the time I did most of the research for this article. The first edition offers several advantages over the second edition, such as the possibility of searching by label, and should not too lightly be discarded as a research tool now that the second edition is available on CD-ROM, too.

3.

It is interesting to note that my approach is the reverse of that of Fischer (1997), who explored the possibilities of analysing the OED as a corpus in order to supplement his limited findings from the Helsinki Corpus.

4.

For a brief description of these corpora, see Aijmer - Altenberg (1991: 315-318).

5.

I am grateful to Meija Kytö for helping me compile the instances from the Helsinki Corpus before the material was generally available.

6.

In the history of English, neither occurs in an enormous variety of different spellings (cf. LALME, Vol. II, 303-308). My search included the following spelling variants: nather, nether, nither, nother, neither, noyther, naither, neother, neyther, nouther, nowther, nowwther, including their variants in -ir and with and without final -e, as well as those with eth and thorn instead of th.

7.

Example (14) occurs in the A, Β and C versions of the text, while (15) is found only in the Β version (see ed. Schmidt 1995). Note also that in Schmidt's edition, the second half of (14) reads ne pouert noper.

8.

This very process is an extremely interesting example of how an early instance of language change may be explained with the help of a research model from modem sociolinguistics, i.e. social network analysis (Milroy 1987). According to this model, Langland would have functioned as the bridge between two social networks, the one he belonged to in his place of origin and the one he moved into in London, along which this particular change would have travelled. The construction would have been picked up by his readers - very likely unconsciously, as it does not seem to have been sufficiently marked to draw any overt attention - of which there must have been quite a few, given the large number of manuscripts

226

Ingrid Tieken-Boon van Ostade

of the text. Langland may thus be characterised as a linguistic innovator, having only weak ties with the group of people he wrote for, and his readers, once they had picked up the new construction, as early adopters. Langland's readers, who presumably belonged to the clergy, would have been of sufficient social standing for their language to be the model of the general population of speakers. As society was generally very tightly organised in his day, it seems quite likely that the language of the clergy functioned as a norm enforcement mechanism. It seems to me that this possibility might veiy profitably be investigated further. 9.

Malory's identity has been much disputed (Tieken 1995: 110). While he is believed to have originated from Newbold Revel in Northamptonshire where the construction may have been in common use at the time, it is impossible to ascertain to what extent Malory's dialect was still that of his place of origin when he wrote the Morte Darthur or how much it had changed by his extensive contacts with his fellow members of the London court. If, indeed, the "Neg ... neither" construction might be taken to distinguish London English from that of the Midlands, we would be able to conclude that Malory's dialect was that of London.

10.

The possibility of the function of the "Neg ... neither" construction being that of a discourse marker was first suggested to me by Terttu Nevalainen.

11.

I have come across example (14) as often as four times in my search for instances in the OED, nor is this sentence the only one which was used as a quotation more than once (see also Fischer 1997:164η).

The origin and development of the "Neg ... neither" construction

227

References Aijmer, Karin - Bengt Altenberg (eds.) 1991 English corpus linguistics. London - New York: Longman. Algeo, John 1990 "The emperor's new clothes: The second edition of the Society's dictionaiy", Transactions of the Philological Society 88: 131-150. Alston, Robin C. (ed.) 1974 English linguistics 1500-1800. Menston: The Scolar Press. Austin, Frances 1984 "Double negatives and the eighteenth century", in: N.F. Blake Charles Jones (eds.), 138-148. Bailey, Richard W. (ed.) 1987 Dictionaries of English [1989] [Reprinted Cambridge: Cambridge University Press.] Blake, N.F. 1969 Caxton and his world. London: Andre Deutsch. Blake, N.F. - Charles Jones (eds.) 1984 English historical linguistics: Studies in development. Sheffield: CECTAL. Chalker, Sylvia - Edmund Weiner 1994 The Oxford dictionary of English grammar. Oxford: Clarendon Press. Duncan, Daniel 1731 A new English grammar. London. [1974] [Reprinted in facsimile by Robin C. Alston (comp.), English linguistics. London: Scolar Press. EL 17.] Fernandez, Francisco - Miguel Fuster - Juan Jose Calvo (eds.) 1994 English historical linguistics 1992. Papers from the 7th International Conference on English Historical Linguistics. Amsterdam: Benjamins. Fischer, Andreas 1997 "The Oxford English Dictionary on CD-ROM as a historical corpus: To wed and to marry revisited", in: Udo Fries - Viviane Müller - Peter Schneider (eds.), 161-172.

228

Ingrid Tieken-Boon van Ostade

Fisher, John H. - Malcolm Richardson - Jane L. Fisher 1984 An anthology of Chancery English Knoxville: The University of Tennessee Press. Fries, Udo - Viviane Müller - Peter Schneider (eds.) 1997 From JElfric to the New York Times. Studies in English corpus linguistics. Amsterdam - Atlanta, GA: Rodopi. Fuami, Kayoko 1991 "Negation in eighteenth-century English - Its use in fictional speech", in: Mac Kawai (ed.), 619-633. Gough, James 1754 A practical grammar of the English tongue. Dublin. [1974] [Reprinted in facsimile by Robin C. Alston (comp.), English linguistics. London: Scolar Press. EL 13.] Herford, C.H. - Percy Simpson (eds.) 1925-1952 Ben Jonson. Oxford: Clarendon Press. Hoad, T.F. 1986 The concise Oxford dictionary of English etymology. Oxford: Clarendon Press. Hoad, T.F. 1987 "Developing and using lexicographical resources in Old and Middle English", in: Richard W. Bailey (ed.), 49-65. Hopper, Paul J. - Elizabeth Closs Traugott 1993 Grammaticalization. Cambridge: Cambridge University Press. [1994] [Reprinted.] Jespersen, Otto 1917 "Negation in English and other languages", Det Kgl. Danske Videnskabernes Selskab. Historisk-filologiske Meddelelser 1: 1-151. Copenhagen. [1962] [Reprinted in Selected writings of Otto Jespersen. London: George Allen and Unwin.] 1940 A Modern English grammar on historical principles. Part V: Syntax. Vol. IV. Copenhagen: Ejnar Munksgaard. Johnson, Samuel 1755 A dictionary of the English language.

The origin and development of the "Neg ... neither" construction

[1968]

229

[Reprinted in facsimile. Hildesheim: Georg Olms Verlagbuchhandlung.] Jucker, Andreas H. 1994 "New dimensions in vocabulary studies: Review article of the Oxford English Dictionary (2nd edition) on CD-ROM", Literary and Linguistic Computing 9: 149-154. Kawai, Mac (ed.) 1991 Language and style in English literature. Essays in honour of Mac Masui. Hiroshima: The English Research Association of Hiroshima. Kytö, Meija - Matti Rissanen - Susan Wright (eds.) 1994 Corpora across the centuries. Amsterdam - Atlanta, GA: Rodopi. LALME = A Linguistic Atlas of Late Mediaeval English 1986 Angus Mcintosh - ML. Samuels - Michael Benskin. Vol. 2. Aberdeen: Aberdeen University Press. LDOCE = Longman Dictionary of Contemporary English. 1978 Harlow, Essex: Longman. [1981] [Reprinted.] Mazzon, Grabriella 1994 "OE and ME multiple negation: Some syntactic and stylistic remarks", in: Francisco Fernandez - Miguel Fuster - Juan Jose Calvo (eds.), 157-169. McMahon, April MS. 1994 Understanding language change. Cambridge: Cambridge University Press. [1995] [Reprinted.] Milroy, Lesley 1987 Language and social networks. (2nd edition.) Oxford: Basil Blackwell. [1989] [Reprinted.] Mizobata, Kiyokazu (ed.) 1990 A concordance to Caxton's own prose. Tokyo: Shohakusha. Nevalainen, Terttu - Helena Raumolin-Brunberg 1989 "A corpus of Early Modern Standard English in a socio-historical perspective", Neuphilologische Mitteilungen 90: 67-110.

230

Ingrid Tieken-Boon van Ostade

OED = The Oxford English Dictionary 1989 Prepared by J.A. Simpson - E.S.C. Weiner. (2nd edition.) 20 Volumes. Oxford: Clarendon Press. Osselton, N.E. 1993 Review of The Oxford English Dictionary (2nd ed.), International Journal of Lexicography 6: 124-131. Quirk, Randolph - Sidney Greenbaum - Geoffrey Leech - Jan Svartvik 1985 A comprehensive grammar of the English language. London New York: Longman. Rissanen, Matti 1994 "The Helsinki Corpus of English texts", in: Meija Kytö - Matti Rissanen - Susan Wright (eds.), 73-79. Schäfer, Jürgen 1980 Documentation in the O.ED. Oxford: Clarendon Press. Schmidt, A.V.C. (ed.) 1995 William Langland, Piers Plowman. A parallel-text edition of the A, B, C and Ζ version. Vol. 1. London - New York: Longman. Tieken-Boon van Ostade, Ingrid 1982 "Double negation and eighteenth-century English grammars", Neophilologus 66: 278-285. 1990a "Drydens versies van The Tempest en Troilus and Cressida: De bewerker als purist", in: Traditie en progressie. Handelingen van het 40ste Nederlands Filologencongres: 161-169 1990b "Betsy Sheridan and Fanny Burney in OED. Notes and Queries n.s. 37: 412-414. 1995 The two versions of Malory's Morte Darthur. Multiple negation and the editing of the text. Cambridge: D.S. Brewer. Ukaji, Masatomo 1979 "Multiple negation in Shakespeare". Studies in English Linguistics 7 (Japan): 100-117. Ussher, George Neville 1785 The elements of English grammar. Glocester. [1974] [Reprinted in facsimile by Robin C. Alston (comp.), English linguistics. London: Scolar Press. EL 27.]

The origin and development of the "Neg ... neither" construction

231

Vinaver, Eugene (ed.) 1977 Malory, Works. (2nd ed.) Oxford etc.: Oxford University Press. Willinsky, John 1994 Empire of words. The reign of the OED. Princeton, NJ: Princeton University Press. Zandvoort, Reinard Willem 1974 A handbook cf English grammar. (13th ed.) Groningen: Tjeenk Willink.

Affixal and non-affixal negation - a case of stable variation over time?1 Gunnel Tottie

It had been a rough day, so when I walked into the party I was very chalant, despite my efforts to appear gruntled and consolate ... I saw her standing alone in a corner ... [S]he was a descript person ... [H]er hair was kempt, her clothing shevelled, and she moved in a gainly way ... [S]he smiled in my direction ... I was plussed. It was concerting to see that she was communicado ... [B]eing corrigible, I felt capacitated ... Jack Winter, "How I met my wife", The New Yorker, Aug. 25, 1994, p. 82.

1. Introduction In my study of negation in spoken and written Present-Day British English (Tottie 1991) I was able to demonstrate the following facts concerning the variation between affixal and non-affixal negation of adjectives (as in e.g. impossiblelnot possible or untrue/not true): (a)

(b)

Although it has often been pointed out that the scalarity of some predicates can make variants with affixal and non-affixal negation semantically nonequivalent (a case in point is not happylunhappy, cf. I'm not happy but I'm not unhappy either), it is still a viable undertaking to make a quantitative study of affixal/non-affixal variation, for two reasons: scalar predicates are unusual, and moreover, the difference between scalar and non-scalar predicates is neutralised in premodifier function, as in the unhappy man. Variation between affixal and non-affixal negation is not free. A large number of constraints, syntactic as well as lexico-semantic, regulate the variation between the two types of negation, and the choice of either variant is de rigueur in the majority of cases.

234

Gunnel Tottie

(c)

Actual variability tends to exist only in a limited number of structural types, the most common being NP-V-Adj, as in The quality is uneven/not even. There is a tendency for spoken English to prefer non-affixal and written English affixal negation, which can probably be ascribed to production contraints inherent in the two varieties and the ensuing trend towards a fragmented style in speech and an integrated style in writing (cf. e.g. Chafe 1982).

(d)

I hypothesised that the facts holding for Present-Day English would, mutatis mutandis, also hold for older stages of the language, and that affixal/non-affixal negation would thus present a case of stable variation over time. I decided to test my hypothesis by comparing affixal and non-afFixal negation in an earlier variety of English and to base my comparison on an investigation of Late Middle English. I used a sample drawn from the Helsinki Corpus, viz. the subcorpus Middle English IV, comprising material from the period 1420-1500. I chose this period because I had reason to believe that it would be fairly uncomplicated with respect to two parameters, namely multiple negation, which was rapidly disappearing from the written language at this time (see e.g. Iyeiri, this volume; Tieken 1995), and the doperiphrasis, which had not yet become obligatory in negative and interrogative contexts (cf. e.g. Ellegard 1953). For obvious reasons, it was not possible to test the relation between spoken and written Middle English; however, it was possible to see differences in usage between various text types, and to base some observations on the likely relation of these text types to the spoken language of the time. In section 2 I will give a more detailed presentation of the material and method used for this study, and in section 3 I will give an overall quantitative comparison of the distribution and use of affixal and non-affixal negation of adjectives in Present-Day English and Late Middle English. Section 4 will be devoted to a survey of constraints on both types of negation in Present-Day English and Late Middle English, covering scope constraints in 4.1, lexico-semantic constraints in 4.2, stylistic constraints in 4.3, and quantitative aspects in 4.4. Section 5 concludes the paper with a discussion of my findings.

Affixal and non-affixal negation

235

2. Material and method From Part IV of the Helsinki Corpus, comprising almost 214,000 words, I extracted all sentences containing the Late Middle English counterparts of the lexical element NOT plus an adjective, all instances of NO plus an adjective, and all instances of adjectives with the prefixes UN-, IN-, and DIS-. I did not include the prefixes Aor NON-. which were not freely used in Middle English.2 I restricted myself to examples containing adjectives, as they constituted the largest category of words capable of taking affixal negation in the Late Middle English sample, just as in the Present-Day English sample. Obviously, this was also the only way of ensuring comparability between the samples - a difficult undertaking in any case, as we shall see. Just as in Tottie (1991), past and present participles were included among the adjectives if the constructions were not obviously passive. Moreover, as in the 1991 study, only true negatives were included, and reversative uses of the prefixes were excluded. (1) and (2) were thus disqualified as both passives and reversatives: (1)

(2)

Wherfor plese hit vnto your souerain lordship, that the seid Beseker may be discharged of the seid Office. (HC, M4, Documents: Petitions) Firste bi bodily werkis besili, vnto we be discharged of this hevy birthen of synne, pe which lettith vs fro goostely wirkynge. (HC, M4, Relig. treatises: Rolle, Prose treatises)

As spelling had not been normalised in Late Middle English, the corpus had to be searched for different spelling variants of the lexemes denoted by capitals above. Thus, NOT subsumes the spellings not, nott, nought, nojt, noghte, and not, with or without preceding ne (there were no instances of naught). NO subsumes the variants no, non, and noon? IN- subsumes in- and jn-, as well as the allomorphs imand ir- (no instances of il- occurred in the sample);4 UN- subsumes vn-, un-, and on-, and DIS- subsumes dis- and dys-. There was a single instance of mys- (your Chyldren many yeris herafter shal be myspreysed and blamed therfore) in the Fiction category, which I simply decided to leave out. It did not seem worthwhile to use sophisticated but probably still insufficient algorithms for searching for more complex spellings (cf. Barnbrook 1992), so I simply used a "Find" command in Word to scan the relevant part of the Helsinki Corpus for spellings found in the OED.

236

Gunnel Tottie

It is perhaps also worth commenting briefly on the inclusion of both main varieties of negation, namely the types called No-negation and Nof-negation in Tottie (1991) (as in I have no money and I don't have any money), in the category referred to here as non-affixal negation. For the purposes of contrasting affixal with non-affixal negation, No-negation and Nof-negation are equivalent - see the classification and the justifications of terminology in Tottie (1991, especially p. 8) and compare the examples in (3) and (4): (3)

a.

(4)

b. c. a. b. c.

... they had no clear intentions. (W.ll.4b. II) 5 ... they did not have any clear intentions. ... they had unclear intentions. * Their intentions were no clear Their intentions were not clear Their intentions were unclear

Notice that the (3a) construction with No-negation is only possible in premodified constructions, as shown by the unacceptable (4a). There were few instances of this type in the Late Middle English sample, only 18/348, or 5% of the total (cf. also Tottie 1991: 258ff, 308f, and 323, where it was shown that NPs with premodifying adjectives tend not to occur with No-negation in Present-Day English). Garden variety examples of affixal and non-affixal negation in Late Middle English are given in (5)—(15): (5) (6)

(7) (8) (9)

God is in him-silf vnchaungeable ... (HC, M4, Relig. treatises: Hilton) Now trewly my lorde pe kynge we had ben vn-hende ['unskilful'] and nevyr non of us Able for to be a knyght (HC, M4, Drama: Ludus Coventriae) thay hqfe othyre vertus unblendyde with pe fylthe of syne and unclene luste. (HC, M4, Relig. treatises: Rolle, The Bee and the Stork) How many ben peifite and how many inperfite? (HC, M4, Handbooks, other: Reynes) It is horde, ye it is impossyble, that a man may have alle joye in this worlde and also in heven ... (HC, M4, Sermons: In die innocencium)

Affixal and non-affixal negation (10)

237

And anone Balyne tolde hys brothir of hys adventure of the swerde and the deth of the Lady of the Laake, and how kynge Arthure was displeased with hym. (HC, M4, Romance: Malory)

Non-affixal negatives: (11)

(12)

(13) (14) (15)

Provided that this acte be not available to eny p~sone for eny entre sen the first day of this p~sent p~liament. (HC, M4, Law: Statutes) And thanne tho iij Mysdoers seiden wan they were so arestid that they were not gylty of the seid wrongfull takyng of the seid ship. (HC, M4, Documents: Petitions) Lord, I am not worthy to heryn pe spekyn & pus to comown wyth myn husbond. (HC, M4, Relig. treatises: Kempe) And yf you spede not I know no knyght 1yvynge that may encheve that adventure. (HC, M4, Romance: Malory) Be not aferd. (HC, M4, Drama: Wakefield)

In order to find the constraints operating on the choice of negative variant, the whole Late Middle English negative corpus was coded for the following factors, which had been shown to influence the selection of negative in Tottie (1991): text type (where Tottie 1991 had speech and writing), category of adjective (adjective proper or present/past participle), syntactic function of the adjective, whether the adjective was simple or co-ordinated, whether there was an adverb present or not (as in totally untrue Φηοί totally true), whether the adjective was in the positive, comparative or superlative form, and occurrence in litotes. As in Tottie (1991), I also tried to code each example according to whether there was a lexical gap or not, i.e. whether there existed a (non-)affixal counterpart of the form actually used, as in gladl*iatglad, intact/*tact, where either the non-affixal or affixal variant is mandatory for lack of a counterpart, or where the variant exists but has a different meaning, as in indifferent&iot different, not easy^uneasy. However, as will appear from section 4.3, this undertaking could not be carried out completely. Furthermore, all examples were checked for such semantic (non-) equivalence with their variants as might arise out of syntactic constructions.

238

Gunnel Tottie

A variable rule methodology cannot be used in the case of comparing affixal and non-aiFixal negation because of the many inherently knock-out constraints. However, for speed and ease of handling, the coding was done according to the system used for Goldvarb, the Macintosh version of variable rule analysis (cf. Rand - Sankoff 1990). For greater ease of comparability, I also recoded the Present-Day English corpus used for Tottie (1991) according to the same system.

3. Overall distribution of affixal and non-affixal negation in Late Middle English and Present-Day English My Late Middle English sample yielded 348 examples of negated sentences with adjectives, 172 with affixal negation, and 176 non-affixal negatives (132 with A b negation, 32 with TVo-negation, ten examples with neither or nor, and two with nothing as the negative adverb.) Table 1 shows a comparison with the Present-Day English spoken and written samples. Table 1. Affixal versus non-affixal negation in the Present-Day English and Late Middle English samples. Present-Day English samples 50,000 words each, HCME4 ca. 214,000 words (row percentages) Affixal

Non-affixal

Totals

Freq. per 1,000 words

PDE spoken

52 (32%)

112 (68%)

164

3.28

PDE written

126 (65%)

68 (35%)

194

3.88

LME (HC, ME 4)

172 (49%)

176 (51%)

348

1.63

Some surprising facts emerge from Table 1. First, the overall frequency of negation of adjectival expressions is much lower in Late Middle English than in Present-Day English - in Late Middle English we have only 1.63 negated expressions with adjectives per 1,000 words compared with 3.88 in written and 3.28 in spoken Present-Day English. The different frequencies in spoken and written Present-Day English were commented on in Tottie (1991: 48) and were related to the generally lower frequency of adjectives in speech than in writing as demonstrated by e.g.

Affixal and non-affixal negation

239

Chafe (1982) and Biber (1988, App. III). But how do we account for the much lower overall frequency in Late Middle English? A possible explanation is that Late Middle English would still be a more "oral" style than later varieties of English.6 This must of course remain a matter of speculation until we know more about the frequency of adjectives (as well as negation of adjectives in different text types) over time, but there is another striking fact which emerges from Table 1 and which supports the hypothesis: the proportion of affixal negation in Late Middle English. Although the Helsinki Corpus obviously consists of written material only, it shows a very different proportion of affixal vs. non-affixal negation than the Present-Day English written sample. The Present-Day English written sample had 65% affixal negation, compared with only 32% in the Present-Day English spoken sample, a highly significant difference. The Late Middle English written sample is half-way between the spoken and written Present-Day English samples with a proportion of 49% affixal negation and thus provides another factor which might support the hypothesis of a more oral style in Late Middle English written texts than in PresentDay English ones. It is of course possible to argue both that the Present-Day English sample was very small - totalling ca. 100,000 words - and could therefore be unrepresentative, and that the text types in the written material represent highly specialised academic prose. Nevertheless, the contrasts between the Late Middle English and Present-Day English samples with regard to frequency of negative sentences containing adjectives as well as affixal/non-affixal negation are striking and require further elucidation in order to be explained; however, this is beyond the scope of the present work. I also made an attempt to find out if there could be any differences with regard to the use of affixal and non-affixal negation between more literate and oral text types in the Late Middle English sample. Intuitively, the text samples that could qualify as "oral" text types in the Late Middle English sample were Romance, which contains a large proportion of text purporting to represent direct speech, Drama (mystery plays), and Private Letters, which together contain 89/348 or 26% of the negative sentences in the material. The distribution of affixal and non-affixal negation in the "literate" and "oral" strata thus defined is shown in Table 2. Interestingly, neither the Romance nor the Drama texts show much difference from the bulk of the material with regard to the proportions of affixal and nonaffixal negation - indeed, the plays have a higher-than-average proportion of affixal

240

Gunnel Tottie

negation, 55% compared with the 51% average. The situation changes radically when we get to the Private Letters: here we have only 12/38 or 32% affixal negation (and a concomitant 68% non-affixal negation), which is exactly the proportion of affixal negation recorded in the Present-Day English spoken sample. Table 2.

Affixal versus non-affixal negation in the Late Middle English sample: Literate versus oral text types (row percentages)

Affixal

Non-affixal

Totals

133 (51%)

126 (49%)

259

Romance

9 (50%)

9 (50%)

18

Drama

18 (55%)

15 (45%)

33

Priv. Letters

12 (32%)

26 (68%)

38

172

176

348

Literate

All

This difference is significant at p