Advances in English Historical Linguistics 9783110804072, 9783110161519

Citation preview

Advances in English Historical Linguistics

W DE G

Trends in Linguistics Studies and Monographs 112

Editor

Werner Winter

Mouton de Gruyter Berlin · New York

Advances in English Historical Linguistics (1996) edited by

Jacek Fisiak Marcin Krygier

Mouton de Gruyter Berlin · New York

1998

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 Advances in English historical linguistics (1996) / edited by Jacek Fisiak, Marcin Krygier. p. cm. - (Trends in linguistics. Studies and monographs ; 112) Includes bibliographical references and index. ISBN 3-11-016151-6 (cloth : alk. paper) 1. English language - Grammar, Historical. 2. English language - History. I. Fisiak, Jacek. II. Krygier, Marcin. III. Series. PE1075.A39 1998 425-dc21 98-27494 CIP

Die Deutsche Bibliothek — Cataloging-in-Publication-Data Advances in English historical linguistics (1996) / ed. by Jacek Fisiak ; Marcin Krygier. - Berlin ; New York : Mouton de Gruyter, 1998 (Trends in linguistics : Studies and monographs ; 112) ISBN 3-11-016151-6

© Copyright 1998 by Walter de Gruyter & 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, Berlin. Printed in Germany.

Preface

This book contains a broad selection of papers presented at the Ninth International Conference on English Historical Linguistics. The conference was held at Adam Mickiewicz University, Poznan, in August 1996. Its participants came from all continents, and it was especially encouraging to see scholars from countries which had not been represented previously. It may be safely said here that, judging by the sheer number of contributions worthy of publication, the 9th ICEHL was as successful as any of its predecessors. The organisers were able to offer a couple of plenary presentations every day, followed by parallel sections. Thus, the conference managed once again to convey the sense of growing interest in English historical linguistics in the scholarly community. Therefore, the task of selecting those papers that ultimately made it into the present volume turned out to be a daunting task indeed. We would like to thank those colleagues who helped us make the final decision concerning the selection. It might be put forward by a concerned reader that the distribution of papers in this volume is rather uneven. Thus, morphological and syntactical studies form by far the strongest contingent in the volume. They cover topics as diverse as word-formation, modality and negation, or clause structure in the history of the English language. A more theoretically-oriented

strain is represented by

contributions discussing issues such as grammaticalization or lexical diffusion in language change. Recent interest in historiography of historical linguistics finds its reflection in a sizable number of submissions, presenting various aspects of works by past grammarians such as Buchanan or Huish, while phonological studies, less numerous for a number of years now, are represented by a few papers only. Among the more fashionable, at least recently, approaches, sociolinguistic studies in the corpus linguistics framework and papers devoted to the development of Early

VI

Preface

Modern English seem to be of particular value. All the same, we believe that the results of our work accurately reflect the main areas of current scholarly interest in English historical linguistics. Poznan, March 1998

Jacek Fisiak Marcin Krygier

Contents

Double prepositions in English Gunnar Bergh

l

Motivations for producing and analyzing compounds in Wulfstan's sermons Don Chapman

15

The degrammaticalization of addressee-satisfaction conditionals in Early Modern English Guohua Chen

23

From unasecendlic to unspeakable: The role of domain structure in morphological change Christiane Dalton-Puffer

33

Anthony Huish: A 17th-century English grammarian Roberta Facchinetti

53

John Bullokar's "Termes of Art" Maurizio Gotti

63

The Dublin Vowel Shift and the historical perspective Raymond Hie key

79

On the ideological boundaries of Old English dialects Richard M. Hogg

107

The spread of -ly to present participles Kristin Killie

119

Inversion after single and multiple topics in Old English Wittern F. Koopman

135

Epenthesis and Mouillierung in the explanation of r-umlaut: The rise and fall of a theory Marcin Krygier 151

On minor declarative complementizers in the history of English: The case of but Maria Jose Lopez-Couso and Helen Mendez-Naya

161

Contents

Bare and to-infinitives in Old English: Callaway revisited BettelouLos

173

The interplay of external and internal factors in morphological restructuring: The case of you Angelika Lutz 189 The origins of long-short allomorphy in English DonkaMinkova and Robert P. Stockwell

211

Modals in past counterfactual conditional protases Rafal Molencki

241

Downsizing the preterite-presents in Middle English Stephen J. Nagle and Sara L. Sanders

253

Social mobility and the decline of multiple negation in Early Modem English Terttu Nevalainen 263 The grammaticalization in Medieval English MichikoOgura

293

Evolution theory and lexical diffusion Mieko Ogura and William S-Y. Wang

315

On nominative case assignment in Old English Masayuki Ohkado

345

Social factors and pronominal change in the seventeenth century: The Civil-War effect? Helena Raumolin-Brunberg 361 Towards an integrated view of the development of English: Notes on causal linking Matti Rissanen 389 Problems of functional structure in some relative clauses Aimo Seppänen

·.. 407

Eighteenth-century linguistics and authorship: The cases of Dyche, Priestley, and Buchanan Robin D. Smith 435

Contents

IX

Adverbialization and subject-modification in Old English

TorilSwan

443

Standardization of English spelling: The eighteenth-century printers' contribution Ingrid Tieken-Boon van Ostade

457

The functional relationship between rules (Old English voicing of fricatives and lengthening of vowels before homorganic clusters) Jerzy Welna 471 Index of subjects

485

Double prepositions in English Gunnar Bergh

1. Introduction The topic of the present paper is double prepositions in English.1 By this term I mean a syntactic structure which — although rather marginal in frequency — attracts a great deal of theoretical interest because it represents a combination of two well-known syntactic phenomena, pied-piping and preposition stranding. Structurally speaking, the construction may be described as containing a prepositional phrase (PP) which is fronted from its logical, usually postverbal2 position and where the moved preposition is repeated in the empty syntactic slot left behind.3 The following sentences illustrate this deployment principle:

(1)

a.

pus it is enpeyringe not oonly ofoone estate of be chirche, but ofallbre, of the whiche I spoke ofinpe bigynnynge. (Wycliffe, 14thc.)

b.

That fair for which hue (Shakespeare, 16thc.)

c.

/ had nobody to whom I could in confidence commit the secrecy of my circumstances to. (Defoe, 18th c.)

gron'd for and would die.

The main reason for using the term "double prepositions" here is, of course, that the target items express the same syntactic function in the sentence (heads of the same PP), as shown by the fact that we can delete one of the prepositions and still be left with a good sentence (or, in modern eyes, an even better sentence). Whereas in Present-day English such constructions are likely to be considered anomalous (Riley—Parker 1986), solecistical (Visser 1963: 407), or downright ungrammatical (Denison 1981: 213), they are in fact attested on a number of occasions in earlier periods of the language, as suggested by the above examples (cf. also Dubislav 1916; Jespersen 1927: 192-193; Ryden 1966: 43-44,139-140). Trying to formulate the problem which this paper will address, it is clear that while sufficient evidence is available for the existence of this otiose construction, few attempts seem to have been made to explain its raison d'etre. And yet when such attempts have been made, it is usually not a question of any detailed analysis

2

Gunnar Bergh

of its origin or syntactic properties but rather of cursory reflections on its use (cf. Sisam 1921: 242; Mustanoja 1960: 347-348; Visser 1963: 407). With this background in mind, we will now consider some of the central issues relating to the use of this construction, in particular as it is represented in w/z-structures (relatives and interrogatives), since this is the theoretically most interesting field. The questions which will guide us through the process can be formulated as follows: What is the relative distribution of the double preposition structure in texts from the different historical subperiods of English? What factors are at work in determining its usage? How can the existence of this construction be explained with reference to its distributional patterns and internal properties?

2. Data In order to be able to answer the first of these questions — what the distributional pattern looks like — one may receive a first hint of the answer by simply consulting some of the major works relating to the syntactic history of English. Thus, e.g., Fischer states in the Cambridge history of English that double prepositions "are quite frequent in Late Middle English" (1992: 390), a claim which is supported by Mustanoja in his Middle English syntax, where he notes that such pleonastic prepositions are "not uncommon in later ME" (1960: 347). Likewise, we may record that Visser (1963: 407), when discussing "prepositions used twice" in his Historical syntax of the English language, gives ten examples of the construction, of which no less than seven are from the period between 1370 and 1500. From these observations, then, one is prone to draw the conclusion that double prepositions are a phenomenon which was particularly favored in Late Middle English. However, since this eclectic method is in itself impressionistic, we cannot rely on it alone in a scientific context, but have to complement it by a systematic search of a collection of authentic texts, preferably one which represents the main subperiods of English in a balanced way. With the Helsinki Corpus as a natural choice here, a computerized investigation was carried out on the basis of its 1.4-million-word collection of Old English, Middle English, and Early Modern English text samples. As the investigation was confined to wA-items, it was possible to limit the search to three lexemes only, viz., the equivalents of the modern pronouns what, which and who — although, of course, realized in a multitude of different spellings. The target sequence which the computer was set to work with was a fronted PP containing any of these items, simply because all instances of double preposition constructions necessarily contain a pied-piped PP. These phrases were then checked to see if the pied-piped preposition happened to co-occur with a stranded preposition representing an identical syntactic function at the end of the same clause.

Double prepositions in English

3

The search produced the results given in Table 1: Table 1. Absolute distribution of wA-constructions with double prepositions in the Helsinki Corpus material. subperiod

number

OE1-3 ME1 ME2 ME3 ME4 EModEl EModE2 EModES

0 2 2 4 7 5 0 0

total

20

As we can see, these results tend to confirm each of the two claims made earlier about the frequency of the double preposition construction: first, the low total indicates that the construction is typically a marginal one in English; second, the figures lend support to the quoted indications in the literature that the construction was more common in the Late Middle English period than in any other period of the language.4 In addition, as suggested by the same data, there seems to be a fairly high proportion in the period immediately following, i.e., in the first subperiod of Early Modem English, where five cases were noted. And, admittedly, this outcome was not altogether unexpected either, since in a detailed study of relative constructions from the 16th century, Ryden (1966: 43-44, 139-140, and passim) makes reference to no less than 28 cases of double prepositions. Thus, it is of interest to note here that out of the total 20 cases identified in the Helsinki Corpus material, no less than 16 turned out to derive from texts representing the 15th and 16th centuries. This clustering suggests that at that time some kind of linguistic development or change was taking place which is related to the noted frequency increase. The main question here, then, is obvious: what did this development consist of?

3. Discussion In the following, three possible explanations will be evaluated, each of which relates to the noted frequency pattern of double prepositions. First, the phenomenon will be discussed in terms of stylistics, specifically the availability of texts from the pertinent centuries and the later rise of prescriptive grammar. Then, a grammatical perspective will be brought in where consideration is given to the possibility of the target development being connected to the rise of preposition stranding with w/z-elements which occurred in Late Middle English. Finally, the

4

Gunnar Bergh

grammatical examination will be continued by a discussion of the idea of different verb units being influential here, and whether the data can be explained with reference to syntactic reanalysis.

3.1.

Informal texts and prescriptivism

With regard to the first of these explanations, in what sense could the availability of texts be relevant here? To begin with, it seems clear that constructions involving preposition stranding generally tend to belong to the informal use of language (cf., e.g., the discussion in Visser 1963: 400-415), and if such a construction is also tautological due to the inclusion of a redundant element — as in double preposition structures — most people are likely to say that it makes the case for informality even stronger. Could it be, then, that the demonstrated increase in frequency is conditioned by the fact that extant informal texts, e.g., private letters and speech-based writings of different kinds, go back only to Late Middle English, and that the noted distribution, therefore, is simply an artifact produced by various historical accidents? To find out, it might suffice to check the material from the Helsinki Corpus again, and specifically the text type classification of the 20 samples containing double prepositions: Table 2. Absolute distribution of text types with double-preposition w/z-constructions in the Helsinki Corpus material. text type

number

fiction law document science, medicine biography handbook, medicine Old Testament philosophy private correspondence religious treatise romance rule

4 4 2 2 1 1 1 1 1 1 1 1

total

20

The natural conclusion, as suggested by these data, would be to claim that double prepositions occurred in all types of text, since obviously a variety of both informal and formal texts are represented here. And this, in its turn, would mean that we

Double prepositions in English

5

cannot subscribe to the idea that the increased incidence of double prepositions would be due primarily to changes in the distribution of different text types, i.e., mainly formal versus informal ones. Still, this argument would only pertain to the rise of double prepositions. What about their decline? In this perspective, the notion of prescriptive grammar might be relevant, with its blanketing effect on all sorts of constructions which appeared to run counter to logic and/or Latin. As is commonly known, prescriptivism started to pluck up courage in the 17th century, not least through the work of Dryden and later Gibbon. An early contribution by the former goes back to the 1680s, as reported by Visser in the following way: Dryden, who openly states: "I am often put to a stand in considering whether what I write be the idiom of the tongue... & have no other way to clear my doubts but by translating my English into Latin", and consequently finds that end-position of the preposition is a non-Latin idiom and therefore inadmissible in English, alters all the sentences with end-position occurring in the 1684 edition of his Essays on Dramatic Poesy published sixteen years before. (Visser 1963: 402) It seems safe to assume that measures of this type must have had a hampering effect not only on preposition stranding proper but also on related pleonastic structures such as double-preposition constructions. The only problem for our purposes is that there is a certain time gap between the noted decline of the target construction and the rise of prescriptivism: the former seems to have occurred around 1600, while the latter did not gain serious weight until after 1700. Accordingly, we conclude that the combination of text availability and prescriptive grammar can not be regarded as a full explanation in this context.

3.2.

Transitional v/h-effect

Let us now turn to the second of our possible explanations — that the increased use of double prepositions represents a transitional stage in the syntactic development of prepositional w/z-phrases. One indication along these lines is provided by Fischer (1992: 390), who argues that sentences with double prepositions "show the development of preposition stranding in ννΛ-structures", thus implying that the two processes are somehow related. To be able to evaluate this implicit hypothesis, we first need to know the basic facts of preposition stranding in w/j-constructions. While the stranding pattern itself was fairly well established in connection with the elements pe (obligatory) and pat (optional) already in Old English (cf. Dekeyser 1990; van den Eynden 1994: 220221), there were apparently no cases of it in w/j-environments at that time, i.e., pied-piping was obligatory with interrogatives, the only wh-type found. Instead, the first sporadic instances seem to have emerged in the early 13th century, after

6

Gunnar Bergh

the extension of w/z-elements into the relative domain; Fischer (1992: 390), e.g., cites the following two examples, the first interrogative and the second relative: (2)

a. nuste nan kempe, whaem he sculde slxn on, (Layamon's Brut, 1205) b. And getenisse men ben in ebron, Quilc men mai get wundren on. (Genesis and Exodus, c. 1250)

It was not until the decades before 1400, however, that the frequency of stranding increased noticeably, and, as noted by Dekeyser (1990), it was not until well into Early Modern English that it became a competitive alternative to pied-piping. Incidentally, by saying that we can also refute the idea suggested by some scholars — in particular Grimshaw (1975: 37) — that pied-piping was obligatory in Chaucerian English. The following examples from The Canterbury tales (late 1380s) should provide sufficient counter-evidence (cf. also Bengtsson 1996): (3)

a.

But to kyng Alia, which I spake of yoore (The Man of Law's tale)

b.

His lady, certes, and his wyf also, the which that law of love acordeth to (The Squire 's tale)

c.

What sholde I teilen ech proporcious of things which that we were hen upon (The Canon 's Yeoman 's tale)

d.

Yet hadde I levere payen for the mare, Which that he rit on (The Manciple 's prologue)

With regard to double prepositions, then, could it be the case simply that the use of these items increased during the transitional stage between obligatory and optional pied-piping with w/z-items? One would be inclined to answer in the affirmative here — although there are other important circumstances which we will bring up in the next section. The main argument for believing so is the almost perfect timewise match between the two processes involved — i.e., both the noted higher frequency of double prepositions and the change from obligatory to optional pied-piping seem to have occurred in the 15th and 16th centuries. Another type of support for this theory comes from odd cases like the following, where one and the same construction appears in two different forms in two different periods: (4)

a.

Wherfore the Mair and aldermen comandeth on the kyngges half and on hire owene half also that no man of what

Double prepositions in English

b.

7

condicioun or degre that he be, priue ne straunge (Proclamations - ME3) The Kyng oure Sov~eigne Lorde I...I hath enacted ordeyned and stablysshed, that ev~y p~sone of what condicion or degree he be of, beyng or herafter be in oure seid Sov~ayn lord the Kyng- wagis (Statutes - ME4)

The first example, which is from ME3, represents an ordinary pied-piping construction, and stands in contrast to the second example, which is from ME4, and which exhibits both pied-piping and stranding, i.e., a double-preposition construction. To this we may add a third type of supportive data which has to do with the extensive language contact — as represented by various Anglo-Norman activities, translation work, etc. — which took place during the Middle English period. Sisam (1921: 242), citing four cases of otiose prepositions in Mandeville's Travels — e.g., fro whom all godenesse and grace cometh fro — provides the following explanation of the phenomenon as it appears in relative clauses (cf. also Denison 1981:215): The pleonasm is explained by the divergence of French and ME. word order. In French, as in modern literary English, the preposition is placed at the beginning of the clause, before the relative (de qui, dont, etc.). ME. writers naturally use the relative that, and postpone the preposition to the end of the clause: e.g., pat all godenesse cometh fro. The translator compromises between his French original and his native habit by placing the preposition both at the beginning and at the end. The French influence would here work as a temporary reinforcement of the previous obligatoriness of pied-piping with w/i-elements at a time when this constraint was being relaxed, thus potentially increasing the difficulty experienced by contemporary speakers/writers in handling these phrases. While it is quite probable, then, that significant influence should be attributed to the rise of preposition stranding in vf/i-structures, it still cannot be seen as the full solution to the present problem. The reason for this is, however, not the fact that there are earlier w/z-cases with double prepositions, e.g., the ones instanced in the corpus investigation: (5)

a.

b.

be muche wlite habbe; nim him of hwas wlite bead awundret of; be sunne & te mone. up-o hwas nebscheft; (Hali Meidhad, c. 1200) sei me hwer bu wunest meast. of hwet cun bu art ikumen of & ti cunde cud me. & burh hwas heaste heane je hali men. (Margarete,c. 1200)

8

Gunnar Bergh c.

'Dame', a seide, 'pat sit aboue, For pat ilche lordes hue, On wham pin herte is on iset: Jewe me to day a meles met!' (The Romance of Sir Beues ofHamtoun, c. 1300) These instances can be accounted for simply by making reference to the parallel odd cases of preposition stranding noted in the 13th century, implying that the two types of construction tend to go hand in hand in their early development. What is more problematic, however, is the fact that there are other structures containing double prepositions which do not involve w/z-elements and which, therefore, cannot be straightforwardly connected to the above explanation. Consider the following examples, which all involve an element of topicalization:5 (6)

a. b. c. d. e.

inne on pssm fsestenne saetonfeawa cirlisce men on (Chronicle A, 884) swa pat on psere rode... sticodon on maenige arewan (Chronicle E, 1083) Of love were liking of to here (Ipomedon, c. 1185) In uncuth land to won ai in (Cursor Mundi, c. 1300) Of many aunters I here of teile (Cursor Mundi, c. 1300)

Constructions of this kind make it necessary, arguably, to turn to our third and last possible explanation, which has to do with the notions of complex verbs and syntactic reanalysis.

3.3.

Verb units and syntactic reanalysis

The significance of complex verbs has been discussed by several scholars. Jespersen (1927: 184-190), e.g., trying to explain the variant use of pied-piping and stranding, notes that there are some prepositions which are more naturally placed at the end of a clause, because they are felt to be less intimately connected with the fronted item than with the verb, e.g., in complex verbs like long for, delight in, wonder at. In contrast, there are other prepositions which exhibit the opposite pattern, i.e., they are more intimately connected with the fronted item than with the verb, e.g., beyond which, during which, except what. These facts would suggest that a preposition is often torn between the attraction of the main verb governing it and the alliance duties it has towards its own complement — thus acting as an important factor in deciding different sentence patterns. Visser (1963: 407) has capitalized on this idea in the context of double prepositions. Specifically, he suggests that these constructions may be conditioned by a verb and a preposition forming what he calls a semasiological unit — i.e., a semantic unit — so that in spite of the fact that the preposition has already been

Double prepositions in English

9

placed before a fronted item, it is resumed and tacked on to the subsequent verb as well. Let us analyze one of his examples: (7)

William Jeney... andyonge Thomas Heigham, /to which personys] [I] [haue spoken to] (Bury Wills, c. 1370)

The preposition to in this sentence would be first fronted together with its complement which personys, and then resumed by the phrase-internal attraction of the verb speak, thereby keeping intact each of the three main constituents of the subclause, as indicated by the added brackets. What are the grounds for subscribing to this theory, then? Well, a crucial factor here seems to be the general change of word order which occurred in the transition from Old English to Middle English, making English an SVO language rather than an SOV language (e.g., Traugott 1992: 273-275; Fischer 1992: 370-372). This is a change which is often associated with the development of preposition stranding in constructions involving NP movement (e.g., prepositional passives, cf. Fischer— van der Leek 1981: 327-329; van Kemenade 1987: 212-213), but arguably the same connection can be established with regard to the movement of ννΛ-items, thus providing a common platform for the notion of syntactic reanalysis (e.g., Koma 1981). Although the issue is likely to be somewhat controversial, not least due to the technicalities of generative (transformational) grammar (cf. Inada 1981; Denison 1993: 144-153), the parallel treatment of the two types of structure is desirable in the sense that it yields a more generalized account of the foundation of stranding. To illustrate the principal idea of this theory, consider the following VP configurations: (8)

a.

[VP[PPPNP]V]

b.

[vpV[ppPNP]]

When, through this change, the (a)-pattem was replaced by the (b)-pattern, V and P became adjacent, which is a necessary condition for syntactic reanalysis to apply. Hereby it is theoretically possible to interpret the structure not only as a simple verb with a PP complement, as in (9a), but also as a complex verb with a regular object, as in (9b), a framework which has turned out to be of great significance in the field of generative grammar (cf, van Riemsdijk 1978: 218226): (9)

a.

[vp [v laughed] [pp at him/whom] ]

b·

[VP tv laughed at} [NP him/whom} ]

10

Gunnar Bergh

Apparently, then, the concept of syntactic reanalysis can be regarded as the basis not only for the general extension of preposition stranding but also for the increasing force of complex verbs in the conditioning of different sentence patterns. While there is much to commend a theory like this, the main question for our purposes would still be how to integrate it into our previous attempt to explain double preposition constructions through the introduction of preposition stranding in w/z-environments. The following is my suggestion. The use of double prepositions seems to be triggered by a compromise between two linguistic forces, one syntactic and one semantic. The pied-piped preposition would be the result of a speaker's desire to keep the w/7-phrase together, something which was more important in those days due to the strong influence of Latin and French, which did not recognize a stranded alternative. The stranded preposition, in its turn, would be a result of a speaker's trying to keep together what was felt to be a complex verb phrase, an interpretation which was facilitated by the change of word order at the beginning of Middle English, and which subsequently relaxed the conditions for end-placed prepositions. Between the two forces at issue, the syntactic one would be the stronger, and as long as it was in full operation the opposing semantic force was kept at bay. However, when the rule of obligatory pied-piping with w/z-items was gradually made redundant, the syntax had to yield partly to the forces working towards semantic unity which were exercised by verbs with prepositional complements, thus paving the way for the double preposition construction. In topicalization structures, in contrast, where no impeding wA-elements occur, these semantic principles would have been able to apply with greater latitude. Incidentally, this also suggests part of the explanation as to why Old English admitted double prepositions in topicalization structures, as indicated in (6) above, but not in wA-interrogatives.

4. Summary To sum up this investigation, then, we may return to the three key questions posed at the beginning of this paper. First, with regard to the relative distribution of double preposition constructions in English, it was shown that they are clearly overrepresented in 15th and 16th century texts, both as indicated subjectively through existing literature on the topic, and as measured more objectively through the data of the Helsinki Corpus. Second, concerning the main factors governing the frequency of the construction, it was suggested that both syntactic and semantic factors are operative in this field, and that stylistic measures in the form of prescriptive grammar are likely to have played a certain part too (but not the availability of informal texts as such). Third, it was argued that the main catalyst of the increased use of double prepositions was the change from obligatory to

Double prepositions in English

11

optional pied-piping in wA-structures and the increased attraction exercised by verbs on prepositions in complex verb constructions, both of which are thought to be related to the underlying general change of word order and, hence, the possibilities of syntactic reanalysis.

Notes 1. Thanks are due to the following colleagues for helpful comments and suggestions on an earlier version of this paper: David Denison, Jennifer Herriman, Masayuki Ohkado, Susan Pintzuk, Aimo Seppänen, Ingrid Tieken-Boon van Ostade, Anthony Warner, Ilse Wischer. 2. There are two main exceptions to this pattern: first, the PP may occur as the subject of a clause in some contexts; second, and more importantly, the PP may be placed before the finite verb when it functions as the object of a clause with SOV word order, e.g., in Old English. 3. When saying that the moved preposition is "repeated" in the slot left behind, it is necessary to add that it is not always the case that the fronted preposition is simply copied there, as suggested by Grimshaw (1975: 41-42). The slot can in fact also be filled by another preposition with the same syntactic function, as shown by the next set of examples (where the deletion test applies equally well): (i)

a.

of him to whom he had most his trust on (Caxton, 1 5th c.)

b.

I fear me, and, I divine, much of doctor Nicolas; a man with whom my fantasy never wrought withal (Latimer, 16th

c.

an occurrence for which they have been... in patient expectation of (Goldsmith, 18thc.)

These examples suggest that we are faced with two subtypes of the double preposition construction — one where there is identity between the two prepositions, which may be referred to as the matched case, and another where there is no identity between these two elements, which may be referred to as the mismatched case (cf. Riley— Parker 1986). While there are several interesting ramifications of this distinction, limitations of space make it necessary to leave them out here. Instead, they will be brought up in a separate article to be published later. 4. Admittedly, the noted frequency pattern is complicated slightly by the fact that the use of w/z-items in general was also on the increase in Middle English. Although this frequency rise was not parallel in time to that of double

12

Gunnar Bergh

prepositions, it might still have influenced the reported figures for that period to some extent. 5. As these examples clearly show, topicalization constructions with double prepositions can be found even in Old English, although some of them are likely to be "imitations of a Latin exemplar containing a compound verb with a spatial prefix" (Denison 1981: 215).

References Adamson, Sylvia—Vivien Law—Nigel Vincent—Susan Wright (eds.) 1990 Papers from the 5th International Conference on English Historical Linguistics. Amsterdam: Benjamins. Bengtsson, Per 1996 Prepositional relatives in English: A diachronic case study of three authors. [Unpublished MS.] Blake, Norman F. (ed.) 1992 The Cambridge history of the English language. 2: 1066-1476. Cambridge: CUP. Dekeyser, Xavier 1990 "Preposition stranding and relative complementiser deletion: Implicational tendencies in English and other Germanic languages", in: Sylvia Adamson et al. (eds.), 87-109. Denison, David 1981 Aspects of the history of English group-verbs, with particular attention to the syntax of the Ormulum. [Unpublished D.Phil, dissertation, University of Oxford]. 1993 English historical syntax: Verbal constructions. London: Longman. Dubislav, Georg 1916 "Studien zur mittelenglischen Syntax I, II", Anglia 40: 263-321. van den Eynden, Nadine 1994 Syntactic variation and unconscious linguistic change. Frankfurt am Main: Lang. Fischer, Olga C. M. 1992 "Syntax", in: Norman F. Blake (ed.), 207-408. Fischer, Olga C. M.—Frederike van der Leek 1981 "Optional vs. radical re-analysis: Mechanisms of syntactic change", Lingua 55: 301-350. Grimshaw, Jane 1975 "Evidence for relativization by deletion in Chaucerian Middle English", in: Ellen M. Kaisse—Jorge Hankamer (eds.), 35-43.

Double prepositions in English

13

Hogg, Richard M. (ed.) 1992 The Cambridge history of the English language. 1: The beginnings to 1066. Cambridge: CUP. Inada, Toshiaki 1981 "Problems of reanalyses and preposition stranding", Studies in English Linguistics (Tokyo) 9: 120-131. Jespersen, Otto 1927 A Modern English grammar on historical principles. Vol. 3. Copenhagen: Munksgaard. Kaisse, Ellen M.—Jorge Hankamer (eds.) 1975 Proceedings of the Fifth Annual Meeting of the North Eastern Linguistic Society. Cambridge, Mass.: Harvard University Press, van Kemenade, Ans 1987 Syntactic case and morphological case in the history of English. Dordrecht: ICG Publishing. Koma, Osamu 1981 "Word order change and preposition stranding in ME", Studies in English Linguistics (Tokyo) 9: 132-144. Kytö, Merja (comp.) 1996 Manual to the diachronic part of the Helsinki Corpus of English Texts: Coding conventions and lists of source texts. (3rd edition.) Helsinki: Department of English, University of Helsinki. Mustanoja, Tauno 1960 A Middle English syntax. Vol. 1. Helsinki: Societe Neophilologique. van Riemsdijk, Henk 1978 A case study in syntactic markedness: The binding nature of prepositional phrases. Lisse: de Ridder. Riley, Kathryn—Frank Parker 1986 "Anomalous prepositions in relative clauses", American Speech 61:291-306. Ryden, Mats 1966 Relative constructions in early sixteenth century English with special reference to sir Thomas Elyot. Uppsala: Acta Universitatis Uppsaliensis. Sisam, Kenneth 1921 Fourteenth century verse and prose. Oxford: Clarendon Press. Traugott, Elizabeth Closs 1992 "Syntax", in: Richard M. Hogg (ed.), 168-289. Visser, Frederikus Th. 1963 An historical syntax of the English language. Vol. 1. Leiden: Brill.

Motivations for producing and analyzing compounds in Wulfstan's sermons Don Chapman

Motivation of compounds, i.e., whether the composite structure of compounds is at least discerned and perhaps even relied upon for interpretation, is essentially a pragmatic issue; the question is not whether a given compound can be analyzed — most can — but whether it is analyzed by a speaker/hearer when it is used. This pragmatic concern can easily be mixed up with synchronic, theoretical concerns of word-formation, i.e., how compounds are originally coined, because both questions examine the patterns that allow a compound to be interpreted or coined. Like syntax, compounding and affixation are systematic enough to invite a search for quasi-generative word-formation rules. But unlike syntax, the output of word-formation is words, not phrases or sentences — words that can enter the lexicon and be retrieved through lexical, not generative processes, words that need never be produced by rules again, words that linger in the lexicon picking up nuances, connotations, and even denotations that could not have been predicted by the word-formation rules. So while every utterance, no matter how many times it is repeated, may be regarded as the output of generative syntactic rules, words, even compounds and affixed words, arise from two separate processes: retrieval (by far more common) and coinage (less common). In this rough model of word-formation and word use, a word may originally be coined by recourse to word-formation rules, but gradually, as it is increasingly used, enters the lexicon, i.e., become lexicalized (cf. Lipka 1981: 120-122). E.g., the compound blackboard has been lexicalized, since black is no longer meaningful — few blackboards, in fact, are black. In any given utterance, then, the individual words will be at different stages of lexicalization, from the recently coined nonce compounds to the fully lexicalized and idiomatized words, like blackboard and cupboard, and even to the obscured compounds like lord < OE hlaf+ OE weard. Simply analyzing compounds by the word-formation rules that originally produced them will likely impose different diachronic states of words on their synchronic use. So will treating all the words as if lexicalized. Since both generative and lexical processes can presumably account for the uses of words, what is needed is some way to gauge the relative lexicalization of a word. This paper is an initial foray into analyzing the lexical status of compounds in the Old English sermons of Wulfstan, Archbishop of York from 1002-1023. Assessing the synchronic state of words in historical states of the language poses considerable challenges. The tools for analyzing historical states of the language are not plentiful or precise, since no native informants are available. But occasionally local clues within a text reassert a compound's original motivation

16

Don Chapman

and suggest that whatever degree of lexicalization the compound had achieved, within its passage it is highly likely interpreted as a compound, not retrieved from the lexicon. Just such clues are found in one of Wulfstan's most prominent stylistic devices, the pairing of compounds with other compounds or simplexes that repeat one of the constituents of the compound, as in the pairs manswican ne mansworan 'evil-deceivers nor evil-swearers' (WHom 8c 162) and wediogan ne wordlogan Oath-liars nor word-liars' (WHom 8c 162-63).! Some 200 such pairs occur throughout Wulfstan's sermons, in varying connections, from the tightly joined pairs like those just cited to more loosely connected pairs like (1)

ac fengon to wuroienne ast nyhstan misllice entas & strece v/oruldmen be mihtige wurdan on v/oruldafelum (WHom 12 36-38) 'But at last they received as objects to worship various giants and severe world-men who were mighty in world-power'

(Cf. Fill (1992) for a further description of this device in later English writing). In general such pairs draw attention to the compound structure of the words simply by the proximity of the repeated constituent. On occasion the pairs more explicitly highlight the compound structure, especially when interpretation of the compound is essential for the rhetorical effect of the larger passage. One such passage is (2)

Ne beon hi xfre... ne aewbrecan, ac healdan heora rihtaswe 'Let them not ever be... adulterers ('marriage-breakers'), but hold their lawful marriage-vow (wife?)' (WHom lOa 11-12)

The pair xwbrecan 'adulterer' and rihtxwe 'lawful wife' highlight the shared element x\v, a resonant word originally meaning law and later specializing to marriage-vow or even wife. These later specialized senses are probably primary within both compounds, but the contrast between brecan 'breaking' and healdan 'keeping' reasserts the earlier, more general meanings, emphasizing that an adulterer (aswebrecd) is literally one who breaks the law and that holding one's wife or keeping a marriage-vow (rihtasw) is a specialized form of more general oath-keeping. The tension between the general and specific meanings in nhtaew emerges even stronger, since the injunction is directed toward monks who would presumably have no wife to hold or marriage to keep. Apparently the general sense is intended (the monks should not be vow-breakers, but should keep their strict vows), except that the clause "that is their monastery" is added as if to explain what the lawful aswe 'wife'? 'marriage'? of a monk could be. But even when the interpretation of the compound is not crucial to the larger meaning, the artificiality of such closely linked compounds cannot help but draw

Motivations for producing and analyzing compounds in Wulfstan 's sermons

17

attention to the compounds and their composite structure, as the following pairs illustrate: (3a) (3b)

Se be \vxre weamod, weorde se gepyldmod 'He who may have been "gloom-minded", let him be patience-minded' Se pe waere hohmod, weoroe se glaedmod (WHom 1 Oc 122-38) 'He who may have been "trouble-minded", let him be glad-minded'

As my translations are meant to show, the lexicalized senses of the words ending with -mod are superseded by the reasserted structure of the compound. Even if weamod would normally be felt to mean nothing more than 'gloomy', its proximity to other words ending with -mod highlight the second constituent and the composite structure. The discussion so far has focused on gauging the composite status of a compound by the likelihood or even necessity of interpreting the compound within a passage. The composite status of a compound would also be asserted in recently coined words, so another method of gauging the relative lexicalization of a word is to assess the likelihood that a given compound was produced or coined for a given situation. The need for new words is often associated with the creative minds of society — e.g., the scientists, who need new words for their inventions and discoveries, or the poets, who seek new words to express their deeper consciousness (cf. Leech 1969: 43-44). Of course, less-creative minds need new words, too, to fill in gaps of performance or individual lexicon (cf. Brekle 1978: 74-75; Downing 1977), but Wulfstan's manipulation of language for rhetorical effect puts him in the same company as the poets. Perhaps he did not need new words to express a new consciousness, but as an orator he would have wanted to escape the ordinary use of words, to confer emphasis by ornamenting the language, to draw attention to the language itself. As Frye notes, the orator and poet are both concerned with figured language (1990: 17). Wulfstan's paired compounds as a figurative device can easily be seen as occasioning the generation of compounds, where generation does not necessarily mean coining — creating a word for the first time in the language — but rather producing a word through generative processes. Throughout his prose Wulfstan favors rhyming and alliterating pairs of words, both Simplexes, like stalu and cwalu 'stealing and killing' (WHom 20.1 52), and compounds, like those already discussed. The value of compounds for creating such rhyming or alliterating pairs is apparent. Once, e.g., the compound \ved-bricas is chosen, one may achieve an echoic pair with a matching compound either beginning with wed- (e.g., wed-logan) or ending with -bricas (e.g., ad-bricas). The ease of forming pairs that compounds provide suggests that Wulfstan could have generated compounds for the express purpose of joining them in pairs. And the structure of the echoic pairs may well have further prompted the generation of compounds. Many of Wulfstan's echoic pairs are tightly joined by parallel syntax, whether a coordinating conjunction (wedlogan ne wordlogan) or

18

Don Chapman

identical placement in parallel syntactic structures (Se pe wxre weomod, weorde se gepyldmod). And most of these pairs occur within extensive lists of similar pairs, like the following: (4a) Se pe wxre weomod, weoroe se gepyldmvd 'He who may have been "gloom-minded", let him be "patience-minded"' (4b) Se pe wxre hohmod, weoroe se glxdmod 'He who may have been "trouble-minded", let him be "glad-minded"' (4c) Se pe wxre idelgeom, weoroe se no/georn 'He who may have been "idle-eager", let him be "useful-eager"' (4d) Se pe wxre /o/georn for idelan wcorftscype, weoroe se carfull hu he swypast mxge gecweman his Drihtne 'He who may have been "praise-eager" for idle honor let him be attentive to how he may most please his Lord' (4e) Se pe wxre ofermod, weoroe se eadmod (WHom lOc 125-30) 'He who may have been proud ("excessive-minded"), let him be humble ("kind-minded")' The tight parallel structure and the incessant repetition act as a formulaic template; Se pe wxre and weoroe se constitute the constant portion of the formula, and the compound pairs that change from pair to pair fill in the slots, i.e., the variable portion. Compounding offers an obvious expedient for filling these variable slots, since it can readily produce pairs of words guaranteed to fit together. The repetition of mod and georn in the example above ensures that the compounds can be paired in parallel, in this instance as antonyms. In effect compounding makes the fixed part of the template larger and the variable part smaller. Thus, the template could be regarded roughly as Se pe wxre mod, weoroe se mod. In templates like this the constituents of compounds could presumably be interchanged with each other to fill in the variable parts of the formula. Indeed Wulfstan mixes and matches the constituents of the compounds, interchanging them in several permutations of rhyming pairs. Just such mixing and matching is seen in the system based on man- as a first constituent, whether with a short vowel meaning 'man' or a long vowel meaning 'evil'. Capitalizing on the obvious pun, Wulfstan blurs the distinction between the long and short vowels, using man- in pairs for which both senses can be appropriate. This is the most complex and productive formulaic system in Wulfstan's writings; all possible permutations of manslagum, manswican, and mansworan occur: (5a)

oyder sculan marmslagan, & older sculan manswican (WHom 7 12829; cf. HomU 41 27424-275'; HomU 34 203.21) 'There must go man-slayers and there must go evil-deceivers'

Motivations for producing and analyzing compounds in Wulfstan 's sermons

19

(5b)

dyder sculan manslagan, & older sculan mansworan (WHom 13 92-93; cf. LawVIAtr 256, sec. 36; HomS 16 148-49.124-25) 'There must go the man-slayers and there must go the evil-swearers'

(5c)

ne beon manswican ne mansworan (WHom 8c 161; cf. WHom lOa 1112) 'Let them not be evil-deceivers nor evil-swearers'

Such a formulaic device not only puts compounds sharing the same constituent close enough together to elicit the recognition of the composite structure of the compounds, but may well also have invited the generation of compounds. So even if Wulfstan did not necessarily coin the compounds in a given pair, he could well have produced them from word-formation, not lexical, processes. And this device may well have invited the coinage, not merely generation, of compounds as well. Several hapax legomena occur among Wulfstan's echoic pairs, such as both terms in freolsbrycas & fsestenbrycas (WHom 20.1 98) 'feast-breakings and fast-breakings'. While a hapax legomenon is not necessarily a coinage, especially given the paucity of surviving texts, many of the hapax legomena occurring in Wulfstan's tightly joined echoic pairs are likely his own creations, such as xlmesriht, aobricas, wedbrycas, manswican, and weddlogan. Some twenty-seven hapax legomena are found in such tightly joined pairs; such a relatively high number suggests the generative capacity of such formulaic systems. The generative capacity of Wulfstan's echoing compound pairs is further seen in the numerous hapax legomena that are synthetic compounds, i.e., compounds formed from a deverbal head and a modifier acting as an argument of the transposed verb, like aöbryca meaning One who breaks (bryca) an oath (a&)\ Marchand designates such compounds "verbal nexus compounds", because they encapsulate the predicate of a sentence (1969: 18). The syntactic relationship between the constituents is explicitly posited in the deverbal constituent for such compounds, in contrast to non-synthetic (or primary) compounds in which the relationship is left unexpressed. In, e.g., the primary compound praslriht, the predicate relationship between constituents is not expressed and must be interpreted as "the right belongs to the slave", or something similar. The predicate relationship in aobryca, on the other hand, is expressly "[one] breaks an oath". Thus, synthetic compounds are among the most transparent, and Marchand claims that such transparency makes synthetic compounds extremely productive (1969: 18). In Present-day English, almost all verbs can be made into nominals denoting either the agent or the action of the verb as in combinations like apple grower and interior decorator. Presumably the same converting processes were also readily available in Old English, and synthetic compounds would have been easy to form and interpret. In fact, Lieber appeals to the productivity of such compounds in Old English to explain why arguments of deverbal constituents still precede rather than follow the deverbal head in Present-day English compounds: "What is remarkable in the history of English is that the synthetic compounding pattern was so

20

Don Chapman

productive that it did not change after the parameter settings for English changed" (about 1200 A.D. by her account) (1992: 63). Among the two most productive formulaic templates (those using -brica and man- respectively) almost all the compounds are synthetic, and most of the -brica compounds appear to have been coined by Wulfstan. It would seem that Wulfstan's template requiring matched pairs of echoing compounds would have provided an impetus and method for creating compounds, and the deverbal compounds would have provided a ready form. Because the compounds are so close to sentence syntax, they are almost as easy to create and understand as syntactic groups. Synthetic compounds veer extremely close to the line separating syntax and word formation. And the preponderance of synthetic compounds in formulaic systems is not limited to new formations, as the man- system illustrates with such long-standing terms as manslaga and manswora. Almost half the compounds that occur in tightly-joined pairs are synthetic, like manswica ne manswora — a far larger proportion than obtains in pairs not so tightly-joined like woruldmen... woruldqfelum cited in (1) above, where the compounds occur far apart, performing separate syntactic functions. Perhaps these synthetic compounds, even though they are not hapax legomena, also suggest the creative impetus of closely-matched pairs So when Wulfstan uses a word like huslgang, a word occurring in earlier writings, perhaps he has generated it anew, at least in part, based on its similarity in type to ingang. In short the demands of creating closely-matched pairs help shape the compounds that are used, and the synthetic compounds that Wulfstan uses show a ready adaptability to such pairs. In summary, these highly artificial compound pairs would likely have reasserted the composite structure of the compounds, because of the striking proximity of the repeated constituents and the occasional necessity of interpreting the compound to gain the full sense of the larger passage. Such pairs may also have occasioned the generation and even coinage of some compounds, which, being new, would almost certainly have been interpreted on the basis of their composite structure. While the lexical status of compounds still remains difficult to gauge for the Old English corpus, at least in these pairs Wulfstan has left us with some striking suggestions that these compounds were interpreted as compounds.

Notes 1. All quotes are cited by the Dictionary of Old English short title; cf. HealeyVenezky(1980).

References Bank, Claudia (ed.) 1992 Language and civilization. Frankfurt: Lang.

Motivations for producing and analyzing compounds in Wulfstan 's sermons

21

Brekle, Herbert E. 1978 "Reflections on the conditions for the coining, use and understanding of nominal compounds", in: Wolfgang U. Dressier—Wolfgang Meid (eds.), 68-77. Downing, Pamela 1977 "On the creation and use of English compound nouns", Language 53: 810-842. Dressier, Wolfgang U.—Wolfgang Meid (eds.) 1978 Proceedings of the 12th International Congress of Linguists. Innsbruck: Institut für Sprachwissenschaft der Universität Innsbruck. Fill, Alwin 1992 '"Visions and revisions': A type of pun based on idiomatized complex words", in: Claudia Bank (ed.), 551-563. Frye, Northrop 1990 Words with power. San Diego: Harcourt. Healey, Antonette diPaolo—Richard L. Venezky 1980 A microfiche concordance to Old English: The list of texts and index of editions (Publications of the Dictionary of Old English 1). Toronto: Dictionary of Old English. Leech, Geoffrey N. 1969 A linguistic guide to English poetry. London: Longman. Lieber, Rochelle 1992 Deconstructing morphology: Word formation in syntactic theory. Chicago: University of Chicago Press. Lipka, Leonhard—Hartmut Günther (eds.) 1981 Wortbildung. Darmstadt: Wissenschaftliche Buchgesellschaft. Lipka, Leonhard 1981 "Lexikalisierung im Deutschen und Englischen", in: Leonhard Lipka—Hartmut Günther (eds.), 119-132. Marchand, Hans 1969 Categories and types of Present-day English word-formation. (2nd edition.) München: Beck.

The degrammaticalization of addressee-satisfaction conditionals in Early Modern English1 Guohua Chen

1. Grammaticalization and degrammaticalization Studies on grammaticalization have shown that a large number of grammatical items have developed from lexical items. On the basis of these studies, Hopper— Traugott (1993: 126) hypothesize that "all grammaticalization involves shifts in specific linguistic contexts from lexical item to grammatical item, or from less to more grammatical item, and that grammaticalization clines are irreversible". Although it is now generally accepted that unidirectionality can not be regarded as an absolute principle, and changes in the opposite direction have been found (cf. Greenberg 1991: 301-314; Harris—Campbell 1995: 337-338), counterexamples, i.e., examples of degrammaticalization, are believed to be rare and "statistically insignificant" (Heine—Claudi—Hiinnemeyer 1991: 5). In theory degrammaticalization should be a common phenomenon, for without it, as more and more lexical items are grammaticalized, the grammatical system of a language will grow more and more complex until it is too complex to serve its purpose effectively. The question is what counts as degrammaticalization. In a narrow sense when a grammatical item has lost its grammatical function and become merely a lexical item, it can be said to have been degrammaticalized. This kind of degrammaticalization is indeed rather rare. More commonly, a grammatical item may acquire a lexical function while retaining its grammatical function. Such a functional diversification should also be seen as a kind of degrammaticalization. Still another kind of degrammaticalization is the disuse or extinction of a grammatical item. This kind of degrammaticalization has so far been treated under grammaticalization. Givon (1979: 208-209) hypothesizes that grammaticalization follows the following cyclic development:

discourse > syntax > morphology > morphophonemics > zero

In my view, the last stage, which is represented by "zero", is actually degrammaticalization. The reason is very simple: if the third-person singular form of the English verb were leveled, other things being equal, the English verb system would certainly be less grammaticalized rather than more grammaticalized. In this paper degrammaticalization is used in a broad sense to cover all three kinds of linguistic change just described.

24

Guohua Chen

2. De-conditionalization and degrammaticalization Prototypical conditional sentences are characterized by a contingency relation between protasis and apodosis, with the truth of the apodosis being contingent on that of the protasis. Very often, however, we find sentences containing a clause clearly marked as a conditional protasis but no direct contingency relationship can be found between it and the rest of the sentence, e.g.: (1)

And yf the prologue bee so small that ye cannot wel reade them, (her is my fathers book. (EModEl CORP RPLUMPT 232)2

Since no direct contingency relationship can be found in (1), the conditional clause in such a sentence is said to express "indirect condition" (cf. Quirk et al. 1985: 1095-1097). For sentences like (1), indirect condition actually means the lack of a direct apodosis to what appears to be a conditional protasis. The missing apodosis, which is omitted by the speaker for one reason or another, is usually understood by the addressee and can be recovered from the context. There are situations, however, in which it is difficult, if not impossible, to recover the implied apodosis, e.g.:

(2)

If it please you (sayes the gentleman), here is a good fellow will goe and attire him in one of his coates. (EModE2 FICT ARMIN 10)

(3)

There's an excellent diapasm in a chain, too, if you like. (1599 B. Jonson, Cynthia's Rev., 5.2)

(4)

As euell as a violent taker, or (if you will) a robber. (1561 T. Norton, Calvin's Inst. Author's Pref.)

In (2)-(4) the /^clauses in fact convey hardly any sense of conditionally. They have become merely conventionalized expressions of either politeness (in (2) and (3)) or linguistic tentativeness (in (4)). Having ceased to function as the protasis of a conditional sentence, they can be said to have become de-conditionalized. By the definition adopted in this paper, they can also be said to have been degrammaticalized. These degrammaticalized /^-clauses have two features in common: (a) they all involve a second-person pronoun or its equivalent,3 and (b) the verb expresses a sense of satisfaction or volition. For the sake of convenience I term them addressee-satisfaction conditionals. In this paper attention will be focused on addressee-satisfaction conditionals involving please, like, will, and list?

The degrammaticalization of addressee-satisfaction conditionals

25

3. The degrammaticalization of addressee-satisfaction conditionals 3.1.

The degrammaticalization of if it please you and its variants

Of all the addressee-satisfaction conditionals, if it please you and its variants have undergone the most thorough degrammaticalization. According to the Oxford English dictionary (s.v. please, v.), please is a descendant of Old French plaisir, which in turn descends from the Latin verb placere 'to be pleasing or agreeable'. With the impersonal pronoun it as its subject and a second-person pronoun or its equivalent as its dative object (preceded by the preposition to), it was first used in the 14th century in the sense 'to seem good to one; to be one's will or pleasure'. In conditional clauses only the form without to survived into Early Modern English, as in (2). Judging from the examples cited in the OED (s.v. please, v. 6. a and b), in the early 16th century please began to be used as an intransitive verb in the sense 'to be pleased, to like; to have the will or desire'. As a result ifland you please emerged: (5) (6)

But tary, I pray you all, Yf ye please. (1530 Jil of Brentford's Test. (Ballad Soc.) 15) / wyll goe, and you please. (1621 Elsing, Debates Ho. Lords (Camden) 58)

In (5) and (6), as in (2), the conditional clause has lost its sense of conditionality and become what the OED (s.v. please, v. 3) calls a "deferential phrase of address". There seems to be no semantic or pragmatic difference between ifland it please you and if and you please. The only difference, judging by their distribution in the Helsinki Corpus, seems to be that the use of if/and it please you was on the decrease while that of iß and you please was on the increase. Where the sense of conditionality expressed by if it please you and its variants was lost, there was of course not much point for these expressions to retain their conditional form. We find and it please you sometimes reduced to an't please you or and please you: (7)

(8)

Your Grandfather of famous memory (an't please your Maiesty) and your great-Vncle Edward the Placke Prince of Wales, as l haue read in the Chronicles, fought a most praue pattle here in France. (1599 Shakespeare, H5, 2622/4.7.92) And please your Maiestie, let his Neck answere for it, if there is any Marshall Law in the world. (1599 Shakespeare, H5, 2760/4.8.46)

The formal reduction of if you please is more difficult to identify. In (9), where the nominative ye is used, there seems to be no doubt that please ye is a reduced form of if ye please:

26

Guohua Chen

(9)

Sir, I shal not be slacke, in signe whereof,/ Please ye we may contriue this afternoone,/ And quaffe carowses to our Mistresse health. (1596 Shakespeare, Shr., 848/1.2.276)

Where you is used, as is more often the case, one can not be certain that please you is a reduced form of if you please. The OED thinks that please you is an elliptical form of please it you. Of course, if we take please it you to be the SV inversion form of if it please you, it is still a reduced form of an addressee-satisfaction conditional. The problem is that the OED does not paraphrase please it you as 'if it please you'. Instead it (s.v. please, v. 3 b and c) defines please it you as 'may it please you', and please you and so please you, with omission of it, as 'may it (so) please you'. All this sounds plausible, because from the very beginning please it to you and please it you could indeed be used in the sense 'may it please you', as in: (10)

Please it, Lorde, to \>e,patpou defende me. (1325 Prose Psalter, xxxix.

(11)

Please youre gracious Hynes to be advertised that [...] (1454 Let. fr. Kildare in Ellis, Orig. Lett. ii. 39 1. 1 18)

The OED (s.v. please, v. 3 b) observes that the infinitive following please you "often lost its to in 16 - 17th c.", as in: (12)

My Lord I cannot be so soone prouided,/ Please you deliberate a day or two. (1591 Shakespeare, TGV, 375/1.3.73)

Sometimes it was the second-person pronoun rather than to that was omitted: (13)

Please to bespeak something else, I have every thing in the House. (EModE3 COME FARQUHAR 7)

When to and the second-person pronoun were both dropped, we get sentences like (14), which is the earliest instance of this use of please cited in the OED: (14)

Please entitle S. only Bart. (1711 Hearne, Collect. (Oxf. Hist. Soc.) III. 147)

The OED (s.v. please, v. 6 c.) concludes that the imperative or optative please was originally short for please you - 'may it (or let it) please you'. This conclusion is supported by the fact that the sense 'may it please you' was sometimes expressed by may it please you itself, as in: (15)

May it please you that I shall aunswer particularly to the matters objected against me. (EModEl TRI THROCKM I,65.C1)

The degrammaticalization of addressee-satisfaction conditionals

27

In some cases, however, the use of please you and so please you differ in important ways from that of may it please you, e.g.: (16)

Imo. Who 's there? My woman: Helene? La. Please you Madam. (1611, Shakespeare, Cym., 905/2.2.1)

(17)

Ros. [... ] Will you heare the letter? Sil. So please you, for I neuer heard it yet. (1600, Shakespeare, AYL, 2186/4.3.37)

In the first place, may it please you is usually followed either by an infinitive (with or without to) introducing a polite request or by a thai-clause (with or without that) initiating a respectful discourse; whereas please you in (16) and so please you in (17) are not followed by an infinitive or a thai-clause. Secondly, in may it please you the referent of it is the following infinitive or thai-clause, without which the sentence is pragmatically incomplete; whereas both please you in (16) and so please you in (17) are pragmatically complete. Finally and most importantly, unlike please you in (16) and so please you in (17), may it please you never seems to have been used as or in association with an answer. In view of these differences, it is unlikely that the please you in (16) and the so please you in (17) were reduced forms of may it so please you. By contrast it is easy to see the similarities between the bare please you and so please you on the one hand and if it please you and its variants on the other. They are all semantically complete and they are all commonly used as, or in association with, answers. In view of these similarities, the bare please you in (16) should be regarded as a reduced form of if you please, and the bare so please you in (17) as a reduced form of if so you please. In principle, when used in making a request or accepting an offer, as in (5), (8), and (17), if it please you and its variants had the potential of being reduced to a bare please: (5')

But tary, I pray you all, (Yfye) please.

(8')

(And) please (your Maiestie), let his Neck answerefor it.

(17') Ros. [... ] Will you heare the letter? Sil. (So) please (you), for I neuer heard it yet. When the bracketed elements in the above sentences are omitted, the bare please will be used in effect as an interjection expressing politeness. Reaching this stage, the degrammaticalization of if it please you and its variants has run through its course. This happened some time after the Early Modern English period. It has to be pointed out that not all uses of if you please in indirect conditional sentences underwent formal reduction. It seems that when it is used in a

28

Guohua Chen

metalinguistic function, it is resistant to formal reduction. The earliest instances of the metalinguistic use of if it please you and if you please cited in the OED are: (18)

We may terme him the Loue-burden, following the originall, or if it please you, the long repeat. (1589 Puttenham, Eng. Poesie, iii. xix. (Arb.) 233)

(19)

The Emperour ana Germans, or if you please the Imperials. (1630 Λ Johnson's Kingdom & Commonwealth 101)

The metalinguistic function of // /'/ please you and its variants was not very common in Early Modem English. No such use is found in the Helsinki Corpus.

3.2.

The degrammaticalization of if it like you and its variants

In Early Modern English if it like you resembled // it please you in every way except it did not go as far in degrammaticalization. According to the OED (s.v. like, v.l 1), the basic meaning of like used as an impersonal verb is 'to please, be pleasing, suit a person'. As like is synonymous with please, if it like you is also synonymous with // it please you. In fact, in the earliest instance of its use in a conditional clause cited in the OED, like is used in conjunction with please: (20)

Depende on me a drope of thy largesse,/ Right in this \vyse if it thee lyke & plesse. (1406 Hoccleve, La male regie, iv. 249)

If it like you shares some of the formal characteristics of if it please you. E.g., it has the variant forms and it like you and like it you: (21)

Than sayed my lord chejfe justyes unto me, "Syr, whate make yow here? are you not a Londynar?" "Yes, andyt lyke your lordshyp". (EModEl ΒΙΑ MOWNTAYNE 206)

(22)

Like it your Grace,/ The State takes notice of the priuate difference/ Betwixt you, and the Cardinall. (1613 Shakespeare, H8, 160/1.1.100)

In (20) the z/-clause functions as the protasis of a conditional sentence, while in (21) and (22) its variants have become merely expressions of deference. When its sense of conditionality was lost, if]and it like you also underwent formal reduction. In (23), e.g., and it like your lordship was reduced to and like your lordship: (23)

Than sayed my lord chyffe justys, "Have you browghte yn your swertyes?" "Ye, and lyke your lordship here they be". (EModEl ΒΙΑ MOWNTAYNE 208)

The degrammaticalization of addressee-satisfaction conditionals

29

Except for if you like, if it like you and its variants did not survive Early Modern English. In the Helsinki Corpus iß and it like you is quite common in EModEl (as common as ifland it please you), rare in EModE2, and non-existent in EModES. As a variant of if it like you, if you like came into existence as early as in the mid 15th century (cf. the OED, s.v. like). Yet it was not until the end of the 16th century that it began to be used non-conditionally, as in: (24)

There's an excellent diapasm in a chain, too, if you like. (1599 B. Jonson, Cynthia 's Rev., 5.2)

Unlike if you please, if you like did not undergo formal reduction. One reason may be that it never lost its function as a true protasis; the other reason may be that its non-conditional use has been restricted to metalinguistic function, which, as we have seen in 3.1, somehow saves the conditional clause from formal reduction. In its metalinguistic function, if you like means "if you wish to phrase or consider something in a particular manner, often used as vaguely intensive expression = 'indeed', 'perhaps'" (OED, s.v. like, v. 6 b.). This metalinguistic function arose rather late. In the Helsinki Corpus not a single instance is found, hi the OED the first instance of its use, which is cited under credulity, is dated 1875.

3.3.

The degrammaticalization of you will

According to the OED (s.v. will, v.l I. 1) the basic meaning of will is 'desire, wish for, have a mind to', which is close to the meaning of please and like used as an intransitive verb. The earliest instance of if you will cited in the OED is (25), which seems to be synonymous with if you please and if you like: (25)

Ich wile be zigge yef bou wylt. (1340 Ayenb. 101)

In Early Modern English it could be used in the same sense and function: (26)

as I am a gentle man, you shall, if you will, enjoy Ford's wife. (1598 Shakespeare, Wiv., 1010/2.2.265)

As will did not begin as an impersonal verb, if you will has had no structural variant. However, it has gradually undergone functional specialization. As the OED (s.v. will, v.l I. 17) observes, ""ifyou will is sometimes used parenthetically to qualify a word or phrase: = 'if you wish it to be so called', 'if you choose or prefer to call it so'", as in: (27)

Gravity [...] depends entirely on the constant and efficacious, and if you will, the supernatural and miraculous Influence of Almighty God. (1696 Whitson, The. Earth, iv. i. § 2. 218)

30

Guohua Chen

The earliest instance of this use of if you will in the OED is (4), which is found under taker. It also turns out that if you will was not merely "sometimes" used in a metalinguistic function. Of the 30 or so citations of Early Modern English if you will under various headwords in the OED, more than half are metalinguistic conditionals. In the Helsinki Corpus there are nine instances of if you will and one instance of and you will. Of these ten instances only (28) has a metalinguistic function: (28)

/ am for venturing one of the Hundreds if you will upon this Knight-Errantry. (EModES COME FARQUHAR 6)

The rest are either expressions of politeness, like (29), or true protases. (29)

/ ham verye sory for yt, beleve me and yow wyll. (EModEl ΒΙΑ MOWNTAYNE214)

In its metalinguistic function if you will is also resistant to formal reduction.

3.4.

If you list — the odd one out

Of all the addressee-satisfaction conditionals, if you list was the only one that seemed to have undergone no de-conditionalization. According to the OED (s.v. list, v.l 1.) the basic meaning of list as an impersonal verb is 'to be pleasing to'. As list in this sense is synonymous with the impersonal verbs please and like, if it list you is synonymous with if it please you and if it like you, though its impersonal use did not survive into Early Modern English. The earliest instance of if you list cited in the OED is (30) below, in which the pronoun you is clearly in the accusative case: (30)

Nu ye reste One while, ef you leste. (al 300 K. Horn, 918)

Though not as common as if you please or if you like, if you list, nevertheless, survived into Early Modern English. By then you had largely taken over the function of ye, and according to the OED (s.v. list, v.l 2 b) list had acquired the sense 'to wish, desire, like, choose'. If you list did not seem to have any variant form except for the switch between you and the occasional ye. It had the potential of being used in a metalinguistic function, as shown in (31), but never seemed to have realized it. (31)

Upon the packsaddels [of an elephant], they haue on euery side a little house, or tower, or cage (If you list so to call it) made of wool. (1533 Eden, Treat. New Ind. (Arb.), 15)

The degrammaticalization of addressee-satisfaction conditionals

31

Only two instances of its use are found in the Helsinki Corpus, both being direct conditionals. Unlike other addressee-satisfaction conditionals, if you list does not seem to have any connotation of deference or politeness. With its last citation in the OED (s.v. list, v.l 2 b.) dated 1823, it failed to make it into Present-day English.

4. Summary In Early Modern English conditional sentences were undergoing grammaticalization in terms of both conditional markers and verb forms (cf. Chen 1996: 37-237). While the use of a conditional conjunction, if in particular, was becoming the standard conditional marker for the expression of open conditions, in the case of // it please you and its variants, however, there was a different trend. Not only did z/fail to squeeze out SV inversion and the substandard and, but the conditional marker itself was being left out, leading eventually to the use of the bare please as an interjection. Grammaticalization and degrammaticalization have been seen as linguistic changes in opposite directions, yet paradoxically, as the case of // /'/ please you and its variants illustrate, they have happened at the same time to the same linguistic item. If we approach the phenomena from the point of view of functional diversification, there is nothing paradoxical about it, for it is common for a linguistic item to develop several functions at the same time. What is puzzling is the fact that metalinguistic function seems to safeguard two of the addressee-satisfaction conditionals from formal reduction.

Notes 1. I am grateful to Sylvia Adamson for her comments on an earlier version of the paper, and to Merja Kytö for kindly arranging for me to use the Early Modem English section of the Helsinki Corpus of English Texts. For reasons that have nothing to do with the purpose of the present paper, I have taken out the Bible samples from the corpus and replaced the samples of Shakespeare's The merry wives of Windsor in EModE2 with samples of Ben Jonson's Every man in his humour. 2. All examples with references beginning with "EModE" are from the Early Modern English section of the Helsinki Corpus, which consists of three sub-periods — EModEl (1500-1570), EModE2 (1570-1640) and EModE3 (1640-1710). For details of the corpus and the key to the abbreviations used in text reference cf. Kytö (1996). 3. When a noun phrase such as your majesty is used in referring to the addressee, it is in effect equivalent to a second-person pronoun.

32

Guohua Chen

4. Of these four verbs, please, like, and list were originally impersonal verbs. The shift of their impersonal use to personal use, although it led to variations in the syntactic structures of conditional clauses involving the three verbs, has little to do with conditionality and is, therefore, not discussed in any detail in this paper.

References Chen, Guohua 1996

Conditional sentences in Early Modern English: A study in grammaticalization. [Unpublished Ph.D. dissertation, University of Cambridge.]

Givon, Talmy 1979 On understanding grammar. New York: Academic Press. Greenberg, Joseph H. 1991 "The last stage of grammatical elements: Contractive and expansive desemanticization", in: Elizabeth Closs Traugott— Bernd Heine (eds.), 1.301-314. Harris, Alice. C.—Lyle Campbell 1995 Historical syntax in cross-linguistic perspective. Cambridge: CUP. Heine, Bernd—Ulrike Claudi—Friederike Hünnemeyer 1991 Grammaticalization: A conceptual framework. Chicago: The University of Chicago Press. Hopper, Paul J.—Elizabeth Closs Traugott 1993 Grammaticalization. Cambridge: CUP. Kytö, Merja (comp.) 1996 Manual to the diachronic part of the Helsinki Corpus of English Texts: Coding conventions and lists of source texts. (3rd edition.) Helsinki: Department of English, University of Helsinki. OED = Murray, James A. H.—Henry Bradley—William A. Craigie—Charles T. Onions 1989 The Oxford English dictionary. (2nd edition.) Oxford: Clarendon Press. Quirk, Randolph—Sidney Greenbaum—Geoffrey Leech—Jan Svartvik 1985 A comprehensive grammar of the English language. London: Longman. Traugott, Elizabeth Closs—Bernd Heine (eds.) 1991 Approaches to grammaticalization. Vol. 1. Amsterdam: Benjamins.

From unasecendlic to unspeakable: The role of domain structure in morphological change Christiane Dalton-Puffer

1. Introduction This paper is about deverbal adjectives of the type translatable, imaginable, washable. In Present-day English the productive formation of these adjectives is carried out exclusively by means of ABLE. This means that of all the deverbal adjectives in the lexicon most are the products of a non-native formative. This is in stark contrast to the other major morpho-syntactic categories of derived English words where non-native formatives play only minor roles at the side of one or more native formatives. For the formation of agent nouns, e.g., we need to distinguish between a core area which is occupied by the native -er, and peripheral areas with formatives like -ant, -ist. The situation is very similar for the other major morphosyntactic categories, i.e., deadjectival nouns and denominal adjectives. The exact nature of the restrictions which define the domains of Present-day English formatives within their category is the subject of a long-standing discussion which is mostly conducted on the phonological and morphonological levels, but these restrictions will not be the concern of this paper. The question which is at the center of attention here is to outline the historical conditions under which ABLE came to be more successful than other non-native formatives entering the English language at the same time and through the same channels. In order to pursue this inquiry we need to determine when the form ABLE entered the language and whether or not it was also a functional-semantic novelty. It will turn out that carriers of an ABLE-like function existed in English prior to the arrival of ABLE so that a comparison between them and ABLE seems the next natural step. This kind of procedure is based on the competition metaphor of language change and we will accordingly have to present arguments in what way(s) ABLE was better fitted for survival in English than the native formative(s) that were there before it.

2. The appearance of ABLE in English It is a well-known fact that ABLE entered the English language during the Middle English period via loanwords from French. Using quantitative data from the Helsinki Corpus covering the period from ca. 1150-1420 we can be more specific about this development.

34

Christiane Dalton-Puffer

Table 1. Quantitative development and morphological patterning of ABLE in Middle English. ABLE total occurrences

types

ME1 tokens

types

ME2 tokens

types

ME3 tokens

1

1

7

10

46

176

1 -

-

1

5 _

8 . 2

30 3 13

111 4 61

morphological category of the base V N Stem/Simplex

Examples:

-

2

V: colourable, deceivable, movable, knowable, wilnable N: contemptible, profitable, honourable Stem/Simplex: amyable, perdurable

Formations prefixed with un- have not been counted as separate types if the material also contains the unprefixed form. This was done on the grounds that negative prefixation with un- is regular and fully productive. Looking at the "total occurrences" line of Table 1 it is easy to see that ME3, the period from 1350 to 1420, accounts for 94 per cent of the 187 ABLE's in this particular corpus. This quantitative explosion is paralleled by all other Romance formatives identified for Middle English (cf. Dalton-Puffer 1996: ch. 6-8) and we can, therefore, exclude the possibility that any peculiarities in the chronology of the borrowing process itself were (co-)responsible for the specially warm reception ABLE seems to have had in the English language. The fact that ABLE was readily analyzed as a possible derivational formative rather than just a recurring element in a number of loans is witnessed by the fact that a small number of formations with native words as their bases starts appearing at an early point in time. (1)

Earliest hybrids with ABLE: de heilest wilnable ping, be whiche is God. (Cloude of Unknowing) sende bee help and cunfort vnspeicable, bat no tunge may teile how myche it is (Hilton)

In the morphological analysis summarized in the second half of Table 1 the double category Stem/Simplex contains not only simplex lexical items featuring the string -able but also those ABLE-words which may well be analyzable within French and/or Latin but whose base does not occur independently in the Middle English lexicon. There will be many morphologists who would exclude such cases from a discussion of Middle English word-formation. Yet the meaning of most of these words (together with part of their form, namely -able) is compatible with

The role of domain structure in morphological change

35

analyzable formations so that we cannot rule out the possibility that they contributed to the overall "strength" of the pattern.

3. Was ABLE a functional novelty? Before answering the question whether ABLE was a functional novelty on top of being a formal one we need to address the question of what the function of ABLE actually is.

3.1.

The fiinction(s) of'ABLE

In considering the Middle English ABLE-figures in Table 1 we noted a clear preponderance of deverbal formations over other types. It, therefore, seems to make sense to look for the main function(s) of ABLE among this large subgroup. The most characteristic deverbal ABLE formations are of the type knowable, unspecable (Middle English), unquenchable, unmatchable (Early Modern English) which are paralleled in Present-day English by, e.g., doable, marketable, untranscribeable, unbearable. The usual paraphrase which is used to render their meaning is something like 'can be Ved'; doable = 'can be done'. This amounts to saying that such adjectives encode the possibility of the noun they modify to be affected by the action expressed in the verb (or its ability/likelihood to carry out the verbal action). Looking at actual data (irrespective of the historical period they are from) it is striking how often negative meanings are involved. So, very often the adjectives actually express the impossibility of the noun they modify to be affected by the action expressed in the verb; i.e., they mean 'canNOT be Ved'. Consequently, formations prefixed with un- make up a sizable part of any set of ABLE data and, once we consider not only prefixal negation but include the phrasal level, the negatives are probably in the majority. I do not think this high incidence of negation is fortuitous1 and I believe we can tease out its implications better if we try to "deconstruct" what is actually encapsulated in the above-mentioned paraphrase. Once we do that, several semantic "building-blocks" can be identified. Besides negation we can identify passivity and some kind of possibility. It seems to me that possibility and negation are more strongly linked to each other than passivity is to either of them, and I would take this as a first indicator that the passivity element is less central to the function of ABLE (and ABLE-type suffixes in general) than the other two elements. Negation and possibility both encode the speaker's view of a particular proposition or a particular portion of reality, which puts them into the circle of modality. Bybee (1985: 176-178) also notes an affinity of negation and other mood meanings, both cross-linguistically and with special reference to affixing. In the case of ABLE (and ABLE-type suffixes in general) the most likely scenario, then, is that the "possibility" meaning attracts negative meanings which can be

36

Christiane Dalton-Puffer

derivational (unVable) or syntactic (not Vable). From this it follows that the modal component represents the core meaning of ABLE-type formatives. Taking this view on ABLE allows us to cover not only the more frequent 'can((not) be Ved' instances but also those derivatives where the passive element is missing, e.g., suitable. It seems to me that a basic modal meaning appears in different guises: in the passive adjectives the modal meaning is one of possibility, in the active ones it tends to be one of ability or likelihood. There are some cases where both readings are possible, e.g., movable which can mean 'able (and prone) to move' but also 'can be moved'. Traditionally, it has been said that ABLE somehow "contains" both an active and a passive meaning and whether one or the other is activated depends on the syntactic properties of the base-verb. This way, derivatives based on intransitive verbs are automatically active and derivatives based on transitive verbs are automatically passive. Confronted with a largish amount of data we have to admit that, while not being entirely false, this claim is not entirely true either, as is shown by the movable example above. Without committing myself to an in-depth discussion of the issue I will simply claim that the passive meaning of ABLE-type suffixes is secondary to the modal meaning, and is generated where appropriate through access to syntactic but probably also referential and contextual information.2 Summing up we can say that the function of ABLE is to derive modal adjectives. The mood which it encodes is Potential, covering both (prob)ability and possibility.

3.2.

LIC as a precursor of ABLE

The question we need to pursue now is how the semantic function I have sketched in the previous section was expressed in English before the arrival of ABLE. This includes the question whether it was expressed derivationally at all and if it was, what the formal exponent(s) looked like. It is evident that, other than the speaker who encodes conceptual into linguistic form, the analyst has to start at the formal end and so the first route of access to the question at hand is often an -able in the gloss of an Old or Middle English word. To cut a long story short, the semantic function encoded by means of ABLE from Middle English onwards was generally expressed less frequently,3 and by more variable formal means in Old and Early Middle English. Nevertheless, there emerges one derivational suffix which served as the main exponent of the said function, namely LIC. In his 1991 article Mclntosh mentions OE -lie in this function and notes some of the problems involved. While not answering his demand that a future, full treatment of -lie should be based on the Toronto material, the data I will present and discuss in the following are based on a corpus rather than on dictionary material.4 The formation of modal adjectives is only one among several functions of LIC in Old English and we will have opportunity to review the others in section 4 of this

The role of domain structure in morphological change

37

paper. In the modal function we find LIC attaching to verb-stems, past participles and present participles. The fact that participles can take on the syntactic role of an adjective has repeatedly led to these derivatives being categorized among deadjectival formations with -lie, but a look at the meanings of the de-participial formations presented here will make it obvious that these are proper ABLE-type, modal adjectives. In the following I present some examples showing LIC in modal function: (2) Modal adjectives in LIC in Old and Early Middle English from

verb-stems: unandwendlic 'inadvertible' V wendan, ungeferlic 'inaccessible' V faran, gedafenliche 'fitting, behooving' V gedafenian, unhierlice 'disobedient, fierce, savage' V hieran, unsehelich 'invisible' V seon

from Past Participles: ungeliefedlice 'unbelievable', unarssfnedlic 'intolerable', untodssledlice 'indivisible', gerisenlich 'suitable', gesewenlic 'visible', unarimedlice 'uncountable', unalyfedlice 'unallowable', unoferswidhedlice 'unconquerable', unbefangenlice 'incomprehensible', unatealledlic 'innumerable' from Present Participles: ungeliefendlice 'unbelievable', unarsefhendlice 'intolerable', untodsslendlice 'indivisible', unassscgendlice 'unspeakable', halsiendlic 'deprecable', onsconiendlice 'abominable, detestable', thearfendlic 'needy, destitute', unadrysnendlic 'unquenchable', unmiltsigendlic 'unpardonable', unwuniendlic 'uninhabitable', \vitnigendlic 'punishable, to be punished', brosniendlic 'perishable', unoferwinnendlic 'invincible, unconquerable' The examples include several doublets (the first four examples of each of the participle paragraphs) and it is interesting to note that the different base forms do not result in different meanings. This is rather puzzling because it destroys a form-meaning symmetry which seems to follow from the meanings of the components involved. If we take the suffix to be responsible for the modal meaning, the past participle can be said to contribute the passive meaning, while the present participle remains [-passive] and activates the ability/probability meaning. Quite according to this principle unalyfedlic, whose base is a past participle, means 'unallowable, cannot be allowed', and styrigendlic, whose base is a present participle, means 'movable, can move'. In the data, however, this form-meaning symmetry clearly exists only in principle as there are more derivatives from present participles with unexpected passive meanings than there are with active meanings. I do not want to speculate at any length on the possible reasons for this state of affairs but I do not think that my aggregate view of the data obliterates a diachronic development here, as the doublets co-occur within the

38

Christiane Dalton-Puffer

same sub-period. What may be involved is dialectal variation but this needs to be checked using a larger database such as the Toronto material. Even so, I have my doubts whether dialectal variation is behind the asymmetry. Contrastive data from Old High German (Schmid forthc.) show the same kind of phenomenon, which means that Old English seems to have inherited it from its ancestor language. Table 2, then, gives a numeric overview of modal adjectives with LIC for the period under investigation. Note that forms exhibiting prefix variation (ge-la-) and the negated «η-forms were counted as only one type. Table 2. Numeric development of modal LIC Old English to Early Middle English.

V-stem PastPart PresPart TOTAL

OE2

OE2

OE3

OE3

OE4

OE4

ME1

ME1

types

tokens

types

tokens

types

tokens

types

tokens

9 12 8 29

15 19 9 43

8 13 24 45

55 49 60 164

4 7 8 19

14 7 17 38

16 2 14 32

22 2 20 44

In interpreting Table 2 we can see that it parallels the impression gained from the examples in (2). The number of formations based on participles by far outweighs those based on verb-stems. There is a certain tendency for decline, visible most clearly with the derivatives from Past Participles (the decline will become much more striking when we move into Middle English in Table 3). The derivatives from verb-stems seem to hang on more tenaciously, even showing the highest number of different types at the latest sub-period under scrutiny here. This is of course in line with the general development of English towards preferring bare stems (later words) as input to its morphology, but the data do not show switches from one form-class to another. I.e., derivatives which used to be from the participle hardly ever show up at a later period featuring the verb-stem as their base. In interpreting tables such as Table 2, it is of course necessary to consider corpus size, because the ratios of type and token figures necessarily vary depending on the overall size of a corpus. In order to enhance comparability between the development of the LIC and ABLE-figures it was, therefore, decided to create sub-corpora of equal size. This is a solution which offers itself since the two adjacent sub-periods OE4 and ME1 taken together contain the same number of words as sub-period ME3, namely ca. 190,000. Table 3, then, shows the quantitative development of LIC and ABLE Table 3, Distribution of LIC and ABLE throughout Late Old English and Middle English. types LIC ABLE

51 1

O4M1 tokens 82 1

types 6 5

ME2 tokens 15 8

types 5 30

ME3 tokens 6 111

The role of domain structure in morphological change

39

The figures show the two exponents of modal deverbal adjectives as being subject to opposite trends. The type-token ratio of LIC in the earliest period looks surprisingly like that of a productive pattern: 82 tokens represent a striking 51 different types. In terms of frequency a role reversal has taken place by ME3 but ABLE certainly cannot be said to have reached the same kind of morphological productivity which LIC appears to exhibit earlier. The type-token ratio speaks otherwise and, of course, most of the types are loans from French at that time. There are, however, formations on native bases and these are hapaxes. Can we say then, that ABLE actively "conquered" the territory of LIC? I do not think we can, and for several reasons. Firstly, the figures (especially the ones of ME2) run counter to such an interpretation: this is a phase where the two "rivals" seems to be at an impasse. Also, in order to bring about some sort of "catastrophic event" ABLE would have had to be extremely vigorous and "superior" to LIC in several ways, on several linguistic levels. In an earlier study which checked the two suffixes against several Naturalness parameters it was concluded that this was not the case (Dalton-Puffer 1993: 159-161). Considering the system-congruity of the two formatives in terms of the morphological status of their input (stems and/or words) the LIC and ABLE are on a par, and in terms of morphotactic transparency LIC even has a slight edge on ABLE. Subsequent events, however, indicate that LIC was experiencing a crisis; something was eroding its status and preparing the ground for the quick flowering of ABLE. The remainder of this paper is dedicated to the examination of what I think was behind the crisis of modal LIC.

4. Comparing the domains of LIC and ABLE 4.1.

Derivational domains

It is standard procedure in paradigmatic morphology to determine for a certain formative what kind of base it attaches to and what it does, i.e., what kind ofthing it "produces" in formal terms and in semantic terms. In principle, form and meaning are considered as being on a par but the realities of morphological study and the traditions which have arisen from it have usually meant that formal descriptions and formal criteria tend to be more refined than their semantic counterparts. Even so, it is usually acknowledged that the connections which hold between derivational forms and derivational functions play an important part in several phenomena derivational morphologists are interested in: establishing suffix taxonomies, delineating diachronic developments, determining productivity, or working out which restrictions govern the applicability of word-formation rules. The information about where an affix goes and what it does is commonly regarded as that affix's domain. The domain thus has a formal and a semantic sub-domain. Each of them can vary in complexity from affix to affix, and so can the connections between the domains. Several word-formation theories which are radically different in outlook and design have acknowledged this fact and have said

40

Christiane Dalton-Puffer

in one way or another that the configuration of derivational domains is important. Structuralist analyses have often worked with the concept of "functional overload" of formal entities when engaged in explaining structural changes in linguistic subsystems. Aronoff in his Unitary Base Hypothesis postulates that the "syntacticosemantic specification of the base... is always unique. A WFR [word-formation rule] will never operate on either this or that" (1976: 48), which leads to a proliferation of homophonous affixes each of which has a satisfactorily simple domain. It will soon be obvious that the approach adopted here is not in keeping with the Unitary Base Hypothesis as it stands, but only with its basic interest in keeping domains simple. In Natural Morphology this general idea takes the shape of a preference for unique (or even bi-unique) form-meaning relations. The assumption behind this is that unambiguous form-meaning relations enhance the iconicity of derived words, thus making them better signs in semiotic terms (cf. Dressier 1987: 111-116,1994). It is the general spirit of the latter theory that the approach adopted here is most sympathetic to without being an instantiation of Natural Morphology in any pure sense. The reasons for this lie in the fact that also Natural Morphology is too form-oriented for what I think is necessary in the present context. Having made this somewhat half-hearted commitment to Naturalness Theory, I will from now on refer to my general assumption about the behavior of derivational suffixes as the "Simple Domain Principle": across theories it is thought preferable for one meaning to be expressed by one and only one form. Conversely, each suffix (form) should have a clearly defined semantic "effect". This entails that each suffix should prefer a certain kind of derivational base. Exactly how this "certain kind of base" is to be defined is a question for which there does not seem to exist a satisfactory solution. Traditionally, morphologists have thought in terms of word-classes or syntactic categories, i.e., mainly in terms of nouns, verbs, and adjectives, but such a coarse-grained classification tends to be unsatisfactory, because it is far too general in many cases. As soon as we transcend strictly syntactic sub-categorization, however, it becomes very hard to constrain the kind and amount of semantic information that needs to come into play. I will make no attempt at solving this problem here but I am aware that we will be confronted with it when we look at the domain structures of LIC and ABLE. 4.2.

Formal domains

4.2.1. Formal domain of LIC In this section our task is to determine which types of bases the form -lic(h) seems to be able to attach to in the Old English and Early Middle English data. Accordingly, our point of view will be (almost) exclusively form-oriented. The discussion will not include any descriptive statistics but the following list has been

The role of domain structure in morphological change

41

ordered according to the frequency of the different formal types, starting with the most frequent ones.5 (3)

a.

N +lic eorblich 'earthly\freondlic 'friendly\flxsclic 'fleshly'

b.

A + lie ieoellic 'idle', blidelic 'blithe', deoplic 'deep', clxnlic 'clean'

c.

Present Part + lie (a)berendlich 'bearable', halsiendlic 'deprecable'

d.

Past Part + lie ungesawenlic 'invisible', unarimedlic 'uncountable'

unasecgendlic

untodeledlich

e.

V + lie unheorlic 'disobedient', unsehelic 'invisible'

f.

stem + lie luflich 'lovable', gelimplic 'suitable'

'unspeakable',

'indivisible',

Group a. represents the majority of occurrences of the form LIC: attaching to nominal bases is a type of behavior most common among derivational suffixes which produce adjectives. Group b. needs a little more comment. It must be stressed that the cases subsumed here are not the cases were -lic/e represents an adverbial ending. These derivatives from adjectives function as adjectives. These are often grouped together with what here appears as groups c. and d., where LIC attaches to participles. As witnessed in (2) above, the data also include a fair amount of doublets with both participles of the same verb involved in LIC-derivation. Then we have group e. where LIC attaches to verb-stems. Other than in Old High German (Schmid forthc.) we find no infinitives as derivational bases, only bare stems. Additionally, we have to recognize a group of //c-derivatives where the sheer form of the stem does not allow conclusions about its syntactic specification which could be either nominal or verbal. Category-membership can only be determined with the help of semantic information about the derivatives (which are always modal). An additional aspect of the topic has been purposefully excluded from the data and left out of the discussion altogether and that is the role of LIC in adverb formation. As there is no way I can do justice to this issue in a few paragraphs I simply would like to point out that adverbial -lice cannot categorically be excluded from the domain of LIC, although in the present article I am following exactly that line. While the adverbial suffix -lice and the adjectival suffix -lie should be

42

Christiane Dalton-Puffer

sufficiently distinguishable in principle, the distinction is not always so clear in real terms (at least with the written evidence we have). The question of whether -lice and -lie are two distinct suffixes certainly merits a separate discussion. 4.2.2. Formal domain of ABLE Earlier on in this paper I presented a quantitative overview of the ABLE formations in Middle English. In order to facilitate comparison I will transform the information contained in Table 1 into the format we used in the previous section and start with the formal pattern which is found most frequently in the data: (4)

a. V + able Receivable, movable, knowable, wilnable, etc. b. Stem/Simplex + able amyable, charitable, credible, horrible, mutable, etc. c. N + able profitable, contemptible, honourable, etc. It is interesting — and possibly not trivial — that the strongest type, namely deverbal ABLE, was the first to be borrowed and that it maintained its head-start throughout the period observed here — and also into Present-day English, it might be added. With the ABLE-data it is also necessary to make explicit how they were divided up into the different formal categories. As we have no access to speaker competences at any one point under investigation we need to rely on dictionary information. It was, therefore, decided that any derivative with -able whose derivational base was also attested in the Middle English dictionary as an independent Middle English verb or noun was classed as verb-based or noun-based. Conversely, if the string preceding -able was not attested independently, the formations were counted among category b. A formation is regarded as stem-based if its base can be found in other Middle English derived words but never on its own (which also includes inflection). The rest of category b. can only be analyzed on the basis of French or Latin but not from within the grammar of Middle English; these words have, therefore, to be regarded as simplex lexical items (e.g., amyable, credible). This means that category b. is heterogeneous but the borderline between the two groups fluctuates over time and the distinction is irrelevant for our immediate concerns. Group c. is atypically small for an adjectival suffix — these suffixes usually have strong denominal branches. In sum, the formal domain of ABLE as reflected in the data under investigation seems reasonably simple.

The role of domain structure in morphological change

4.3.

43

Semantic domains

Before embarking on a discussion of the semantic domains of the two suffixes we need to address the question of what kind of semantic description we are going to aim for. What we want is clearly a semantic description which operates on a level of generality which is comparable to that of the formal domain. The classic way of doing things is to find natural-language paraphrases for the meaning of the derivational process under discussion. Such paraphrases are abstractions from an array of existing lexemes, their smallest common denominator in other words. They are useful and flexible and do not require an elaborate formal apparatus. We have used this technique in the introductory sections of this article when we said that the function of modal adjective suffixes was to produce adjectives meaning 'can be Ved'; i.e., doable means 'can be done'. What can be a problem with paraphrases is that there are no explicit constraints on what goes into them so that one may end up with "alternative meanings" of one and the same suffix which are, intuitively, on totally different levels of generality. This problem can be illustrated by means of Marchand's treatment of -ly (1969: 329-331). Among several other paraphrases for -ly Marchand mentions "periodic recurrence" as the meaning of -ly in words like daily, yearly, etc. This paraphrase is totally different in character from other paraphrases such as "characteristic of N, connected with N" as in bodily, earthly, timely. Also, one feels that timely (which is subsumed under another paraphrase) and daily must somehow be "the same" and the question is how one might grasp this sameness. In short, our interest is in finding and using a constrained vocabulary for paraphrasing the meaning of derived lexemes. Ideally we should be able to identify a limited number of deeper-level semantic operators from which we can generate all the meanings we find expressed through derivation. The questions which this raises clearly deserve an in-depth treatment which cannot be accomplished here. For the time being, however, I will point out that the descriptive vocabulary used by me in the following is heuristic in character and will ultimately have to be shown to be integrable into a coherent theory of derivational meaning. Connections to cognitive approaches (e.g., Szymanek 1988; Zbierska-Sawala 1993) are obvious.

4.3.1. Semantic domain of LIC Let me first turn to the semantic function of LIC in which we are particularly interested here, namely the one which produces modal adjectives. In section 3.1 I have outlined in detail what I think is the configuration of this particular derivational function so that I will only restate my conclusions. We said that there seem to be three main semantic building blocks involved in this pattern: Potentiality, Negation, and Passivity, and that these seem to be ordered hierarchically. The primary function of ABLE-type suffixes is the modal meaning of Potentiality which subsumes Possibility and (Prob)Ability. As there is a special

44

Christiane Dalton-Puffer

affinity relation between negative and modal meanings, Potentiality and Negation are fairly likely to occur together. Like Negation, Passivity is secondary, too, and whether it is realized seems to be partly a function of the internal structure of the derivational base, partly also of pragmatic factors. This account of the semantic function of modal adjective suffixes, which was established with ABLE in mind, works equally well with regard to the Old and Early Middle English data with LIC, as it helps us to account for the significant number of negative formations and for some notable form-meaning asymmetries. As mentioned in section 3.2 we would expect past participles to appear in passive derivatives (e.g., unalyfedlic 'unallowable') and present participles to appear in active derivatives (e.g., giliorendlic 'transitory'). This, however, is not die case. While the past participle derivatives meet our expectations almost without fail (note exceptions like awendedlic 'mutable, changeable' which can be active and passive), the present participle derivatives most certainly do not. Close to one half of these formations have to be interpreted passively. (5)

arefnendlic 'tolerable', unadrysnendlic 'unquenchable', unasmeagendlice 'inscrutable', unaspyriegendlich 'uninvestigable'

Let me now turn to other functions which LIC performs in the Old English and Early Middle English data. LIC often encodes a possessional relation when it attaches to nouns. In such cases it is predicated of the head-noun which the derived adjective modifies that it "has X" (X being the base of the derived adjective). (6)

lustlic 'with pleasure', gedyldelic 'with patience, patient', egeslic 'fearful'

It is evident that the base-nouns share the characteristic of being abstract nouns. This relatively straightforward fact opens up a whole spate of questions regarding the generation of derivational meaning. Is it really LIC which triggers the possessional meaning, or is it also or even exclusively certain semantic characteristics of the base noun? If it is exclusively the base noun, what is the role of LIC? The last question is particularly pertinent as there is a considerable number of doublets featuring other suffixes such as FUL, IG, and SOM. Possessionality, however, is not the only function encoded in denominal LIC-derivatives. There are denominal formations where the meaning relation predicated about the derived adjective and its head-noun is rather vague but definitely not possessional. Here are some examples: (7)

lichomlic 'bodily', flassclic 'fleshly', eorplic 'heavenly', deaplic 'deadly'

'earthly',

heofonlic

To describe their meaning Marchand suggested (1969: 329-331) something like "characteristic of N, belonging to N" where N is the derivational base of the

The role of domain structure in morphological change

45

derived adjective. If we look closely enough we realize that there is also a relation of possession encapsulated here but that it works in the opposite direction to the previous group of denominals: there, it is predicated that the modified noun "has" the base-noun (lustlic), here it is predicated that the modified noun "belongs to" the base-noun (heofonlich). In other words, the relation of possession is reversed, and we might speculate how much this has to do with the semantic characteristics of the base-nouns involved. Alternatively, we might take an entirely different interpretive route and go for the first option in Marchand's paraphrases, "characteristic of. If an attribute is characteristic of an entity this attribute will figure prominently when we compare that entity to another. In this indirect way the element of similarity comes into play. While I am not going to embrace this solution for heavenly-type words for the moment, it has to be noted that the notion of similarity is necessary elsewhere in the semantic description of LIC, and this is what we will turn to next. There is a group of denominal LIC adjectives which encode that their phrasal head noun is like their base-noun. Such words in the data are manlic 'manly', cynlich 'kingly', hierdelic 'shepherd-like'. In these cases LIC can be said to have a clearly similitudinal function, a function which is also motivated in its etymology.6 The concept of similarity might also be embodied in another group of LIC-adjectives, where LIC attaches to adjectives without any apparent effect on their meaning. Present-day English goodly, lowly are isolated survivors of this kind of derivative which was not infrequent in Old English. It has sometimes been claimed that these words show no discernible meaning differences from the bare adjectives, which made the presence of the suffix rather puzzling. Mclntosh (1991) shows very convincingly, however, that a contextual analysis does point to a well-defined, if subtle, function of LIC in these cases. Investigating pairs of bare and suffixed adjectives (e.g., OE deor : deorlic) he finds that the unsuffixed member typically modifies animate nouns, while the //c-bearing member of the pair is found in front of what he calls "nouns which are only secondarily human": e.g., deor of a person; deorlic of a deed (1991: 299). What counts as primarily human and what counts as secondarily human may be negotiable,7 but the principle of the animacy hierarchy at work can also be seen from the following examples where 'voice' can be said to be more concrete than 'steadfastness': (8)

mid unablinnende stemne 'with unceasing voice' ther wxs unablinnendlic staöolfxstnys 'there was unceasing steadfastness'

In these de-adjectival LIC-adjectives, therefore, Similarity is the principle according to which predications are percolated down the animacy hierarchy in the process of metaphor. Finally, Similarity can also be said to lie at the bottom of LIC/LY turning into the obligatory adverbial marker of English. After all, adverbs indicate that a verbal action is carried out in a particular manner. Put differently, a

46

Christiane Dalton-Puffer

relation of similarity is established between what is predicated of a noun and what is predicated of the action(s) which this subject-noun carries out. Summarizing the semantic domain of LIC we can say that we have identified four core semantic areas: Potentiality, Possessionality, Possessivity, and Similarity.

4.3.2. Semantic domain of ABLE In discussing the semantic domain of ABLE we can build on the categories introduced for LIC and be accordingly succinct, especially since ABLE uses only a subset of the functions identified for its predecessor. In a very limited number of formations, ABLE can be said to express a possessional relation between head-noun and base (profitable, honorable), although alternative modal interpretations are also feasible (honorable "can (and should) be honored"). In all other cases derivatives with ABLE are modal adjectives expressing primarily Potentiality but usually in combination with either Passivity or Negation or both.

5. Summary of domain-structure discussion The numerous points discussed in section 4 can be summed up by means of a diagram which should facilitate direct comparison of the domain structures of the two suffixes LIC and ABLE. The top left-hand box contains a list of formal input-options and enumerates different formal-morphological categories which can serve as bases in derivation. The list is only a subset of the list of all possible inputs to derivation but it represents a fairly large portion of that finite set. The top right-hand box, on the other hand, contains the set of semantic functions which the two suffixes have been found to represent. Whether these functions are semantic or cognitive in nature is a matter that needs to be decided in the context of theory-building and will be left open for now. Whether semantic or cognitive, the categories are to be understood as part of a longer list. They are to be understood as a subset of all possible semantic functions. Whether this list is language-specific or universal, closed or open are questions which we cannot hope to solve in the context of this paper. The two boxes on the next level, finally, restate which form-function combinations the two suffixes have been found in and summarize the types present in the data.

6. Conclusions The line of argument which I have been pursuing in this paper basically says that the differences in their domain structures were a decisive factor in the conflicting history of the two English suffixes LIC and ABLE. The discussion in section 4 has

The role of domain structure in morphological change

47

shown that that each individual sub-domain of LIC is more complex than that of ABLE, and Figure 1 demonstrates that this is certainly the case for its overall formal-semantic domain. LIC has more than twice as many form-function bundles as ABLE.

formal categories A ADJECTIVE B NOUN VERB PRESENT PARTICIPLE

LIC 1A IB ID IE 2B 3B 4C 4D 4E 4F

deoplic 'deep' cynlic 'kingly' ungeliefendlic 'unbelieving' unarimedlic 'uncounted' lustlic 'with pleasure' lichomlic 'bodily' unheorlic 'disobedient' halsiendlic 'deprecable' untodeledlich 'indivisible' luflich 'lovable'

functional categories t; 2. 3. 4.

SIMILARITY POSSESSIONALITV POSSESSIVITY POTENTIAL (NEGATION)

ABLE 2B profitable, honourable 4C knowable, unspeakable 4F mutable, amyable

Figure 1. The domain structures of LIC and ABLE.

I introduced the term "Simple Domain Principle" in order to capture what I regard as a general tendency or preference in derivational systems (though not a "rule" or a "law" or an "absolute prediction"). An advantage of the Simple Domain Principle over other related statements (like Aronoff s Unitary Base Hypothesis) is that it integrates the formal and semantic aspects of derivation.8 According to the Simple Domain Principle, a formative is more likely to be historically successful if it tends towards having a unitary base and uses a small set of bases. Also, a formative is likely to be diachronically more successful if it encodes a limited set of meanings, perhaps only one.

48

Christiane Dalton-Puffer

With regard to the question at hand we have to say that even if we allow for the principle to be an idealization which can be stretched by the requirements of "real life", LIC seems to constitute a problem. If the Simple Domain Principle is to be of any theoretical and explanatory value, LIC cannot be "allowed" to continue in the way it presents itself in Late Old and Early Middle English. Mclntosh (1991: 304) formulates his evaluation of Old English -lie in the following way: The variety of possible implications of -lie and the complexity of the conventions governing its correct interpretation hi different contexts may well have produced an unmanageable state of affairs in post-Conquest England. While our judgments about the complexity of Old English and Early Middle English LIC are certainly compatible, I would not necessarily agree with the implications of dysfunctionality encapsulated in the phrase "unmanageable state of affairs". Such scenarios are popular among historical linguists but should not be misread as statements of immediate causality in language change. Complicated setups may function over long periods of time. LIC itself must have been functioning for a while and there is no reason why it should not have gone on functioning.9 It may just have been the case that ABLE offered a more economic way of rendering a subset of LIC's semantic functions (and only function 4 actually took off!) simply because all the other functions could be disregarded completely in the interpretive process. The basic contention is, then, that simpler derivational domains have pay-offs in the amount of syntactic context and co-text that needs to be considered during language processing.

Notes 1. Compare Mclntosh's suggestion that "more consideration needs to be given to such matters as why adjectives which terminate in OE in -endlic, -edlic, and -enlic should... be much commoner hi their negative than in their positive forms..."(1991: 304). 2. A full discussion of this issue throws up a host of questions regarding our conception of how the meaning of derived words constitutes itself and what kinds of information feed into the process; whether, e.g., we are allowed to look beyond the boundaries of the derived word and may consider referential meaning. In other words, such a discussion asks for a coherent theory of derivational meaning. 3. On possible reasons for that cf. van Marie (1988b). Van Marie discusses the history of Dutch modal adjectives but there is no reason to doubt that his observations on their involvement in the development of elaborate written styles can be transferred to the English situation. 4. The LIC-data used in this article are drawn from the Helsinki Corpus, Old English sections 2-4 and Middle English section 1, i.e., up to 1250. The reader

The role of domain structure in morphological change

5. 6.

7.

8. 9.

49

may remember from Table 1 that the earliest Middle English period shows only one instance of -able. It, therefore, seemed reasonable to include this part of the corpus in a survey of the pre-ABLE state of affairs. Compare Dalton-Puffer (1996) for detailed quantitative information about the Middle English period. The OED states that "The original Teut. adjectives were compounds of the sb *likom "appearance, body, form...". Thus *mannlike ('manly') means etymologically 'having the appearance or form of a man' (OED s.v. LY)". This leaves one wondering, though, how a N+N compound like mann+ *likom came to be an adjective. I hope the reader will bear with a little bit of speculation at this point. We can reasonably claim that persons, i.e., concrete, animate entities, are more central to and, therefore, primary in human cognition. Some languages may, therefore, decide to mark overtly those cases where predications about primary entities are extended to secondary ones such as non-animate, or non-concrete entities. The "convention remains fully alive in Icelandic right down to the present" (Mclntosh 1991:302). For a recent appreciation of the problems concerning the role of different grammatical levels in word-formation compare Kastovsky (1995: 385-386). At least in Dutch and German, though, the cognates of LIC also lost their modal functions to new exponents, if only slightly later than in English. It would be extremely interesting to have a comparative historical survey of all the Germanic languages.

References Aronoff, Mark 1976

Word-formation in generative grammar. Cambridge, Mass.: MIT Press.

Baayen, Harald 1989 A corpus-based approach to morphological productivity: Statistical analysis and psycholinguistic interpretation. Amsterdam: University of Amsterdam. Booij, Geert E.—Jaap van Marie (eds.) 1988 Yearbook in Morphology I. Dordrecht: Foris. Bybee, Joan L. 1985 Morphology: A study of the relation between meaning and form. (Typological Studies in Language 9.) Amsterdam: Benjamins. Dalton-Puffer, Christiane 1993 "A diachronic view on morphological productivity: The case of English de-verbal adjectives", in: Wolfgang U. Dressier—Livia Tonelli (eds.), 157-168.

50

Christiane Dalton-Puffer

1996

The French influence on Middle English morphology. A corpus-based study of derivation. (Topics in English Linguistics 20.) Berlin: Mouton de Gruyter. Dressier, Wolfgang U. 1982 "On word formation in natural morphology", Wiener Linguistische Gazette 26: 3-14. 1987 "Word-formation as a part of Natural Morphology", in: Wolfgang U. Dressler et al. (eds.), 99-126 1994 "Interactions between iconicity and other semiotic parameters in language" in: Raffaele Simone (ed.), 21-37. Dressler, Wolfgang U.—Willi Mayerthaler—Oswald Panagl—Wolfgang U. Wurzel (eds.) 1987 Leitmotifs in natural morphology. Amsterdam: Benjamins. Dressler, Wolfgang U.—Livia Tonelli (eds.) 1993 Natural morphology. Perspectives for the nineties. Padova: Unipress. Hogg, Richard M. (ed.) 1992 The Cambridge history of the English language. 1: The beginnings to 1066. Cambridge: CUP. Hoinkes, Ulrich (ed.) 1995 Panorama der lexikalischen Semantik: Thematische Festschrift aus Anlaß des 60. Geburtstages von Horst Geckeier. Tübingen: Narr. Kastovsky, Dieter 1992 "Semantics and vocabulary", in: Richard M. Hogg (ed.), 290408. 1995 "Wortbildungssemantik: Ein historischer Lagebericht", in: Ulrich Hoinkes (ed.), 385-398. Kytö, Merja (comp.) 1996 Manual to the diachronic part of the Helsinki Corpus of English Texts: Coding conventions and lists of source texts. (3rd edition.) Helsinki: Department of English, University of Helsinki. Marchand, Hans 1969 Categories and types of Present-day English word-formation. (2nd edition.) München: Beck, van Marie, Jaap 1988a "On the role of semantics in productivity change", in: Geert E. Booij—Jaap van Marie (eds.), 139-154. 1988b "A case of morphological elaboration: the history of Dutch -baar", Folia Linguistica Historica 9.1: 213-234. Mclntosh, Angus 1991 "Old English adjectives with derivative -lie partners: Some semantic problems", Neuphilologische Mitteilungen 92: 297310.

The role of domain structure in morphological change

51

Schmid, Hans U. forthc. Althochdeutsches lih und Verwandtes: Untersuchungen zu einem germanischen Wortbildungstyp. Vandenhoek: Göttingen. Simone Raffaele (ed.) 1994 Iconicity in language. (Current Issues in Linguistic Theory 110.) Amsterdam: Benjamins. Szymanek, Bogdan 1988 Categories and categorisation in morphology. Lublin: Wydawnictwo Katolickiego Uniwersytetu Lubelskiego. Zbierska-Sawala, Anna 1993 Early Middle English word-formation: Semantic aspects of derivational affixation in the AB language. (Bamberger Beiträge zur Englischen Sprachwissenschaft 32.) Frankfurt am Main: Lang.

Anthony Huish: A 17th-century English grammarian Roberta Facchinetti In the immenseßeldof... knowledge, innumerable are the paths, and GRAMMAR is the gate of entrance to them all. (Cobbett 1832 [1986]: 10)

1. Introductory note Anthony Huish is a relatively little-known seventeenth century English scholar who wrote two books, Priscianus Nascens (1660) and Priscianus Ephebus (1668), aimed at "clearing and smoothing the way to the SYNTAX, both English and Latin, of the usual grammar, commonly called LILIES GRAMMAR" (Huish 1668: frontispiece). The books appear to be detailed presentations of Latin grammatical categories and syntactic notions, supported by a great variety of examples of translation from and into Latin. Indeed, Huish writes at a time when English is only just appearing on the threshold of schools as a subject for curricula, while Latin is still of extreme importance. Reflecting the situation of his time, like a great many other contemporary linguistic writers, Huish does not attempt to overshadow Latin; in contrast, he emphasizes the profitability of learning it for anyone wishing to become a real English gentleman. Having done so, however, he also appears to point out the steadily expanding role of English, by remarking on the necessity of using his mother tongue as a medium for learning Latin and, most of all, by stating that "the English dialect" is a language whose structures and rules do not always coincide with the Latin ones. Although recognizing Huish's valid contribution to the field of translation,1 the present paper will attempt to better qualify his role as an early English grammarian, by focusing on the sections of his two books which deal directly with the presentation of English grammar and syntax. To do so, it will be necessary to remark on Huish's close dependence on a previous book, A perfect survey of the English tongue, published in 1624 by John Hewes.2

2. Anthony Huish and John Hewes: A close relationship Nothing is known of either author; in his Athenx Oxonienses Anthony Wood simply mentions an Alexander Huish, a biblical scholar who lived between c. 1594

54

Roberta Facchinetti

and 1668, son of John Hewish or Huish. Judging from the similarity of the two surnames, the same historical period, and finally, as I will highlight later on, strong interrelationships between the texts of the two authors, one might surmise that Huish may well have been Hewes's son, or some other relative of his. Huish himself acknowledges that he is greatly indebted to Hewes, particularly for the compilation of the second part of his Priscianus Nascens, which he claims to be "a Revew, better ordering, augmentation and exemplifying of Mr. John Hewes his Survey of the English tongue, comprehending rules which shew a new and more easy and safe way of Translating English into Latine, then was formerly used" (Huish 1660: f4v). In the preface to the same text he even adds that the reader "may sufficiently serve himself in Mr. Hewis his own examples, by him fitted to his second and third tables of Rules" (Huish 1660: |6V). Huish also owes to Hewes's intervention a section of Priscianus Ephebus dealing with a number of considerations deemed useful for any pupil who is to translate from and into Latin; indeed, the rhyming epilogue of this part is even signed as "J.H.". Besides the open declarations of dependency quoted above, a close analysis brings to light more aspects which Huish has apparently drawn from his predecessor. Both writers overtly state that their books are simply better expositions and exemplifications of Lily's grammar; both address their books to teachers, students, or anyone willing to improve their knowledge of Latin; both deal with the so-called rules of the three concords, respectively between subject and verb, substantive and adjective, and antecedent and its relative pronoun. However, Hewes occupies most of his book with a mere series of examples; in contrast, Huish also provides a great many other rules, together with their exceptions and exemplifying clarifications. Finally, and most significantly, both authors state that their books also include some remarks on the structure of English. Hewes's intention is to enable the reader to learn Latin more easily. Huish has no different objective; he is firmly convinced that for a student "it must be far more easie and rational to translate English into Latine by rules drawn from his own tongue (which in that Language will carry along with them some evidence and demonstration of the doing) then by rules drawn from Latine, which as yet he understands not" (Huish 1660: |7r). To achieve their aim, both writers dedicate a whole section to the parts of speech as they should be classified in English.

3. The parts of speech In the table reported at the beginning of his Survey Hewes mentions two basic parts of speech, those undeclined — conjunctions, adverbs, prepositions, and interjections — and those declined — noun substantives, adjectives, relatives, persons, verbs, and participles.3 The author employs the system of signs inherited from Lily, according to which specific English words are in a biunivocal relationship with as many Latin

Anthony Huish: A 17th-century grammarian

55

constructions. So, e.g., in Hewes's classification articles are not a separate class, but simply signs of substantives, in the same way as to is the sign of an active verb and to be, when compounded, is the sign of a passive verb. Huish also employs the system of signs in his two books and mentions articles simply as words used to precede and introduce other parts of speech. He dedicates a whole section of his Priscianus Nascens to English grammar under the heading "A Brief Note of the Parts of Speech, according to the English", together with "A Brief note of Moods" and "A Brief note of the TENSES". His indications of English grammar, however, are also scattered throughout Priscianus Ephebus while presenting the rules of translation from and into Latin, although no specific section is devised for them in the book. He mentions the same parts of speech as Hewes, but does not distinguish between declined and undeclined classes, yet he is more precise in the analysis of each part. More specifically, Hewes simply mentions "noun substantives", while Huish distinguishes between proper and common nouns qualifying the former as "Noun Substantives Proper" and the latter as "Noun Substantives Common". In the same class he also introduces the adjectives as "Nouns Adjectives"; even in previous sections of Priscianus Nascens, however, he anticipates the qualification of adjectives and nouns: (1)

Mast. Put these words who or what after the Adjective, and then ask the question who or what by the Adjective, and the word that answers to the question shall be the Substantive to the Adjective. As in the English, A clear speech, Clear you know is the Adjective, because you may put man or thing after it. Now put who or what after clear, thus A clear who or a clear what? and then the word that will in this English answer to the question made by the Adjective clear, is the word speech, thus, a clear who, or a clear what? the answer is, a clear speech. And so you see that the word speech is the Substantive to clear. (Huish 1660: sect. 8,p.l4). 4

As in Hewes's Survey, adjectives are defined as a part of speech which cannot stand on its own and may be followed by "man or thing"; hence, interestingly enough, both authors define this category according to the English patterns rather than the Latin ones; this is further testified in (1), where "a clear speech" is the example provided.5 In the category of adjectives Huish also mentions "adjectives used substantively": "Remember here, that Adjectives are frequently put Substantively; that is, without Substantives: And that either in the Masculine gender, when the word Man is understood; Or in the Neuter, when the word thing" (Huish 1668: 7) Moreover, faithfully following Lily's grammar (Lily 1567 [1945]: the second concorde), he affirms that pronouns and participles can be used adjectivally: "The Adjective, whether it be noun, pronoun, or participle, agreeth with his Substantive in three things, in Case, Gender, and Number" (Huish 1660: sect. 8, p. 15).

56

Roberta Facchinetti

Undoubtedly, the presentation of case, gender, and number agreements between adjective and substantive leads the reader to conclude that the author is still so greatly Latin-oriented as to be unaware of the inconsistency of his terminology. An unclear categorization is provided also for pronouns; on one hand personal pronouns are qualified as "persons", further distinguished into first, second and third pronouns of singular or plural number; hence, the category appears to include only /, thou, he, we, ye, and they, although you is also employed throughout the texts, as (2) testifies: (2)

You write that a you will [or a thou wilt come to me.]

Scribis a te ad me venturam. (Huish 1660: sect. 12,p.3)

On the other hand Huish claims that in the Latin example Tu tibi venerabilis, thou is "a Pron.subst. of the first declen. and 2. per. tis nom. sing.". (Huish 1660: sect. 3, 60). Consequently, he also makes pronouns a subclass of substantives. The Latin focus of his texts strongly affects his definitions and leads him to qualify the possessives like meus, tuus, and suus as "pronouns", "persons", or "adjectives", according to their function in the Latin sentences illustrated.6 As far as "relatives" are concerned, rather than qualifying them with a separate definition, Huish mentions them in more than one group, mainly according to their different translations into Latin: "These words, who, which, whose, whom, whosoever, whomsoever, what, whatsoever, and that when it may be turned into which, are Relatives, to be made by quis, qui, quicunque [sic] &c" (Huish 1668: 25). The same occurs for such, so great, so many, when followed by as, and for whether when indicating whether of these or whether of the both, or even neither meaning neither of the both. Then, Huish mentions conjunctions, but not as a class; rather he presents them in an inaccurate way: "If you doubt of any word in the beginning of any clause of speech in the English, it is usually a Conjunction" (Huish 1668: 83). Adverbs, too, are simply qualified as words ending in -ly and joined to verbs, or as words which are not nouns, verbs, or participles: (3)

|

32. That ye doubt of in the middle or end of any clause. 33 Of Time ^

All WOrds

> f

34ofp/£7ce

*

not 5e n

' ' g Nouns, Verbs or Participles, as ffv.r.^r. H*re,There._

adverbs

35. That end in ly & which joyn with verbs. (Huish 1660: sect. 12, p. 9)

Anthony Huish: A 17th-century grammarian

57

No doubt there is a lack of professionalism, which also accounts for the fact that Huish acknowledges the existence of prepositions and interjections, but is not completely able to provide an adequate definition, so they simply must be studied by heart so as not to use them incorrectly. The same occurs for participles, which are clearly distinguished from other parts of speech — in the example above from nouns and verbs — but no further clarification is provided for their description. Despite such uncertainty of classification, example (3) leads us to acknowledge that Huish exhibits a certain etymological knowledge allowing him to point out that words ending in -ly are usually adverbs, without failing to recognize, later on, that if the above words in -ly are followed by a substantive they are no longer adverbs but adjectives; that same knowledge of word-formation leads him to produce examples of zero-derivation as regards adjectives and adverbs: (4) And so before Adverbs, derived from Adjectives; as No man regards how well, Nemo quam bene vivat, sed quam but how long he may live. diu curat. Sen. (Huish 1668: 31) A similar approach is followed when Huish moves on to verbs. He does not provide a descriptive definition of this part of speech and simply remarks that the learner must be able to identify "the several kinds of verbs... their Voices, Moods, and tenses" (Huish 1660: f6r); he distinguishes between active, passive, neuter, impersonal, and deponent verbs. While the last two categories clearly testify to the Latin influence on his categorization, with regard to the first three it is interesting to notice that his lack of clear definition leads him to introduce verbs in different sections and in various ways, thus indicating Huish's concern both for Latin and for English. Let us consider, e.g., his presentation of active and neuter verbs: (5)

a. Spoliaf, robs, is a verb. Active because it ends in ο and signifies to_do, may take r to make it a Passive. (Huish 1660: sect. 3, p. 66) b. Nocet, hurts, is a verb neuter because it endeth in o, and may not take r to make it a passive. (Huish 1660: sect. 3, p. 69)

The formal features adopted for the descriptions above are directly derived from Lily, who also pointed out that an active verb "endeth in o, and betokeneth to do: as Amo, I love: and by putting to r, it may be a passive", and that "a verb neuter endeth in o, or m, and can not take, r, to make him a passive" (Lily 1567 [1945]: Bii). The standpoint appears to be different, however, in other sections of Priscianus Nascens, where, enlarging the definitions previously provided by his predecessor Hewes, Huish remarks that: "To before a verb is the signe of the Infinitive mood Active" (Huish 1660: sect. 12, p. 3) and that "Am, is, art, are, was, wast, were, wert, and other Englishes of sum, when they are double words, or come before the English of a Participle of the preter tense, are signes of the Verb

58

Roberta Facchinetti

Passive" (Huish 1660: sect. 12, p. 5). He does not qualify the passive verbs with clear definitions, but views the verbal forms from an English perspective, avoids the Latin translation, and simply identifies this class by means of the formal aspect of the verb to be followed by the participle of the verb in question. Indeed, this is a further hint of Huish's gradual, steady shift towards English as the reference language for his grammatical descriptions.

4. Moods and tenses In the qualification of tenses and moods Huish is greatly indebted to Hewes, although, once more, he enlarges and better qualifies his definitions. Influenced by the Latin categories, both authors exhibit five tenses all qualified by formal features allowing them to be identified in English by specific "signs", which also enable the reader to choose the exact corresponding form in Latin. So, the preterimperfect is identified by the signs did, was, and were, but also by might, would, and should; the preterperfect is identified by the signs have, hast, and hath, the preterpluperfect by the sign had, and the future tense by shall and will. As regards the present tense, Huish simply states that it is employed in all other unidentified cases: "In all other your Speeches your verb is ever of the Present tense" (Huish 1660: sect. 12, p. 12). The same imprecision is to be retraced in the analysis of moods. Following Hewes's example, Huish identifies six moods: imperative, indicative, optative, potential, subjunctive, and infinitive. The indicative mood is qualified after all other moods, simply by saying that "in all other speeches, your Verb must be the Indicative Mood" (Huish 1668: 106). All other moods are identified by formal features; more specifically, the infinitive mood is always to be retraced by means of the particle to before the verb, or by the fact that in two adjacent verbs the second one must be in the infinitive, despite the absence of to. The optative mood is identified by means of the expressions Would God, I pray God, and God grant that, while may, might, would, and should are not recognized as auxiliaries yet, but simply as signs of the potential mood. The imperative mood is also categorized semantically as the mood employed "if you bid, or command" (Huish 1668: 98), and let is an unquestionable sign of its English form. The subjunctive mood, in turn, is identified in English by words or phrases like although, as though, and as soon as, or even by a relative: "(Rule 55.) A Relative usually will have after it, a Subjunctive Mood" (Huish 1668: 104). In the above quotation, at first sight the adverb usually leads the reader to judge Huish's definitions as inaccurate and faulty, but a deeper analysis shows that this and other words like frequently and often, as well as the modal will, are quite possibly employed by the author with a rhetorical purpose, i.e., to anticipate and introduce all the exceptions to the rules exhibited, particularly in Priscianus Ephebus. Hence, rule 55 is immediately followed by rule 56: "But yet, here also

Anthony Huish: A 17th-century grammarian

59

you shall frequently find an Indicative Mood, following the Relative, but always if a Question be asked" (Huish 1668: 105).

5. Huish's teaching methodology Huish's method of presentation allows him to lead his pupil step by step, uncovering a pedagogical imprint which is hardly recognizable in Hewes's Survey. Most of the words employed hi the rules and in the examples are collected as entries of indexes and dictionaries, which is very useful for the reader, who is thus saved from having to buy any other dictionary. Huish insists on the graduality of presentation, on the fact that the words employed in exemplifications do not include any rule which has not been presented, so as not to confuse the reader; he emphasizes the importance of the pupil's role in the planning of his book, he focuses on the importance of learning a number of rules by heart, on some suggestions for the teacher, and finally on the morality of explanatory examples, which would allow even "a childe of middle capacity" to improve both his knowledge of Latin and his good manners. This pedagogic sensitivity is totally lacking in Hewes's work, which is structured in a detached way and does not include any comments on the presentation of moods, tenses, or cases. Unlike Hewes, Huish is greatly concerned with the learning and teaching aspects of his books. He even remarks on the different ways of teaching the rules presented, by means of dictations and frequent translation exercises; indeed, his two books, as even the titles suggest, are purposefully devised in such a way as to be complementary, since all the rules of syntax presented in the second part of Priscianus Nascens are further developed, exemplified, and specified throughout the volume Priscianus Ephebus, i.e., Priscianus "grown a youth". Furthermore, in Priscianus Nascens the rules of the three concords are presented in the form of a dialogue between pupil and teacher, as examples (6a) and (6b) testify: (6)

a.

Schol.

What must I do, when I have found the principal verb?

Mast.

You must carefully look out the Nominative case to the verb.

Schol.

But how shall I know which word in the English is the Nominative case to the verb?

Mast.

Put these words who or what before the verb, and ask the question who or what by the verb. And then that word which in the English answereth to the question

60

Roberta Facchinetti

so made by the verb, shall be the Nominative case to the verb. (Huish 1660: sect. 7, p. 3)

b. Schol. What must I do that I may find out the Antecedent to the Relative? Mast,

Put who or what after the Relative, and then ask the question who or what by it; and then the word that answers to the question made by the Relative shall be the Antecedent to it.

Schol.

I pray, shew me how.

Mast.

You shall see it in this English. (Huish 1660: sect. 7, p. 20)

The imperative form and the frequent use of the modal verb must disclose Huish's intent to lay down specific rules in a somewhat basic grammar; moreover, the dialogical pattern, the use of the personal pronouns / and you, as well as some expressions like / pray and you shall see, and the frequent use of co-ordinating conjunctions like and and but create a colloquial environment which is also typical of Priscianus Ephebus, where the rules are separated and numbered, but at the beginning of many of them we often read: "Remember here", "note also that", "you shall meet", "but yet", "yet shall you find", "yea, you shall also find".

6. Conclusion In conclusion, we cannot overlook the fact that Anthony Huish's two books, Priscianus Nascens and Priscianus Ephebus, owe many of their structures and contents to Lily, to John Hewes, and, more generally, to the pedagogic and linguistic tradition common in 17th-century England; the author himself remarks that the rules presented "are no innovations upon, nor contradictions to the former and received wayes of teaching to make Latine; but a clearer illustration of them onely to the English learner" (Huish, 1660: f6v-t7r)· Although we cannot claim that Huish distinguished himself as an original thinker, nonetheless, he skillfully exploited his predecessors' ideas; the linguistic insights scattered through both books, together with Huish's pedagogic concern, and also, most of all, his manifest insistence on the necessity of writing an English book to learn Latin, allot him a dignified place in the linguistic environment of Renaissance England.

Anthony Huish: A 17th-century grammarian

61

Hence, I believe we can confidently claim for Huish's volumes what Michael states with regard to early English grammar books in general, namely that, although they "were unimportant works individually, together they formed part of the educational history of the country" (Michael 1970: vii).

Notes 1. For further details cf. Facchinetti (forthc.). 2. Not all scholars acknowledge the importance of Hewes's book among the early English grammars; indeed, Michael regards it as "the first textbook to attempt a basic grammar" (Michael 1970: 155), while Poldauf does not even mention it in his bibliography of the early grammars and grammatical treatises published before 1800 (Poldauf 1948: 21-39). Besides Michael, for further reference concerning A perfect survey cf. Algeo (1985: 198-199); Padley (1985: 66-68, 158-160, 165, 168-169); Salmon (1988: 89-92). 3. Among verbs, the gerund and the supine are individually identified. 4. Huish appears not to be wholly systematic in the qualification of the various sections of Priscianus Nascens; consequently, to avoid confusion, in the present paper the bibliographic details of the quotations from this work faithfully follow the index provided by Huish at the very beginning of his book under the heading "The several heads handled in the Book". 5. Quite different, as remarked by Michael (1970: 155), was Lily's position, whom both authors claim to follow; Lily simply affirms that an adjective is a noun which requires to be followed by another word, and, biased by his Latin standpoint, brings the examples bonus and pulcher. 6. The Latin tradition itself is unstable in the categorization of this part of speech and appears at times to qualify it as a separate class, while at other times it mentions them among nouns. 7. For further reference cf. Facchinetti (1996).

References Algeo, John 1985

"The earliest English grammars", in: Mary-Jo Arn—Hanneke Wirtjes (eds.), 191-207. Am, Mary-Jo—Hanneke Wirtjes (eds.) 1985 Historical and editorial studies in Medieval and Early Modern English. Groningen: Wolters-Noordhoff. Cobbett, William 1832 [1986] A grammar of the English language. A facsimile reproduction with introduction by Charlotte Downey and Floor Aarts, Delmar. New York: Scholar's Facsimiles and Reprints.

62

Roberta Facchinetti

Facchinetti, Roberta 1996 "Bilingual lexicography in 17th-century England: The case of Anthony Huish", Linguistica e fllologia 3: Quaderni del dipartimento di linguistica e letterature comparate (Bergamo: Universita degli Studi di Bergamo): 209-229. forthc. "Anthony Huish's Priscianus Ephebus: A seventeenth-century handbook of translation", in: English diachronic translation. Proceedings of the VII National Conference on History of the English Language, 2-4 October 1995, Gargnano, Italy. Hewes, John 1624 A perfect survey of the English tongue, taken according to the use and analogie of the Latine. London: Printed for William Garret. Huish, Anthony 1660 Priscianus Nascens or a key to the grammar school. London: Printed for William Garret. Priscianus Ephebus or a more full and copious explanation of 1668 the rules of syntax. London: Printed for William Garret. Lily, William 1567 [1945] A shorte introduction of grammar, with an introduction by Vincent J. Flynn. New York: Scholar's Facsimiles and Reprints. Michael, Ian English grammatical categories and the tradition to 1800. 1970 Cambridge: CUP. Padley, G. A. Grammatical theory in western Europe 1500-1700: Trends in 1985 vernacular grammar I. Cambridge: CUP. Poldauf, Ivan On the history of some problems of English grammar before 1948 1800. (Prague Studies in English 55.) Prague: Näkladem Filosoficke Fakulty University Karlovy. Salmon, Vivian Studies in the history of the language sciences: The study of 1988 language in 17th-century England. Amsterdam: Benjamins. Vorlat, Emma The development of English grammatical theory 1586-1737. 1975 Leuven: Leuven University Press. Wood, Anthony Athenx Oxonienses, edited by Philip Bliss. (The Ecclesiastical 1848 History Society Publications 3.) Oxford: For the Ecclesiastical History Society.

John Bullokar's "Termes of Art" Maurizio Gotti

1. Introduction The contents of John Bullokar's An English expositor (1616) are clearly identified in the second part of the title of his work, where he specifies that the dictionary aims to teach "the interpretation of the hardest words vsed in our Language".1 In the introductory part of his dictionary, Bullokar distinguishes the hard words2 that he has listed into three categories. The first includes "the great store of strange words, our speech doth borrow, not only from the Latine, and Greeke, (and some from the ancient Hebrew) but also from forraine vulgar Languages round about vs"; this category, therefore, refers to the many foreign loans that had become so common in the English language in Bullokar's times and that had caused the outburst of the "inkhorn controversy". The second category of hard words listed in An English expositor consists of "sundry olde words now growne out of vse". These are words mainly appearing in literary works written in previous centuries — particularly by Chaucer — which had been included in the dictionary so as to help its users to interpret these archaic terms correctly.3 To underline the fact that such words are reported to aid the decoding of literary texts but in the writer's opinion they should not be used in everyday speech, Bullokar marks each of them with a preceding asterisk. The third category of hard words listed in An English expositor contains "diuers termes of art, proper to the learned in Logicke, Philosophy, Law, Physicke, Astronomie, etc. yea, and Diuinitie itselfe, best knowen to the seuerall professors thereof. The author's aim in his adoption of these terms is to "open the signification of such words, to the capacitie of the ignorant, whereby they may conceiue and vse them as well as those which haue bestowed long study in the languages". Bullokar, indeed, is convinced that as more and more specialists have turned to the English language for the writing of their books, the meaning of the latter should be made accessible to the readers for whom such works are intended; in his opinion, a dictionary of hard words should fulfill such a function.4 In favoring a more general acceptance of such new terms — mainly of foreign origin — Bullokar is directly involved in the "inkhom controversy" in support of the borrowers of words taken from other languages, as he considers their behavior "sometime necessary by reason our speech is not sufficiently furnished with apt termes to expresse all meanings". The three categories of hard words identified in the introductory part do not have identical coverage in Bullokar's dictionary. Indeed, the first category — i.e., the one including many of the foreign loans adopted by the English language — is the

64

Maurizio Gotti

most widely represented (and is approximately two thirds of all the entries); this is due to the fact that this category does not only include the many "strange words" commonly used in the general language, but also several terms of a semi-technical nature which are common to various specialized branches. The second category of hard words — i.e., the one referring to archaic words — is the smallest of the three, only including 140 words out of a total of over 4,000. The third group — i.e., that of "termes of art" — contains about one third of the entries of the dictionary.5 It is this third category of hard words that will constitute the object of the analysis of this paper. The types of Bullokar's "termes of art" will be taken into consideration, as well as their origin and field of specialization. A further element of investigation will be the way the author deals with the various entries and the techniques employed in the explanation of their meanings. Another part of the paper will be devoted to the appreciation of the degree of representativeness of Bullokar's entries of the specialized English lexis of his time, hi particular when compared with the contents of other contemporary dictionaries of hard words; for this purpose a list of specialized terms belonging to the medical field will be made use of. An analysis of how Bullokar deals with the explanation of the semantic value of the main affixes of foreign origin will conclude the paper and will point out the great value of this dictionary in terms of creating greater awareness in its user of the processes of derivation found in the foreign loans included in An English expositor.

2. The specialized branches covered by Bullokar's dictionary As a first step in our analysis we shall take into consideration the specialized fields from which Bullokar's "termes of art" are drawn. The range of subjects covered by this dictionary is very wide. In his definitions of the single entries, however, Bullokar does not usually specify the branch to which they belong; only in a very limited number of cases does he make such a specification; this is the reason why fewer categories than are actually dealt with are mentioned in this work. Moreover, the terms used in the dictionary to identify the specialized fields of some of the words may sometimes be misleading for the modem reader, as in referring to such branches Bullokar employs the terminology in use in his time, which does not always correspond to what is used nowadays. The main disciplines mentioned in Bullokar's definitions are: Lawe, Physicke (term referring to what we nowadays call "Medicine"), Hunting, Philosophie (which then also included the area of Natural Philosophy, i.e., of the physical sciences), Astronomy, Heraldry, Logicke, Cosmographie. Our analysis has shown that many more branches are actually covered by Bullokar's terms, which would correspond to the following modern disciplines: Religion, Rhetoric, Linguistics, Botany, Zoology, Geography, Mineralogy, Physics, Politics, Military Tactics, and Geometry. Although terms specific to these fields of knowledge were beginning to appear in English books, such sectors had not yet become completely independent

JohnBullokar's"TermesofArt"

65

from the traditional and officially recognized specialized disciplines, whose names were, therefore, retained in Bullokar's definitions. It is interesting to note, however, that the great progress made in such fields during the sixteenth and early seventeenth centuries is not only well attested by the many recently coined words included in Bullokar's dictionary, but also by the appearance in that work of several terms referring to the new fields of knowledge which were growing so rapidly in those years and becoming independent disciplines. A confirmation of the great expansion of new specialized branches typical of that period can also be found in Bullokar's dictionary; a perusal of the entries beginning with the letter A shows several terms denoting new specialized sectors, not belonging to the range of disciplines used by the author in his definitions: Alchymie, Anatomie, Architecture, Arithmetike, Astrologie, and Astronomie. Some of these terms might not sound very innovative for Bullokar's times,6 but the fact that they are included in his text means that — although some had been coined even a few centuries before — they had not become very popular among English speakers. As regards the space given to each discipline in the dictionary, greater coverage was still offered to the traditional fields (such as Law, Religion, Rhetoric, and Medicine); however, a wide coverage was also provided for subjects such as Botany, Zoology, Geography, and Mineralogy, due to the great number of new words referring to plants, animals, minerals and geographic terms concerning far-off places, particularly those located in the New Worlds being discovered and colonized in those years.7 Here are a few examples of such words: Armadillo. A beast in India of the bignesse of a young pigge, couered ouer with small shels like vnto armour [...]. China. A hard knotty roote brought out of the East Indies, of a reddish colour

Coca. An hearbe in India [...]. Besides terms referring to new discoveries, the dictionary also contains several geographic proper names (such as Africa, Alps, Asia, Europe) denoting places which were undoubtedly well-known in early seventeenth-century England, but which were probably included by Bullokar so as to give a more encyclopedic character to his dictionary. In order to make an approximate assessment of the different coverage assigned to the various branches in An English expositor, a statistical analysis of the "termes of art" listed under the letter A of the dictionary has been carried out; the results are the following: Law (52 terms), Religion (27 terms), Rhetoric and Linguistics (10 terms), Botany (9 terms), Medicine (8 terms), Zoology (7 terms), Geography (6 terms), Mineralogy (6 terms), Astronomy (3 terms), Heraldry (3 terms), Physics (3

66

Maurizio Gotti

terms), Politics (3 terms), Hunting (2 terms), Military Tactics (2 terms), Geometry (1 term), and Logic (1 term).8 These numbers, however, should only be considered as indicative of the areas covered rather than as precise estimates of the importance acquired by the various specialized branches in Bullokar's times or as an index of productivity of each discipline in that period. Indeed it should be remembered that the words included in An English expositor were not collected meticulously after a thorough examination of all the specialized texts available; instead, in the selection of his entries, the author mainly borrowed the terms which he thought suitable from the dictionaries and specialized glossaries he had come across in his unsystematic search for appropriate sources for his dictionary.9

3. Bullokar's definitions As we have seen, John Bullokar's aim in writing his dictionary was to provide a tool for the correct interpretation of several new terms, foreign loans, and archaic words. It was, therefore, justified on his part that he should be so careful in the preparation of his definitions of the lexemes he had chosen and that he should provide them with "svndry Explications, Descriptions, and Discourses". In the wording of his definitions Bullokar adopted several techniques. The simplest definitory procedure corresponds to the provision of a mere synonym in English. This method, however, was particularly suitable for the hard words belonging to the first two categories mentioned in the introductory part of our paper — i.e., those including foreign loans or archaic words — but was not very adequate for the explanation of the meaning of specialized terms, as most of them had been borrowed from other languages since no correspondent equivalents existed in English. There were, however, a few cases in which it was possible to provide an English synonym for a "terme of art"; in such cases the foreign term usually provided a specialized connotation, while the native equivalent was to be commonly employed for general reference. Here is an example: Angle. A corner. However, the most frequent technique employed in the explanation of Bullokar's "termes of art" was usually the recourse to a definition, as can be seen in the following example: Abbett. To helpe or assist one in euill. In contrast with the procedure adopted by Cawdrey in his dictionary of hard words, Bullokar usually provided very lengthy definitions for his entries, so as to strengthen the decoding function of his work. The following quotation, e.g., provides a very accurate explanation of the meaning of the entry:

John Bullokar 's "Termes of Art"

67

Balme. A precious iuice or liquor, otherwise called Balsamum, or Opobalsamum. It droppeth by cutting out of a little lowe plant (about a yard high) hauing leaues like Rue, but whiter, which plant groweth in Egypt, and some places of the holy Land. This iuice is somewhat like to oyle, but more clammie, and inclining to a certaine rednesse. It hath a strong smell, and is not pleasant in taste: Being put into a vessell of water, it will sinke downe to the bottome like a round pearle, without breaking, and may bee taken vp againe with the point of a knife. It is an excellent medicine to take any scar out of the body, and for diuers other purposes, but very costly and rarely gotten, Saladinus writes that there was but one vineyard of these in the whole world, and that belonged to the great Türke. To make his explanation of the entry more clearly understandable, Bullokar often added exemplifications, as the following case shows: Allegorie. A sentence consisting of diuers tropes which must be vnderstood otherwise then the litterall intepretation sheweth; as when Saint lohn Baptist speaking of our Sauiour, Matth. 3 said: Whose fanne is in his hand, and he shall make cleane his floore, and gather the wheat into his borne but the chqffe he shall burne with vnquenchable fire: The meaning whereof is, that Christ being supreme ludge of all, shall separate the good from the euill, rewarding the one in heauen, and punishing the other in hell fire. If the entry had more than one meaning (e.g., a general meaning and a specialized meaning), Bullokar was careful to point out the two semantic values of the entry: Abate. To make lesse: In our common Law it signifieth, to enter into any inheritance, before the right heire take possession, with intent to keepe the said heireout of it. This procedure seems to constitute a contradiction to one of the principles specified by Bullokar in his "Introduction to the reader", which states that "if a word bee of different significations, the one easie, the other more difficult, I onely speake of interpretation of the hardest". There is, indeed, no contradiction on the author's part, as his statement mainly concerns polysemic words of Anglo-Saxon origin, as the examples he provides of his principle (i.e., Tenne, Girle, and Garter) seem to suggest. In some cases Bullokar was not certain of the true meaning of a term; it is interesting to note that on such occasions the author honestly admitted his inability to provide a definite explanation of the entry and, therefore, expressed any possible interpretation with great caution. His definition of the lexeme beauer provides an excellent instance of the author's behavior:

68

Maurizio Gotti

Beauer. In armour it signifieth that part of the hehnet which may bee lifted vp, to take breth the more freely: It is also a beast of very hotte nature, liuing much in the water. His two forefeet are like the feete of the beast called Gattus, (as Joannes de Sancto Amanda writeth:) but what this Gattus is, I doe not well vnderstand, only I suppose it to be an Otter [...]. However, before providing his own interpretation, it was customary for Bullokar to report the meanings commonly given to that term — even when he did not find them very convincing — as can be seen in the following case: Boras. A white substance like vnto saltpeter, wherewith goldsmiths vse to solder gold and siluer: some write it is the gumme of a tree, which is very vnlikely; others affirme it to bee made of old lees of oyle, by art and drying in the sunne brought to be white; notwithstanding I suppose it rather to be a minerall. In his desire to strengthen the didactic function of his dictionary, Bullokar sometimes also added the mention of the origin of the concept or procedure to which the term he was explaining referred: Abiure. To sweare or forsweare: a terme sometime vsed in Lawe, when one hauing committed a capitall offence flyeth to a Church, or Churchyard, and chooseth rather perpetuall banishment: viz. to abiure the Realme, then stand to try all of Lawe. This Law was instituted by S. Edward the Confessour in fauour of life, but now is not in vse. At times, besides providing the correct specialized value of the term he was defining, Bullokar also added the metaphorical connotation that such a term had acquired. Here is an example: Chameleon. A little beast like a Lizard, hauing a rough scaly skin, straight legs, sharpe clawes, a slow pace like a Torteyes, and a long wreathed taile: Hee changeth himselfe quickly into any colour that he fitteth vpon, except white & red: wherefore men that are inconstant and fickle, are sometime called Chameleons [...]. On other occasions Bullokar explained the origin of the term; an example of such comments can be found in the following definition of the word canker: Canker. A hard swelling in the veines, being ouercharged with hot melancholy humors. It is called a Canker, because the veines so swollen are like vnto the clawes of a Crab [...].

John Bullokar 's "Termes of Art"

69

Bullokar often made comments also on the linguistic form of the term he presented, and at times he provided its English translation: Apocalypse. A diuine book written by Saint lohn Euangelisf, while he was banished in the He Pathmos: so called because it conteineth many profound mysteries there reuealed vunto him. In English it signifieth a Reuelation. Only in very few cases did Bullokar specify the language from which the term had been taken; this occurs, however, for almost all the terms deriving from Hebrew and Greek and belonging to the religious semantic field. Example: Chrisme. A Greeke word, signifying an Oyntment: Sometime it is taken for a white linnen cloth, wrapped about an infant after it is newlie christened. Apart from these two languages, the words included in Bullokar's dictionary had been drawn from various others — as the author himself confirms in his "Introduction" — the sources of his terms were Latin and French, as well as other modern languages such as Spanish and Arabic.

4. The representativeness of Bullokar's entries The criteria of selection adopted in the choice of terms and the contents of their definitions may offer us a good chance to interpret the degree of evolution of the world of knowledge as expressed in Bullokar's dictionary. As a matter of fact, the picture that can be drawn from a reading of this work is that the terminology of many of the branches dealt with in it was involved in a process of rapid growth, yet it still needed some form of systematization. This impression is backed up by the existence of doublets within the dictionary to express identical concepts. The following examples — drawn from the medical field — confirm the existence of two different words in the English language to refer to the same human organ: Beades of Saint Elline. [...] They are of great vertue against griefs of the stomack, as also of the kidnies or reines. Capers. [...] The roote hereof is much vsed in Phisicke, against obstructions of the spleen or milte [...]. The impression that specialized disciplines in this period were still in need of a process of systematization and should retain only scientifically-based terms and eliminate unsound traditional explanations is confirmed by the presence in the definition of some entries of a reference to mythological or legendary interpretations of their meanings and origins. Here is an example:

70

Maurizio Gotti

Aconitum. A venemous herbe, hauing a root much like to a Scorpion, and thining within like alabaster. Poets faine that Cerberus the three headed dogge of hell, being dregged vp in a chaine of Adamant by Hercules, did cast some of his fome vpon this herbe, whereby it became so venemous. Even when legends and myths were not used in the explanation of certain concepts, the inadequacy of the adoption of old-fashioned definitions is confirmed by the quotation of classical or medieval naturalists, physicians or philosophers (such as Galen, Pliny, and Avicenna) as sources. Here is an example of one such case: Camphire. A kinde of Gumme, as Auicen writeth. But Platearius affirmeth it to be the iuice of an herbe. It is white of colour, and cold and dry in operation. Moreover, the impression of the scientific world that emerges from the reading of this dictionary is still of one that devotes most of its efforts to the processes of observation and description rather than experimentation. Indeed, in this dictionary there are very few terms referring to instruments, and those that are recorded are still linked to traditional activities, such as building and military campaigns. Here are a couple of examples: Calthrope. An instrument vsed sometime in Warre [...]. Capstand. An instrument to wind vp things of great weight: some call it a Crane. Also the terms expressing properties reported in An English expositor usually concern general concepts, and are, therefore, not very highly specialized, but rather common to several disciplines, as the following examples show: Capacitie. Aptnesse to receiue and hold. Carnalitie. Fleshlinesse. Cauitie. Hollownesse. Celeritie. Swiftness, speed. Similarly, the entries referring to processes consist mainly of semi-technical terms, and are suitable to be employed by specialists belonging to various disciplines, as the following examples confirm: Calcination. A burning, a turning into ashes. Calculation. An account, a reckoning. Calefaction. A making warme.

John Bullokar 's "Termes of Art"

71

It must be pointed out, however, that the picture offered by An English expositor is outdated with respect to the actual state of specialized research of that time, as the author often made use of glossaries and dictionaries published a few decades earlier. Indeed, if we look at several terms belonging to the specialized disciplines dealt with by Bullokar and coined in the decades preceding the publication of his dictionary, we can see that many of them are not included in that work. Here below is a list of terms belonging to the branch of medicine, with the date of their first publication as derived from Murray et al. (1933) and Burchfield (1980);'° each word is followed either by a + or - sign: the + sign indicates the presence of that entry in the dictionary mentioned at the top of each column, while the - sign indicates its absence:

spleen (c. 1300) palate (1382) genitals (1382) sperm (1386) artery (1398) colon (1398) esophagus (1398) semen (1398) thorax (c\400) trachea (c!400) wvM/a(cl400) testicle (cl 425) embryo (1477) fracture (1525) nerve (1531) muscle (1533) abdomen (1541) cartilage (1541) cavity (1541) rectum (1541) tendon (1541) ulna (1541) cranium (1543) vulva (1548-77) larynx (157S) pancreas (1578) scapula (1518) skeleton (1578) intestine (1597) scrotum (1597)

Bullokar

Cawdrey

Cockeram

+ + + + - ll .

+

+

+ + -

+ + +

.

.

+

+ + +

.

.

+ + + + + + .

-

-12

72

Maurizio Gotti

This list shows the unsystematic nature of Bullokar's selection of items; we can rightly suppose that the author neglected all the words marked with a minus sign not because he did not know of their existence — as in the title-page of his dictionary he prides himself on being a "Doctor of Physicke" — but because they were not in the sources he had made use of. A comparison of the items in the list above with those included in the first hard word dictionary — i.e., Robert Cawdrey's A table alphabetical! (1604), which was used as one of the sources in the compiling of An English expositor — shows that in the case of medical "termes of art" Bullokar did not stick to his predecessor's policy; indeed, of the ten medical terms included in An English expositor only three are present in A table alphabeticall, while only one (genitalles) is reported in Cawdrey's text but not in Bullokar's. This quantitative discrepancy might be explained by the different number of entries in the two dictionaries: 2543 in Cawdrey versus 4249 in Bullokar; another reason may be found in a greater interest shown by the author in the inclusion of "termes of art" in his dictionary. This second explanation finds greater support in the comparison of Bullokar's work with the third hard word English dictionary, i.e., Henry Cockeram's (1623); in spite of the fact that in The English dictionary the number of entries is even greater (5836 words), that of the medical terms selected in our list is smaller than in An English expositor: only seven words are reported instead of the ten of Bullokar's dictionary. This means that Bullokar's interest in "termes of art" was more prominent, as is shown also by the lengthier explanations that accompany the various items. The following comparison between the definitions of the same term in the first three hard word English dictionaries provides a very clear example: splene, milt (Cawdrey 1604) Splene. The milte of man or beaste: which is like a long narrow tongue, lying wider the shorte ribbes on the left side, and hath this office of nature, to purge the liuer of superfluous melancholicke blood: sometime it signifieth anger or choler. (Bullokar 1616) Spleene, The milt of man or beast. (Cockeram 1623)

The quotation above shows the brevity of Cawdrey's definitions, which usually consisted of a single equivalent or very few words of paraphrase; on the contrary, Bullokar's dealing with the entry splene did not only provide a native equivalent for the term of foreign origin, but also offered an indication of its position in the human body and an explanation of its function; moreover, it expressed its metaphoric meanings; Cockeram's definition, instead, reflects his usual behavior, providing shorter explanations than Bullokar's, often consisting in the mere simplification or shortening of those reported in An English expositor.

John Bullokar 's "Termes of Art"

73

5. Affixation in Bullokar's "termes of art" The reading of the definitions of some of the entries in Bullokar's dictionary shows that there was still a gap between the specialized terminology of other languages — particularly Latin — and the English tongue; the latter still lacked some of the "termes of art" which already existed in other foreign codes. The following quotation provides a confirmation of this state of inferiority; in dealing with a medical term, Bullokar was not able to make exclusive use of English words to express his definition, and had to borrow two Latin expressions to refer to the two types ofmeninges he was describing: Meninges. Thinne skins in which the braine is contained. There are two such skinnes: one called by Phisitians, Dura mater, which is the stronger of the two, and next vnto the scull. The other named Pia mater, is within this first, being more tender and fine, and close wrapping the braine it selfe. If any of these skinnes bee wounded, it causeth speedy death. This lack of terminology, however, was not generalized; indeed, there were disciplines which were very rich in specialized words: this was particularly true for the oldest branches — such as Law, Religion, Medicine, and Rhetoric — which possessed not only the lexemes to refer to specialized concepts, but also all the words relating to a particular semantic area. A couple of examples drawn from the legal field illustrate this point: to express the concept of "the disposing of a dead mans goods, that made no will" Bullokar did not only report the appropriate legal term (Administration), but also its related verb (Administer), and the name to refer to the person "to whom the Ordinary committeth in charge the goodes of a man dying without will" (Administrator). The same lexical completeness can be seen in the semantic area of legal agreements, where five terms were mentioned in An English expositor, the name to refer to the agreement itself (Arbiterment), the adjective pertaining to that concept (Arbitrary), the verb to indicate the making of the agreement (Arbitrate), and two words to designate the judge proposing the agreement (Arbiter and Arbitratour). The mention of these groups of words pertaining to the same semantic field highlights several examples of the affixes most commonly used in the classical languages to form derived forms and imported into English by means of "inkhorn terms". The process of adoption of these affixes by the English language usually consisted of two phases: at first the various loans were adopted as single words and commonly employed without any specific awareness of the semantic value of the various parts forming the borrowed terms. Little by little the same affixes appearing in different words were isolated, semantically identified, and consciously re-employed by English scientists in the creation of their new terms. In the formation of this awareness of the semantic value of various affixes of foreign origin, Bullokar's dictionary was to play an important role, as its definitions gave a systematic rendering of the various derived forms. Bullokar's

74

Maurizio Gotti

approach to this process of specification of the semantic value of the main affixes of foreign origin was indirect, in the sense that his dictionary did not deal systematically with word-formation rules; only rarely did he specify the meaning of the main affixes of classical origin that appeared in the terms he was explaining, and when he did so he dealt with them in an implicit way. The following example — containing the indication of the semantic value "contrary or against" for the Greek prefix anti — demonstrates Bullokar's method: Antichrist. An aduersary to Christ: It is compounded of the Greeke preposition Anti, and Christus, which signifieth contrary or against Christ. However, although there were no explicit rules of word formation, the user of An English expositor was gradually able to become accustomed to the semantic value of the different affixes by reading the explanation of the entries he was analyzing. Indeed, in expressing the definitions of his terms, Bullokar followed a regular pattern of wording, thus building up a series of semantic equivalences. Here are a few examples concerning various affixes of foreign origin found in An English expositor. (a)

(b)

(c)

(d)

(e)

(f)

The meaning 'together' attributable to the prefix con-: Congregate. To gather together. Coniunction. A ioyning or coupling together. Connexion. A knitting together. The meaning 'contrarie, against' attributable to the prefix counter-: Countermand. To giue commandement contrarie to that wich was commanded before. Countermine. To mine or dig in the earth against another. Counterpoise. Any thing laid in waight against another thing. The equivalence of the prefix in- with the English prefix un-: Inaccessible. Which cannot bee come vnto, vnapprochable. Inauspicious. Vnluckie, vnfortunate. The meaning 'before' attributable to the prefix pre-: Precede. To goe before. Predestinate. To appoint before hand, what shal follow after. Predecessor. Hee that was in place or office before another. The meaning 'again' attributable to the prefix re-: Reassume. To take againe. Reedifie. To build again, to repaire. Reenter. To enter againe. The use of the suffix -ation to derive a noun from a verb: Consummate. To finish, to make an end. Consummation. An end, a finishing of a matter. Cooperate. To worke together, to helpe. Cooperation. A working with another, a helping.

John Bullokar 's "Termes of Art"

(g)

(h)

75

The use of the suffix -ible to derive an adjective from a verb: Corrigible. That which may be corrected or amended. Credible. That which may be beleeued. Edible. Which may bee eaten, The use of the suffix -or to derive the noun indicating the agent from a verb: Compositor. He that composeth or setteth a thing in order. Mediator. He that maketh meanes or speaketh for another.

In providing these series of equivalences, Bullokar increased the pedagogic value of his dictionary: not only did it fulfill his desire to provide a tool useful in promoting a correct interpretation of several hard words commonly found in English texts by providing precise and detailed definitions of the various entries; it also formed a basis for a systematic patterning of the explanation of derived words, thus enabling the user of An English expositor to realize the semantic value of many common affixes of foreign origin encountered in the "termes of art" reported in the dictionary.

Appendix - Law: Abate, Abatement, Abbett, Abbeftour, Abiure, Abiuration, Abrogate, Accessory, Acquitall, Action, Addition, Adiournement, Adiure, Adjuration, Administer, Administration, Administrator, Adopt, Adoption, Aduscate, Aduouson, Amerce, Amercement, Annuitie, Annull, Appariter, Appeale, Appellant, Appelation, Appropriate, Appropriation, Arbiter, Arbiterment, Arbitrary, Arbitrate, Arbitratour, Arrerages, Articulate, Assets, Assumpsit, Astipulation, Attainder, Attaynt, Attestation, Atturnie, Atturnment, Auditor, Auerre, Auerment, Auowable, Auow, Auowrie. - Religion: Abba, Abbot, Ablution, Absolue, Absolution, Adore, Adoration, Aduent, Anathema, Anathematize, Anchoresse, Anchorite, Antheme, Antiphone, Antichrist, Apocalypse, Apocrypha, Apostasie, Apostata, Apostaticall, Apostle, Apostolicall, Arrian, Assoile, Atheisme, Atheist, Azymes. - Rhetoric and Linguistics: Allegorie, Allegoricall, Amphibolic, Amphibologie, Anagramme, Aphorisme, Apologie, Apologicall, Apophthegme, Aspiration. - Botany: Abrahams Baume, Acatian, Aconitum, Agarick, Alkagengi, Aloes, Ambrosie, Auime, Asafoetida. - Medicine: Aloesuccotrina, Ambia, Anatomie, Anatomize, Antidote, Apoplexie, Apozeme, Arterie. - Zoology: Aerie, Alcion, Ambergrise, Anchouie, Apocynon, Armadillo, Aspe. - Geography: Africa, Alps, Antartike Pole, Antipodes, Articke Pole, Asia. - Mineralogy: Abeston, Adamant, Alabaster, Amber, Amethist, Antimonie. - Astronomy: Apogeon, Aspect, Astrolabe.

76

Maurizio Gotti

- Heraldry: Aduentaile, Atcheuement, Attired - Physics: Ammoniacke, Antiperistasis, Auripigmentum. - Politics: Anarchie, Aristocratic, Arisiocraticall. - Hunting: Allay, Antlier. - Military tactics: Ambuscado, Artillerie. - Geometry: Angle. - Logic: Axiome.

Notes 1. All quotations in this paper are taken from the facsimile copy of the British Museum original text, published in 1971 by Georg Olms Verlag, HildesheimNew York. 2. For an analysis of the concept of "hard words" and their inclusion in the earliest English dictionaries cf., among others, Gotti (1997). 3. Most of these archaic words had been taken from Thomas Speght's glossary of "old and obscure words" appended to his edition of Chaucer's works (1598, revised and enlarged in 1602). A few had been drawn from E. K.'s explanatory notes to The shepheardes calender. 4. A confirmation of the fact that Bullokar has mainly conceived his dictionary as an aid for the decoding of difficult words can be found in the title of the work: An English expositor, indeed, in his dictionary Bullokar defines the term expositour as "an expounder or interpreter". 5. This different volume of the three categories is based on a statistical analysis of the words listed under the letter A of the dictionary: out of 408 words, 261 (i.e., 63.97%) are general words (or semi-technical terms) of foreign origin, 4 (0.98%) are archaic words, while 143 (35.05%) are specialized terms. The extension of these percentages to the whole dictionary is of course risky, as the contents given under the various letters of the alphabet might vary; our estimate is, therefore, only indicative of a general attitude and does not aim to provide exact statistical figures for the whole book. 6. These are the dates of first appearance of these words in English texts according to Schäfer (1989): alchymie (1362), anatomie (1503), architecture (1563), arithmetike (1250), astrologie (1375), astronomic (1205). 7. Most of the terms referring to the new plants recently found in exotic places were derived by Bullokar from John Frampton's loyfull newes out of the newe founde worlde (1577), a translation from Spanish of Nicolas Monardes' Dos libros ...de nuestras Indios Occidentales (1569). 8. The "termes of art" listed under the letter A — grouped according to their different disciplines — are listed in the Appendix. 9. The main sources of Bullokar's text were Thomas Cooper's Thesaurus linguae Romanae et Britannicae (1565), John Frampton's loyfull newes out of the newe founde worlde (1577), Thomas Thomas's Dictionarium linguae Latinae et

JohnBullokar's"TermesofArt"

77

Anglicanae (1587), Robert Cawdrey's A table alphabetical! (1604), and John Co well's The interpreter (1607). For a more detailed analysis of the origin of Bullokar's entries cf. Hayashi (1978); Kerling (1979); Landau (1984); Riddell (1974,1983); Schäfer (1970, 1989); Starnes—Noyes(1946). 10. The dates of first appearance reported in the Oxford English dictionary and A Supplement to the Oxford English dictionary have been corrected according to the indications of Schäfer (1989). It is well known that the Oxford English dictionary is unreliable as a source of the dates of first occurrence, as very frequently previous appearances in other texts have been pointed out for the various entries. However, the dates suggested here are not meant to provide with any certainty the date of the first occurrence of these words in the English language, but rather indicate that such words already existed in the English language at the date shown. 11. Colon is included only as a term of punctuation. 12. Although Intestine is not recorded as a noun, it is recorded as an adjective (= "Bred in the bowels").

References Burchfield, Robert W. (ed.) 1980 A supplement to the Oxford English dictionary. Oxford: OUP. Gotti, Maurizio 1997 "The "hard words" of Levins' dictionary", in: Raymond Hickey—Stanislaw Puppel (eds.), 483-501. Hayashi, Tetsuro 1978 The theory of English lexicography, 1530-1791. Amsterdam: Benjamins. Hickey, Raymond—Stanislaw Puppel (eds.) 1997 Language history and linguistic modelling: A festschrift for Jacek Fisiak on his 60th birthday. (Trends in Linguistics: Studies and Monographs 101.) Berlin: Mouton de Gruyter. Kerling, Johan 1979 Chaucer in early English dictionaries. Leiden: Leiden University Press. Landau, Sidney I. 1984 Dictionaries: The art and craft of lexicography. Cambridge: CUP. Murray, James A. H.—Henry Bradley—William A. Craigie—Charles T. Onions 1933 The Oxford English dictionary. Oxford: Clarendon Press. Riddell, James A. 1974 "The beginning: English dictionaries of the first half of the seventeenth century", Leeds Studies in English n.s. 7: 117-153.

78

Maurizio Gotti

1983

"Some additional sources for early English dictionaries", The Huntington Library Quarterly 46: 223-235.

Schäfer, Jürgen 1970 "The hard word dictionaries: A re-assessment", Leeds Studies in English n.s. 4:31-48. 1989 Early Modern English lexicography 1-2. Oxford: Clarendon Press. Starnes, De Witt T.—Gertrude E. Noyes 1946 The English dictionary from Cawdrey to Johnson, 1604-1755. Chapel Hill: The University of North Carolina Press.

The Dublin Vowel Shift and the historical perspective Raymond Hickey

1. The background The present paper is concerned with examining a major change in the English spoken in the city of Dublin, a change currently in progress and which has become evident to the author in investigations of Present-day Irish English over the last decade or so. I have given it the working title of the Dublin Vowel Shift and will present details in the course of this paper which will hopefully show that it is of comparable status to other major attested vowel shifts in the history of English and that it has a significance which goes beyond any local interest in the pronunciation of Dublin English. To all appearances the Dublin Vowel Shift is something which started less than 20 years ago and has not yet reached phonological stability. In essence this externally motivated change involves a retraction of diphthongs with a low or back starting point and a raising of low back vowels. Specifically, it affects the diphthongs in the TIME and TOY lexical sets and the monophthongs in the COT and CAUGHT lexical sets. (1)

Dublin Vowel Shift, principal movements a.

b.

retraction of diphthongs with a low or back starting point time [taim] > [taim] toy [toi] > [toi], [tot] a raising of low back vowels cot [koj] > [koi] caught [ko:U > [foxti, [ko:)J 01 t

RAISING RETRACTION

ai

->

o: t

DI t

D t

Dl

D

o:

αϊ

As a change in progress the Dublin Vowel Shift is of particular interest to historical linguists as it offers evidence for the mechanisms of actuation and propagation in sound shifts. Because it has not yet become stable one can observe how and to what extent it is being picked up by speakers from various groups

80

Raymond Hickey

within Ireland and in this respect it is parallel to, but of course not identical with, current changes among the short tensed vowels in the northern cities of the eastern United States (cf. various discussions in Labov 1994). Observations concerning the change show that it has a distinct social distribution: it would appear to emanate from those social groups in Dublin which choose not to identify themselves linguistically with popular Dublin English; more on this presently.

1.1.

Historical remarks on Irish English

The history of English in the south of Ireland can be divided into two periods. This is justified on both language-internal and external historical grounds. The first period dates from the late 12th century to 1600 and the second from the latter date to the present day. The division between the two periods rests on the external events at the end of the 16th century and in the 17th century, when a vigorous policy of forced settlement took place in Ireland, establishing the dominance of English which has increased ever since, much to the detriment of the native Irish language. The centuries after the coming of the Normans in 1169 were characterized by Gaelic resurgence which led to the decline of English (and Anglo-Norman) in the entire country by the early 16th century with the exception of Dublin and its immediate surroundings, known as the Pale from the fortifications which separated it from the rest of the country under the control of the native Irish. The latter borrowed many words from the foreigners in this early period. Most of these are from Anglo-Norman but there are some from English, e.g., whiting gavefaoifin in Irish which shows clearly a pre-Great Vowel Shift pronunciation of / in Irish English, because ao in the borrowed word was definitely pronounced I'd in Irish. The result of the Gaelic resurgence in rural Ireland before the 17th century is that Dublin English today is the only substantial variety of English in Ireland which can claim continuity from the time of the original settlers. This fact is of no sociolinguistic relevance today but it is responsible for certain features of Dublin English which are not found in the rest of the country such as the lengthening of many short vowels and the breaking of long vowels; more on this in a moment.

1.2.

Documentation of Dublin English

The documentation of Dublin English is unfortunately quite scanty. What little material there is can be found in documents such as city records. These offer a glimpse of some archaic features of Dublin English, e.g., the deletion of post-sonorants stops as in /ρευη(?)/ pound with a glottal stop at most as a trace of the former alveolar plosive (still a feature of popular Dublin English). The Dublin Records from the 15th century show this deletion in words like stone 'stand', sir one 'strand'.

The Dublin Vowel Shift and the historical perspective

81

Apart from non-fictional prose documents there is the language of authors from Dublin. However, this is not necessarily Dublin English. Certainly for authors of the early modern period (from the mid 17th century onwards) one cannot assume that their language was by any manner or means a reflection of the popular speech of the capital. Authors like Jonathan Swift (1667-1745) or Richard Brinsley Sheridan (1751-1816) wrote in standard English. Nonetheless, there are glimpses of what Irish English, if not Dublin English, can have been like in previous centuries. Swift wrote two small pieces which purport to represent the English of the native Irish and the English planters of the early 18th century (Irish eloquence and Dialogue in Hybernian stile; cf. Bliss 1976: 557). Quite a number of dramatists of the Restoration period (after the return of Charles II to the throne in 1660) parodied Irish figures and their speech in their plays and this casts some light on Irish English of the late 17th century, hi addition, the father of the playwright Richard Brinsley Sheridan, one Thomas Sheridan (1719-1788), was something of an authority on language matters in the late 18th century and also a well-known elocutionist who traveled widely. Sheridan is the author of A rhetorical grammar of the English language (1781), which includes an appendix on the language of educated Dubliners, which he examines and corrects (laying out a series of rules to be observed by the Irish in order to speak English properly). This brief treatment is a valuable source of information on the state of upper-class Dublin English two centuries ago. Among the features remarked on by Sheridan are the following: a) b)

c)

d)

1.3.

The pronunciation of α was /a:/ and not /e:/ in words like patron, matron. A pronunciation of English /ai/ from ME I'd was found as [si] (assuming a correct interpretation of Sheridan's spellings), and this tallies with what is known from Present-day Dublin English, e.g., mine [main] for [main]. There was an unshifted realization of ME /ε:/ and /e:/ as [ε:] in words like beat, leave, meet, which is in agreement with local Dublin English usage up to the beginning of the present century. This vowel is still found as a caricature of a former Irish pronunciation of English, e.g., in Jesus [cfceiziz], decent [de:sm(?)]. Today unshifted ME /ε:/ is a deliberately archaic vowel realization in Dublin English. A realization of /a:/ before former liquids as /o:/ as in psalm [so:lm], balm [bo:m] appears to have been current. This pronunciation is unknown nowadays, though the back vowel before velarized /!/ probably resulted in the pronunciations /baul/ and /aul/ still found for bold and old colloquially.

Super imposition of more standard varieties

The various comments on Irish English since the 16th century and the limited non-fictional documents which are available from the 15th century onwards show that the Great Vowel Shift was only partially carried out in Ireland. This is of

82

Raymond Hickey

course due to the fact that the varieties brought to Ireland as of the late 12th century obviously could not have had the shift and that it was later only adopted in part. In particular the lack of diphthongisation of I'd to /ai/, the lack of raising of /a:/ to /e:/, and the absent shift with both /ε:/ and /e:/ to /i:/ are noticeable. Of these shifts only that of /ε:/ and /e:/ to I'd is still missing in local Dublin English, though here the lack of shift is recessive. This situation points clearly to the super-imposition of more standard varieties at various stages in the development of Dublin English. The peculiarities which Sheridan referred to concern educated Dublin English pronunciation, so that in the course of the 19th century ME /a:/ must have receded entirely in favor of /e:/ in words like take, matron. Furthermore, the hypercorrect forms such as prey [pri:] and convey [kan'vi:], noted by Sheridan, died away, being replaced by pronunciations with /e:/. Other features of Dublin English are recorded by Sheridan, but not found anywhere today. E.g., there seems to have been a general lowering of ME /e/ to /a/ before /r/ as seen regularly in English words like barn, dark from earlier bern, derk. Sheridan notes pronunciations like

(2)

sarch for search, sarve for serve, etc.

Again the superimposition of more standard layers of English has been responsible for their disappearance since.

7.4.

Lexical extinction

Words such as tea [te:] or please [ple:z] may serve as fossilized retentions of an early stage in the variety which has been superseded by another. But the frequent claim that [ε:] is used for id betrays an undifferentiated perspective on Dublin English. With Hogan (1927) one already has references to the recessive nature of the mid-vowel realization and its confinement to keywords in lower registers. The number of forms which exhibit the earlier vowel realization has been constantly reduced up to Present-day English, and if the process of lexical extinction of these realizations is carried through there will be no record of this earlier vowel value left in Dublin English.

1.5.

Divisions in Dublin city

Like any other modern city Dublin shows areas of high and low social prestige. The city lies at the mouth of the river Liffey in the center of the east coast, and

The Dublin Vowel Shift and the historical perspective

83

spreads along the shores of the horseshoe shape of Dublin bay. The suburbs, which have increased dramatically since the sixties, reach down to Bray and beyond into Co. Wicklow in the south, to the West in the direction of Maynooth, and to the north at least to Swords, the airport, and beyond. The Dublin conurbation now encompasses about a third of the population of the Republic of Ireland, i.e., over one million speakers. Within Dublin there is a clear divide between the north and the south side of the city. The latter is regarded as more residentially desirable (with the exception of Howth and its surroundings on the peninsula which forms the north side of Dublin bay). Within the south there is a cline of prestige with the area around Ballsbridge, Donnybrook (and, somewhat further away from the center, Foxrock) enjoying highest status. This is the area of certain key complexes like the Royal Dublin Society (the most important exhibition and event center in the capital) and the national television studios RTE (Radio Telefis Eireann, 'Irish Radio and Television'), and of the national university (University College Dublin) in Belfield. This entire area is known by its postal number, Dublin 4. Indeed this number has given its name to a sub-accent within Dublin English known as the "Dublin 4 Accent" which shows the vowel shift in its more extreme form. The less prestigious parts of the city are known by their district names such as the Liberties in the center of the city, immediately north of the river Liffey, and Ballymun, the only suburb in Ireland with high-rise flats, which is associated with adverse social conditions.

1.6.

The supra-regional variety of English

In the Republic of Ireland, i.e., excluding the north which, because of its different demographic history, is quite separate from the rest of the country, there is something like a supra-regional standard which is characterized by the speech of middle-class urbanites. This can be classified into different sub-varieties on the basis of features which are found in one and not the other. E.g., the urban speech of Cork may show a tendency to raise Id before nasals and that of Dublin may raise and lengthen /D/ before voiceless fricatives. Nonetheless, there is a core of common features which can be taken as characteristic of general middle-class speech of the south, and it is these which the non-Irish use as clues for identifying an Irish accent, e.g., rhotacism (with a velarized [i]), dental stops for dental fricatives, fricativisation of /t d/ in open position (intervocalically and word-finally before a pause as in butter and cut), monophthong equivalents to the Received Pronunciation diphthongs /ei/ and /au/, and the lack of any significant lexical distribution of long and short vowels in the BLAND and GLANCE lexical sets (as the length and quality difference in these vowels is so slight) to mention just some of the more prominent examples.

84

1.7.

Raymond Hickey

The Status of Received Pronunciation

For an anglophone country like Ireland the relationship to other larger English-speaking countries plays a significant role, one which naturally has a linguistic dimension as well. In this connection the question arises whether the less regionally-bound varieties of any of these larger countries act in any sense as a guideline for the Irish. In practice the only two countries which are involved here are the United States and Britain. There is considerable exposure to forms of English from America in the media, above all through films, but whatever accents may be represented here they have no effect on language behavior in Ireland with the exception of some lexical items which may be adopted. The situation with Britain is different given the geographical proximity and the close economic and cultural ties which it has with Ireland. This is a complicated issue as the history of Ireland often dictates a skeptical and critical attitude to Britain while contemporary social conditions in Ireland are such that there is much contact with Britain due to Irish people working there and the general orientation towards this country, though there has been a certain re-orientation as a result of membership in the European Union in the past decade or so. However, linguistically Britain is and will remain Ireland's powerful and dominant next-door-neighbor. Hence the valid question concerning the status of standard forms of British English in Ireland needs to be asked. It may well be part of the colonial legacy in Ireland that speakers are very sensitive to the danger of being seen to have a thick, boorish Irish accent, particularly because a condescending attitude to the Irish has a considerable tradition in Britain. However, here as in other spheres of life, one notices the ambivalence of the Irish relationship with Britain. On the one hand, the Irish do not want to be all too readily regarded by the British as having an unacceptable accent, on the other hand, they would regard it as pandering to the British to adopt an accent approaching Received Pronunciation. Hence, it is true to say that speakers from the Irish Republic do not emulate Received Pronunciation. Instead the supra-regional standard of Dublin origin provides an orientation for the southern middle-class Irish. Certain characteristics of this speech, such as rhotacism, alveolar /!/ in all positions, monophthong long mid vowels, centralized /a:/, plosive equivalents to /Θ 8/, and the retention of /M/ (Hickey 1984) are so obvious that no possibility of confusion with the southern British standard is possible. A major reason for the lack of influence of Received Pronunciation on the supra-regional standard of the south is the attitude of the population to England and English society. It would not befit any nationalist-minded Irishman to imitate an English accent. Although the vast majority of the Irish speak English as their native language, using an English accent approaching Received Pronunciation is regarded as snobbish, if not to say unpatriotic, and usually evokes derision from one's fellow countrymen. It is this which is scorned as a grand [grcr.nd] or posh or

The Dublin Vowel Shift and the historical perspective

85

lah-di-dah accent, an outward manifestation of pretentiousness and social condescension.

1.8.

How can you tell a moderate Dublin accent?

The supra-regional accent of English in the south of Ireland is derived ultimately from a conservative southern urban pronunciation as found in Dublin. However, as with any metropolitan accent one finds certain popular features represented to a limited degree in educated non-localized forms of the city variety. This is what is termed a "moderate Dublin accent" and the term could be equally applied to other capitals such as London where, e.g., a diluted form of the Cockney vowel values is typical of the speech of educated non-working-class Londoners. Dubliners can be distinguished from urbanites from the rest of the Irish Republic by characteristic pronunciations. E.g., there is a general fronting of the /au/-diphthong which results in realizations like house [haeus] for [haus]. Another trait is the lengthening of low back vowels. Thus, words like lost are pronounced [b:st] rather than [lost]. This feature can lead to a merger of pairs like horse and hoarse under [ho:rs] which are normally kept apart in general urban Irish English. A further characteristic of Dublin English is its rhotacism. Unlike England there is no prestige variety of Irish English which is non-rhotic. The fact that syllable-final /r/ should be maintained so consistently, especially in Dublin English, is not an accident. Lower-class Dublin English is non-rhotic as pronunciations like [pAUta] for porter testify. In this respect Dublin English is similar to New York. In both cities rhotacism is prestigious as lower class speech lacks syllable-final III. Dublin has, however, departed from the general velarized r of the rest of the country and developed a retroflex [4], so that a word like^br [fba] is pronounced as [fo:.i]. This realization is connected with the raising of back vowels, again characteristic of Dublin English. Apart from such specific features, Dublin English shares the typically Irish traits of pronunciation such as the weakening of alveolars in an intervocalic or word-final position: night [naijj, fighting [faijiq]. It also shows ubiquitous yod-dropping after alveolar sonorants: news [nu:z], neuter [nuija1·]. However yod-deletion after non-sonorants, e.g., /stuipid/ would be understood as a deliberate imitation of an American accent. Note that the features just mentioned do not have any significance as social markers. They show a minimum of variation stylistically and generally enjoy only low awareness with speakers. Equally variables such as (h) and (ae) are not markers of high or low social position, as they frequently are in British English: Λ-dropping is unknown in Irish English, and there is no differential use of /ae/ versus /a:/ (as with RP bland and glance mentioned above), because these vowels are only weakly distinguished. Both are realized as a long low central vowel, [a:]: [bla:nd] and [glams] respectively.

86

Raymond Hickey

Apart from segmental features such as those just discussed one should mention that Dublin English shows a number of allegro phenomena which spread across groups of words. They are phonologically non-systematic but nonetheless contribute to the impression of a slurred pronunciation which the non-Irish sometimes associate with Irish English; a case in point would be the procope of pre-stress syllables or the reduction of consonant clusters as in (3)

1.9.

procope of unstressed syllables 'member ['memba^J (remember), 'mhere ['mia·-] (= come here) consonant cluster simplification recognize ['reksnaiz], months

How do you recognise a lower class Dublin accent?

Moving down the social scale in the capital facilitates the task of recognizing Dublin speech. There are obvious features in morphology and syntax such as the agglutinative plural for second person pronouns, either youse or yez, the latter consisting of the inherited archaic ye and the productive plural suffix {S}. Another indicator is the use of unstressed do for a habitual aspect as in She does be worrying about the kids all the time. The pronunciation of Dublin English has equally unambiguous features. In the area of vowels the clearest of these are the centralization of the /ai/ diphthong, the fronting of/au/, the over-long realization of phonemically long vowels, often with disyllabification as a result, the realization of historically short vowels before /r/, and that of Early Modern English short /u/. (4)

Centralization of/ai/ time [taim] [tojsm] Fronting of /au/ down [deun] [deun] Over-long vowels with frequent disyllabification school [skurl] [skuial] mean [mirn] [mi:an] Historically short vowels before /r/ circle [se:k|] first [fu:s(t)] Early Modern English short /u/ Dublin [dublan]

[skuiwal] [mi:jan]

In the area of consonants there are equally clear indications of popular Dublin English. Some are unique to lower registers and others are extensions of features found in more accepted forms of Dublin English. Unique features include the simplification of consonantal syllable codas, particularly of stops after fricatives or

The Dublin Vowel Shift and the historical perspective

87

sonorants. Intermediate registers may have a glottal stop as a trace of the stop in question. (5)

pound last

[peun(?)J [lae:s(?)]

Extensions include the lenition of III in a weak position beyond the initial stage of apico-alveolar fricative to /r/ then to /h/ with final deletion as in the following instance. (6)

/t/ motorway

[t] -> [moj^we]

[a] -» [moaa^we]

[h] -> [moh^we]

0 [mo:(j)we]

Finally, mention should be made of the merger of dental and alveolar stops in lower-class Dublin English. Although it may seem to non-Irish ears that ambi-dental fricatives are always realized as alveolar stops this is by no means the case. There is a clear distinction between the dental stops [jj and [dj, which are the equivalents of English [Θ] and [o] respectively, and the alveolar stops [t] and [d] which correspond to those in non-Irish varieties. Irish ears are tuned to this difference and the retraction of the dental stops to an alveolar position is immediately noticeable and hence stigmatized as typical of low-prestige speech. (7)

educated speech thinker [jirjka1] tinker [tirjka^]

lower-class speech thinker, tinker [tinka-]

breathe [bri:d_]

breathe, breed

breed

[bri:d]

[bri:d]

7.70. The lower class Dublin community Looking at lower-class Dublin English from a broader perspective one immediately recognizes its conservative character. Its salient features are those which were in all probability established in the first period of English in Ireland, and it has, generally speaking, survived the later superimposition of more standard forms of English in the last two centuries which eliminated many of the characteristics of middle-class speech in the city. The conservative profile is in keeping with the notion of network strength within a closely-knit community which inhibits change, and the relative homogeneity of lower-class Dublin English should be seen within the context of shared values in this stratum of society. Furthermore, the focused nature of the lower-class community is attested in the amount of synchronically opaque forms which represent fairly complex variation which is not readily accessible to those outside the community. E.g., the realization

88

Raymond Hickey

of historic short vowels before /r/ is unpredictable for middle-class speakers, e.g., one has (8)

circle [se:k|] but bird [bu:d]

This group has the merger of HI and Id before /r/ and so the distinction is no longer recoverable for them whereas it is an integral part of lower-class speech. The speech of the middle-classes, on the other hand, is more indicative of supra-local varieties which tend towards koines and are simpler in structure than more local varieties because, for the higher social groups, variation no longer shows any in-group function; this could go part of the way in explaining why the merger of short vowels before /r/ was accepted by socially prestigious, but weakly networked groups, which originally did not have the merger (as Sheridan shows), during the later superimposition of standard varieties.

2. What is the Dublin Vowel Shift! For the remainder of this article I will be concerned with describing the Dublin Vowel Shift in some detail and analyzing it within a broader linguistic framework. Let me begin however with some remarks on the present position of the capital city Dublin. Dublin is a modem metropolis with over a million inhabitants and shows a clear social structure, stretching from the poor, lower-class city-center and northside to the considerably more affluent middle-class south side. Local varieties of Dublin English are associated with the less well-off parts of the city, so that the adoption of a sophisticated accent in Dublin serves the dual purposes of hiving oneself off from the poorer elements and associating oneself with the more affluent sections of the capital's population.

2.1.

Fashionable and local Dublin English

For the linguist describing different kinds of Dublin English the question arises: how does one refer to the participants in the modern urban setting of Dublin? One could start with traditional terms which refer to class. Class definitions typically involve education, occupation, and relative wealth. The middle classes would have all these attributes to a positive degree and one could say that they are the group which sees itself as separate from the lower classes. However, the matter is not that simple. E.g., air hostesses, bank clerks, company secretaries, people from the world of film and fashion are not necessarily regarded as belonging to the middle class, but they certainly come from a section of the population which does not want to be identified with an all too localized form of

The Dublin Vowel Shift and the historical perspective

89

Dublin English. One could say that it is the group of those aspiring upwards — the socially ambitious — which is the motor behind the changes in Dublin English. Certainly this group belongs to those partaking in the changes. However, social movement is not in my opinion the defining factor for assignment to the group which is the driving force behind the Dublin Vowel Shift. Rather it is an attitude of condescension and snobbishness towards the low-prestige and linguistically salient sections of the capital city. These are certainly "below" in a vertical interpretation of social structure and the former group can be thought of as "above", at least in a metaphorical sense of those who look down on others. A common means of referring to middle class speakers is to speak of them as "educated"; again this term is not accurate enough for the matter at hand. The determining factor for active participation in the Dublin Vowel Shift is the extent to which speakers espouse urban sophistication. This can be seen as a rejection of an all too local identification with Dublin and a conception of self as a player on a (fictional) international stage. Such an understanding of the motivation explains why the Dublin Vowel Shift is found among groups which have not enjoyed tertiary education and who are not necessarily among the more prosperous — air hostesses, company secretaries, up-market shop assistants. It further accounts for why many established professionals — genuinely educated speakers in any sense of the term — such as doctors, teachers, lecturers, do not necessarily show the Dublin Vowel Shift or only the weakened form with lexical keywords (more on this in a moment). Indeed, if there is a strong sense of local bondedness among such speakers then they may evince a high degree of popular Dublin features and none of the signs of the vowel shift.

2.2.

The variable (ai) in Irish English

Let me now rum to the description of the Dublin Vowel Shift. The first point to note is that a conservative pronunciation of (ai) in Dublin is maintained in lower-class speech as [ai]. There is historical documentation of this realization which is particularly revealing as it shows that it was typical of the middle classes in late 18th century Ireland. E.g., Fanny Burney (1752-1840), in her reminiscences of famous individuals she knew, imitates the Irish accent of the playwright Richard Brinsley Sheridan (1751-1816) by referring to his pronunciation of kind as [kaind], indicated orthographically as koind. This pronunciation seems to have become part of the stereotype of an Irish accent and authors such as Kipling used the oi spelling to indicate this as in woild Oirland. Now the supraregional variety of the south has for (ai) a diphthong which has a low-mid or low-front starting point, i.e., either [ai] or [aei]. This realization tallies with that in many varieties of Irish, although the position in Irish is of no relevance to the Dublin Vowel Shift. What is significant here is that a non-central starting point is the commonest one for non-regional varieties of Irish English. This pronunciation would seem either to have developed independently in the capital or

90

Raymond Mickey

to have been adopted from the large influx of rural speakers, most of whom would have had the [ai] realization. Recall that in the latter half of the 19th century at a time when the population of Ireland sank by several millions, that of Dublin actually increased by almost 10%. If one now considers local Dublin English one finds that its realization for (ai) as [ai] is quite stigmatized in Dublin. One can maintain that the greater the phonetic separation of middle-class Dublin English from more local forms in the capital grew, the more the corresponding forms of the lower social classes became stigmatized. However, the matter does not end there. For middle-class Dubliners the [ai, aei] pronunciations sufficiently delimit them from popular Dublin English. But increasingly a back starting-point is being used with this diphthong, i.e., for a word like style the pronunciation is not [stall] but rather [stall]. This retracted starting-point is particularly noticeable before /r/ so that the name of the country is realized as [aubnd] rather than [auland]. The social group which most clearly shows this pronunciation is that referred to above and their variety is that which I choose to term "fashionable Dublin English", as this term best captures the element of vogue which is associated with pronunciations within Dublin which are maximally distinct from the conversative and strongly local forms.

2.3.

Distribution of the shift

The fact that the Dublin Vowel Shift is not that old offers the linguist the opportunity of observing a change in its initial stages and provides evidence for how a change begins to spread through the variety in which it occurs. In this connection it is relevant to consider the distribution of the segments affected. The most noticeable aspect of the shift is that is does not apply to all possible inputs as can be seen from the following words with (ai). (9)

rice tight life

[rais] [taij] [laif]

rise tide lives

[raiz] [taidj [laivz]

The generalization here is that retraction to [αϊ] only occurs before voiced segments. This makes phonetic sense: the retracted onset of the diphthong requires that the tongue travel a longer distance from a rest position than for the unshifted realization [ai]. In this respect the restrictions on the diphthong shift are similar to those on diphthong realization in Canadian English, commonly known as Canadian Raising. Here a centralized onset is used for the diphthongs /ai/ and /au/ before voiceless segments, before voiced ones an onset in the region of/a/ is found. (10)

Canadian Raising

house [hsus]

houses [hauziz]

The Dublin Vowel Shift and the historical perspective

91

This distribution shows that the Canadian phenomenon is not an instance of raising but rather of vowel lowering (from /o/ to /a/) before voiced segments, i.e., the tongue can travel from a central to a low position before those segments which are relatively long phonetically. The argumentation used here is phonetic and would appear to offer the best chance of accounting for apparent counter-examples to the distribution assumed. Thus, many speakers have the shift in words like crisis [kraisiz] /krai.siz/. On closer consideration the explanation would seem to be that the shift is found before voiced segments and in open syllables. This also accounts for its occurrence in final position, e.g., in bye [bar] and July [djuilai], as well as in isolated instances where /ai/ occurs before a voiceless segment in a closed syllable but in a strongly stressed position at the end of a phrase, e.g., in Your man is so nice ['nais]. Stress positioning plays a further role: the shift is not in evidence in unstressed syllables or those with secondary stress because of their inherent phonetic shortness. (11)

headline nationwide

['hedlain] ['neijan.waid]

It is difficult to predict whether this distribution will remain typical for the Dublin Vowel Shifi. It may very well be that it is only characteristic of an initial phase and that the shift will spread to all instances of/ai/, masking the present distribution. Or it may freeze at this stage, as has been the case with Canadian Raising. And of course the shift may peter out and an unretracted [ai] pronunciation may be re-instated for all instances of the diphthong.

2.4.

General shift of low vowels

The Dublin Vowel Shift is not just confined to the realization of (ai). As might be expected other vowels in the area of this diphthong are affected. The elements in question are the diphthong /oi/ and the low and mid vowels /D/ and /o:/. Now it should be noted that in Irish English these vowels have a generally lower realization than in southern British English.

(12)

boy pot law

/oi/ /D/ /D:/

-> -> -»

[boi] [pot] - [pa)J [lo:]

These realizations show that the change has the characteristics of a chain shift, a major shift, far more than anything found currently in Received Pronunciation, e.g., the fronting of/u:/ or the lowering of/ae/. It can be compared in its breadth to the Scandinavian vowel shift of the late middle ages which led to a shift backwards, upwards, and forwards for all vowels from /a:/ to /u:/.

92

Raymond Hickey

2.5.

Lack of stylistic variation

The Dublin Vowel Shift is a set of changes which do not appear subject to any appreciable stylistic variation. The reason is probably that, as a set of changes (relaxed jaw setting, retraction of low vowels, except /a:/, and raising of back vowels), showing stylistic variation would involve altering the whole set. Stylistic variation with (ng) /iq/ or /in/ is to be expected as it involves a single feature which is easily reversible. The same applies to the agglutinative plural second person pronoun, youse oryez, instead of you.

2.6.

Fashionable Dublin: How to avoid local features

The retraction of low vowels is the most acoustically salient feature of fashionable Dublin English, and it is this which constitutes the core of the Dublin Vowel Shift. However, the story does not end there: other avoidance strategies used to maximally differentiate fashionable forms from local forms of Dublin speech are found. (13)

Fashionable Dublin features

a.

Strict avoidance of retraction of III before /r/ in third, first, i.e., [i^:d] [fa-:st] and not [tu:d] [fu:s(t)j. The local back rounded /u/ is replaced by an unrounded front vowel which is almost /i/, as in Sunday. Local Dublin English has a distinction between historic back and front short vowels before /r/, [ε:] and [u:]. But because the open front realization is so stigmatized (so typical of local Dublin English), there is a migration in fashionable Dublin English of historic front long vowels to the central rhotic type as seen in examples like care [ka^:] and pear [pa-:]. A retroflex /r/ is used which has the advantage of marking the /r/ even more clearly vis a vis the popular forms of Dublin English which, if at all, only have a weak syllable-final /r/.

b. c.

d.

2.7.

Relative chronology

Given the lack of work on Dublin English it is difficult to pinpoint when the vowel shift began. The use of central rhotic vowels for historic long mid vowels before III (in words like care, pear) has been a feature for a few decades as has the occurrence of general lengthening and raising of/a/ as in lost [lo:st] and of b'J as in morning [mo:.miq] (cf. the rather loose set of comments in Wells 1982: 418-428).

The Dublin Vowel Shift and the historical perspective

93

Wells (1982: 426) mentions that there is an educated Dublin variant of the /ai/ vowel [αϊ]. He refers to Bertz who sees this realization as typical of Dublin. But Bertz has no inkling of the Dublin Vowel Shift and does not attempt any systematic distinction between local Dublin English and fashionable Dublin English despite his long taxonomies of every conceivable vowel quality which he thinks he has heard in Dublin. The remarks by Bertz stem from research carried out in the early 1970s (1975, 1987), and those of Wells from work in the late 1970s (1982), so that the value of their fleeting comments lies in the apparent existence of retracted realizations of /ai/ twenty years ago. The reports are tantalizingly slight so that it is difficult to say whether the Dublin Vowel Shift had started by the 1970s. Certainly by the mid to late 1980s when the present author began his own scrutiny of Dublin English, the shift had reached considerable proportions with definite patterning and distribution. There is a further reason for rejecting too early a dating, apart from its going unrecognized. This is that its distribution is scant among middle-aged or older speakers in present-day Dublin. If these were exposed to it in their youth then it would be reasonable to assume that they would retain it. Furthermore, many speakers of middle age and onwards have the vowel shift in lexical keywords and not with the general distribution one would expect if they were motivated participants; e.g., among the author's recordings, a number of speakers over 40 had the vowel shift in Irish and Ireland but not in the numerals^/ve and nine.

3. Arguments for and against the shift 3.1. Is there a shift at all? At this point it is sensible to consider the arguments against the Dublin Vowel Shift: it could be maintained that all one has is a gradual approximation to more standard forms of southern British English due to the strong influence of England on Ireland through travel and the media. This looks like the simplest and most convenient explanation for developments in Dublin. After all there is a lexical influence on Irish English, in the use of many buzz-words and semantic shifts such as the use of joy in the sense of success. However, the imitation view can be dismantled very quickly. Bear the following facts concerning the Dublin Vowel Shift in mind.

a.

If British influence was making itself felt, then one would expect other features to be adopted, such as /αϊ/ for long a. But this is out of the question. Words of the GLANCE lexical set have [a:], i.e., [kla:s]. Indeed, the low back realization is used by the Irish to ridicule a plummy British accent by referring to someone as having "a [gja:nd] accent" (although the Received Pronunciation form of the word is [grand]). The normal form of

94

Raymond Hickey the word is [graend]). The normal Irish pronunciation is [gaaind] with a low central vowel.

b.

There is no tendency in Irish English to drop syllable-final /r/. If British English influence was operative then one would expect non-rhotacism to be spreading into fashionable Dublin speech. One could of course say that conversative local Dublin English was, and to some extent still is, non-rhotic; the middle classes cling to rhotacism as a demarcative feature.

c.

Most importantly, however, if southern British influence were operative in Dublin one would expect a generalized, or at least sporadic, retraction of /ai/ to [αϊ]. But the retraction shows a phonetically determined distribution: only before voiced segments and in open syllables.

d.

Lastly, many Irish involved in the shift push it further than the retracted vowel values typical of general southern British English. There are speakers who have, say, [anoi] and others who have [pnoi] for annoy. This point is of theoretical significance and leads to the next matter to be considered.

3.2.

Propagation of the shift

Now for the present analysis of the current change there is one essential premise which is made and that is that speakers are unconsciously aware of minute phonetic shifts which are taking place in the language around them, and they themselves carry out such changes. Furthermore, speakers recognise unconsciously the direction in which a change is moving and can thus force the change beyond the stage at which is it currently. E.g., if a speaker intuitively grasps that back vowels are being raised then he/she can actively participate in this process, e.g., by raising /a/ beyond bl to /o/. In order to do this, of course, speakers must first of all realize that there exist gradations along a cline of pronunciations with divisions like conservative, moderate, advanced. Speakers are unconsciously aware of such a spectrum by exposure to variation in the community of which they are members and by noticing the relative frequency and the conditions of occurrence (situation, speaker group, etc.) of sets of pronunciations for phonological segments. They build up an awareness of variation and change which is part and parcel of their knowledge of their native language. For the present author postulating such unconscious awareness on the part of speakers is an article of faith. Without this there is no principled and coherent way of explaining the origin and course of shift like the current one. Assuming unconscious awareness accounts for the non-random nature of change; it sees speakers as aware of the direction in which their language is moving. Speakers can jump on the band-wagon so to speak, they can put their foot on the accelerator or

The Dublin Vowel Shift and the historical perspective

95

on the brake as they feel inclined. And they can do this without verbalizing their linguistic behavior for themselves or others.

3.3.

Pushing the vowel shift

Speakers of fashionable Dublin English would seem to be aware of the trajectory on which the vowel shift is located even though their own personal realization of key vowels may not be at the most innovative end of this track. Furthermore, this accounts for why young speakers are seen to push the vowel shift. The trajectory for the shift is unconsciously recognized by speakers, and they can not only move within a degree of personal variation on this curve but they can also shift their range of realizations in the direction of innovation, in this case backwards and upwards. This assumption of speakers' unconscious perception of the sound shift has one great advantage for the linguist grappling with the mechanism of the change. It releases him/her from the necessity of imputing to the language a knowledge of the drift of things in Dublin English. Of course it is a convenient abstraction to talk of "Dublin English" as if it were an independently existing entity, to reify it so to speak. But this is just a convenience as we tend to conceptualize phenomena, which we recognise as belonging together, as parallel to objects in the physical world as when we talk of the state, society, etc. As mentioned above the vowel shift in the capital has gained a certain momentum and is moving beyond · height values which are found in southern British English for corresponding vowels. This is particularly clear with the diphthong /DI/.

(14)

[DI] -> [01] —»[01]

boys [boiz]

noise [noiz]

An important point in this connection is that the shift for speakers with the above realizations at this level "overshoots" its goal so to speak. There would appear to be a certain awareness of this behavior in contemporary Dublin English, as a term has emerged in recent years for a kind of exaggerated accent which is putatively typical of one of the more prestigious areas of Dublin, what is called a "Dublin 4 accent" after the postal code number for a fashionable district on the south side of the city. Speakers with this accent are recognized as having more extreme vowel values for the vowel shift and are often ridiculed by more mainstream speakers. However, with time, such extreme values may come to be regarded as possible realizations for other groups of speakers if the latter no longer come to regard the speech of a small minority as unduly exaggerated.

96

3.4.

Raymond Hickey

Downward percolation

A change rarely remains restricted to one layer in a society. For the Dublin Vowel Shift a phenomenon can be observed, albeit embryonically, among local varieties of Dublin English. This is what I term "downward percolation". It denotes the adoption of the shift by speakers who would not normally show it, as they have come to realize that it is typical of more prestigious speech in the city, e.g., in pronunciations like [stail], [taim], [maild], etc. If this happens on a broad scale, the ultimate fate of the shift is then uncertain. What has started as a feature unique to socially pretentious speakers in Dublin may well spread vertically in the city (as it is beginning to do regionally for many younger-generation urbanites) and lose the significance it has at the moment as a delimiting factor vis a vis the lower classes in the capital.

4. Phonological interpretation 4.1.

Vowel shifts and the notion of vowel space

The Dublin Vowel Shift can be seen to be taking place within a constellation of vowel values which have been affected by the re-shuffling of the initial shift, most probably the retraction of /ai/. Now there is historical support for the notion of vowel space, in the movements which constitute the English vowel shift, as has been discussed exhaustively. Recall that the articulatory positions for many vowels conflict with the traditional vowel quadrangle used in phonetics, the vowel Id is slightly higher than /u/, /i/ much more so, while lol and /o/ are on the same level, the latter somewhat more forward than the former (Lass 1984: 119). However, from an acoustic point of view there is more justification for the vowel quadrangle. If one plots the first formant on a vertical axis against the difference between the second and first formant on a horizontal axis then the result looks reasonably like a vowel quadrangle (again cf. Lass 1984: 120-121); above all the correspondences in height seem to be what historical changes and synchronic alternations would lead one to expect, e.g., Id and lol are on the same level, which is in accord with umlaut rules like that of modem German (Sohn - S hne), which is a fronting rule with no other change in the parameters for the vowel affected. This is true historically for the former English umlaut rule as in foot —feet and for changes such as the retraction of Id to /o/ in Irish (later unrounded to /Λ/) as in eochair 'key' /exsr,/, later /oxsr,/, now /ΛΧΟΓ.,/. The upshot of these considerations is that there is a certain validity to the claim that speakers are unconsciously aware of acoustic phonological space (irrespective of articulatory implementation).

The Dublin Vowel Shift and the historical perspective

4.2.

97

Where can low vowels go?

One can say that if there is a shift in the onset of a low diphthong then one of three situations occurs:

1) 2) 3)

centralization fronting retraction

/ai/ /ai/ /ai/

->· -> —»

hi/ /ei/ /ai/

It is not possible to predict what track will be taken. Retraction of diphthongs and of low vowels in general is well attested. There is the retraction of /a:/ and raising of low back vowels with West Germanic /a:/ which traveled on a back track in English but on a front one in German and Scottish English, e.g., OE ham /ha:m/ to modern Received Pronunciation /haum/ from earlier /ho:m/ but Scottish /hem/ and Modern German /haim/ from earlier /heim/. Furthermore, there may be a change in direction within a short period of time. If the observations on /ae/ in Received Pronunciation for the present century are correct (Bauer 1994: 120-121), then Received Pronunciation would seem to have gone through a raising phase at the beginning of the century, and in the post-war period to have been subject to a tendency to lowering (going on evidence from Jones and Ward on the one hand, and from Wells, Gimson, and Bauer on the other). In the opinion of the present author the motivation for shifts is to be found externally and the direction taken is also due to external stimuli, bearing in mind language-internal constraints, such as the occupation of vowel space, at the same time. The vowel constellations in Dublin English and the shifts taking place at the moment show that there is no necessary correlation between the short and long low monophthongs /a/ and /a:/ and the starting point of the diphthongs /ai/ and /au/. These happen to be the same in the conservative supra-regional standard of the south, but in Dublin English the /au/ has a rather fronted starting point ([seu] or [eu] in more extreme forms), and the /ai/ diphthong has a centralized starting point in popular varieties of Dublin English and a retracted low back onset in more fashionable varieties. The low monophthongs show no noticeable difference on a front-back axis, except that there is a retracted conditioned allophone before /r/ as in card [kaad]. In local Dublin English, if anything, there is a fronting of /a:/ especially before /r/ as in He parked the car [hi paeikda kae:]. This in fact runs counter to the statistically more usual retracted realizations of long /a:/, e.g., in all European languages, bar Hungarian and Dutch. Educated speakers avoid such a fronted variant for the phonemically long low vowel, i.e., a tense, raised realization such as [lae:s(t)] would be less acceptable than [la:st] for last.

98

Raymond Mickey

5. Who is involved in the vowel shift? All the remarks so far have concerned the linguistic behavior of Dubliners, seeing as how it is their sound shift, so to speak. But in a country as capital-city-oriented as the Republic of Ireland the question must be asked to what extent speakers outside of Dublin are affected by changes in the English of the capital. The behavior of speakers from other urban centers outside of Dublin — Cork, Waterford, Limerick, Galway, etc. — is particularly relevant to the issue at hand. There is strong motivation on the part of urban dwellers outside Dublin to imitate features of the capital city. After all these people could choose to remain non-participants as they have done with the retraction and rounding of /ae/ in northern Irish English (as in family [fomli]). To handle the contrast of capital-city and other speakers descriptively a distinction, for which there would appear to be considerable justification, would seem necessary, that between motivated and detached participants in a sound change.

5.1.

Motivated participants

Motivated participants are those who are taking part in the shift in order to hive themselves off from the lower-classes in Dublin. Their behavior provides the reason for the shift and this group comprises the initiators of the shift.

5.2.

Detached participants

The second group is that of detached participants. They can be characterized as those speakers who have realized that there is a shift taking place but have decided to participate for a different motivation. Their motivation is secondary: it is to associate with the initiating group of speakers, and not, in the case of the Dublin Vowel Shift, to separate themselves from lower-class Dubliners in their speech. They are at one remove from the initiating group as they do not share their motivation for participation. It is a moot point whether detached participants are by definition outside of the geographical area where a change is taking place (e.g., those Dubliners who do not share the pretentiousness of many would-be sophisticated urbanites and are, hence, not among the motivated participants may nonetheless become infected by the change). This is the case, however, with the Dublin Vowel Shift as the detached participants are those speakers outside of Dublin who are showing signs of the shift. It is important to stress that the detached participants are reacting to a shift which has already been initiated, i.e., they have not started it. Furthermore, the prototypical detached participant is an adult. However, it may be that the children among the detached participants (here the urbanites outside of Dublin) are

The Dublin Vowel Shift and the historical perspective

99

innovative themselves as they show vowel realizations which are frequently in the range of those of advanced motivated speakers (here fashionable Dubliners). What may very well be the case is that the children have a better grasp of the thrust of the shift and, as opposed to the adults who perceive it in a lexically diffusional manner, recognise the trajectories of the shifting vowels and can thus push them further than the adults from whom they have picked up the change. This would be another instance of children in language acquisition overtaking their parents. The distinction between groups of participants is important in itself but it is of further significance with respect to the linguistic manifestation of the shift which they display.

6. Propagation of sound change 6.1.

Neogrammarian advance and lexical diffusion

The Neogrammarian advance of a sound change implies that any input which matches the structural description for the change is affected, i.e., the change is phonetically gradual and lexically universal. In the present case this means that all instances of /ai/ are retracted. If this were the entire story then there would be no exceptions: we would have the much maligned Ausnahmslosigkeil. This is not quite the case, however, as there is the question of phonetic recalcitrance with potential input to a change. 6.2.

Phonetic recalcitrance

This acts as a brake on the Neogrammarian advance. If the input to a change has some feature which acts against the change then the input form is recalcitrant. It would seem to be a frequent occurrence that a sound change in its initial period affects those segments which phonetically present a most natural input, or put the other way around, those words where some phonetic aspect would militate against a change are affected last, if at all. A good example of this is the 17th century lowering and unrounding of /u/ to /Λ/. This did not occur in those words where the following segment had inherent lip rounding (a feature in common with /of), e.g., pull, bull, push, bush versus pun, bun where the [f] and the [1] prevented the shift (put and but are conflicting examples, however). The change later lost momentum and subsequent shortenings of /u/ as in took, cook did not undergo it. With reference to the Dublin Vowel Shift the cases of phonetic recalcitrance are those where /ai/ occurs before a voiceless segment which, because of its fortis quality, results in a somewhat shorter vowel preceding it so that the tongue has less time to move down and back to the position for [αϊ]. Essentially the lexical diffusion hypothesis claims that a change starts with some words and spreads to others, encompassing the entire vocabulary of a language,

100

Raymond Hickey

given the important proviso that the change does not lose momentum, i.e., that it is carried through along the S-curve to 100%. With the lexical diffusion model, a question arises which is not of relevance with the Neogrammarian model, i.e., what words are affected and if there is any generalization which can be made as to those which first undergo the change, i.e., how does the change progress through the lexicon.

6.3.

Different participants and the progress of the sound change

Before continuing I should note that there is something like a standard wisdom on the occurrence of Neogrammarian advance versus lexical diffusion. Labov (1981: 304) maintains that "low-level output rules" typically show gradual change across the board (Neogrammarian advance), whereas "changes across subsystems", e.g., long to short vowels, proceeds by a process of lexical diffusion. The latter contention would seem to be upheld by the many vowel shortenings in the history of English, e.g., that of/ε:/ before alveolars as in red, dead, and the later shortening of/o:/, again before alveolars (blood, flood), and still later that of/u:/ before velars (look, took, cook). These are all instances of lexical diffusion as they have not encompassed every possible input. The question is whether Neogrammarian advance always applies to Labov's "low-level output rules". The level one is referring to here is that of phonetic realization — call it low-level, allophonic, post-lexical, or whatever. For a sound change in progress this is probably the only level which can be examined, as a change in progress will not have reached phonological stability. Now the current investigation of the Dublin Vowel Shift shows quite clearly that motivated participants — fashionable Dubliners — display the Neogrammarian advance for the shift whereas detached participants — socially conscious urbanites from outside Dublin — exhibit lexical diffusion. With the group of detached participants the first word to show the Dublin Vowel Shift is Ireland and its derivative Irish. This is almost a test case, a keyword, for those speakers who are beginning to participate in the shift. Note that the word does not belong to the core vocabulary of the language like parts of the body or verbs denoting basic activities or whatever. But in an Irish context it stands to reason that the name of the country and its people, as it has a vowel which is a potential input to the vowel shift, is something of a keyword for the adoption of the change. The keyword view of lexical diffusion is closely linked to the notion of salience of certain words. Often the words are used as carrier forms for a characteristic pronunciation of a group. In this respect one can cite the high back /u/ of local Dublin English which is frequently used when pronouncing the name of the city, i.e., as [dublan], by speakers who normally have [dAbbn], particularly in colloquial registers.

The Dublin Vowel Shift and the historical perspective

101

The process of lexical diffusion for the diphthong shift among detached participants can be taken as applying to keywords and particularly common lexical items, e.g.: a. b. c.

6.4.

the keywords Irish and Ireland the numerals/zve and nine various commonly occurring adjectives like wild, mild, kind; nouns like time, mind, side; verbs like rise, drive, hide, etc.

Types, tokens, and lexical diffusion

The theory of lexical diffusion implies, furthermore, that not only does a certain change — a new vowel value — spread gradually through the lexicon of the variety/language affected, but also not all tokens of a given type (lexeme) exhibit the new pronunciation immediately. This is clear in the group of detached participants and, if the author's observations on the Dublin Vowel Shift are correct, it would seem that the older members of this group show the new pronunciation for given lexemes only for a low percentage of tokens, e.g., the realization [aubnd] rather than [auland] is found with only some tokens of the country's name. Any situation like this with co-variants occurring alternatively implies that there is external conditioning on their occurrence. The circumstances for the use of the retracted diphthong in Irish English are something which is certainly sensitive to social factors in discourse settings.

6.5.

Why is lexical diffusion typical of the detached participants?

The question arises from the observations made above as to why lexical diffusion is typical of detached participants. The answer lies in the lack of motivation. For urbanites outside Dublin there is no reason to use a different realization from that which they acquired natively, i.e., [ai]. Hence, they do not grasp the motivation among their metropolitan counterparts actively involved in the shift. They adopt the shift as they are confronted with it in words with high salience (Ireland and Irish) and/or high statistical frequency (numerals, common adjectives, nouns, verbs, etc.).

6.6.

How long are there two types of propagation?

It goes without saying that a vowel shift must pass a perception threshold before it will be noticed by non-participants, and these then become detached participants if

102

Raymond Hickey

the conditions are right — as they are in urban centers outside Dublin vis-a-vis the capital. So how long are there two types of propagation? The distinction in the course of a sound change — Neogrammarian advance or lexical diffusion — would appear to hold most clearly while the shift is taking place. Furthermore, depending on such aspects as the quantitative relationship of motivated participants to detached participants, the latter may be marginalized by the former, and the shift takes place fully. Or the level of exposure may lead to detached participants adopting an even increasing number of words with the new pronunciation, and this may eventually lead to the demise for them of their old pronunciation as it does not survive anywhere in their lexicon. Both these situations would mask the stage of lexical diffusion and make the sound change appear to have proceeded by Neogrammarian advance.

6.7.

Lexical diffusion or later superimposition of standard layers

Apparent later survival of lexical diffusion can have a variety of causes. One of these is seen in Dublin English and is a pitfall for the linguist. There was a raising of /ae/ to /ε/ in Dublin English in the early modern period as in /keif/ for catch (Sheridan 1781: 144). This pronunciation has receded but is occasionally present in local Dublin English. The danger here is that one might be tempted to posit a general shift of /ae/ to Id which took place by lexical diffusion, encompassing words such as catch, but petering out before the entire inventory of lexical items with /ae/ was subject to the change. The warning which must be made here is to distinguish identifiable cases of lexical diffusion (vowel shortenings, e.g.) from those of an earlier change being masked in its extent by the later superimposition of more standard varieties.

6.8.

The termination problem

Concluding this paper it would seem advisable to address briefly an issue which is called the termination problem. It really involves two questions. The first is: when is lexical diffusion complete? The answer is simple: when there is no instance of the original sound value left, e.g., the shortening of English long /u:/ is terminated when there is no instance of it left, though new ones may later arise, just as new instances of long /a:/ arose in Early Modern English due to lengthening before voiceless fricatives and before /r/ prior to it being lost. The second question is when is a change, which is taking place by Neogrammarian advance, complete. The answer here is that there is no termination point. There is simply a stage when speakers regard the change as having crossed a phonological threshold, i.e., when a pronunciation is assigned to one phoneme as

The Dublin Vowel Shift and the historical perspective

103

opposed to another. This is particularly clear when phonemic contrast either arises or is lost. But when the new pronunciation is moving towards a value not already present in the phonological system of the variety concerned — as is the case with the new [αϊ] of the Dublin Vowel Shifi — then there is no given end-point. Furthermore, apparent end-points can be passed through with or without mergers — witness on the one hand the merger of /o:r/ (court) with /o:/ (caught) and their later raising in advanced RP to [o:] and, on the other hand, recall the diphthongisation of ME /i:/ and the raising of ME /ai/ with different results in modern English (though whether they intersected is a matter of much debate). It is too soon to say what will happen to the Dublin Vowel Shifi. For its continuation the behavior of the motivated participants in the change is most important. Assuming that the unarticulated goal of these speakers is to evolve a form of speech phonetically distinct from that of local Dublin English, then that goal is all but reached. The shift has created new allophones [αϊ] for former [ai], and [01], [01] for former [01], but there are no threatening mergers so that there is no system-internal pressure to continue on a shift cycle and to re-align phonemic oppositions which are under threat at the moment. The present situation is one in which the change could simply fizzle out — the firework type of sound change. If this happens now or in the near future then we will be left with a situation like that in Canadian English: phonetically recalcitrant examples — /ai/ before voiceless segments — will not be subject to the shift. On the other hand, the change may continue and regularize the allophony of the segments involved. But there is no internal reason for this; in the final analysis it rests in the unconscious judgment of the speakers who are the motor of the change to decide whether they have crossed the finishing line, indeed to decide if a finishing line exists at all.

References Aalen, Frederick H. A.—Kevin Whelan (eds.) 1992 Dublin city and county: From prehistory to present. Dublin: Geography Publications. Bauer, Laurie 1985 "Tracing phonetic change in the received pronunciation of British English", Journal of Phonetics 13: 61—81. 1994 Watching English change. London: Longman. Bertz, Siegfried 1975 Der Dubliner Stadtdialekt. Teil I: Ph nologie. [Unpublished Ph.D. dissertation, University of Freiburg.] 1987 "Variation in Dublin English", Teanga 7: 35-53.

104

Raymond Hickey

Bliss, Alan J. 1976

"The English language in early modern Ireland", in: Theodore W. Moody—Francis X. Martin—Francis Byrne (eds.), 546-560. Chambers, Jack K. 1973 "Canadian raising", Canadian Journal of Linguistics 18: 113135. Chen, Matthew—William S-Y. Wang 1975 "Sound change: Actuation and implementation", Language 51: 255-281. Clery, Arthur E. 1921 "Accents: Dublin and otherwise", Studies 10.40: 545-552. Cosgrove, Art (ed.) 1988 Dublin through the ages. Dublin: College. Edwards, John 1977a "Students' reactions to Irish regional accents", Language and Speech 20: 280-286. 1977b "The speech of disadvantaged Dublin children", Language Problems and Language Planning 1: 65-72. 1979 "Judgements and confidence in reactions to disadvantaged speech", in: Howard Giles—Robert St.Clair (eds.), 22-44. Giles, Howard—Robert St.Clair (eds.) 1979 Language and social psychology. Oxford: Blackwell. Gimson, Alfred C. 1994 Pronunciation of English, revised by Alan Cruttenden, (5th edition.) London: Arnold. Hickey, Raymond 1984 "Syllable onsets in Irish English", Word 35: 67-74. 1997 "Change from above and the development of Dublin English", in: Ernst H. Jahr (ed.). Jahr, Ernst H. (ed.) 1997 Historical sociolinguistics. Berlin-New York: Mouton de Gruyter. Jones, Daniel 1975 An outline of English phonetics. (9th edition.) Cambridge: CUP. Kelleher, Terry 1972 The essential Dublin. Dublin: Gill & Macmillan. Labov, William 1981 "Resolving the Neogrammarian controversy", Language 57: 267-308. 1994 Principles of linguistic change: Internal factors. Oxford: Blackwell. Lass, Roger 1984 Phonology. Cambridge: CUP.

The Dublin Vowel Shift and the historical perspective

105

McMahon, April 1994 Understanding language change. Cambridge: CUP. Moore, Desmond F. 1965 Dublin. Dublin: Three Candles. Ossory-Fitzpatrick, S. 1977 [1907] Dublin, a historical and topographical account of the city. Cork: Tower Books. Sheridan, Thomas 1781 A rhetorical grammar of the English language calculated solely for the purpose of teaching propriety of pronunciation and justness of delivery, in that tongue. Dublin: Price. Trench, William F. 1936 "English in Dublin", Times Literary Supplement 184. Wang, William S-Y. 1969 "Competing changes as a cause of residue", Language 45: 9-45. Ward, Ida C. 1948 The phonetics of English. (4th edition.) Cambridge: Heffer. Wells, John C. 1982 Accents of English 1-3. Cambridge: CUP.

On the ideological boundaries of Old English dialects Richard M. Hogg

Most papers on Old English dialectology are scarcely distinguishable from papers on Old English dialects. Yet this is not a situation which we should view with untroubled equanimity. For, whenever anyone writes on Old English dialects they are making assumptions about the nature and structure of Old English which should not necessarily go unchallenged. This paper, therefore, reverses the usual scene, perhaps to an extent with which not everyone will feel comfortable. For the intention of this paper to is examine and attempt to understand the most common assumptions held today about the nature and structure of Old English dialects. One consequence of this is that there are here virtually no references to Old English textual evidence, and it is certainly the case that there are no attempts to assign any texts to any particular dialectal location. Nevertheless, I believe that what I have to say will be of relevance to specialists in Old English and will be capable of application to empirical studies of Old English. The purpose of this paper, therefore, is to consider the kinds of approaches to Old English dialectology which are current both today and in the past, and consider why it is that the dominant approach to dialect variation in the Old English period is as it is (and has been for many decades). I want to argue, in effect, that the kind of approach to Old English dialects which is most widespread and most generally accepted has both linguistic and ideological underpinnings which militate against a proper understanding of the Old English dialect situation. This is not to deny that good work is being done, some of it, such as Kitson (1990), influenced by parallel work on Middle English as in The linguistic atlas of Late Medieval English by Mclntosh—Samuels—Benskin (1986). But in general the dominant paradigm remains a quite different one, and requires a more skeptical investigation than it has usually received. It may be agreed that the two standard works on the language of the Old English period which present a fairly recent view of the dominant approach are, for morphology and phonology, Campbell's (1959) Old English grammar, and, for syntax, Mitchell's (1985) Old English syntax. Mitchell has nothing substantial to say about dialects per se, although obviously there are substantial remarks about syntactic differences between dialects. In essence, Mitchell simply follows Campbell's presentation of the fundamental dialectal characteristics of Old English. Let us, therefore, look to see what Campbell himself says. Two quotations from Campbell (1959) may be enough to give a flavor of what this dominant view (which I shall call the "modern traditional" view) of Old English dialects might be. The first is from § 6:

108

Richard M. Hogg

In the extant Old English monuments four well-marked dialects are to be traced, Northumbrian, Mercian, West-Saxon, and Kentish.1 Northumbrian and Mercian are associated as the Anglian dialects. As you will have noticed, there is a footnote attached to the first sentence. We shall come back to that footnote quite shortly. But before we do that, let us link the second quotation to the first. This quotation comes from much later in the book, in a subsection entitled "Dialectal summary", and, more specifically, from § 256: It must first be emphasised that the dialectal names are in this book used practically without claim to territorial significance. I have argued elsewhere (cf. Hogg 1988) that such statements make it hard to comprehend the nature of the ontological commitment which Campbell is making when he talks of, say, a West-Saxon dialect. It would appear that in § 256 he is, at the least, explicitly denying any geographical commitment. It is this denial which is at the heart of this paper. I have for a long time worried about its significance and purpose. Why should Campbell have made such a statement, and what is its intention? Neither question seems to me to be easy, but both seem to me to be important. Here I should stress that nothing in this paper is to be seen as a criticism of Campbell. Rather, what I am trying to do is understand him. For it is only when we fully understand what precisely Campbell is doing that we can use his insights to develop a fuller understanding of Old English dialect geography. And towards the end of this paper I shall indeed argue that Campbell made significant steps forward in helping us to an understanding of Old English dialectology, even if we have, eventually, to continue in an entirely different direction from the one that he proposed. The footnote from Campbell which I mentioned earlier gives a brief historical survey of when the Old English dialects were first identified by philologists. A careful reading of that footnote will make it clear that the most significant steps in establishing the now traditional view of Old English dialect variation were taken in the ten years either side of 1875. Indeed, it is exceedingly tempting to suggest that almost the whole of the traditional view of Old English dialects was formed in the period bound on the one side by Henry Sweet's 1871 edition of the Cura Pastoralis and on the other side by his 1885 edition of The oldest English texts for the Early English Text Society. Sandwiched in between these is his 1876 paper to the Philological Society entitled "Dialects and prehistoric forms of Old English", his Presidency of the Philological Society, and, of course, Oxford's recognition of his qualities by the award of a Fourth Class Honors B.A. degree. And in Germany, we must not forget, there was a contemporaneous linguistic event of the greatest moment, namely the Neogrammarian revolution of the mid-1870s. I do not want, by the above remarks, to imply that there were no contemporaries of Sweet making substantial contributions to the study of Old English dialects. That is clearly untrue, and one need only think of Walter W. Skeat's comparative

On the ideological boundaries of Old English dialects

109

edition of the Four Gospels (1871-1887) to see that others were making critically important contributions. On the other hand, the evidence that Sweet was the central figure in the establishment of the fundamental assumptions underpinning the traditional study of Old English dialect variation seems to me indisputable. At every point there is ample evidence to show that Sweet was pre-eminent either as a reflector of contemporary linguistic thinking or as a leader of such thought. In fact, the latter alternative appears to be more probably the case. And Sweet's leadership is not confined to Britain, for Eduard Sievers, the greatest of the German anglicists of the time, wrote in his 1882 Angelsächsische Grammatik: Die ersten nachhaltigen anregungen zu einem historischen Studium des angelsächsischen und die erste grundlage einer angelsächsischen dialektkunde verdanken wir Henry Sweet. And this quotation is repeated with enthusiasm in the 1888 Altwestsächsische Grammatik of the Dutch scholar Peter J. Cosijn. But the primary intent of this essay is not to create or further strengthen any cult of the personality of Henry Sweet. All I intend by the above remarks is to establish, if it were in any doubt, the centrality of Henry Sweet in the study of Old English dialects. Furthermore, this is essentially a matter of convenience. It is helpful to have an agreed source of authority for particular views on the relevant topic. What I want to argue is that the kinds of views of Old English dialects which were propounded by Sweet and others in the late nineteenth century, and which remain more or less intact in modern traditional accounts, were created by the conjunction of two quite different trains of thought. On the one hand, and from the scientific point of view, there was the burst of theoretical innovation in linguistics, perhaps above all exemplified by the Junggrammatiker movement in Leipzig, but where Sweet was by no means the least of the new linguists. On the other hand, there were beliefs about nationhood and the structure of society, of which Sweet, like others, was no doubt little more than the unconscious sharer. These elements seem at first sight to be quite unconnected, and it is probable that linguists such as Sweet rarely if ever saw any connections. Let me now try to persuade you otherwise. Sweet's first important work was his 1871 edition for the Early English Text Society of Alfred's translation of the Cura Pastoralis. That there was a double motivation for undertaking that work can scarcely be disputed. Linguistically, the principle had long been recognized that older specimens of a language were more likely to be archaic and show older stages of development than younger specimens. In parenthesis we might note that, although this principle looks tautologous, it is demonstrable that it is not always true; on the other hand, it is at worst an extremely useful heuristic for the historical investigation of language. Consequently, Alfred's text is of particular linguistic importance, for it is one of the earliest substantial works in any of the varieties of Old English. As Sweet himself writes (1871: xix):

110

Richard M. Hogg

A curious feature in the history of Old English philology is the neglect of the older documents of the language: not only are the forms that appear in our grammars and dictionaries West-Saxon, to the almost entire exclusion of the equally important Anglian and Kentish dialects — they are to an equal extent, late, as opposed to early West-Saxon. A quick reading of the preface to his edition of The oldest English texts (Sweet 1885) makes it clear that the edition of the Cura Pastoralis was only one part of a larger project to provide reliable editions of all of the earliest texts. But there is another relevant factor. Outside linguistics Alfred is one of those semi-legendary figures in the pantheon of English history, someone often described, in terminology which I find alternately ludicrous and alarming, as one of the founding fathers of the English nation. In this way, the purely linguistic choice of the Pastoral Care is reinforced and confirmed by contemporary cultural predispositions. E.g., March in his 1870 Comparative grammar of the Anglo-Saxon language, which Sweet knew and praised, said of the Anglo-Saxon period: its Augustan age was the reign of Alfred the Great, King of the West-Saxons. What higher praise could a Victorian give to an earlier age? None of this, however, need have had any interesting, let alone dangerous, consequences, if it had not been for the presence of a catalyst which caused these two elements to combine in a mixture of explosive force. To adopt the unfortunately patronizing tone of too many later students of Old English, the problem with Sweet's edition was that it was the first "modem" edition of an Old English text, an edition which in its methodology and skill set the standard for any future editions of historical texts. Consequently, and indeed as Sweet had hoped, the first grammars which were produced took as their basis the Early West-Saxon of the Cura Pastoralis and the other apparently Alfredian texts. This is most evident in Cosijn (1888), but also, and more significantly, in Sievers' Altenglische Grammatik (1882, 1886), for that was the grammar which was to set out the parameters for all (or nearly all) subsequent grammars. By as early as 1913 Sweet's position was being unconsciously parodied: There can hardly be any doubt that all practical teachers of the subject will agree that it is far better and easier for the student to take early West Saxon as the standard for Old English, and to group around it the chief deviations of the other dialects, than to start with a grammar which treats all the dialects as being of equal importance. (Wright—Wright, 1913: vii) This kind of statement, which directly contradicts the remarks of Sweet which we quoted earlier, is all the more remarkable coming from the pens of Joseph Wright, the author of the English dialect dictionary, and Elizabeth Wright, the author of one of the most substantial works on Northumbrian (Lea 1894).

On the ideological boundaries of Old English dialects

111

What were Sweet's own views on Old English dialects? His initial views are given in great detail in his 1876 paper "Dialects and prehistoric forms of Old English". Here he recognizes the existence of the four dialects cited in the quotation from Campbell (1959) at the beginning of this paper, although with some differences in the dialectal attribution of particular texts, most notably the Vespasian Psalter. By 1888, however, his account of Old English dialects is similar to that found in almost every modern textbook, unremarkably so, of course, since they all tend to take as their basis Sweet's views. Why four dialects? Why are these four dialects Northumbrian, Mercian, West Saxon, and Kentish? Yet again, I believe, we have to come back to the twin factors of linguistic theory and cultural pre-occupations, and see how they combine together. By the time Sweet was writing, and the map of Old English dialects was being drawn by him and others associated with the Philological Society, the concept of the Stammbaum had already been developed by August Schleicher in Germany. This sees the development of language as like a family tree, so that Sweet's views of 1876 can be represented as follows: Proto-OE

Anglian I Northumbrian

Kentish

West Saxon

I Mercian

Stages in the development of a language into dialects (and further languages) are represented by branches developing from nodes. The above diagram is a by no means unusual specimen, and might appear relatively harmless. Consider, however, the results presented in a paper "On the classification of Modern English dialects" presented to the Philological Society in March 1875 by Alexander Ellis, with the co-operation of Prince Louis Lucien Bonaparte and James Murray: The dialects are distributed into 3 families, 7 branches, 13 dialects, 42 sub-dialects and numerous varieties and sub-varieties (Ellis 1875) Two points should be clear. Firstly, given the authorship of this paper, this represents the generally accepted view of the leading British linguists of the time. Secondly, the results are the consequence of an overwhelming ambition to taxonomize which carries the writers beyond the bounds of plausibility. But this is the inevitable result of the Stammbaum method, which, with its concentration on evolution through division, emphasizes splits and irrevocably separates out those which are not clearly distinguishable.

112

Richard M. Hogg

Evidence of the influence of the Stammbaum theory is easily obtained. Old English dialectologists have consistently searched for linguistic features which would confirm the existence of particular branching nodes. Thus, e.g., the sound change known as Anglian Smoothing is commonly held to distinguish Anglian texts from all other Old English texts, and similarly the pronunciation (or, rather, the spelling) of the word for deed as dsed demonstrates a West Saxon text as opposed to the form ded which shows a non-West Saxon text. On the other hand, dialectal distributions which run contrary to Stammbaum theory, such as the various monophthongizations, have usually been considered only atomistically and without reference to their impact on the general theory of Old English dialects. Without going into any more detail than that, it is quite clear that such procedures would be wholly unacceptable to present-day dialectologists. But rather than considering in detail the theoretical shortcomings of the Stammbaum method, I want to consider briefly a further aspect of the Ellis paper. For it is interesting to note that the overwhelming majority of the 42 "sub-dialects" turn out to be based on counties, i.e., on basic inherited English political divisions. This gives us a clue as to the four Old English dialects which we inherit from Sweet: Northumbrian, Mercian, West Saxon, and Kentish. For these dialect names correspond to four of the seven kingdoms of the heptarchy, of which the OED says: The term appears to have been introduced by 16th c. historians, in accordance with their notion that there were seven Angle and Saxon kingdoms so related that one of their rulers had always the supreme position of King of the Angle-kin (Rex gent is Anglorum) 'so that in the heptarchy itself there seems always to have been a Monarchy' (Camden)). The correctness and propriety of the designation have been often called in question, but its practical convenience has preserved it in use. (OED: Heptarchy) The naming of the four dialects, therefore, is based upon political divisions: the four dialects correspond to the four kingdoms of the heptarchy from which texts are believed to originate. Thus, we see fulfilled here the double aim of having a taxonomic description into discrete dialects, as directed by the methodology of the Stammbaum, and the alignment of these dialects into political divisions which are the Anglo-Saxon counterpart of the Victorian counties in their contemporary descriptions of the Modern English dialects. Note, however, that there is a level of sophistication here which seems too often to have eluded later scholars, for the OED entry makes it clear that the concept of the heptarchy is dubious. We may, therefore, assume that the division into four dialects is a primarily a matter of convenience rather than of dogma. Later studies, which have demonstrated quite clearly that ecclesiastical regimes were probably of more importance than political regimes in determining the linguistic features of individual texts, should, in these circumstances, have caused a serious re-examination of the nature of Old English linguistic variation.

On the ideological boundaries of Old English dialects

113

But alas, that has generally not been the case. Indeed the dominant approach has been, in the words of Wrenn (1933: 65): ...the popularising and cheapening of the road first broken by Sweet, through his less accurate and less scientific successors amongst whom he seems to include, fairly or unfairly, the Wrights, Ritchie Girvan, and Ferdinand Holthausen. An example from over half-a-century later is to be found in Mitchell—Robinson (1992: § 217), which on the back cover is described in an extract from a TLS review as "The best available introduction". The following is a somewhat abbreviated quotation: Thus by c. 700... the four dialects mentioned in § 2 — Northumbrian, Mercian, West-Saxon, and Kentish — can be distinguished in a language which at the time of the invasions appears to have been spoken in much the same way by all those who came to England... There was certainly much fighting between the various kingdoms, with now one, now another, temporarily 'top-dog'. Then came the Danes... One by one, the kingdoms of Northumbria, East Anglia, and Mercia, ceased to exist... and in 878 Wessex too was nearly extinguished... But Alfred was equal to the challenge. His grasp of the principles of war... and his activities in education, learning, and administration... are such that, for some people at any rate, his only rival for the title 'The greatest Englishman of all' is Sir Winston Churchill. Even, or especially, as an introduction to Anglo-Saxon language and history, such a statement has little to commend it. The virtual deification of Alfred, whose Pastoral Care Wrenn (1933: 67) described as "a certain very dull translation", has, in the present context, the virtue of confirming my earlier remarks on his semi-legendary status. The apparently unquestioning acceptance of the Heptarchy is in marked contrast to its carefully-weighted description in the OED. And linguistically, the notion that Old English arrived in this country as a virtually undifferentiated and homogeneous language merely reflects Sweet's own remarks over a hundred years previously (1876: 560-561): What were the dialectal distinctions in English during the first few centuries of the conquest of Britain? The answer is that they were very slight. There seems to me little point in discussing this passage in more depth. Instead I wish to turn my attention to a rather different and more sophisticated analysis of Old English dialects, namely that presented in Campbell (1959), with which we started this paper. The two quotations from Campbell which I gave earlier, when set alongside one another, make extremely puzzling reading. Now, however, I think we are in a

114

Richard M. Hogg

position to conjure some sense out of them. The first quotation is clearly a restatement of the position taken by Sweet in 1876. But having made that statement, Campbell later recognizes that such a position is apparently irreconcilable with the facts. In this light consider the following quotation from Campbell (1959: § 11): The great phonological feature which connects Mercian and Northumbrian is the Anglian 'smoothing'. When a text exhibits this, but lacks the peculiar features of Northumbrian, we are justified in regarding it as Mercian. By this means we are able to add to the Mercian material the interlinear glosses on the Vespasian Psalter and the Rushworth Gospels..., although they represent in many respects highly divergent dialects. I have puzzled over this statement for a very long time, trying to work out what it might mean. If these two glosses represent highly divergent dialects, how can they both be Mercian? Does Campbell simply mean that they are different sub-dialects of Mercian? Current terminology usually follows this interpretation, and we find regular references to the Vespasian Psalter as West Mercian and the Rushworth Glosses as North Mercian. But closer inspection and deeper reflection suggests that this is not what Campbell intended. After all, the first part of the quotation distinguishes Mercian and Northumbrian in a way which suggests that they are not "highly divergent", and that can be confirmed by the frequent use of the term Anglian to cover both Mercian and Northumbrian. Is it, then, simply a reluctance to invoke a further dialect beyond the four canonical dialects? That might seem to be at least part of the answer. However, such an assumption, which I, for one, have not been loathe to make in the past, now appears to me to be mistaken and also uncharitable to Campbell. My reason for thinking this stems from that second quotation from Campbell, wherein he denies the territorial significance of the dialectal names. For why should he do that, and why should he do that explicitly? The answer to both those questions is, I think, the same. Unlike almost every other upholder of the tradition, Campbell has, with considerable acuity, recognized that that tradition cannot be defended on the traditional grounds. As far as I know, no one has realized this, and, therefore, it has either been thought that Campbell was simply restating the traditional view or, as I have claimed (cf. Hogg 1988 and my earlier remarks) that what he was saying was incoherent and incomprehensible. Both these attitudes now seem to me to be unsustainable. In order to justify this statement I shall have to quote at length from Campbell (1959: § 256). This quotation follows immediately upon his remarks about the lack of territorial significance, and this is how he continues:

Northumbrian means the agreement of Lindisfarne, Durham Ritual and Rushworth 2, and the early names and fragments, or of such of all these sources as offer evidence on the point under discussion. Mercian means

On the ideological boundaries of Old English dialects

115

similarly the agreement of Vespasian Psalter, Corpus, Epinal, and the majority forms ofRushworth 1, or of such as these as afford evidence. Anglian means the agreement of Northumbrian so defined with Mercian so defined. Kentish means such forms in the charters of Kentish origin as are shown to be genuinely Kentish by their reappearance in Kentish Glosses, or the appearance of their obvious reflexes... there. Non-West-Saxon is the agreement of Anglian so defined with Kentish so defined. West-Saxon means the agreement of the majority forms of the four generally accepted early West-Saxon manuscripts... with a large body of later West-Saxon. Late West-Saxon means forms prevailing in a considerable body of the later West-Saxon manuscripts, but not prevailing in the four manuscripts accepted as early West-Saxon.

This passage is as remarkable for its carefully repetitive phrasing as it is for anything else. And it is that phrasing which allows us to understand the passage as a whole. What Campbell is presenting us with here is a set of definitions. And he presents them with an almost mathematical precision, as, e.g., in "Non-West-Saxon is the agreement of Anglian so defined with Kentish so defined". The obvious question to ask is why Campbell should have written this passage. It is, after all, sufficiently unlike his normal style as to attract attention. The answer, I believe, is that Campbell has adopted a theory of dialects, which delineates them as what we might call, along with Lass (1980: 125), "pure structure". In such a theory, which if not Saussurean is certainly Hjelmslevian, there is no ontological commitment at all, except to the abstract structure. This type of theory is common enough in linguistics: Lass (1980), e.g., points to Harris (1951: 18), who says of the constructs of linguistic theory that they are "purely logical symbols, upon which the... operations of mathematical logic can be performed". Indeed, the 1950s, when Campbell was writing his Grammar, was the heyday of so-called "hocus-pocus" linguistics, both in the US with American Structuralism as led by Harris, cf. Householder (1952), and in the UK in the Firthian school of linguistics. For most linguists of the time, therefore, such an approach to the study of language would have been unsurprising. But in Campbell's case there is a real difficulty, in that he was clearly hostile to most linguistic theory — note, e.g., his total avoidance of the concept of the phoneme. But perhaps Campbell was simply breathing the general atmosphere. It would be helpful to know more. One suggestion which I have received, and which it might be extremely fruitful to pursue, is that Campbell was influenced by the early Wittgenstein. This is most plausible, for Wittgenstein's philosophical theories were not, in the 1950s, without influence at Oxford. Whatever the source of Campbell's strategy, to understand it as a form of hocus-pocus linguistics enables us to understand a good deal more of what Campbell is saying. Thus, recall the earlier quotation from the same section of Campbell, "that the dialectal names are in this book used practically without claim

116

Richard M. Hogg

to territorial significance". That, of course, must be the case if the dialect names fail to refer to anything outside the structure which contains them. The earlier quotation from Campbell, which classifies the Vespasian Psalter and the Rushworth Gospels as Mercian "although they represent in many respects highly divergent dialects" now becomes equally clear. The decision to classify them both as Mercian is a structure-internal decision, which follows directly from the enumeration and definition of the Old English dialects as elements of the linguistic structure of Old English. That the Vespasian Psalter and the Rushworth Gospels might on other grounds be classified as different dialects is irrelevant, although it may be worthy of note. Now the introduction of the concept of dialects as pure structure must seem to many present-day linguists to be at best idiosyncratic, at worst ridiculous. I have to say that I find it neither. Rather it seems to me to be an extremely elegant maneuver. At a stroke the cultural presuppositions which we have seen underlying the approach to the Old English dialects disappear. No longer do terms such as Mercian refer to dubious conceptions concerning Anglo-Saxon political structure: their meaning is derived solely from the definitions provided by Campbell. It is of the essence of this type of approach that any other structure may be applied to the varieties of language apparent in the Old English period. The only tests that need to be applied are those of internal coherence, descriptive adequacy, and relevance. And it may well be that that more than one structure may pass such tests, according to the purpose for which each structure is devised; cf. here the important comments byChao(1934). This is not to say that I believe that this kind of approach is correct. Paradoxically, it seems to me that such an approach is not only a significant advance on earlier work but also fundamentally mistaken. It is a significant advance because it lays bare the presuppositions of an earlier age and shows how they can be removed from the argument. On the other hand, it is mistaken because it removes the linguistic study of a language from the political, social, and cultural contexts in which that language is situated. A much harder task is to determine how that context can be restored into the description without the cultural presuppositions of our own age, concerning which we should not attempt to mimic a Victorian confidence. Some scholars, I know, would argue that such a task is impossible. But if that is so, then surely we shall need a mid 21st-century Campbell to demonstrate that the only safe, albeit temporary, retreat is back to hocus-pocus. Yet, we can argue that that is fundamentally wrong, because of the embedded nature of dialectal variation. Rather, what the Old English dialectologist must do is firstly, along with Campbell, recognise the tainted nature of the traditional approaches. Then, however, he or she must look to modern conceptions of linguistic variation and the methodologies used to account for them. To a much greater extent than is, perhaps, realized, such methodologies offer the opportunity for a more adequate description of variation in Old English, in which that variation will be treated to the greatest extent possible in exactly the same ways as variation in current living languages.

On the ideological boundaries of Old English dialects

117

References Campbell, Alistair 1959

Old English grammar. Oxford: Clarendon Press.

Chao, Yuen-Ren 1934 "The non-uniqueness of phonemic solutions of phonetic systems", Bulletin of the Institute of History and Philology, Academica Sinica IV.4: 363-397. (Reprinted in Joos 1957.) Cosijn, Peter J. 1888 Altwestsächsische Grammatik, I. The Hague: Nijhoff.

Ellis, Alexander J. 1875 "On the classification of Modern English dialects" [summary only], Transactions of the Philological Society 16 (Minutes of meetings): 15-16. Harris, Zellig S. 1951 Structural linguistics. Chicago: University of Chicago Press. Hogg, Richard M. 1988 "On the impossibility of Old English dialectology", in: Dieter Kastovsky—Gero Bauer (eds.), 183-203. Householder, Fred W. 1952

Review of Harris (1951), International Journal of American Linguistics 18, 260-268. Joos, Martin (ed.) 1957 Readings in linguistics I. Chicago: University of Chicago Press. Kastovsky, Dieter—Gero Bauer (eds.) 1988 Luick revisited. Tübingen: Narr.

Kitson, Peter R. 1990 "On Old English nouns of more than one gender", English Studies 71: 185-221. Lass, Roger 1980

On explaining language change. Cambridge: CUP.

Lea, Elizabeth M. 1894 "The language of the Northumbrian gloss to the Gospel of St. Mark",AngIia 16: 62-206. March, Francis A. 1870

Introduction to Anglo-Saxon. New York: American Book

Company. Mclntosh, Angus—Michael L. Samuels—Michael Benskin (eds.) 1986

A linguistic atlas of Late Medieval English. Aberdeen: Aberdeen

University Press. Mitchell, Bruce 1985 Old English syntax 1-2. Oxford: Clarendon Press. Mitchell, Bruce—Fred C. Robinson 1992 A guide to Old English. (5th edition.) Oxford: Blackwell.

118

Richard M. Hogg

OED = James A. H. Murray—Henry Bradley—William A. Craigie—Charles T. Onions 193 3 The Oxford English dictionary. Oxford: OUP. Sievers, Eduard 1882 Angelsächsische Grammatik. Halle an der Saale: Niemeyer. 1886 Angelsächsische Grammatik. (2nd edition.) Halle an der Saale: Niemeyer. Skeat, Walter W. (ed.) 1871-1887 The Holy Gospels in Anglo-Saxon, Northumbrian, and Old Mercian versions. Cambridge: CUP. Sweet, Henry 1876 "Dialects and prehistoric forms of Old English", Transactions of the Philological Society 16: 543-569. Sweet, Henry (ed.) 1871 King Alfred's West-Saxon version of Gregory's Pastoral Care 12. (Early English Text Society Old Series 45, 50.) London: OUP. 1885 The oldest English texts. (Early English Text Society Old Series 83.) London: OUP. Wrenn, Charles L. 1933 "'Standard' Old English", Transactions of the Philological Society 32: 65-88. Wright, Joseph—Elizabeth M. Wright 1913 An Old English grammar. (2nd edition.) Oxford: OUP.

The spread of -ly to present participles Kristin Killie

1. Aims and background The present paper is a pilot study to a larger investigation of adverbs derived from present participles, i.e., adverbs such as cunningly and knowingly (Killie forthc. b). Adverb derivation from -ing has been shown to be very productive in Present-day English (cf. Kjellmer 1984; Baayen—Renouf 1996: 82, 88-89; Killie forthc. a); however, current productivity does not necessarily imply productivity in the past. Swan (1990: 49-51) and Breivik—Swan (forthc.) note that present participles in adverbial function are not normally suffixed in Old English.1 Furthermore, it is a fact that languages like Norwegian, Swedish, and German never started suffixing their participles, even though the verbal and adverbial morphology of these languages used to be quite similar to that of earlier English. This, according to Swan (1990: 51), suggests that the unsuffixed form represents the unmarked case in Germanic, and that, consequently, something must have happened to English which did not affect its Germanic sister languages.2 My main objective in this paper is to answer the question of when -ing became productive as a base for adverbial derivation. I argue that the 14th century is the transitional period in this respect. I also briefly discuss the syntax and semantics of the new adverbs, and finally I try to show that the spread of the adverbial -ly suffix to present participles does not represent an isolated fact in the history of English, but is part of a more comprehensive process.

2. Tracing the emergence of the -ingly pattern: Methods and results To trace the emergence (and spread) of -ingly adverbs, I have done two things: I have checked (1) when the adverbs in question are first attested according to the Oxford English dictionary, and (2) their frequencies in real text. A closer description of these procedures are given in the next two sections, together with an account of the results.

2.1.

First occurrences according to the Oxford English dictionary

Most of the -ingly adverbs currently in use are derived from French and Latin verbs (indeed, in my material from the 20th century only 32 of the 150 types recorded can be safely traced back to Old English [cf. Killie forthc. a]). The greater part of

120

Kristin Killie

these verbs entered the English word stock relatively late (i.e., in the centuries following the Norman Conquest or in the Renaissance), and, thus, their adverbial derivatives naturally occur at an even later stage. I suspected that this might in many cases be a stage much later than the time when participles began to serve as a base for adverb derivation. In order to trace the beginnings of the -ingly pattern it thus seemed wise to concentrate on the adverbial history of some of the oldest verbs in English, i.e., those which are recorded in Old English. I, therefore, made a list of 134 Old English verbs. Invaluable to this work were the verb lists in Mitchell—Robinson (1986: 46, 49-52, 152-158); however, I have also included verbs taken from elsewhere. I selected only those verbs which are still in use (with their Old English meanings more or less intact), and whose history can thus be traced by using the Oxford English dictionary.31 ended up with a list of 134 verbs. My next step was to check which of these verbs have (or have had) adverbial derivatives, and when these adverbs are first attested. I let no preconceptions influence my work, but looked up all potential adverbs (however odd they seemed), i.e., I checked if there was a breakingly, a divingly, a swimmingly, etc. The number of adverbs was indeed surprisingly high; in fact, 84 of my 134 verbs had a corresponding participial adverb.4 However, this does not mean that the total of adverbs is 84; it is actually 94, because ten of my verbs have, thanks to prefixation, yielded more than one adverb, e.g., from last we have both lastingly and everlastingly, and from know we have knowingly and unknowingly. Table 1. First occurrences of suffixed present participles.5 pre-12th century willingly (a 1000)

a!205 +hyingly

1150-1250 cunningly +overhyingly

14th century bitingly burningly feelingly holdingly knowingly lastingly +likingly

lovingly lyingly playingly shiningly soakingly stealingly stirringly

understandingly unknowingly unwittingly weepingly wittingly wooingly

longingly meaningly

sparingly unsparingly

laughingly leapingly

unwillingly whisperingly

15th century everlastingly +faringly livingly 16th century +bowingly creepingly

The spread of-ly to present participles

121

Table 1. First occurrences of suffixed present participles (continued). +drawingly gropingly hangingly + haltingly

melfingly mourningly singingly +slrewingly

windingly wonderingly yieldingly

17th century choosingly daringly fightingly floatingly flowingly forgivingly

+goingly helpingly sleepingly slidingly soothingly speakingly

^standingly swearingly swimmingly threateningly workingly

1 8th century flyingly freezingly glidingly

growingly heedingly strikingly

unheedingly unmeaningly unthinkingly

losingly misleadingly shakingly shrinkingly springingly

tellingly thinkingly unseeingly warmingly yearningly

19th century achingly bindingly bodingly drivingly flickeringly grindingly 20th century bruisingly seekingly unforgivingly weavingly

The data in Table 1 show that English adverbial grammar has indeed undergone a change; from not as a rule deriving -ly adverbs from participles, the language has started doing this on a large scale. The 14th century apparently represents a watershed in this respect. Before this century there are only sporadic adverb formations from participles: willingly is coined before the year 1000, cunningly somewhere between 1150 and 1250, and hyingly before 1205 (i.e., most probably in the 12th century). In the 14th century, however, 20 of my Old English verbs yield adverbial derivatives, and from this time on there is a steady growth of new derivations. If we look closely at the adverbs in Table 1, we see that among them there are some which have become firmly established in the language, e.g., willingly, knowingly, and lovingly. By contrast, adverbs like goingly, hangingly, strewingly in (1) below, and several others, never became permanent members of the English

122

Kristin Killie

word stock. They are attested only once or twice and are considered by the Oxford English dictionary to be "nonce-uses", or "dead" to all practical purposes (marked with a cross in Table 1 above and in the entries from the Oxford English dictionary given in (1) below). What is more, some of these highly creative adverbs, exemplified in (1) by hangingly, occur only in dictionaries. It has been noted that dictionaries may in fact list words which are not really in use (Baayen—Lieber 1991: 83), and I believe this to be the case with hangingly and similarly creative-sounding derivations; thus, a number of my adverbs are not really representative of the written English of the time. (1)

f 'goingly adv., at a walking pace. ? nonce-use. 1651 Bedell in Füllens Abel Rediv., Erasmus 73 He can run but goingly, who ties himselfe to another mans footsteps.

•f 'hangingly adv., in a hanging manner. 1548-67 Thomas ltd. Gram., In pendents, hangeyngly, or in double.

f 'strewingly, adv. Obs. rarel. [f. strewing pr. pple. of strew v. + Dispersedly. 1578 Banister Hist. Man vii. 92 b, The flesh of Muscles..is for the most part one kynde [of fibre], and those more strewyngly set.

It is difficult to know how to relate to these odd-looking formations, which appear in some cases to be nothing more than a play with the morphological resources of the language. However, even though they never took root in English, I believe they are quite telling of what the linguistic situation must have been like at the time. The mere fact that writers felt at liberty to coin such odd adverbs must mean that present participles had acquired a status as potential bases for adverbial derivation; in other words, the writers in question were capitalizing on a linguistic trend. The fact that most of these creative adverbs occur in Early Modem English and not in Middle English may be further evidence that the pattern was still new in the Middle English period, if we assume that normally speakers have to get used to a construction before they start to play around with it.

2.2.

Frequency of use in Middle English and Early Modern English

My next step was to study the frequencies of -ingly adverbs in real text. I focused on the Middle English and Early Modern English periods because the data in Table 1 had confirmed my suspicion that the Old English period played no decisive role in the establishment of the -ingly pattern. What I did was to scan a number of texts for present participle + -ly derivations.6 The texts used were the Middle English

The spread of-ly to present participles

123

and Early Modern English parts of the Helsinki Corpus of English Texts as well as a number of texts downloaded from the Internet (cf. the list at the end of the paper). The idea behind this was that the frequencies of -ingly adverbs in the different periods would be indicative of the age of the pattern. I assumed that if suffixed present participles turned out to be rare in Middle English texts, this could be seen as evidence that the class was just beginning to emerge at this stage. A high frequency in this period, by contrast, would seem to indicate that the adverbs in question must be older than I had assumed on the basis of the data from the Oxford English dictionary. Table 2 below gives the rates of suffixed participles in the Middle English and Early Modern English parts of the Helsinki Corpus. It appears that the Middle English part contains only 16 of the adverbs in question, while the Early Modern English sections contain 80.7 These figures seem to support the idea of the Middle English era as the period when the pattern became established, and the Early Modern era as the period when it began to be used.8 Table 2. Frequencies of suffixed participles in the Helsinki Corpus.9 ME1 (1150-1250) ME2 (1250-1350) ME 3 (1350-1420) ME4( 1420-1500)

6 0 6 4

Total

16

EModEl (1500-1570)

17

EModE2 (1570-1640)

32

EModE3 (1640-1710)

31

Total

80

As for the texts taken from the Internet, the 16 Middle English ones contain 15 suffixed participles. In the Early Modern English texts there are more tokens — 21 all told — even though this corpus comprises only nine texts at this stage of my work.10 This may be another indication that it is not until the Modern English period that the adverbs in question are really beginning to be used.

3. The syntax and semantics of the new adverbs It has been generally assumed that present participles can be either verbal or adjectival (cf., e.g., Poutsma 1923: §§ 7-27; Hedlund 1992: ch. 7). Examples of verbal participles are given in (2) and (3) below. In (2) the participle forms part of a regular progressive, i.e., it occurs in collocation with be. In (3) it occurs without be in an adverbial participial subclause. The presence of a manner adverbial in

124

Kristin Killie

these two clauses makes the verbal nature of the participles unquestionable. In (4) we have a participle in adjectival function. It is in attributive position to a noun and is premodified by an adverb. (2)

Jane was smiling broadly.

(3)

Smiling broadly, Jane came towards him.

(4)

Jane has a really smiling face.

These are the two functions which are most commonly focused on in accounts of the present participle.11 A less discussed function is that of smiling in (5) below. Like Willis (1972: 27, 47, 287), I consider this use of the participle to be adverbial. The one thing all the suffixed participles in my corpus have in common is that they occur in a similar function as the participle in (5),'~ i.e., they do not stand in direct relation to a noun phrase, as adjectives do,13 and they are not followed by complements or adverbials as is commonly the case with verbs. They thus appear to be neither prototypical verbs, nor prototypical adjectives, but are functionally more like adverbials. It thus appears to be no coincidence that these participles increasingly appear with the adverbial suffix. (5)

Come here and I'll help you, Jane said smiling.

Apart from this, the new recruits for the adverbial stock show a rather irregular behavior both semantically and syntactically. As regards semantic type, the adverbs in question are not recruited into one particular category, but are of different semantic types. A large number of them function as pure manner adverbs. Examples are given in (6) and (7) below, where bityngly and connandly, like typical manner adverbs, can be paraphrased 'in a biting/cunning manner'. Others are so-called "subject adjuncts"; i.e., their primary task is to impute some characteristic to the subject while s/he is involved in some situation (cf., e.g., Quirk et al. 1985: 574). Secondarily some of them modify the manner of performing the action, too. Louyngly in (8), e.g., describes the subject's mental state (he is feeling full of love), as well as his manner of advising his wife (which is, of course, a result of the love he feels). The adverbs in (9) and (10) are unequivocal subject adjuncts: they characterize the subject's mental mode or volitional inclinations and say nothing about the manner of doing something. Lastingly in (11) is a time adjunct. The function of some other adverbs is more difficult to pin down. Stelendelich in sentence (12) qualifies neither the subject nor the manner of doing something, but seems to say something about the circumstances around the activity; the men drink the water when circumstances are such that nobody notices it. Similarly, the scope of the adverb in (13) is vague; the meaning of the sentence seems to be that the maiden's act itself is a lie; i.e., she mourned outwardly but not

The spread of-\y to present participles

125

in her heart ('falsely' would probably be the best modem translation here). All the examples below are from the Oxford English dictionary. (6)

Pat oper man ans\vered[e] ajeinful bityngly (c!374 Chaucer Boeth. ii. vii. 59) 'That other man answered full [very] bitingly' (7) Hymefiil connandly scho gret (1375 Sc. Leg. Saints, Theodera 402) 'Him very cunningly [politely] she greeted' (8) Louyngly the man auyseth his wyfe yf she doo amys (1398 Trevisa Barth. De P.R. vi. xiii. (1495) 197) 'Lovingly the man advises his wife if she does something wrong' (9) He truly pat knawyngly & wilfully fallis in-to pe lest [sin]... (1435 Misyn Fire of Love 103) 'He truly that intentionally and readily yields to the least sin' (10) The gude king said... it \ves in his awnfoly, For he raid sa vnvittandly... (1375 Barbour Bruce xvi. 248) 'The good king said... it was his own folly, for he spoke so unwittingly' (11) / sail sehe pi face lastandly til my ded (a 1340 Hampole Psalter XXVi. 13) Ί shall seek your face continually till my death' (12) Many of his men... Agein kyng Alisaunder hestes, Stelendelich dronken of this lake (13.. K. Alis. 5080) 'Many of his men... against king Alexander's commandments, stealthily drank from this lake' (13) For that the je maden leiyyngli the herte of the iust man to mourne, whom Υ made not sorewful (1382 Ezek. xiii. 22) 'That the maid lyingly mourned the heart of the just man, whom I did not make sorrowful' On the basis of the data in (6)-{13), as well as the rest of my material, I conclude that the new adverbs represent a whole range of semantic types. It is further impossible to generalize about their verbal collocates and the types of verb from which they are derived: a great number of types are represented. As far as these new adverbs are concerned, they simply defy generalization.

4. Present participles in earlier English One can hardly discuss the history of suffixed present participles without mentioning the history of the participles from which they are derived. To my knowledge, there exists no diachronic study of English present participles in adjectival function, so there is not much for me to build on. What I have done is simply to check the first occurrences of the participles from which my adverbs are

126

Kristin Killie

derived. Again I had to rely on the Oxford English dictionary. The relevant data are given in Table 3 below. Table 3. First occurrences of (unsuffixed) present participles.l4 Beowulf 800s 900s 1000s 1100s 1200s

1 6

5 8 2 12

1300s 1400s 1500s 1600s 1700s 1800s

23 7

15 7 5 1

Table 3 shows that only 34 of my 84 verbs are attested in participial form prior to the 14th century. 23 new participles are derived in the 14th century alone, and 45 in the Late Middle English period (cl300-1500) as a whole.15 Thus, it appears that the use of present participles in adjectival function is modest until the 1300s, when it virtually explodes.16 Possibly the data in Table 3 can be seen in connection with those in Table 1. It may well be the case that it was the sudden profusion of participles which drew attention to them as candidates for adverbial derivation.

5. Suffixed present participles in an adverbialization framework Finally I would like to draw attention to the fact that the change I have been discussing has not taken place in a vacuum but can be placed in a context. Throughout the history of (written) English the adverbial -ly suffix appears to have steadily gained ground. From its adoption as an adverbial suffix in Old English, this suffix has spread to a whole range of new items, syntactic environments and functions, showing itself to be extremely versatile. One example is epistemicity, which has been increasingly expressed through -ly adverbs. Hanson (1987: 137-143) claims that there are no adverbs in this function until the Middle English period. According to Swan (1988: 90-91, 131), Old English does have a few -ly adverbs expressing high probability. However, she agrees with Hanson that it was not until the Middle English period that the class of epistemic or modal adverbs really began to expand, and it was also in this period that low-probability adverbs began to occur (1988: 295-299). Swan also shows that the whole category of sentence adverbs (or disjuncts) has undergone a similar development. Sentence adverbs are defined by Swan as adverbs which express the speaker's evaluation of the content of the clause. In addition to modal adverbs, Swan's sentence adverb class contains evaluative adverbs (fortunately, regrettably), speech act adverbs (frankly, briefly), and subject disjuncts (wisely, cleverly). All these classes, according to Swan, have expanded and diversified enormously through the times,

The spread of-\y to present participles

127

especially in the twentieth century (1988: ch. 5). Another function which has become increasingly dominated by -ly adverbs is subject modification. In sentences like Johnny merrily/determinedly walked up the stairs to Jenny 's apartment, Old English makes much more extensive use of adjectives to describe the subject's state of mind than does Present-day English (Swan 1984: 4.4). Killie (1993: Ch. 5) shows that there is a decrease of adjectives in this use in the Late Middle English and Early Modern English periods. Like subject modifiers, intensifiers have become increasingly realized as -ly adverbs. While in Shakespeare's time exceeding sorry and extreme ill were perfectly normal and acceptable phrases, suffixless intensification is today restricted to a few adverbs and used mainly in oral, informal discourse (e.g., real good, cf. Quirk et al. 1985: 406). With respect to the change in question, the Early Modern English period has been claimed to represent a watershed (Nevalainen 1994: 245). Manner adverbs are another case in point. According to Dormer (1991: 2), adding the -ly suffix to adjectives "evidently constituted common practice with the general run of adverbs from no later than the closing decades of the twelfth century on". Finally, Killie (forthc. a) shows that stative adjectives like red and damp increasingly appear with the -ly suffix attached. There seems to be substantial evidence, then, that the -ly suffix has, throughout the written history of English, spread to ever new items and taken over new functions; in Swan's terminology, English has been (and still is) going through an "adverbialization process" (Swan 1988). I suggest that the suffixation of present participles is yet another instantiation of this process.17

6. Summary and conclusion In this paper I have drawn attention to the fact that English has developed a tradition for deriving -ly adverbs from present participles in adverbial function. The language did have a few such formations in Anglo-Saxon times; however, the pattern did not become really productive until the 14th century, and it was not in common use until the Early Modern English period. The new adverbs do not behave in any way similarly, but have been proven to be extremely variable both syntactically and semantically. The habit of attaching the adverbial -ly suffix to present participles is probably special to English. It appears to be a part of a more comprehensive process whereby more and more functions are becoming increasingly filled by -ly adverbs. Finally, it should be noted that the conclusions drawn on the basis of my material must be made with a reservation. To be able to draw conclusions about the history of suffixed participles with absolute certainty, it would be necessary to include all Old English verbs, i.e., also those which have not survived. Nevertheless, I do believe my data to be representative of the adverb class as a whole. After all, even if the verbs included represent only a selection, they do represent a rather large selection, and, furthermore, there is no reason to suspect that the verbs not included

128

Kristin Killie

behave any differently from those investigated. I am thus reasonably confident that my study reveals the major trends in the adverbialization of present participles.

Notes 1. This is not to say that the adverb type in question is totally absent from Old English texts. The Oxford English dictionary lists unaberendlice, unasecgendlice, and willendlice (cf. under willingly and under un-, suffix). Swan (1984: 52) reports the use of welwyllendlice in Kline's Lives of Saints (a text from c. 1000), and Visser (1963-1973: 1077) mentions gewemmendlice, scynendlice, spowendlice, and twxmendlice. However, although there are a few such adverbs, they are apparently much more rare in Old English than they are today. 2. I do not intend to imply that all English participles in adverbial function have become suffixed; structures like X said smiling are quite common, so what we see is not a complete shift. 3. In Killie (forthc. b), however, I map the history of a substantial number of both older and newer -ingly adverbs and discuss their syntactic and semantic properties. 4. The following verbs do not have adverbial derivatives: be, bear, beat, bid, bide, blow, break, bring, burst, buy, come, dive, do, drink, eat, fall, fear, find, flee, fold, get, give, have, hear, hew, let, load, lock, may, own, ride, rise, say, sell (with the original meaning 'give'), ship, shoot, shove, sit, sow, span, swallow, tread, wade, walk, wring, write, and yell. 5. Because so many of these adverbs are not used today (most of them having been replaced by other morphological types), their meanings are not obvious. I have, therefore, provided a list of glosses below. bowingly 'in a curving manner or direction' drowingly 'drawingly' hyingly 'hastily', 'quickly' likingly 1: 'in a pleasing manner, pleasantly, daintily, attractively, to one's liking, with pleasure' 2: 'in a probable manner, probably' sleepingly 'sleepily' speakingly 'eloquently, strikingly' stealingly 'stealthily, furtively' strewingly 'dispersedly' The fact that the list of 20th century adverbs is so short should not lead anyone to believe that our century is unproductive with respect to adverb formation from present participles. Indeed, as shown in Killie (forthc. b) the derivation

The spread of-ly to present participles

129

and use of the adverb type in question has really exploded in the last two centuries. The reason why there are so few adverbs in the 20th century list is simply because it comprises only verbs attested in Old English, most of which had apparently produced adverbs before our century. As stated at the beginning of this section, most of the suffixed participles used in Present-day English are of Latinate origin. 6. I checked for the various realizations of the present participle which are known to have existed in Middle English and Early Modern English, i.e., -ende, -inde, -ande, and -ing (cf. Scheffer 1975: 232). 7. The figures include all suffixed present participles, not only those with a Germanic origin, as I wanted to check the use of suffixed participles in general. 8. These figures are "raw counts" and, thus, say nothing about the relative frequencies of the adverbs in question, i.e., as seen in relation to the number of words in the corpus. To estimate relative frequencies, we need statistical measures, like those employed by Baayen—Renouf (1996) and Baayen— Lieber (1991). However, raw counts can be used to reveal trends in the data. 9. The scanning of the Helsinki Corpus and the Internet texts largely confirmed the datings of the Oxford English dictionary. There are only two exceptions: playingly and cunningly. The former is dated by the OED to 1680, but actually occurs in The Cloud of Unknowing, which was, according to the Cambridge companion to English literature, written towards the end of the 14th century. Cunningly, which according to the OED first occurs in c!374, appears in The Lyfe oflpomydon, from the Middle English period 1, i.e., somewhere between 1150 and 1250. These adjustments are included in Table 1. 10. Shakespeare is kept out of this count as he is apparently spearheading the use of suffixed participles with his 40 tokens (cf. Killie forthc. b). 11. Another recognized function of "-ing forms" is the nominal function of the gerund (and similar nominal forms). However, the gerund is not normally counted as a participle as it has a different origin, being derived from nominal -ingl-ung forms, while participles originally ended in -endel-indel-ande (cf. e.g., Wik 1973: 106; Scheffer 1975: 242). 12. I.e., with the exception of those which function as regular intensifiers, as in // is bitingly cold, where the adverbial nature of the suffixed participle is obvious. 13. This is not to say that they occur in any particular position; on the contrary, they appear everywhere in the clause. The point here is the looseness of the relation between the participle and the noun phrase(s) of the clause. Cf. Killie (forthc. b) for a more thorough discussion of this relation. 14. Two of my adverbs, strewingly and heedingly, are unattested in unsuffixed form. 15. These data should be taken with some reserve. As shown by Scheffer (1975: Ch. 11), it is sometimes extremely difficult to distinguish between adjectival and verbal uses of the participle. Scheffer gives a number of examples of Old English participles which have been interpreted as and translated by an adjective by some and by a verb by others.

130

Kristin Killie

16. Interestingly, this use parallels the use of the progressive, which according to Scheffer (1975: 216) really gained momentum in the 14th century after having been out of use for a few centuries. It is not unlikely that there may be a connection between the increase in the verbal and adjectival uses of the participle. 17. It is important to recognize that we are talking of not one but (at least) two different processes. With respect to the disjuncts and subject-modifying adjuncts mentioned above, e.g., the addition of the -ly suffix turns an adjective into an adverb, and the function becomes unequivocally adverbial. Nevalainen (1994: 243-244) refers to this type of change as a "functional-semantic shift". The growing use of -ly adverbs as intensifiers and manner adverbs, by contrast, involves no such functional shift, but is the result of a morphological regularization process taking place within the adverb categories themselves (cf. Nevalainen 1994: 243-244). In this process formerly "flat adverbs" (i.e., the leveled remnants of the Old English -e adverb) take on the -ly suffix to increase homogeneity with the rest of the class; in other words, one (morphological) type of adverb is ousting another. It seems plausible that the two processes may have reinforced each other, thus giving even stronger momentum to the "-ly tide" which has been — and still is — sweeping the English language. References Primary sources Middle English corpus:

Anon. 1993 Anon. 1993 Anon. 1993 Anon. 1993

The alliterative morte Arthure. Everyman. The owl and the nightingale. Pearl.

Anon. 1996 (5)

Pierce the Ploughmans crede.

Anon. 1993

Sawles warde.

Anon. 1993

The siege of Jerusalem.

Anon. 1993

Sir Gawain and the green knight.

Chaucer, Geoffrey 1993

The Canterbury tales.

The spread of-ly to present participles

131

Gower, John 1993 Confess io amantis. Henryson, Robert 1993a Orpheus and Eurydice. 1993b The testament of Cresseid. Langland, William 1993 The vision of Piers Plowman. Layamon 1993 Layamon 's Brut. Paston family 1993 Paston letters and papers of the fifteenth century, part I. Rotelande, Hue De 1996 (3) The lyfe oflpomydon. All the Middle English texts (except those from the Helsinki Corpus) are taken from: The

Humanities Text Initiative. Middle English http://www.hti.umich.edu/english/mideng/bibl.html: University of Michigan.

Collection. The

Early Modern English corpus: Bacon, Francis 1994

Behn, Aphra 1996(5) Behn, Aphra 1996(5)

The new Atlantis. Http://www.hti.umich.edu/english/pdmodeng/bibLhtml: UM Humanities Text Initiative.

Oroonoko: or, The royal slave. http://lang.nagoyau.ac.jp/~matsuoka/UK-authors.html. [no indication of publisher]. The unfortunate happy lady: a true story. http://curly.cc.utexas.edu/~churchh/bhnufhpl.html [no indication of publisher].

Dunbar, William 1993 The tretis of the two mariit women and the wedo. http://www.hti.umich.edu/english/mideng/bibl. html: UM Humanities Text Initiative. Johnson, Ben 1993 (6) Volpone. http://www.umich.edu/english/pd-modeng/bibl.html: Jeffrey Triggs, Bellcore & the Oxford Text Archive. Johnson, Ben 1995 The alchemist. http://www.umich.edu/english/pdmodeng/bibl.html: the Oxford Text Archive.

132

Kristin Killie

Milton, John 1993 (2)

Paradise lost. 1993, Feb http://www.hti.umich.edu/english/pdmodeng/bibl.html: Jeffrey Triggs at the Oxford Text Archive. More, Sir Thomas 1996(3) Utopia. http://zippy/dct.ac.uk/www/books: Kirk Crady,

Shakespeare, William 1989 William Shakespeare the complete works. Macintosh version. Oxford: OUP. Wroth, Lady Mary 1992(12) Pamphilia to Amphilanthus: a sonnet sequence from the Countesse of Mountgomeries Urania. http://www.umich.edu/english/pd-modeng/bibl.html: Richard Bear, University of Oregon/Humanities Text Initiative

1996 (3) indicates that no publication date was given for the Internet version of the text, but that I used it in March 1996, so that it must have been created before this date.

Secondary sources Baayen, Harald—Rochelle Lieber 1991 "Productivity and English derivation: A corpus-based study", Linguistics 29: 801-843. Baayen, Harald—Antoinette Renouf 1996 "Chronicling the Times: Productive lexical innovations in an English newspaper", Language 72: 69-96. Breivik, Leiv-Egil—Toril Swan forthc. Subject orientation and adverbialization in Old Norse and Old English. Donner, Morton 1991 "Adverb form in Middle English", English Studies 72: 1-11. Franz, Wilhelm 1893 "Zur Syntax des älteren Neuenglischen", Englische Studien 18: 191-219. Hanson, Kristin 1987 "On subjectivity and the history of epistemic expressions in English", in: Barbara Need—Eric Schiller—Anna Bosch (eds.), 133-147. Hedlund, Cecilia 1992 On participles. Stockholm: Department of Linguistics, Stockholm University.

The spread of-ly to present participles

133

Kastovsky, Dieter (ed.) 1994 Studies in Early Modern English. Berlin: Mouton de Gruyter. Killie, Kristin 1993 Early Modern English subject modifiers. Oslo: Novus. forthc. a Stativity and adverbial derivation: A synchronic and diachronic study. forthc. b. The history of suffixed present participles in English. Kjellmer, Goran 1984 "Why great: greatly but not big: "bigly?", Studia Linguistica 38: 1-19. Kytö, Merja (comp.) 1996 Manual to the diachronic part of the Helsinki Corpus of English Texts: Coding conventions and lists of source texts. (3rd edition.) Helsinki: Department of English, University of Helsinki. Mitchell, Bruce—Fred C. Robinson 1986 A guide to Old English. (4th edition.) Oxford-New York: Blackwell. OED = Murray, James A. H.—Henry Bradley—William A. Craigie—Charles T. Onions 1989 The Oxford English dictionary. (2nd edition.) Oxford: Clarendon Press. Mustanoja, Tauno F. 1960 A Middle English syntax. Part I. Helsinki: Societe Neophilologique. Need, Barbara—Eric Schiller—Anna Bosch (eds.) 1987 Papers from the 23d Annual Regional Meeting of the Chicago Linguistic Society, Part one: The general session. (CLS 23.) Chicago: Chicago Linguistic Society. Nevalainen, Terttu 1994 "Aspects of adverbial change in Early Modern English", in: Dieter Kastovsky (ed.), 243-259. Ousby, Ian (ed.) 1988 The Cambridge guide to literature in English. Cambridge: CUP. Peters, Hans 1993 Die englischen Gradadverbien der Kategorie booster. Tübingen: Narr. Poutsma, Hendrik 1923 The infinitive, the gerund and the participles of the English verb. Groningen: Noordhoff. Quirk, Randolph—Sidney Greenbaum—Geoffrey Leech—Jan Svartvik 1985 A grammar of contemporary English. London: Longman. Scheffer, Johannes 1975 The progressive in English. (North-Holland Linguistic Series 15.) Amsterdam: North-Holland Publishing Company.

134

Kristin Killie

Swan, Toril 1984

"Adverbial usage in Äilfric's Lives of Saints", Nordlyd: Troms0 University Working Papers on Language and Linguistics 9: 7— 70. 1988 Sentence adverbials in English. (Tromsa-studier in spräkvitenskap X.) Oslo: Novus. 1990 "Subject-oriented adverbs in 20th century English", Nordlyd: Tromse University Working Papers on Language and Linguistics 16: 14-58. Visser, Frederikus T. 1963-1973 An historical syntax of the English language 1-3. Leiden: Brill. Wik, Berit 1973 English nominalizations in -ing. Synchronic and diachronic aspects. (Studia Anglistica Upsaliensia 12.) Uppsala: Almqvist & Wiksell. Willis, Hulon 1972 Modern descriptive English grammar. San Francisco: Chandler.

Inversion after single and multiple topics in Old English Willem F. Koopman

1. Introduction Main clauses in Old English show some important differences from subordinate clauses. They do not have a subordinating conjunction, verb-final syntax is rare, but topicalization is frequent. Topicalization (or "fronting") occurs in subordinate clauses only in such limited contexts (cf. van Kemenade forthc.) that it is in fact almost exclusively a main-clause phenomenon, and as such can be seen as an important signal for main clause syntax. I use the term "topicalization" loosely to describe a sentence element in first position in the clause. Topicalization is often accompanied in some modern Germanic languages, such as Dutch and German, by VS word order (traditionally called "inversion"), thus giving the characteristic Verb Second effect. I shall use the term "inversion" in this paper as a convenient descriptive term for VS word order without making claims about elements actually switching places. While inversion is absolutely regular in Dutch and German when main clauses begin with a non-subject, this is not the case in Old English. Personal pronoun subjects (henceforth called "pronominal subjects") always invert in questions and in clauses beginning with the negative ne 'not', and practically always when there is an initial adverb such as pa 'then', but with a different first element they do not invert, as in (1) where there is an object in first sentence position: (1)

t>xt an lifhe xteowde mid his deade. "That one life he manifested with his death' (JECHom i.224.18)

This behavior has been accounted for by van Kemenade (1987) and Pintzuk (1991) by claiming that personal-pronoun subjects are syntactic clitics. As a clitic does not count for the syntax, (1) can then still be analyzed as a regular Verb Second sentence. If pronominal subjects are indeed clitics, we can expect a fairly straightforward distribution of pronominal subjects and other subjects in main clauses with topicalization: (2)

pronominal subjects do not invert, nominal subjects do1

However, it is not as simple as all that. Contrary to what we might expect, pronominal subjects do sometimes invert even after adverbs such as pa, while nominal subjects, on the other hand, occasionally fail to show inversion, as in (3) after an initial object:

136

(3)

Wittern F. Koopman

pa medonnesse dxre strengio se salmscop ongeat 'the excellence ofthat strength the psalmist acknowledged' (CP 85.22)

Mitchell (1985: § 3922) notices that the presence of another topic seems to lessen the force of pa, so that inversion is less regular. In this paper I will describe first in section 2 the effect of different first elements (objects, prepositional phrases, and some selected adverbs) on the position of the subject and the verb in a range of Old English prose texts. This will show to what extent the prediction formulated in (2) holds. In section 3 double topicalization involving pa and ponne 'then' will be investigated, to see whether inversion is significantly less regular than in main clauses with just pa or ponne as the first element. Theoretical implications are briefly discussed in section 4, followed by the conclusion in section 5.

2. Inversion after initial objects, prepositional phrases, and (certain) adverbs In this section I will test (2) on a corpus of some of the major Old English prose texts. This will provide the necessary statistical data with which the effect of double topicalization can be compared. I will deal first of all with clauses where there is a fronted object, then look at fronted prepositional phrases, then contrast the effect on inversion of a few selected fronted adverbs to the effect adverbs such as pa and ponne have.

2.1.

Inversion after initial objects

Main clauses with initial objects in my corpus show the distribution given in (4):

(4)

Inversion after initial objects2

CP Orosius ByrM jECHomi [ WHom

pronominal subjects +inv -inv 48 l 18 30 87 l 100 31 10

nominal subjects +inv 19(51%) 15(58%) 5(50%) 49(91%) 52(90%) 28(93%) 12(57%)

-inv 18(49%) 11(42%) 5(50%) 5(9%) 6(10%) 2(7%) 9(45%)

Inversion after single and multiple topics in Old English

137

We can see from (4) that pronominal subjects behave as expected from (2). Only two exceptions were found with inversion of the pronoun, given in (5) and (6): (5)

Daet wille ic gecypan, past... that want I reveal that... Ί want to reveal this, namely that...' (Or 37.2)

(6)

Micel mssg heo set hire Bearne abiddan... Much may she from her Child obtain... 'Much may she obtain from her child...' (/ECHom ii.22.27)

Both are to some extent unusual. In (5) we have a provisional object and in (6) a quantifier. Only two exceptions in so many (long) texts is very little indeed. They could be mistakes, but it seems to me more likely that we have non-clitic subject pronouns here. Just as some object pronouns at least cannot be clitics in the way defined by van Kemenade (1987), as I have argued in Koopman (1992), it seems reasonable to assume that some subject pronouns too are non-clitics. The behavior of nominal subjects is not at all what we would expect from (2). All the seven texts investigated show variable inversion of nominal subjects after initial objects, though cases of non-inversion, as in (3), do not have equal frequency in the texts investigated. Inversion is the norm in ^Ifric (a Late Old English prose writer), who has a low percentage of non-inversion, whereas CP and Orosius (both Early Old English prose texts) have a much higher percentage. This might point to some diachronic development, but the (admittedly low) figures from Wulfstan (also a Late Old English writer) seem to contradict this, and individual variation is more likely. 2.2.

Inversion after initial prepositional phrases

The figures for fronted prepositional phrases are given in (7): (7)

Inversion after fronted prepositional phrases

CP Or ByrM jECHomi ^CHomii .ELSvoll WHom

pronominal subjects + inv -inv 47 50 1 16 4 72 2 68 28 1 1

nominal subjects +inv 75(69%) 154(52%) 41(85%) 186(93%) 193(93%) 64(94%) 9 (45%)

-inv 34(31%) 143(48%) 6(15%) 13(7%) 14(7%) 4(6%) 11 (55%)

138

Willem F. Koopman

Inversion of pronominal subjects is again rare (as with fronted objects). The norm is non-inversion (to be expected if the subject pronoun is a clitic). (7) shows seven exceptions which are given in (8)-(14): (8) (9)

(10)

(11)

(12)

(13)

(14)

buton tweon ne mihte he beon &lmihtig God without doubt not could he be Almighty God 'without doubt he could not be Almighty God' (yECHom i.276.19) On eallum dinum \veorcum beo du gemyndig bines endenextan dasges In all your works be you mindful of-your last day 'Be mindful of your last day hi all your works' (^ECHom i.408.32) On eallum öingum beo du gemyndig bines endenextan dasges In all things be you\mindful of-your last day 'Be mindful of your last day in all things' GECHom i.482.6) On Cristes naman neforhtige icfor dinum tintregum In Christ's name not-fear I for your torments. 'In the name of Christ I fear not for your torments' (JECHom i.428.26) jEfter disonfeotton hi eft ealle xtforan bam casere After this fell they again all before the emperor 'After this they all again fell before the emperor' (^CHom ii.426.20) For dyssere twynunge nolde we hreppan his orowunge For this doubt not-wanted we touch his passion 'For this doubt we would not touch his passion' (JECHom ii.520.16) Be bsere sunnan cwede we bus About the sun speak we thus 'Let us speak about the sun' (ByrM 122.8)

Three have negated verbs ((8), (11), and (13)), and it may be the negative which is responsible for the inversion rather than the initial prepositional phrase. In (9), (10), and (14) there are subjunctives, and inversion is normal there. This leaves only (12) as a genuine exception in this corpus. Inversion of nominal subjects after initial PPs varies considerably from text to text. CP, Orosius, and WHom have more non-inverted cases than JElfric and initial prepositional phrases are quite numerous in Orosius. Inversion of nominal subjects in the texts investigated varies after initial objects as well as initial prepositional phrases, and is not equally frequent in each text. However, the type of initial element makes no significant difference, as each text has practically the same rate of inversion after initial objects as after initial prepositional phrases.

2.3.

Inversion after some initial adverbs

Initial adverbs occur often in Old English prose texts, and there is a wide range of them. As not many are frequent enough to make separate figures for each text

Inversion after single and multiple topics in Old English

139

meaningful, I present information for five adverbs in (15) for the whole corpus, which in this instance consists of 11 texts.3 The data were complied with the help of the microfiche concordances (Healey—Venezky 1980; Venezky—Butler 1985). (15) Inversion/non-inversion after some initial adverbs (figures taken from Koopman 1997)

witodlice 'truly' xfre 'ever' oft Often' eft 'again' nu 'now'

pronominal subjects + inv -inv 19 6 9 38 105 (56%) 81 (44%)

nominal subjects +inv 1 6 6(46%) 31(62%) 132 (99%)

-inv 78 7(54%) 19(38%) 2 (1%)

Witodlice 'truly' triggers no inversion to speak of with either type of subject. Only one case of inversion came to my attention (JECHom ii.58.14), and one of the other manuscripts does not have inversion here at all. When we look at the meaning of witodlice, the absence of inversion is not really surprising. More often than not it functions as some sort of attention getter, almost an interjection, probably best interpreted as being outside the syntax of the clause. This is supported by the fact that there are indeed very few instances where witodlice is not in clause-initial position. After xfre 'ever' we find the expected distribution with nominal subjects showing inversion and pronominal subjects not. With oft Often' and eft 'again', nominal subjects have variable inversion. With nu 'now' there is inversion as well as non-inversion with nominal subjects. This can be understood if we take into account that nu need not necessarily act as a temporal adverb. A comparison with Modern Dutch may be instructive. There nu 'now' sometimes does not cause inversion, which otherwise in Dutch is obligatory after topicalized elements. This suggests that in these cases nu cannot be within the clausal syntax. Compare: (16)

(17)

Nu wil Jan dot niet meer doen Now wants John that not more do 'John does not want to do it any longer' Nu, Jan wil dot niet meer doen Now, John wants that not more do 'Well now, John does not want to do it any longer'

Notice that there is an intonation break in (17) and that the adverb does not mean 'at this point in time', but something like 'the next thing I want to discuss is...'. I interpret this as an element like Old English witodlice, some sort of connecting discourse element which is outside the syntax of the clause.

140

Willem F. Koopman

A similar interpretation seems to me to be possible for some of the Old English nu cases as well. This can be illustrated in (18) for one of the two clauses with a nominal non-inverted subject. Here nu does not have a temporal reading either. (18) Nu se rica mann ne mxg her habban be ma be ure asnig pa orsorgan and pa unateorigendlican blysse Now the rich man not can here have the more than of-us any the sorrowless and untiring bliss 'Now the rich man cannot have here, any more than any of us, sorrowless and untiring bliss' GELS 1 268.110) If this is a reasonable approach to take, we would have an explanation for the variable inversion with nominal subjects after nu. Inversion occurs when nu is a temporal adverb, non-inversion when it is a discourse element. However, it is not always easy to distinguish the purely temporal meaning from the "discourse" meaning in Old English, as the two shade into each other. Whatever the reading of «M, the pronominal subject should not invert. After "discourse" nu, which is not part of the clausal syntax, the pronominal subject has clause initial position and naturally precedes the verb. When nu is a topic, the pronominal subject does not count as it can be interpreted as a clitic. I do not think that we can say that non-inversion of nominal subjects necessarily shows that we have "discourse" nu. As the figures presented in this paper show, inversion is not as regular as it is in Dutch and German. It would, therefore, be circular to use inversion/non-inversion as a diagnostic test in this case. It is possible that eft can also have such a "discourse" function, and perhaps oft too. If so, that could help explain variable inversion with nominal subjects. Some support can be found for this, I think, by the combination eft pa which functions often as an interjection. (19)

Eft Oase casere sende to bam cwellerum... Afterwards then the emperor sent to the executioners... 'After that the emperor sent word to the executioners...' (^LS 2.162.74)

It remains to be seen whether all non-inversion of nominal subjects after initial adverbs can be attributed to the "discourse" meanings of those adverbs.

2.4.

Inversion after initial pa/ponne

We have seen in section 2.3 that inversion of pronominal subjects after initial nu is quite frequent. This is very different after fronted objects, prepositional phrases, and the other adverbs of section 2.3, where there is no inversion of pronominal subjects to speak of. Some other adverbs have similar effects as nu. The most

Inversion after single and multiple topics in Old English

141

prominent are pa 'then' and ponne 'then', though xr 'before' and sippan 'afterwards' should also be included here (cf. Mitchell 1985: § 2536). I will concentrate on pa and ponne. Both pa and ponne are a great deal more frequent in Old English texts than the other adverbs investigated in section 2.3, and I can give figures in (20) only for some texts (TECHom i, Bo (pa and ponne), Orosius (pa), and CP (ponne)). Nevertheless, they should give us a good idea of the inversion effect: (20)

Inversion after pa/ponne:4

pa

+ inv

-inv

Bo Orosius

431(98%) 431(99.5%) 276 (99%)

8(2%) 2(0.5%) 2 (1%)

ponne

+inv

-inv

jECHomi Bo CP

124(98%) 112(100%) 210(96%)

2™

In (20) inversion is very regular after the adverbs pa and ponne, whereas (15) shows that it varies after nu. This could perhaps be related to their overall frequency and in particular to how often they are used in correlative constructions, where pa is quite common and nu rare.5 It has often been remarked that inversion after initial adverbs such as pa is extremely frequent, and that all are also conjunctions. Mitchell (1985: §§ 25362560) devotes considerable space to a discussion of the "ambiguous adverb/conjunction". It is sometimes difficult, he says, to determine whether pa and ponne are adverbs or conjunctions when they introduce a clause, and, therefore, whether the clause is main or subordinate. This is because the word order is not a completely reliable guide, though the "rule" is that main clause word order is VS after these initial adverbs, whereas subordinate clause order commonly is S(...)V. Mitchell remarks that the prose is a great deal more regular than the poetry (which is not my concern in this paper) in showing different word orders for main and subordinate clauses in these cases, but even so exceptions to the "rule" can be found. We need to rely on other clues sometimes, such as whether pa is doubled, the use of a subjunctive verb, the presence of S...V word order, and of course the context, to recognize the subordinate clause. In (14) S.. .V order and the context are decisive:

142

(21)

Wittern F. Koopman

Donne hi leorniaö midfulre estfulnesse da sodan god to secanne, donne hi mid Julie gesceade ongietad öaet... Then they learn with full affection the true good to seek when they with full understanding perceive that... 'Then they learn to seek the true good with full affection, when they perceive with full understanding that...' (CP 441.17)

Pragmatically, there is a good reason for inversion after certain adverbs. Topicalization is really a main clause phenomenon and can be said to signal that a main clause follows. This would be sufficient in the case of "normal" topics as there is no genuine room for a different interpretation. The function of word order is, therefore, less important. The normal syntactic rules can then apply: non-inversion with pronominal subjects (because of their clitic status) and variable inversion (up to a point) with nominal subjects. It would be different with adverbs that can also function as subordinating conjunctions. It would not be so easy for speakers (or readers) to identify the adverb and, therefore, the main clause in the case of a pronominal subject, unless the subject inverts. The other clues available, such as the use of the subjunctive and S...V order are less reliable, we could say, than the order of subject and verb. The subjunctive is not always clearly marked as such, and S.. .V word order also sometimes occurs in main clauses. Looked at from this perspective, inversion with pronominal subjects is functional, as there is a greater need to signal the presence of the main clause by the order of subject and verb.

3. Multiple topicalization involving jto andponne Mitchell (1985: § 3922) notices that the presence of another element after the initial adverb may be responsible for some of the non-inverted main clauses he discusses. In other words, we could say that the presence of another topic lessens the impact of ba or bonne. The other topic sometimes seems to have the effect that the initial adverbs is as it were forgotten. It is as if a clause with, e.g., initial ba followed by a prepositional phrase gets treated as if there was only the prepositional phrase as topic. The potential ambiguity of pa or bonne would be considerably less when there is another topic to indicate the presence of the main clause. On the other hand, when ba or bonne is not the first topic, there is no possibility for ambiguity at all, as it cannot be taken for a conjunction. If I am right, we expect to see in combinations of ba or bonne and another topic more of the distribution we found after initial objects, prepositional phrases, and other adverbs (non-inversion for pronominal subjects, variable inversion for nominal subjects). I have investigated multiple topics involving ba and bonne in some texts. The results are given in (22) (pa/bonne as first element) and (23) (pa/bonne following another topic):

Inversion after single and multiple topics in Old English

143

(22) pa/ponne followed by (an)other topic(s) pronominal subjects + inv -inv

nominal subjects +inv -inv

pa

jECHom ii Orosius WHom ponne Orosius CP WHom ByrM (23)

2 2 5 _

23 14 14 7_

2 2 3 _1

-

1 1

7 1

2 1 -

Multiple topics with pa/ponne not as first element

Orosius ,ECHomii CP ByrM

6 4 4 1 1

pronominal subjects + inv -inv 1

nominal subjects +inv 1 7 -

-inv 11 3

1

There were 11 non-inverted pronominal subjects, given in (24)-(34): (24)

(25) (26) (27)

t>a purh deofles swicdom and Adames gylt weforluren pa gesaelox ure sowie Then through devil's treachery and Adam's guilt we lost the happiness of-our soul 'Then through the treachery of the devil and the guilt of Adam we lost the happiness of our souls' (AiCHom i.20.2) Da set nextan he dihte pisne pistol to pxre halgan wydewan Paulam 'Then at last he composed this letter to the holy widow Paula' (^ECHom i.436.18) Pa under ptem gewinne hie genamonfriö wid da wxpnedmen... Then under the war they took peace with the men... 'Then during the war they made peace with the men...' (Or 29.28) Pa xt nihstan he wearp selfbesiered? ofslaegen 'Then at last he was himself deceived and killed' (Or 89.9)

144

(28) (29)

(30)

(31)

(32)

(33) (34)

Willem F. Koopman

Pa sefter pasm wordum hie budon him past he... 'Then after those words they offered him that he...' (Or 95.31) pa set paem priddan cirre hie sendon χ hiera ieldstena wietena Then at the third occasion they sent 10 their of-oldest counselors 'Then at the third occasion they sent ten of their oldest counselors' (Or 97.17) t>a on mergen hie waeronpset ilce donde Then in morning they were that same doing 'Then in the morning they were doing that same thing' (Or 121.25) Pa sippan on xfentiman hi setton hine on cweartern Then afterward in evening they put him in prison 'Then afterwards in the evening they put him in prison' (^LS 1.324.58) pa xfter sumumfyrste he wearo on swefrie gemynegod Then after some time he was in dream admonished 'Then after some time he was admonished in his dream' (JELS 2. 228.145) pasnne binnan nigontyne wintrum hyt getimad past... 'Then within nineteen years it comes to pass that...' (ByrM 76.24) Donne eac xfter dysum he bio wel husles wyroe Then also after this he is well housel worthy 'Then also after this he will be well worthy of the housel' (WHom 180.96)6

The figures presented in (23) (low though they are) support the idea ihatpa/ponne after another topic is not decisive for inversion, but (22) shows that the influence of the initial adverb remains quite strong, and that in most texts there are still relatively few non-inverted cases. We must, however, compare inversion/non-inversion after multiple topics, with inversion/non-inversion after single initial pa/ponne, to see whether the difference is significant. (35) compares single and multiple topics in ^CHom i. and Orosius (for pa): (35)

Inversion after pa and after pa as first topic inverted subjects

i) pa (single) pa (first topic) (Orosius) pa (single) pa (first topic)

non-inverted subjects

431 29

(98%) (88%)

8 4

(2%) (12%)

276 8

(99%) (57%)

2 6

(1%) (43%)

The χ2 test shows that the difference is statistically significant for both texts.7

Inversion after single and multiple topics in Old English

145

4. Theoretical discussion In this section I want to discuss briefly some of the theoretical implications of the data presented in this paper. In the theoretical frameworks developed by van Kemenade (1987) and Pintzuk (1991) pronominal subjects are clitics and as such are "invisible" for the syntax. A Verb Second analysis can, therefore, still be maintained for what appear to be "Verb Third" clauses. As the data in this paper show, pronominal subjects only follow the verb after adverbs such as pa, while after other adverbs, objects, and prepositional phrases they regularly precede the verb. The few counterexamples are perhaps best explained as non-clitic pronouns, the result of an innovation. In a formal tree structure the pronominal subject would occupy a position higher than the finite verb in topicalized sentences (giving SV order), but the verb would be higher in questions, in clauses beginning with ne 'not', and after adverbs such asjja (giving VS order). Why such adverbs have this effect is not easy to explain. When nominal subjects occupy a lower position than the verb we get inversion, when they are higher there is non-inversion. With initial prepositional phrases there are what appear to be minimal pairs (one with and without inversion), as in (36) and (37): (36)

Be dam Paulus cuxd 'about this Paul said' (CP 97.11) (37) Be daem cuaeo Paulus about this said Paul 'about this Paul said' (CP 93.25) The existence of such pairs makes it hard to see what "triggers" the different positions the nominal subject (or the verb) occupies. I will discuss two possible (non-technical) explanations for the variable behavior of nominal subjects: (1) Verb Second could have failed, and (2) the initial element could be outside the clausal syntax. Both can be invoked, but it is difficult to prove in how far they apply (if at all), and they cannot take care of all the non-inverted clauses. The problem of the theoretical motivation for the variable inversion of nominal subjects remains at least for some cases. The data from multiple topicalization provide problems for the theory as well.

4.1.

Failure of Verb Second

As we do not expect the nominal subject to appear left of the finite verb in main clauses when Verb Second has applied, it is possible that non-inversion is simply due to the failure of Verb Second. In other words we could be dealing with main clauses that are verb-final. Such clauses occur in all the major prose texts, with a low percentage (cf. Koopman 1995). Topicalization does not necessarily trigger

146

Will em F. Koopman

Verb Second, as can be seen from (38), where there is a prepositional phrase in initial position but still a nominal object before the verb: (38)

On oaere ilcan tide Hannibal his agnum willum hine selfne mid atre acwealde On the same time Hannibal his own will himself self with poison killed 'In the same period Hannibal killed himself with poison' (Or 110.1)

However, it is hard to prove conclusively that failure of Verb Second is involved, and sometimes this explanation will not work, as for (39), where the finite verb is no longer in clause-final position, and must have been moved: (39)

4.2.

Mid bam pater nostre man sceal to Gode gebiddan With the pater noster one must to God pray One must pray to God with the pater noster' (WHom 176.17) Initial element outside the clausal syntax

The initial object could be left-dislocated. If that is indeed the case, there is no topicalization at all and the subject would naturally precede the verb. We would expect a resumptive pronoun with left-dislocation (obligatory in Modern English). Perhaps Old English did not require a resumptive pronoun in such cases, as objects did not have to be present (cf. Ohlander 1943-1944) when they were recoverable from the immediate context, as we can see from (40), where the direct object his lac is not repeated in the second coordinate clause. (40)

and bxr him sylfhis lac and lede uppon bxt weofod and carried him self his offering and laid upon the altar 'and carried the sacrifice himself and laid it upon the altar' (/ELS 1.162.236)

However, the presence of cases with a resumptive pronoun, such as (41), makes it very unlikely that the non-inversion of nominal subjects after an initial object can be attributed to left-dislocation:8 (41) and pa spell be ic secge ic hi sceal gescyrtan and the stories which I say I them shall curtail 'and I shall shorten the stories that I shall tell' (Or 27.23) Initial prepositional phrases could likewise be outside the clausal syntax. Such an interpretation might work for (36) by reading a comma after Be dam, but it is not an option for (39), (42), or (43) where the initial PP is an integral part of the clause.

Inversion after single and multiple topics in Old English (42)

Durh da earan oa word biod anfangen through the ears the words are received 'the words are received through the ears' (CP 97.2)

(43)

On Greciscre spraece steorra ys aster genemned In Greek language star is aster called Ά star is called aster in the Greek language' (ByrM 184.23)

4.3.

147

Multiple topicalization

The data presented in this paper show that it is not necessarily the first initial element that determines what is going to happen in terms of inversion. We have seen that the (almost) universal inversion after initial adverbs such as pa is explained in the theory as the result of the different position the verb occupies (it will be higher than whatever type of subject). With multiple topics with initial pa or ponne there is another topicalized element (often a prepositional phrase). We could argue that non-inversion is the result of a "fresh" start after pa and that the following topic is decisive. On the other hand, when there is inversion of pronominal subjects after multiple topics as in (24)-{34), the verb must be higher than the subject, while a prepositional phrase as second topic is even higher than the verb. It is difficult to see why a prepositional phrase can have such a high position only as second topic and not when it is the sole initial element.

5. Conclusion In this paper I have investigated the position of the subject in relation to the verb when there is a fronted first element and looked at what happens with multiple topicalization. Pronominal subjects generally do not invert after initial objects, prepositional phrases, and adverbs (only a few exceptions were found in the texts investigated), whereas nominal subjects on the whole invert, but certainly not always. Both types of subjects invert after initial adverbs such as pa and ponne. I have also shown that few exceptions occur, but that when such an adverb is the first of two (or more) topics, non-inversion was significantly more frequent. It is difficult to find a satisfactory theoretical explanation for the data presented here. The two possible reasons for non-inverted nominal subjects (failure of Verb Second, and initial elements outside the clausal syntax) cannot account for all occurrences of non-inverted nominal subjects.

148

Wittern F. Koopman

Notes 1. All subjects, except personal pronoun subjects are called "nominal" subjects in this paper. 2. The abbreviations used are the ones proposed by Mitchell—Ball—Cameron (1975): ^CHom vol i and ii GElfric's Catholic Homilies: Thorpe 1844-1846), ,ELS (Alfric's Lives of the Saints: Skeat 1881-1890); Bede (The Old English version of Bede's Ecclesiastical History: Miller 1890-1898; CP (Pastoral Care: Sweet 1871); Or (Orosius: Bately 1980), Bo (Boethius: Sedgefield 1899), WHom (Wulfstan's Homilies: Bethurum 1957), ByrM (Byrhtferth's Manual: Crawford 1929), ChronA and ChronE (Parker Chronicle and Peterborough Chronicle: Plummer 1892-1899). 3. The texts investigated were: ^ECHom i and ii, ^LS, WHom, CP, Bo, Or, Bede, ChronA, ChronE, and ByrM. 4. Non-inversion after fro/bonne involved nominal as well as pronominal subjects. 5. 76 (18%) out of the 431 main clauses with initial ba in ^CHom i are in correlative constructions. The figures for Bo are: 57 (13%) out of 431. Only one of 61 main clauses with initial nu in jECHom ii is part of a correlative construction. 6. Two other manuscripts have inversion here. 7. jECHom i: χ2=13.139, ΡΟ.001. Orosius: χ2=88.816, ΡΟ.001. 8. I am grateful to Susan Pintzuk for pointing this out to me.

References Bately, Janet (ed.) 1980 The Old English Orosius. (The Early English Text Society, Supplementary Series 6.) London: OUP. Bethurum, Dorothy (ed.) 1957 The homilies of Wulfstan. Oxford: OUP. Colman, Fran (ed.) 1992 Edinburgh studies in the English language 2. Edinburgh: Donald. Crawford, Samuel J. 1929 Byrhtferth's Manual. (The Early English Text Society 177.) London: OUP. Healey, Antonette DiPaolo—Richard L. Venezky 1980 A microfiche concordance to Old English. (Publications of the Dictionary of Old English 1.) Toronto: Pontifical Institute of Medieval Studies. Hickey, Raymond—Stanislaw Puppel (eds.) 1997 Language history and linguistic modelling: A festschrift for Jacek Fisiak on his 60th birthday. (Trends in Linguistics: Studies and Monographs 101.) Berlin: Mouton de Gruyter.

Inversion after single and multiple topics in Old English

149

van Kemenade, Ans 1987 Syntactic case and morphological case in the history of English. Dordrecht: Foris. forthc. "Verb second and embedded topicalization in Old and Middle English", in: Ans van Kemenade—Nigel Vincent (eds.). van Kemenade, Ans—Nigel Vincent (eds.) forthc. Parameters of morphosyntactic change. Cambridge: CUP. Koopman, Willem F. 1992 "Old English clitic pronouns: Some remarks", in: Fran Colman (ed.): 44-87. 1995 "Verb-final main clauses in Old English prose", Studio Neophilologicaol: 129-144. 1997 "Topicalization in Old English and its effects: Some remarks", in: Raymond Mickey—Stanislaw Puppel (eds.), 307-321. Miller, Thomas (ed.) 1890-1898 The Old English version of Bede's Ecclesiastical history of the English people. (Early English Text Society 95, 96, 110, 111.) London: Trübner. Mitchell, Bruce 1985 Old English syntax 1-2. Oxford: Clarendon Press. Mitchell, Bruce—Christopher Ball—Angus Cameron 1975 "Short titles of Old English texts", Anglo-Saxon England 4: 207221. Ohlander, Urban 1943-1944 "Omission of the object in English", Studio. Neophilologica 16: 105-127. Pintzuk, Susan 1991 Phrase structures in competition: Variation and change in Old English word order. [Unpublished Ph.D. dissertation, University of Pennsylvania.] Plummer, Charles (ed.) 1892-1899 Two of the Saxon Chronicles parallel 1-2. Oxford: OUP. Sedgefield, Walter J. (ed.) 1899 King Alfred's Old English version of Boethius. Oxford: The Clarendon Press. Skeat, Walter W. 1881-1890 Alfric's Lives of the Saints. (The Early English Text Society 76, 82, 94, 114.) Oxford: OUP. Sweet, Henry (ed.) 1871 King Alfred's West-Saxon version of Gregory's Pastoral Care. (Early English Text Society 45, 50.) London: OUP. Thorpe, Benjamin (ed.) 1844-1846 The sermones catholici or homilies ofALlfric 1-2. London: /Elfric Society.

150

Willem F. Koopman

Venezky, Robert L.—Sharon Butler 1985 A microfiche concordance to Old English: The high-frequency words. (Publications of the Dictionary of Old English 2.) Toronto: Pontifical Institute of Medieval Studies.

Epenthesis and Mouillierung in the explanation of /-umlaut: The rise and fall of a theory Martin Krygier

1. Introduction In discussing English historical phonology one tends to adopt a very broad perspective. Changes such as the Great Vowel Shift, Homorganic Cluster Lengthening, or Open Syllable Lengthening captivate the imagination of a researcher, discouraging him from tackling less popular, seemingly well-researched subjects. Changes operating earlier, in the less dramatic period of the history of the language, are sometimes taken for granted, as something explained and dissected a long time ago. However, this need not always be the case. One such problem is that of the nature of /-umlaut. Major textbooks seem to agree that all the basic facts about this process have already been established, and a neat formula has been formulated to capture all the peculiarities of the change. Given this peaceful picture presented by all the most recent textbooks, the amount of disagreement concerning all aspects of /-umlaut may come as a surprise to an unprepared reader. This paper is devoted to the various attempts at explaining the very nature of /-umlaut. More precisely, it will trace the development of two very closely related theories concerning the process: the epenthesis and the Mouillierung theory. One may ask what reason if not purely historiographic considerations one may have to study this particular subject, as both these theories have disappeared from standard textbooks, and Hogg's A grammar of Old English considers them to be "discredited" (1992: 123). For one thing, they have occupied minds of Germanic scholars for over a century, giving rise to one of the most heated debates concerning /-umlaut. Even today this discussion is far from being over, and though structuralists claimed to have successfully defeated the old school once and for all (cf., e.g., Penzl 1949), their claims seem to have been somewhat premature. Moreover, the last decade has witnessed a revival of these theories in a number of papers questioning the generally accepted view of /-umlaut as a case of distant assimilation. Alternative explanations of some well-known problems in Old English phonology, such as the relative chronology of /-umlaut and consonant palatalisation to name just one, can be offered once these theories are accepted for /-umlaut itself. Consequently, a closer look at the history and arguments used to substantiate them seems to be wholly justified.

152

Marcin Krygier

2. The epenthesis and Mouillierung theories The early research into the origin and nature of /-umlaut was to a large extent built on excessive reliance on written data. The realization that umlauted vowels, regardless of their spelling, were monophthongal in their nature, took over two centuries to sink in. The diphthongal interpretation was, however, not entirely forgotten. Based on digraphic spellings, which could be found in almost all early Germanic languages, the diphthong stage was removed in time and postulated to have constituted a necessary intermediate step in the development of umlauted vowels. In order to explain the emergence of these transitional umlaut diphthongs, earlier palatalisation of the intervening consonants as the vehicle for the transference of the umlaut quality was suggested. The term "Mouillierungstheorie" was first applied to the hypothesis formulated by Wilhelm Scherer in 1868. In his opinion /-umlaut started with the umlaut factor in the unstressed syllable palatalizing immediately preceding consonants. The next stage was the development of an epenthetic [i] glide, which then combined with the root vowel to form a new, fronted vowel. The diphthongal stage, a necessary prerequisite, was to be found, according to Scherer, in early digraphic spelling for umlauted vowels: Der germanische Umlaut ist ursprünglich derselbe Vorgang: das bezeugen die ahd. Schreibungen airin, aigi..., ailliu..., muillen..., suinta, zuinta, troistet, guita..., scoina... und das altnordische aw für das durch u umgelautete a. Eben diese Schreibungen bezeugen zum Theil, was wir auch ohne sie vermuthen müssten, dass das vorklingende j oder / erst deutlich vernommen werden konnte, als das nachklingende verschwand und das schwache e oder an seine Stelle trat. Aber auch dann gehörte ein besonders feines Ohr dazu, um das / als einen eigenen Laut abzusondern. Die Majorität der Sprechenden besass ein so feines Gehör nicht, und so entstand ein neuer Laut, der zwischen den beiden sich vermischenden lag. (Scherer 1868: 144-145) This statement was to remain the standard interpretation of Germanic /-umlaut for the next seventy years. From an early date, however, some scholars believed that digraphic spellings were just a way of indicating the palatal nature of the root vowel, using for this purpose the inadequate resources of the Latin script. Consequently, they rejected the existence of the diphthongal stage and suggested that the assimilation proceeded directly from the palatalised consonants onto the root vowel. The foremost proponent of this hypothesis was Eduard Sievers. He first formulated his views in 1873 and expressed them fully in his contribution to Paul's Grundriss:

Seltener wirkt (...) der umlautende Vokal direkt (ahd. säjen = mhd. ssejen), gewöhnlicher treten Konsonanten als Vermittler auf, indem sie die spezifische

Epenthesis and Mouillierung in the explanation ofi-umlaut

153

Stellung des umlautenden Vokals durch Artikulationsmischung in sich aufnehmen und so mit der des umzulautenden Vokals in Kontakt bringen. (Sievers 1891:296)

Sievers's theory found a strong supporter in Wilhelm Braune and a number of other prominent neogrammarian scholars. By the second decade of the twentieth century, the Mouillierungstheorie, the epenthesis theory, or a combination of these two were widely accepted by the scholarly community. Accordingly, they were employed in a number of major handbooks of various Germanic languages, which were to remain standard textbooks for many years to come. This explanation of i-umlaut was offered for Old High German (Behagel 1916), Old Norse (Kock 1888), and Old English (Luick 1914-1940). On the basis of further evidence adduced by, e.g., van Dantzig, von Essen, Götlind, Förster, Rooth, or Kranzmayer, these theories still found their way into more recent textbooks such as, e.g., Schwarz (1951), Gutenbrunner (1951), or Campbell (1959), who stated that "/-umlaut was, in fact, begun by epenthesis of / into an accented syllable, and such an epenthetic could hardly be anything but a glide anticipating a palatal consonant" (1959: 72). Later, however, no major book on the subject has accepted epenthesis or Mouillierung as the primary explanation for /-umlaut. Mam arguments for the epenthetic stage and/or the palatalised consonants as the vehicle of /-umlaut can be summarized as follows: (a) Digraphic spellings in a number of Northwest-Germanic languages, at an early stage of their development, reflect diphthongal pronunciation; this is attributed to the emergence of an epenthetic glide (in all likelihood, though not necessarily, before a palatalised consonant), (b) The fact that in Old High German and Middle High German (and, to a lesser extent in other Northwest-Germanic languages) certain consonantal clusters generally associated with the [+velar] quality hindered the speed and/or extent of /-umlaut shows the process to proceed by way of palatalizing intervening consonants, (c) Experimental phonetics confirms the existence of fronting and/or palatalizing influence of an [i] or [j] in an unstressed position on the preceding consonant(s) and root vowels, (d) A number of Northwest-Germanic languages and/or their dialects still exhibit palatalised consonants (Old English, Old Frisian, Upper German) or epenthetic glides (e.g., Swedish),

154

Martin Krygier

(e) A number of Old Irish loanwords in Old English with the original structure identical with the suggested /-umlaut diphthongs developed in the same way as original mutated vowels; moreover loanwords from Upper German into Slovene, in which Old Bavarian sequences of a consonant adjacent to an [i] were realized as palatal consonants, prove the existence of the same in early German.

3. Criticism of the epenthesis and Mouillierung theories It must be admitted that over the years an impressive list of counterarguments has been formulated. From the very beginning a considerable number of scholars expressed doubts about the validity of the Mouillierung and epenthesis theories. Moreover, one has to remember that even among the proponents of these theories differences abounded. Some of them accepted epenthesis while rejecting consonant palatalisation, some believed in Mouillierung and denied the existence of an epenthetic stage, while still others supported an explanation of /-umlaut combining both of the above. A perfect example of this behavior may be Eduard Sievers himself, who in the course of his life at least twice changed his mind about the correct explanation of z'-umlaut. On the other hand, those who saw umlaut primarily as a case of distant assimilation voiced their opposition to these theories from the moment they had first been formulated. To begin with, the diphthongal interpretation of digraph spellings was questioned. An alternative hypothesis was put forward, namely that they constituted imperfect attempts on the part of native scribes to represent new umlaut vowels using combinations of letters available to them in the Latin script. This also found support in studies devoted to the development of the runic alphabet. The possibility that the existence of palatalised consonants in Germanic dialects is not connected with /-umlaut was first presented by Victor (1890). He observed that in attested Germanic languages evidence for palatalised consonants was limited mainly to velars. Moreover, if they did appear, they were not always linked with /-umlaut. Van Haeringen (1918) in his detailed study of umlaut and breaking listed a number of arguments against the Mouillierung and epenthesis theories. Thus, he stated that a conditio sine qua non of all umlaut changes in Germanic was the weak stress on the inducing syllable, while in the languages which exhibit consonant palatalisation both stressed and unstressed vowels could act as triggers. He also offered a salient observation against the epenthesis theory, namely the need for explaining the exceptionless monophthongisation of the diphthongal stage after the emergence of the /-glide. This question has been left unanswered until the present day. In the 1920s the Mouillierung and epenthesis theories started losing popularity. In a number of papers and handbooks they were treated with suspicion, either openly rejected altogether or questioned and presented as only one of a number of

Epenthesis and Mouillierung in the explanation ofi-umlaut

155

possibilities (cf. Schatz 1927: 39; Hirt 1931-1934; H fler 1955). This move culminated with the appearance of a series of very influential papers on the subject of /-umlaut (Penzl 1949; Samuels 1952; Schweikle 1964). The arguments formulated therein remain valid until the present day and cam be summarized in the following fashion: (a) The digraphic spellings represent new umlaut vowels, not epenthetic glides or palatalised consonants. Scribes did not use a narrow phonetic transcription; not only did they omit nondistinctive features but also they left certain phonemic contrasts not signified in writing, (b)The Mouillierungstheorie cannot explain double umlaut (e.g., OHG edili 'noble', fremidi Ί performed', etc.) or cases of umlaut across word boundaries (e.g., OHG werfiz Ί threw', meg ih Ί may') (Penzl 1949: 235); the existence of consonantal clusters blocking /-umlaut is only apparent, as the mutated vowels appeared even in those positions, though they were represented in writing later than those in other environments (Penzl 1949: 235). Moreover, in Old Norse, Old English, and Old Frisian /-umlaut took place largely irrespective of the quality of the intervening consonants. Consequently, if one believes in what is now unequivocally agreed upon, namely that umlaut processes in all Germanic languages were of the same nature, this constitutes a major counterargument to the Mouillierungstheorie (Samuels 1952: 38-40). (c) The use of loanword evidence has only a limited application in establishing phonetic details (Penzl 1949 contra Kranzmayer 1937), (d) If the epenthesis theory were correct, we should expect similar behavior in all three umlauts (Andersen 1946), which is not true. All this, however, seems not to have discouraged a number of scholars from reviving the Mouillierung and particularly epenthesis theories in recent years. Thus, Stefan Sonderegger declared himself a supporter of the epenthesis theory writing in 1979: "The oldest umlaut graphemes of Old High German show not infrequently digraphic spellings ae, ei, ai for the /-umlaut reflex of a. In this we see an indication of an earlier, intermediate stage a1, ae > e1, ee > e" (1979: 302-304). Likewise, Samuels in his Linguistic evolution listed the Mouillierungstheorie as one of the possible explanations of /-umlaut: "Some doubt attaches to whether the feature spreads through intervening consonants... or whether the change is psychological, resulting from a simplificatory habit ('rule') adopted in the coordinating mechanism; probably either or both factors may apply in different languages" (Samuels 1972: 15-16). This is the more remarkable as the same author in an earlier paper took a radically anti-Mouillierung stance (Samuels 1952). Similar sentiments have been expressed more recently by, e.g., Wehia (1987),

156

Mar ein Krygier

where it is given equal status with distant assimilation: "The above phonetic explanation of /-umlaut can hardly be refuted since velar consonants [k, g] in intervocalic position must have been influenced by the mutation-causing segments, and they appear in Old English as [δ] and palatalized [g] respectively. In addition, certain consonants followed by [i] in the next syllable were even capable of preventing breaking" (Wema 1987: 60). Other scholars associated with this trend are Basb011 (1993), who draws extensively on theories of Walde (1900), and Liberman, who maintains that "There is sufficient historical and dialectal evidence that umlaut also passed through the diphthongal stage, and a diphthong is a proper basis to accommodate two morae in place of one" (Liberman 1991: 132).

4. Conclusions In conclusion we might ask how much is left of the evidence for Mouillierung and epenthesis, and whether there are any grounds for reviving them. Perception of digraph spellings relies heavily on the interpretation of the data. The evidential value of umlaut-hindering consonantal clusters is diminished when one remembers that velar obstruents are very susceptible to palatalisation and, therefore, should not act as inhibitors in this case. The putative existence of palatalised consonants and the testimony yielded by loanwords is limited by the impossibility of exact reconstruction of phonetic details (it is enough to mention here the ongoing debate over the distribution of palatal velars in Old English to see the scope of the problem). Experimental phonetics, finally, though some caution must be retained in projecting its findings diachronically, seems generally to confirm the possibility of Mouillierung and epenthesis in the /-umlaut environment. On the other hand, none of the crucial arguments for Mouillierung and epenthesis has been decisively defeated. There is no doubt that if the Mouillierung and epenthesis theories are to be once again seriously considered, answers to two crucial questions must be provided: (i) why is there almost no trace of the postulated palatalised consonants (especially non-velars) both in old texts and in modern dialects?, and (ii) why is the behavior of the root vowel and epenthetic glide clusters different from that of regular diphthongs? At the same time there is little justification in dismissing them as categorically as it has been done over the last few decades. Much still remains to be researched, and only after much more is known about the phonetics of Primitive Old English the epenthesis and Mouillierung theories will have their judgment day.

References Andersen, Harry 1946 "Er brydnigen et omlydsfaenomen?", Arkiv for Nordisk Filologi 61: 157-170.

Epenthesis and Mouillierung in the explanation of i-umlaut

157

Andersen, Henning (ed.) 1995 Historical linguistics 1993. Selected papers from the llth International Conference on Historical Linguistics, Los Angeles, 16-20 August 1993. Amsterdam-Philadelphia: Benjamins. Antonsen, Elmer H.—Hans H. Hock (eds.) 1991 Stxfcraeft. Studies in Germanic linguistics. Select papers from the First and the Second Symposium on Germanic Linguistics, University of Chicago, 24. April 1985, and University of Illinois at Urbana-Champaign, 3-4 October 1986. Amsterdam-Philadelphia: Benjamins. Basb011, Hans 1993 "The Nordic /-umlaut and natural principles of syllabification: a possible scenario?", North-Western European Language Evolution 21/22: 37-52. Behagel, Otto 1916 Geschichte der deutschen Sprache. Strassburg: Trübner. Campbell, Alistair 1959 Old English grammar. Oxford: Clarendon Press, van Dantzig, Branco 1929 "Der Einfluss des hetero-syllabischen Jots auf einige vorangehende Vokale in der niederländschen Sprache", Archives neerlandaises de phonetique experimentale 4: 60-66. von Essen, Otto 193 5 "Über das Wesen der Assimilation", Vox 21: 10-23. Förster, Max 1935 "Zur i-Epenthese im Altenglischen", Anglia 59: 287-298. Götlind, Johan 1927 "Ett modernt omljud av a till ", Arkiv for Nordisk Filologi 43: 69-71. Gutenbrunner, Siegfried 1951 Historische Laut- und Formenlehre des Altisländischen. Heidelberg: Winter, van Haeringen, Coenraad B. 1918 De germaanse inflexieverschijnselen ("umlaut" en "breking") phonetics beschouwd. Leiden: van Nifterik. Hirt, Hermann 1931-1934 Handbuch des Urgermanischen. Heidelberg: Winter. Höfler, Otto 1955 "Stammbaumtheorie, Wellenthorie, Entfaltungstheorie", Beiträge zur Geschichte der deutschen Sprache und Literatur 77: 30-66. Hogg, Richard M. 1992 A grammar of Old English. 1: Phonology. Oxford: Basil Blackwell.

158

Mar ein Krygier

Kock, Axel 1888

"I-omljudet och den samnordiska förlusten af andelsevokaler", ArkivförNordiskFilologH: 141-162, 185. Kranzmayer, Eberhard 1937 "Die Geschichte des Umlautes im Südbairischen", Zeitschrift für Mundartforschung 14: 73-100. Liberman, Anatoly 1991 "Phonologization in Germanic: umlauts and vowel shifts", in: Elmer H. Antonsen—Hans H. Hock (eds.), 125-137. Luick, Karl 1914-1940 Historische Grammatik der englischen Sprache. Stuttgart: Tauchnitz. Paul, Hermann (ed.) 1891 Grundriss der germanischen Philologie. Strassburg: Trübner. Penzl, Herbert 1949 "Umlaut and secondary umlaut in Old High German", Language 25: 223-240. Rooth, Erik 1935 "Primära i-omljudet och frägan om muljerade konsonanter och i-epenteser i vast- och nordgermanskan", Vetenskapssocietetens i Lund Ärsbok 1935: 5-34. Samuels, Michael L. 1952 "The study of Old English phonology", Transactions of the Philological Society: 15-47. 1972 Linguistic evolution. Cambridge: CUP. Schatz, Josef 1927 Althochdeutsche Grammatik. Göttingen: Vandenhoeck & Ruprecht. Scherer, Wilhelm 1868 Zur Geschichte der deutschen Sprache. Berlin: Duncker. Schwarz, Ernst 1951 Goten, Nordgermanen, Angelsachsen: Studien zur Ausgliederung der germanischen Sprachen. Bern-München: Francke/Lehnen. Schweikle, Günther 1964 "Akzent und Artikulation. Überlegungen zur ahd. Lautgeschichte", Beiträge zur Geschichte der deutschen Sprache undLiteratur 86: 197-265. Sievers, Eduard 1873 "Ueber den Umlaut im Deutschen", Verhandlungen der achtundzwanzigsten Versammlung deutscher Philologen und Schulmänner in Leipzig 1872. Leipzig. 1891 "Phonetik", in: Hermann Paul (ed.), 266-299.

Epenthesis and Mouillierung in the explanation of {-umlaut

159

Sonderegger, Stefan 1979 Grundzüge deutscher Sprachgeschichte. Diachronie des Sprachsystems. Band I. Einführung — Genealogie — Konstanten. Berlin-New York: de Gruyter. Vie'tor, Wilhelm 1890 "A review of Henry Sweet's A history of English sounds", Phonetische Studien 3: 79-96. Wehia, Jerzy 1987 A critical survey of a historical phonology of English vowels. Warszawa: Wydawnictwa Uniwersytetu Warszawskiego.

On minor declarative complementizers in the history of English: The case of but Maria Jose Lopez-Couso and Belen Mendez-Naya

1. Introduction Throughout the history of English, that and zero have been used as the major links introducing finite declarative complement clauses. However, a number of minor links in complementizer function have also been on record since Old English times. These include how (thai), but (that), (as) if, (as) though, and lest, among others. While much attention has been devoted to the diachronic variation between that/zero, the study of these minor declarative complementizers has been almost wholly neglected in the literature. In this paper we shall be concerned with one of these, namely but.1 Section 2 will be devoted to the uses of the complementizer but in the history of English and to the description of the different construction-types in which this item occurs. In section 3, we shall focus on the origin of but as a declarative complementizer. Finally, a summary of the main conclusions will be given in section 4. For our purposes we have examined all the occurrences of but in the computerized Helsinki Corpus of English Texts (Kytö 1996). Additional material has been drawn from other various sources, among them the standard historical dictionaries.

2. The complementizer use of but in the history of English As a subordinating conjunction, but has been typically associated in different stages of the English language with adverbial clauses of exception (cf. Quirk et al. 1985: § 15.44) and condition, with the meanings of 'except that' and 'if not' respectively (cf. OED s.v. but II.8-11). Along with these subordinating uses of but, examples can be found in the history of English where this item introduces complement clauses. As a complement clause link, but can be characterized as an originally negative complementizer, roughly equivalent to 'that... not', and found, according to the OED (s.v. but II.12ff), "after negative and questioning constructions". This negative use of but, henceforth but\, is illustrated in example (1), where the meaning of the 6w/-clause can be glossed as 'that they were not Portuguese': (1)

Yet being asked, what they were, they said, they were (^Moores*), and shewed vs their backes all written with Characters; and when we affirmed them to be (^Portugals*), they then told vs the (^Portugals*)

162

Maria J. Lopez-Couso and Belen Mendez-Naya were not circumcised. But to conclude; our Company would not be perswaded but that they were (^Portugals*) (1612, R. Coverte, A Trve and Almost Incredible Report of an Englishman, 16)

However, in the course of time but developed a second use, henceforth referred to as but^ Contrary to but\, but2 does not carry any negative implication in itself, thus being an equivalent to that, the declarative complementizer par excellence. (2) below is a particularly telling example, in that it illustrates the apparent interchangeability between that and but2 in complements to the predicate not doubt. (2)

(*Ph. *) Then thou doubtest not that those who deserve Punishment are miserable. (ΛΒο.Α) No, I cannot.(*Ph.*) If therefore thou wert to be Judg, to which dost thou think thou wouldst adjudg Punishment, to him who hath done, or to him who hath suffered the Injury? (ΛΒο.Λ) I doubt not but that I should adjudge Satisfaction to the Sufferer, by punishing the Doer of Wrong (1695, R. Preston, Boethius Or The Consolation of Philosophy, 183-184)

The most obvious difference between the two uses of but as complementizer is that but\ affects the polarity of the subordinate clause as a whole, marking the complement as negative, whereas but2 leaves it unaltered. Given that but} is a marker of sentence negation, a second negative on the verb is regarded as unnecessary, and even redundant (cf. Jespersen 1917: 131). In fact, none of the but} examples in the Helsinki Corpus and additional material show a negated verb in the sub-clause.2 The status of but as sentence negation marker becomes evident in the following examples taken from two different translations of the Bible (Luke xvii.l), quoted in the OED (s.v. but 11.16). Note that the version with that in example (3b) shows a negated VP in the sub-clause. By contrast, but2, like that, is found introducing both affirmative complements, as in (2) above, and negative complements, as in (4): (3a) // is impossible but that offences will come (1582 Rhem.) (3b) // is impossible that sclaundris come not (Wyclif) (4) we beseche the same to pardone us: not doubting but in the same and all other your Highnes maters we shal not faile to endevoyre our selfs according to our most bounden dueties as shalbe, we hoope, to your Graces satisfaction and contentement (1539, A Letter by the Lords of the Council, ΙΛ25) The overall number of examples of but as a declarative complementizer in the different sections of the Helsinki Corpus amounts to 64.3 Table 1 shows the distribution of these examples, and the breakdown for occurrences of but] and but2 as defined above:

On minor declarative complementizers in the history of English

163

Table 1: Distribution of complementizer but in the Helsinki Corpus. Ml

but,

but2

. -

2 -

M2 _ -

M3

M4

El

E2

1 -

2 1

1 21

1 16

E3 _ 19

As the figures in Table 1 suggest, the use of but as a complement-clause link consolidates in the Early Modern English period. Our data thus corroborate Warner's assertion that occurrences of the complementizer but are rare before the late 14th century (1982: 250, note 9), since only a few scattered examples have been recorded in the Middle English section of the Helsinki Corpus. According to Warner (1982: 222-223), the spread of but as a complementizer is paralleled by the decline of another negative link formerly available in English, namely ME bat ne, its obsolescence apparently deriving from the loss of the negative adverb ne in the late 14th and early 15th centuries.4 In most of our early examples, but is used in its original function as a negative complementizer, i.e., with the meaning 'that... not'. Likewise, the only two Old English examples that we have been able to trace in our additional sources are liable to be interpreted as cases of but\ These are (5) and (6) below: (5) (6)

Fordam be se bisceop wxs bysig mid bxm cynincge. and nyste butan hi sungon lofsangforö on (JELS 21. 236) Ne gedafhad cristenum menn buton bset he on duste swelte (MLS 31. 1355)

In order to provide a more detailed analysis of the structure under study, we have classified the matrices taking oid-complementation in the corpus according to two criteria. The first criterion is whether the matrix clause is negative or interrogative (cf. the OED characterization of the owi-construction that we cited at the beginning of this section).5 The second relates to a semantic feature of the matrix predicate, namely whether it is inherently negative6 or not. The following list shows the distribution of our corpus examples according to the above-mentioned criteria. Predicates taking iw/pcomplements have been marked with an asterisk. I. Negative matrices (59) (a) Inherent negatives (49): not doubt (25), be no doubt (4), have no doubt (2), be not doubted (2), not deny (8), be not denied (2), not question ('not doubt'; cf. OED s.v. question v.4), nothing hinders, nothing lets or prohibits, be not unlikely, be not impossible, be not improbable, (cf. examples (2) and (4) above). (b) Predicates with no inherently negative import (10): be not possible (2), *be not likely, *not know (4), *not say, not hear, *not be persuaded, (cf. example (1) above). II. Interrogative matrices (3) (cf. example (7) below), (a) Inherent negatives: doubt, deny.

164

Maria J. Lopez-Couso and Belen Mendez-Naya

(b) Predicates with no inherently negative import: be possible. III. Matrices which are neither negative nor interrogative (2): (a) Inherent negatives: *be impossible, (cf. example (3a) above). (b) Predicates with no inherently negative import: *be a great shame (cf. (8) below). (7)

(8)

This industry used infourminge litel infantes, who shall dought, but that they, (not lackyng naturall witte,) shall be apt to receyue lerninge, whan they come to mo yeres? (1531, Thomas Elyot, The Boke Named the Gouernour, 23) 'Iffthys be trew,' seyde Arthure, 'hit were grete shame unto myne astate but that he were myghtyly withstonde' (a 1470, Malory, Morte Darthur, 45)

As can be easily gathered, the matrices in group III do not conform to the OED description, whereby owf-clauses complement negative or interrogative matrices. Yet, curiously enough, the OED (s.v. but 11.17) does quote examples parallel to (8) above in connection with the matrix (*Tis) pity, as in (9): (9)

// were pity but thou were hanged before (1573 N. C. Il.iii. in Hazl. Dodsley 111.34)

The most salient feature of the 6w/-clauses in examples (8) and (9) is the presence of a so-called commentative or factive predicate7 in a matrix which is neither negative nor interrogative. In order to account for examples of this kind, we believe that a characterization of 6wf-constructions wider than that given in the OED is required. This problem can easily be solved by describing the matrices of £w/-clauses as "non-assertive". As is well known, the label "non-assertive" covers not only negative and interrogative environments, but also those involving both commentative predicates, as in (8) and (9), and predicates of negative import, as in (10) (cf. Quirk et al. 1985: § 10.61): (10)

Therfore, al the while that a man hath in hym the peyne of concupiscence, it is impossible but he be tempted somtime and moeved in hisflessh to synne (c!390, Chaucer, The Parson's Tale, 297.C1)

Besides, the predicate list above yields relevant information on the distribution of but] and bu(2. In the corpus, but\ never occurs with group La, i.e., after negated matrices with inherent negatives (in this connection cf. Jespersen 1917: 131). Thus, it is only selected by the predicates not know (2), not say, be a great shame, be not likely, and not be persuaded. By contrast, it is with group I. a, which is precisely the context from which but\ seems to be excluded, that buh appears to be most generally associated. In fact, 51 out of the 57 examples ofbut2 can be characterized in this way.

On minor declarative complementizers in the history of English

165

However, even in such cases the selection of but2 as complementizer is by no means obligatory in the corpus. Other links, such as that or zero, have also been attested in the environment in question. Although the number of examples satisfying this description is too small to warrant definite conclusions, some tentative observations can be made. Thus, there are a few predicates that show a clear preference for but·,. Such is the case of not doubt, which occurs with finite declarative complements in 31 examples in the Early Modem English section of the Helsinki Corpus. But2 is selected in 25 cases, while zero and that occur in 4 and 2 examples respectively. This is, however, not the general tendency for other predicates of this kind, since, as noted by Fanego (1990: 19), complement clauses in non-assertive contexts select but much less frequently than that and zero. (2) above shows the alternation ofthat and but2, while (11) illustrates the use of zero in a similar context. (11) for I double not Sir (^Thomas Wyat*) hath bin examin'd of me, and hathe sayde what he could directly or indirectly (1554, Trials of Sir N. Throckmorton, I,68.C1) As often happens with syntactic variation, it could well be the case that the selection of but over that/zero just described is not totally unmotivated. However, at this stage of our research, we have not been able to identify any factor underlying this complementizer choice. Further work is required in this respect.

3. A hypothesis on the origin of the complementizer use of but According to the OED, complementizer but was modeled on the Latin conjunction quin. From a morphological point of view, quin derives from the old instrumental case of the interrogative, indefinite, and relative stem qui- plus the enclitic negative -ne (cf. Woodcock 1959: 141; Lakoff 1968: 135), In the course of time quin came to be used as a complementizer under highly specific circumstances, namely, after negated matrices with inherent negatives and in subordinate constructions having negative force or implications, with the meaning 'that... not' (cf. OLD, s.v. quin conj. B.4; Ernout—Thomas 1951: §§313-314; Lakoff 1968: 130-137). From there quin spread to structures of other types, such as non-assertive contexts with utterance or knowledge predicates, and also to constructions where the complementizer acquires the meaning 'that'. As becomes evident, there is a clear parallelism between English but and Latin quin, and this raises the possibility that the former may have been influenced by the latter.8 If one assumes, following the OED, that but is modeled on Latin quin, then our earliest examples of the 6«/-construction would be expected to occur in Latin-related texts. Table 2 gives the distribution of our-clauses in the corpus as regards relation to a Latin original.9 Since in the Helsinki Corpus Latin-related texts are poorly represented in comparison with texts that have no relationship to a

166

Maria J. Lopez-Couso and Belen Mendez-Naya

Latin original, the table provides both raw numbers and normalized frequencies for a text-length of 10,000 words. Table 2: Distribution of ouf-clauses as regards relationship to Latin. (LR = "Latin-related texts"; MR = "no relationship to Latin").

LR NR

Ml

M2

M3

M4

El

E2

E3

(-) 2(0.17)

-

1 (0.07)

3(0.15)

6 (6.06) 16(0.88)

2 (2.9) 15 (0.82)

7 (7.93) 12 (0.73)

As can be seen in Table 2, and contrary to our expectations, none of the earliest instances of but in the Helsinki Corpus occur in Latin-related texts. Likewise, the two ^Elfrician examples found in our additional sources, quoted above as (5) and (6), are not recorded in a translation from Latin either. It seems clear, therefore, that the evidence available does not allow us to confirm the claim that the complementizer use of but is modeled on Latin. At first sight, the figures given for the Early Modern English period in Table 2 would seem to suggest a much closer connection between Latin-related texts and the occurrence of but since 1500. However, detailed examination of the evidence shows that this is far from being the case. The only Latin translations represented in sections El to E3 are the various Early Modern English renderings of Boethius' De Consolatione Philosophiae. This text, a debate between Philosophy and Boethius, abounds in non-assertive contexts which, as seen in section 2, are the typical environments in which 6w/-complementation occurs (e.g., Do you doubt + clause?, / do not deny + clause). Thus, the very nature of the text under consideration may go a long way towards explaining the relative frequency of 6«/-complements in the Early Modern English sections of the Helsinki Corpus. hi the light of the evidence presented so far, there seem to be no grounds to posit a direct dependence of but on Latin. This is also corroborated by the fact that, even in Latin-related texts, but may occur independently of quin. In other words, ^-complements not only translate