a comparative grammar of brtiish english dialects

384
A Comparative Grammar of British English Dialects

Upload: seda-kndmr

Post on 08-Nov-2014

167 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: A Comparative Grammar of Brtiish English Dialects

A Comparative Grammar of British English Dialects

Page 2: A Comparative Grammar of Brtiish English Dialects

Topics in English Linguistics50.1

Editors

Elizabeth Closs TraugottBernd Kortmann

Mouton de GruyterBerlin · New York

Page 3: A Comparative Grammar of Brtiish English Dialects

A Comparative Grammarof British English Dialects

Agreement, Gender, Relative Clauses

by

Bernd KortmannTanja HerrmannLukas PietschSusanne Wagner

Mouton de GruyterBerlin · New York

Page 4: A Comparative Grammar of Brtiish English Dialects

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

�� Printed on acid-free paper which falls within the guidelinesof the ANSI to ensure permanence and durability.

Library of Congress Cataloging-in-Publication Data

A comparative grammar of British English dialects : agreement, gen-der, relative clauses / by Bernd Kortmann … [et al.].

p. cm. � (Topics in English linguistics ; 50.1)Includes bibliographical references and index.Contents: The Freiburg English Dialect Project and corpus / Bernd

Kortmann, Susanne Wagner � Relative clauses in English dialects ofthe British Isles / Tanja Herrmann � “Some do and some doesn’t” :verbal concord variation in the north of the British Isles / LukasPietsch � Gender in English pronouns : southwest England / Su-sanne Wagner

ISBN 3-11-018299-8 (hardcover : alk. paper)1. English language � Dialects � Great Britain. 2. English lan-

guage � Great Britain � Grammar. 3. English language � Relativeclauses. 4. English language � Agreement. 5. English language �Gender. I. Kortmann, Bernd, 1960� II. Series.

PE1721.C66 2005427�dc22

2005001607

Bibliographic information published by Die Deutsche Bibliothek

Die Deutsche Bibliothek lists this publication in the Deutsche Nationalbibliografie;detailed bibliographic data is available in the Internet at �http://dnb.ddb.de�.

ISBN 3-11-018299-8

” Copyright 2005 by Walter de Gruyter GmbH & Co. KG, 10785 BerlinAll rights reserved, including those of translation into foreign languages. No part of thisbook may be reproduced or transmitted in any form or by any means, electronic or mecha-nical, including photocopy, recording, or any information storage and retrieval system, with-out permission in writing from the publisher.Cover design: Christopher Schneider, Berlin.Printed in Germany.

Page 5: A Comparative Grammar of Brtiish English Dialects

Preface

Bernd Kortmann

Since the 1980s, but especially over the last ten years or so, the study of the grammar of English dialects has been very much on the rise after more than a century of neglect in English dialectology and dialectology, in general. Witness, in particular, Trudgill and Chambers (1991), Milroy and Milroy (1993), and, on a global scale, Kortmann and Schneider (2004). Apart from these and several other publications related in spirit, however, the vast majority of publications on the grammar of English dialects concentrates on just one particular phenomenon in one particular dialect or dialect area, is based on a very small database and purely descriptive. Moreover, the small size of the available databases often makes it very difficult to formulate valid descriptive generalizations. Virtually non-existent in English dialectology are systematic comparative studies of individual grammatical subsystems across a selection of dialects (like comparative studies of the tense and aspect systems, pronominal systems, relativization or complementation patterns, etc.). Exceptions in this respect form the sociolinguistic studies by Tagliamonte and her research team (e.g. Tagliamonte 1999, 2002, 2003), and the contributions, especially the regional and global synopses, in Kortmann and Schneider (2004). However, useful as the synopses are in providing general orientation, they can be no more than very useful starting-points for systematic comparative analyses of individual phenomena of dialect grammar.

The present volume, the first in a series of volumes which will be published at irregular intervals, tries to set an example as to how this gap in English dialectology can be filled. Secondly, it will do away with another problem that has beset the study of English dialect syntax for many decades, namely the lack of a sufficient amount of reliable data. The Surveyof English Dialects, for example, compiled in the 1950s and serving as the most important data source for English dialectologists and dialect geographers ever since, was simply not geared to the systematic collection of data on grammar. Just a fraction of the more than 1300 questions in the SED questionnaire was explicitly designed to collect morphological and syntactic information. Only since the late 1980s have efforts been made at compiling large data collections, such as the Survey of British Dialect

Page 6: A Comparative Grammar of Brtiish English Dialects

vi Bernd Kortmann

Grammar (Cheshire, Edwards and Whittle 1989), the Newcastle Electronic Corpus of Tyneside English (NECTE; see Allen et al., forthcoming), and, largest of all, the computerized Freiburg English Dialect Corpus. It is the latter, FRED for short, which will take centre stage in this volume. Thirdly, all studies in this volume are informed by a typological approach to English dialect grammar (apart from the fact that two of the dialect phenomena investigated here, namely the Northern Subject Rule and pronominal gender, are typologically very rare). This approach is the hallmark of the Freiburg research group on English dialect syntax, initiated and coordinated by Bernd Kortmann, and will be outlined in the scene-setting paper by Kortmann and Wagner. It is in this paper, too, that the nature and design of FRED, and its advantages for both qualitative and quantitative analyses of dialect phenomena will be discussed in some detail.

The subject matter of the three studies forming the backbone of this volume can briefly be characterized as follows: Tanja Herrmann examines (adnominal) relative clauses in six dialect areas of the British Isles (Central Midlands, Central North, Central Southwest, East Anglia, Northern Ireland, Scotland). The results of this cross-dialectal study she relates to typological hierarchies, particularly to Keenan and Comrie’s (1977) Noun Phrase Accessibility Hierarchy. The Accessibility Hierarchy is largely verified for all relative clause formation strategies found in the data, including the zero relative marker strategy (as in The man ___ called me was our neighbour).From a diachronic perspective, the Accessibility Hierarchy also helps to reveal the pattern underlying the way individual relative markers (e.g. the relative particles as and what) enter or exit an existent relative marker system.

Lukas Pietsch investigates, synchronically as well as diachronically, the so-called Northern Subject Rule (NSR), a feature found in the Northern dialects of England, but also in Scotland, Northern Ireland and the Republic of Ireland. This rule is concerned with subject-verb agreement, and can roughly be formulated as follows: every verb in the present tense can take an s-ending unless its subject is an immediately adjacent simple pronoun. (Third person singular verbs always take the s-ending, as in Standard English). In other words, the NSR involves a type-of-subject constraint (pronoun vs. common/proper noun) and a position constraint (+/- immediate adjacency of pronominal subject to verb). Thus, in NSR-varieties we get the following examples: I sing vs. *I sings, Birds sings, andI sing and dances.

Page 7: A Comparative Grammar of Brtiish English Dialects

Preface vii

Susanne Wagner, finally, provides a comprehensive account of a special semantic system of (pronominal) gender marking, which is distinctive of the traditional dialects in Southwest England. What we encounter in Somerset, in particular, is pronominal gender that is primarily sensitive to the mass/count distinction and only secondarily to the animate/inanimate and human/nonhuman distinction. It is only used for mass nouns. Count nouns take either he or she: she is used if the count noun refers to a female human, and he is used for count nouns either referring to male humans or to nonhuman entities. Thus we get a contrast as in Pass the bread – it’s over there. (bread = mass noun) and Pass the loaf – he’s over there. (loaf = count noun). Most gendered pronouns are masculine pronominal forms (he,him and Southwestern un, en < OE hine) referring to inanimate referents.

Acknowledgments

All authors most gratefully acknowledge the generous support by the Deutsche Forschungsgemeinschaft. Without the funding of the Projects KO 1181/1-1,2,3 over a five-year period (2000-2005) the studies published here and the compilation of FRED, the Freiburg English Dialect Corpus, would have been impossible.

References

Allen, Will, Joan Beal, Karen Corrigan, Warren Maguire, and Hermann Moisl forthc. Taming Unconventional Digital Voices: The Newcastle Electronic

Corpus of Tyneside English. In Using Unconventional Digital Language Corpora. Vol. I: Synchronic Corpora, Joan Beal, Karen Corrigan and Herman Moisl (eds.). Basingstoke: Palgrave Macmillan.

Cheshire, Jenny, Viv Edwards, and Pamela Whittle 1989 "Urban British dialect grammar: The question of dialect levelling.

English World-Wide 10: 185-225. Kortmann, Bernd, Edgar W. Schneider in collaboration with Kate Burridge,

Rajend Mesthrie, and Clive Upton (eds.) 2004 A Handbook of Varieties of English, Vol. 2: Morphology and Syntax.

Berlin/New York: Mouton de Gruyter.

Page 8: A Comparative Grammar of Brtiish English Dialects

viii Bernd Kortmann

Milroy, John, and Lesley Milroy (eds.) 1993 Real English. The Grammar of English Dialects in the British Isles.

London/New York: Longman. Tagliamonte, Sali 1999 Was/were variation across the generations: View from the city of

York. Language Variation and Change 10: 153–191. 2002 Variation and change in the British relative marker system. In

Relativisation in the North Sea Littoral, Patricia Poussa (ed.), 147–165. Munich: LINCOM EUROPA.

2003 ‘Every place has a different toll’: Determinants of grammatical variation in cross-variety perspective. In Determinants of Linguistic Variation, Günter Rohdenburg, and Britta Mondorf (eds.), 531–554. Berlin/New York: Mouton de Gruyter.

Trudgill, Peter, and Jack K. Chambers (eds.) 1991 Dialects of English. Studies in Grammatical Variation. London/New

York: Longman.

Page 9: A Comparative Grammar of Brtiish English Dialects

Table of Contents

Preface............................................................................................... v Bernd Kortmann

The Freiburg English Dialect Project and Corpus (FRED) .............. 1 Bernd Kortmann and Susanne Wagner

1. Comparative dialect grammar from a typological perspective.................12. The Freiburg English Dialect Corpus (FRED).........................................43. Linguistic consequences of using oral history material .........................13

Relative clauses in English dialects of the British Isles.................. 21 Tanja Herrmann

Abstract ...............................................................................................................211. Introduction ............................................................................................212. Data ........................................................................................................223. Overall distribution of relative clauses and relative markers .................244. Previous investigations of relative markers............................................285. Restrictiveness/nonrestrictiveness..........................................................386. Personality/nonpersonality .....................................................................417. Preposition placement ............................................................................458. Accessibility Hierarchy ..........................................................................489. Resumptive pronouns.............................................................................7010. Which as ‘connector’? ............................................................................8711. Conclusion..............................................................................................94Appendix 1 ...........................................................................................................97Appendix 2 .........................................................................................................105

“Some do and some doesn’t”: Verbal concord variation in the north of the British Isles ......................................................................... 125 Lukas Pietsch

Abstract .............................................................................................................1251. Introduction ..........................................................................................1252. The Northern Subject Rule: Descriptive problems ..............................1283. Data from twentieth-century northern dialects.....................................1324. The history of the Northern Subject Rule ............................................1735. Theoretical accounts of the Northern Subject Rule..............................1796. Discussion: Variation and usage-based theories ..................................190

Page 10: A Comparative Grammar of Brtiish English Dialects

Table of Contents x

Gender in English pronouns: Southwest England ........................ 211 Susanne Wagner

1. Introduction ..........................................................................................2112. Gendered pronouns ..............................................................................2153. Gender in English and elsewhere .........................................................2214. The corpora ..........................................................................................2355. Special referent classes.........................................................................2516. Non-dialectal studies of gender assignment .........................................2617. Persistence of gendered pronouns ........................................................2758. SED – Basic Material ...........................................................................2859. The SED fieldworker notebooks data...................................................29210. Southwest England oral history material..............................................31911. Material from Newfoundland ...............................................................33912. Overall summary ..................................................................................346Appendix ............................................................................................................353(Additional) corpus material...............................................................................353

Index.............................................................................................. 368

Page 11: A Comparative Grammar of Brtiish English Dialects

The Freiburg English Dialect Project and Corpus (FRED)

Bernd Kortmann and Susanne Wagner

1. Comparative dialect grammar from a typological perspective

The Freiburg project started in the late 1990s and has received major funding from the Deutsche Forschungsgemeinschaft from spring 2000 until spring 2005.1 Its basic approach to the study of dialect grammar is informed by the theoretical and methodological framework of functional (or: Greenbergian) typology, which is primarily concerned with the patterns and limits of morphological and syntactic variation across the languages of the world. The basic idea of the Freiburg project is to adopt functional typology as an additional reference frame for dialectological research that fruitfully complements existing approaches. Among other things, this means that, in a first step, we determine the cross-dialectal variation observable in individual domains of grammar (in the present volume: negation, relative clauses, pronominal and agreement systems) before, in a second step, judging it against the cross-linguistic variation described in typological studies. Both dialect syntacticians and typologists are bound to profit from this kind of approach (cf. Kortmann, ed. 2004 for a collection of studies on dialects in Europe conducted in this spirit). On the one hand, dialectologists can draw upon a large body of typological insights in, hypotheses on, and explanations for language variation. Dialect data can thus be looked at in a fresh light and new questions be asked. On the other hand, typologists will get a broader and, most likely, more adequate picture of what a given language is like if they no longer ignore dialectal variation. In fact, non-standard dialects (as varieties which are almost exclusively spoken) are bound to be a crucial corrective for typological research, which is typically (especially for languages with a literary tradition) concerned with the written standard varieties of languages. Standard British English, for example, is anything but representative of the vast majority of English dialects if we think, for instance, of the absence of multiple negation or the strict division of labour between the Present Perfect and the Simple Past.

Page 12: A Comparative Grammar of Brtiish English Dialects

2 Bernd Kortmann and Susanne Wagner

In the present volume, the typological perspective is most prominent in Herrmann’s study on relativization (not surprisingly so, given the fame of Keenan and Comrie’s NP Accessibility Hierarchy in language typology).

(1) Accessibility Hierarchy (AH) subject > direct object > indirect object > oblique > genitive > object of comparison

According to the AH, if a language can relativize any NP position further down on the hierarchy, it can also relativize all positions higher up, i.e. to the left of it. This constraint applies to whatever relativization strategy a language employs. For the relativization strategy known as zero-relativization (or: gapping) there is thus a clear prediction that the relativized NP is most likely to be gapped if it is the subject of the relative clause, next most likely if it is the direct object of the relative clause, etc. However, this is clearly not the case for Standard English: the direct object position can be gapped (2a), whereas the subject position cannot (2b):

(2) a. The man I called _____ was our neighbour. (direct object) b. *The man _____ called me was our neighbour. (subject)

English dialects, on the other hand, conform to the AH prediction. Examples like (2b) are nothing unusual, at all; in fact, gapping of the subject position is an extremely widespread phenomenon in non-standard varieties of English in and outside the British Isles:

(3) a. I have a friend ____ lives over there. b. It ain’t the best ones ____ finish first.

So here we have a striking instance of the situation where the non-standard varieties of English conform to a typological hierarchy whereas the standard variety does not. For further ways in which the AH is relevant when looking at other relativization strategies used in English dialects compare Herrmann’s comparative study of relative clauses in six English dialect areas in this volume.

Another rewarding area for anybody investigating dialects from a typological perspective is the study of negation markers and strategies. This subsystem of English dialect grammar has been investigated in depth by Lieselotte Anderwald, a senior member of the Freiburg research team (cf.

Page 13: A Comparative Grammar of Brtiish English Dialects

The Freiburg English Dialect Project and Corpus 3

especially Anderwald 2002a, 2003). For one thing, multiple negation (or: negative concord) is another striking proof of the typological “well-behavedness” of non-standard varieties of English (and other Germanic languages), since multiple negation is the rule for many standard languages in Europe. Only the standard varieties of Germanic (e.g. Standard English, Standard Dutch, Standard German) are the exceptions. Furthermore, the invariant supraregional negation markers don’t (i.e. also for he/she/it don’t)and ain’t are in full accordance with the powerful typological concept of markedness: as Greenberg found for many languages, morphological distinctions tend to be reduced under negation. As Anderwald (2002b, 2004) has also nicely shown, the alleged amn’t gap (Hudson 2000) in almost all varieties of English (*I amn’t vs. I am not, aren’t I) is anything but a gap and can indeed be considered an extreme case of local (or: reversed) markedness. Whereas for all auxiliary verbs negative contraction (e.g. haven’t, hasn’t, won’t) is vastly preferred over auxiliary contraction (e.g. ’ve not, ’d not, ’ll not), we get the reverse picture for be. Even isn’t(12.5%) and aren’t (3.5%) are used very rarely in the British Isles, so that the near absence of amn’t in standard as well as non-standard varieties is not a striking exception, but simply the tip of the iceberg.

The motivation for this striking preference of be-contraction over negative contraction for all other auxiliaries is most likely a cognitive one, namely the extremely low semantic content of be. This leads on to another typological concept which can be usefully applied to the interpretation of dialect facts: iconicity. In the case of be-contraction we find an extreme formal reduction of a semantically near-empty auxiliary. In other words, the amount of coding material matches the semantic content to be coded. Another case in point is the fact that quite a number of non-standard varieties in the British Isles and, in fact, around the world have made new use of the number distinction for Past Tense be, i.e. the was-weredistinction (cf. Anderwald 2002a). These varieties use was for all persons in the singular and plural in affirmative sentences, while using weren’t for all persons in singular and plural in negative sentences, thus remorphologizing the number distinction of Standard English as a polarity distinction. What we have here is a showcase example of iconicity: a maximal difference in form (was vs. weren’t) codes a maximal semantic and cognitive difference (affirmation vs. negation). The relevant non-standard varieties of English have clearly developed a more iconic polarity pattern than Standard English has.

Page 14: A Comparative Grammar of Brtiish English Dialects

4 Bernd Kortmann and Susanne Wagner

Having outlined and illustrated the typological approach to comparative dialect grammar established by the Freiburg research group, we need to mention at least briefly that there is of course also a generativist perspective from which especially the dialect grammars of Italian, Dutch and German (much less so English dialects) have been investigated. As a matter of fact, generativists discovered the significance of dialect syntax for linguistic theorizing and models of syntax much earlier than typologists. This generativist interest in microparametric syntax began with the advent of the Principles and Parameters approach in the 1980s and has steadily increased ever since (cf., for example, Black and Motapanyane 1996 or various contributions to Barbiers, Cornips and van der Kleij 2002), finding its way even into generative theories based on Optimality Theory (cf., for example, the Stochastic OT account of morphosyntactic variation in English by Bresnan and Deo 2001). In this volume, the generativist perspective will come in only where relevant publications on the dialect phenomena investigated here exist.

2. The Freiburg English Dialect Corpus (FRED)

Given the aims of the Freiburg project it was first of all necessary to compile a database which would allow to conduct serious qualitative and quantitative morphosyntactic research across English dialects. The result is the computerized Freiburg English Dialect Corpus (FRED), which has been compiled over a period of roughly five years (including the digitization of some 120 hours of audio material). FRED consists of approximately 2.5 million words, with representative subsamples for all English dialect areas including data from Scotland and Wales. The data in FRED are orthographically transcribed interviews collected for the most part during the 1970s and 1980s in the course of oral history projects all over the British Isles. The majority of the informants are born between 1890 and 1920, i.e. are roughly a generation younger than the generation of informants who were recruited for the Survey of English Dialects (SED).

2.1. Principles of compilation

Firstly, the corpus was designated to permit the investigation of phenomena of non-standard morphosyntax (rather than analyses of phonetic or

Page 15: A Comparative Grammar of Brtiish English Dialects

The Freiburg English Dialect Project and Corpus 5

phonological details). Features of syntax are – almost by definition – much rarer than features of phonetics and phonology and very large quantities of text are therefore necessary. (According to some estimates, about 40 times the amount of text is needed for a syntactic analysis as opposed to a phonetic one.) This considerably restricted the practicality of collecting our own corpus from scratch. Instead, we decided to try to compile a corpus from materials that were already available. We decided against collecting material with the help of questionnaires in the first phase of the project. Questionnaires were however designed and distributed in the second phase of the project when, on the basis of extensive corpus analyses, interesting, transitional or rare phenomena became apparent that could not be further investigated with the help of FRED alone.

Secondly, we decided to collect material that would best be classified as traditional dialect data. This means that we explicitly tried to find material from speakers who grew up before the Second World War, as this date seems to be the major cataclysmic event after which wide-ranging social and economic changes (with concomitant linguistic changes) came into effect. For example, highly increased mobility after WWII led to dialect levelling on a hitherto unknown scale (see for example Williams and Kerswill 1999: 149); mass affluence resulted, amongst other things, in television sets becoming easily available and spreading at least passive knowledge of the standard language; increased public spending made sure that education changed not only qualitatively but also quantitatively, such that children leaving school at age 11 or 12 – not unusual for lower class children only 60 or 70 years ago – is no longer possible, and so on. Only by concentrating on speakers born before WWII could we at least have a chance that our data would still be “dialectal” in a regional sense, and be comparable to older dialect descriptions and dialect data (on the background of speaker selection for the SED, see Orton 1962: 14). There are a number of other arguments and a priori considerations which contributed to this decision: We had established contact with various researchers, research groups and private individuals who were either in possession of similar materials or were already working with such data, and who had kindly offered us access to them. Moreover, the only existing sources on variation in morphosyntax are based on traditional material, most importantly the SED (Orton et al. 1962–1971). To guarantee comparability between these materials, it was essential that FRED should also consist of traditional dialect material without having to take into account factors like mobility or the influence of mass media.

Page 16: A Comparative Grammar of Brtiish English Dialects

6 Bernd Kortmann and Susanne Wagner

Due also to time constraints, but mainly for the reasons detailed above, it was considered impossible from the outset to record, digitise and transcribe all data that should make up FRED ourselves. Based on our research objectives, we were looking for large quantities of traditional regional speech, preferably by older local speakers with strong family affiliations in the area, that would record the use of speakers who grew up before WWII, or even better, before WWI. This meant that we were looking for material preferably from the 1970s and 1980s, recording older speakers, or from the 1990s, if these recorded very old speakers. Our material had to be recorded in acceptable quality for linguistic analysis, ideally even including transcripts that were reliable on a word-by-word basis, and – most important of all – the material had to be more or less freely available to us as researchers who had not originally been part of the research design. These criteria suggested a new source that has so far not – or hardly – been used for dialectological purposes, namely tape recordings and transcripts from oral history projects.

2.2. The role of oral history

As defined by the Oral History Society, “[o]ral history is the recording of people’s memories. It is the living history of everyone’s unique life experiences” (Oral History Society at http://www.oralhistory.org.uk). Oral history collections sometimes originate from projects (short- or long-term) undertaken by an individual (sometimes also a group of individuals or an institution), typically lay persons, not professional historians, with an interest in a specific theme or topic, often just recording life memories. Such a focus has certain implications concerning the content2 and general circumstances of an interview. Interviewees are generally pensioners in their 60s or older, and only rarely do we find projects that have as many female as male speakers.3

The recording situation makes oral history material ideal for linguistic investigation. The interviewers were usually true insiders, coming from the area, often still speaking the dialect themselves, which tends to relax the interview situation considerably. A second advantage is that the speaker’s attention was genuinely on what was being said, rather than on how it was being said. Fortunately, the Oral History Society advises all potential interviewers to give a copy of the tapes to their local library or archive4,

Page 17: A Comparative Grammar of Brtiish English Dialects

The Freiburg English Dialect Project and Corpus 7

and these are the places where oral history material can be found today across Great Britain.5

2.3. From original recording to text in FRED – the steps and processes

Members of the Oral History Society are advised to at least “[w]rite a synopsis of the interview which briefly lists in order all the main themes, topics and stories discussed”;6 verbatim transcripts are not explicitly mentioned. But, of course, for anyone thinking about long-term work with the material, a transcript is a very good way of allowing people from outside to get an impression of the content of the interview without actually having to listen to the tapes, which is a very time-consuming business. The intentions for the future use of transcripts largely determine how the interview was transcribed, “how” here referring particularly to the (unfortunately very common) practice of “normalizing” the speakers’ language. Since oral history projects as a rule do not involve the employment of a professional transcriber, this is the usual course of events, which is of course perfectly justified for oral history purposes. Just to give one example consisting of several actually occurring utterances, consider (4) which could end up as (5):

(4) That pot? Oh, I, I don’t know, I don’t remember what I made he for. I don’t collect no pots now.

(5) I don’t remember what I made that pot for. I don’t collect pots now.

“Normalization” here has eradicated three morphosyntactic dialect features (he = pot; he here used in an oblique context; double negation don’t ... no),not to mention all the “superfluous data” (repetitions and so on) that are simply left out. This kind of standardized re-written text is of course much more useful to the general public than a transcript that uses so many instances of “eye-dialect” to represent non-standard pronunciation as well as dialect that it is difficult to follow the line of argument.

Despite the obvious linguistic drawbacks, the Freiburg research group was very glad to have transcripts of at least some of its material. Although these were highly deficient from a dialectological point of view, they at least solved such difficult problems for us as deciphering correctly some specialist vocabulary, unusual place names, personal names and so on, such as the names of different apple sorts used for cider making. We then

Page 18: A Comparative Grammar of Brtiish English Dialects

8 Bernd Kortmann and Susanne Wagner

carefully compared existing transcripts with the original tapes and re-inserted all morphological, syntactic and discourse features, taking out irrelevant phonetic or phonological features and features of pure eye-dialect (compare 6 with 6'):

(6) And the farmer wot my gran used ter ee used ter have a white high healed collar.

(6') #And the farmer what my gran used to, he used to have a white high healed collar. (FRED Wil_011)

For the rest of the material where no transcripts were available, we transcribed the original tapes ourselves, mostly with the help of native speakers who either worked on the project or were associated with it in related research projects. In addition, all transcripts were carefully checked by dialectologically trained research assistants.

As a result, the actual transcripts used for FRED are verbatim equivalents of the spoken versions: hesitations, repetitions, false starts of the same sentence and so on are all included.

(7) #Oh well now, I tell you, when I first made my will, Mr (gap ‘name’) my lawyer, (unclear) oh yes (/unclear) he’s still alive, (trunc) I-(/trunc) I told him I I I says, I want to leave the Salvation Army a bit of money, and I have done. (FRED Nott_016)

In addition, as stated above, and most importantly for our research purposes, all morphosyntactic dialect features have been reinserted (indicated in bold print below).

(8) … there used to be a Ginnet what we used to call was a Ginnet, hewere a nice eating apple, a nice sweet apple and a good apple for cider. #When them apples were ripe you could pick them up and could press them like that and you’d see your thumb mark in them or any apple really when he’s ripe, wadn’t it, but when he’s not ripe he’s hard, isn’t he, (unclear) you ain’t gonna, (/unclear) well,anything at all. (FRED Som_001)

Among the features likely to have been “corrected” in the original transcript are, as in (8) for example, a what-relative, demonstrative themand “gendered” pronouns.

Page 19: A Comparative Grammar of Brtiish English Dialects

The Freiburg English Dialect Project and Corpus 9

A variety of phonological features were also kept, either if they were already represented in the original transcripts, or if we suspected that they might interact with morphosyntax, for example contracted forms like wanna, gonna, s’pose and so on. It should be noted that we use the semi-phonetic form mi for /mi:/ used as the possessive pronoun not as “eye-dialect” but in order to facilitate searches. (For severe criticism of gratuitous eye-dialect, see for example Preston 1985, 2000). The orthographic form me, although widespread in other corpora, not only suggests a certain etymology for this form (at worst, a “substitution” of the object form of the personal pronoun for the possessive function), but also complicates computer-based searches considerably, as all instances of the object case of the personal pronoun (He saw me) have to be manually excluded, at least as long as the corpus is not tagged for word class yet.

We also represent certain paralinguistic features like laughter, long pauses, or indistinct stretches of conversation (marked as gaps, unclear passages or truncated words; see also the examples above). All these features are indicated in the transcripts by specific tags to minimize the risk of ambiguities.7 This opens up the possibility for analyses on a pragmatic or discourse level. In this way we have tried to remedy the linguistic shortcomings of the original oral history material as far as possible.

As mentioned above, extralinguistic variables in FRED are constrained by intention – FRED is not designed to be a representative sociolinguistic corpus, but a regionally representative corpus of as broad a dialect speech as possible. As has already been pointed out, our oral history projects concentrated on interviewing older people. These older people are typically very local, that is they still live in the place where they were born, without having moved outside the region for any considerable stretch of time. Also, typical FRED speakers usually left school about the age of fourteen, often much earlier, certainly not progressing to higher education. Finally, most of our speakers are male – as is well known, women tend to use more prestigious, in many cases more standard forms of speech where these are available to them (see for example Chambers and Trudgill 1998²: 30). In other words, most of our speakers would qualify in dialectology as typical NORMs (see Chambers and Trudgill 1998²: 29), that is non-mobile old rural male speakers with little education. Although this restricts the range of investigations that can be conducted with the help of FRED in sociolinguistic terms, it represents exactly the same bias as in earlier dialectological work, where we find a preponderance of NORM speakers as

Page 20: A Comparative Grammar of Brtiish English Dialects

10 Bernd Kortmann and Susanne Wagner

well, so that results from work on FRED will be comparable to earlier studies or to material from earlier investigations.

2.4. Advantages and disadvantages of orthographic transcripts

FRED is transcribed orthographically, as are most computerized corpora. A number of factors – besides a simple realistic evaluation of our resources – had made it clear from the beginning of the project that orthographically transcribing the dialect material would be the only viable (short-term) procedure. First, we had been granted only a restricted amount of time (and funding) to complete the compilation and transcription of the corpus, and there has to be a natural trade-off between the detail of a transcription (depth) and the coverage (breadth). Our aim was a large corpus that would cover a number of dialect regions, and so we had to trade in phonetic detail. Moreover, since our explicit focus was on morphosyntactic variability, for all relevant features of dialect grammar that we expected to investigate and that are discussed in the dialectological literature, a phonetic or phonemic transcription would not only have been unnecessary, but even counterproductive in many cases. For example, one major drawback of a non-orthographic transcript concerns comparability. A non-orthographic transcript would dramatically hamper the feasibility of searching for all tokens of a certain type (for instance be, personal pronouns, and so on), as the researcher would have no clue which forms to look for without knowing which realisations actually occur in a given interview (or even across all interviews). As a result, one would have to return to the procedure that was common in corpus linguistics before the advent of computers: reading through the texts and marking all forms of interest in the process – certainly not an ideal situation. Finally, only an orthographic transcription of the data met the other requirements of our corpus: the finished corpus was intended to be machine-readable, enabling easy access, a variety of searches with various tools, and, most importantly, comparability with other corpora/projects.

As has been mentioned above, research ties between the Freiburg team and similar projects had been established. Since most of these projects were working with orthographic transcripts, this lent additional support to the decision to use orthographic transcription for FRED. Moreover orthographic transcription would allow us to compare our data with older collections and enable us to make comparisons between different speakers,

Page 21: A Comparative Grammar of Brtiish English Dialects

The Freiburg English Dialect Project and Corpus 11

different dialects, different dialect areas, and different corpora. A further advantage of orthographic transcription is the concentration on real (morphosyntactic) dialect features, as phonetically exceptional forms do not distract the analyst’s eye from the task at hand.

All transcription conventions have of course been documented. Thus in many cases phonetic peculiarities may be traced from the transcription and the additional databases alone without having to return to the sound files. An alignment of sound and transcripts is planned for the near future.

2.5. FRED – corpus design and area coverage

2.5.1. Word counts and areal distribution

FRED consists of 370 texts, which total roughly 2.5 million words of text or about 300 hours of speech, excluding all interviewer utterances (see Table 1). The FRED material is broadly subdivided to cover nine major dialect areas, following Trudgill’s “modern dialects” division of Great Britain (see Trudgill 1999²: 65).

Table 1. FRED word counts and areal distribution

dialect area size (in thousands of words) % of total

Southwest (SW) 571 23

Southeast (SE) 643 26

Midlands (Mid) 359 15

North (N) 434 18

Scottish Lowlands (ScL) 169 7

Scottish Highlands (ScH) 23 1

Hebrides (Heb) 151 6

Isle of Man (Man) 10 1

Wales (Wal) 89 4

total 2,449 100%

Page 22: A Comparative Grammar of Brtiish English Dialects

12 Bernd Kortmann and Susanne Wagner

Each dialect area is subdivided into different counties. A detailed breakdown of counties can be found on the project website (http://www.anglistik.uni-freiburg.de/institut/lskortmann/FRED/).

2.5.2. Speakers

FRED contains data from 420 different speakers (excluding interviewers): 268 (63.8 per cent) are male, and 127 (30.2 per cent) are female (gender is unknown for the rest). In all, 77.2 per cent of the textual material in FREDis produced by male speakers, and 21.4 per cent by female speakers.

The age of speakers included in FRED ranges from six years to 102 years, with a mean age of 75.2 years. A breakdown of age groups, according to the amount of text produced by them, is given in Table 2. As can be seen, about three quarters of the textual material in FRED is produced by speakers older than 60 years.

Table 2. FRED – speakers’ ages

age group number of speakers % of textual material in corpus produced

0 – 14 years 9 0.5%15 – 24 years 14 1.2%25 – 34 years 2 0.2%35 – 44 years 2 0.1%45 – 59 years 14 3.8%60+ years 233 74.8%unknown 145 19.4%

The oldest of FRED’s speakers was born in 1877. Overall, 14 speakers (3.3 per cent) were born between 1880 and 1889, 60 speakers (14.3 per cent) were born between 1890 and 1899, 96 speakers (22.9 per cent) were born between 1900 and 1909, and 64 speakers (15.2 per cent) were born between 1910 and 1919. This means that 89 per cent of all speakers in FRED were born before 1920.

Page 23: A Comparative Grammar of Brtiish English Dialects

The Freiburg English Dialect Project and Corpus 13

2.5.3. Recordings

The material included in FRED was recorded between 1968 and 1999. A detailed breakdown of recording dates can be found in Table 3. Over two thirds of all interviews were thus conducted in the 1970s and 1980s, guaranteeing comparability with much dialectological work conducted at that time.

Table 3. FRED – interview recording dates

recording date number of speakers % of all speakers

1961–69 2 0.5%1970–79 122 29.1%1980–89 163 38.9%1990–99 61 14.6%unknown 71 16.9%

3. Linguistic consequences of using oral history material

The decision to base the FRED corpus predominantly on sources of oral history projects has had a range of linguistic consequences, some of them foreseen, others not predictable at the outset. Perhaps the most clearly predictable linguistic consequences stem from the fact that oral history material necessarily involves the speaker talking about his or her life story at great length – very often, in fact, the speakers are actively encouraged to talk almost exclusively about their past. In the realm of tense and aspect, a predominance of past time narratives implies a predominance of past tense contexts (although not infrequently, of course, stretches of past time narratives are narrated in the historical present tense as well). This is an advantage for studies concentrating on past tense paradigms (for example Anderwald in progress), but a clear disadvantage for any investigation into the present tense, as the data typically yields too few examples to make a regional comparison reliable (see Anderwald 2004). It also means that any features that are linked to the present tense domain can be expected (and indeed shown) to be underrepresented: for example use of the (present) progressive vs. the simple form; forms for the “recent past” (for example the “after”-perfect in Hiberno-English); uses, if any, of a habitual present and so on.

Page 24: A Comparative Grammar of Brtiish English Dialects

14 Bernd Kortmann and Susanne Wagner

A second feature one would expect, considering the fact that FREDspeakers tend to tell their own life stories, is a skewing in pronoun frequencies. Based on the monologic nature of many of the interviews in FRED, we might expect first person singular and first person plural contexts to be over-represented. However, a comparison with the more balanced demographic part of the British National Corpus (BNC) that records everyday spontaneous conversations reveals that this is not the case (see Table 4).

Table 4. Personal pronouns in FRED and the BNC

FRED BNC spoken

Pronoun occ. % of total occ. % of total I 61,458 23.4 309,797 26.8he 29,733 11.3 75,442 6.5she 9,418 3.6 42,879 3.7it 41,776 15.9 254,049 22.0we 27,240 10.4 108,698 9.4you 54,163 20.6 268,642 23.2they 38,608 14.7 96,672 8.4total 262,396 1,156,179

Despite the impression one gets when reading through FRED transcripts, first person contexts are not over-represented in the corpus, but account for roughly one third of all personal pronoun contexts in both FRED and the spoken part of the BNC. Although there are slight deviations in frequencies for individual third person contexts (which can easily be explained on the basis of the nature of the recording situations), the overall frequency of first versus third person contexts is surprisingly similar at 33.8 per cent versus 45.5 per cent in FRED and 36.2 per cent versus 40.6 per cent in the BNC(spoken). Based on these figures, we expect that comparative analyses of FRED and other corpora of spoken English involving the category person will produce representative results.

Finally, in the realm of discourse, it has to be stressed that FRED does not contain genuinely spontaneous, everyday conversations, as for example the BNC does. In the worst case, some (but fortunately only a tiny minority of) speakers actually read from prepared notes, as witnessed by pages rustling in the background and distinctive pauses where pages are turned. Although this worst case is mercifully rare, many interviews are

Page 25: A Comparative Grammar of Brtiish English Dialects

The Freiburg English Dialect Project and Corpus 15

nevertheless monologic – understandably, the interviewers tried to keep in the background most of the time. FRED for this reason would probably not lend itself well to the investigation of discourse strategies. However, this limitation is probably not specific to FRED, but applies to dialectological and sociolinguistic interviews alike, as the main objective is always to record the speakers’ speech, rather than one’s own (see Feagin 2002).

The nature of the data in FRED influenced the choice of the phenomena which have been investigated in, so far, four Ph.D. theses and about a dozen Masters theses. The former include those by Herrmann, Pietsch and Wagner, which are presented in revised and shortened versions in the present volume. In all these studies the focus has been on high-frequency morphosyntactic phenomena. Moreover, the machine-readability of FRED,which allows analyses via automatic text retrieval programmes like TACTor WordSmith, has also influenced the methodology, in that for the first time it is possible to conduct not only qualitative, but also quantitative studies of dialect morphosyntax applying established corpus-linguistic techniques.

Comparisons across the whole FRED material have not been possible for very long yet, so most truly comparative projects by members of the Freiburg research team are currently still work in progress. These include cross-dialectal comparisons of multiple negation (Anderwald to appear 2005), past tense paradigms (Anderwald in progress), pronoun systems (Hernández in progress), complementation patterns (Kolbe in progress), and for several areas of dialect morphosyntax the phenomenon of priming (Szmrecsanyi 2004 and 2005). In addition, a whole range of Masters theses have been completed or are currently under way on the basis of material from FRED. For further information on FRED, the Freiburg project and future perspectives of (English) dialect syntax, especially from a typological perspective, compare Kortmann 2002, 2003, 2004; Anderwald and Wagner 2005.

Notes

1. During the funding period (2000-2005) of Project KO 1181/1-1,2,3 a dozen Masters theses and four Ph.D. theses were completed. Two doctoral theses, two postdoctoral theses and several Masters theses based on FRED are well under way. A selection of studies which has grown out of this project is given in the references (marked with a superscript * preceding the publication year).

Page 26: A Comparative Grammar of Brtiish English Dialects

16 Bernd Kortmann and Susanne Wagner

For further information on the Freiburg project and FRED, please consult: http://www.anglistik.uni-freiburg.de/ institut/lskortmann/index.html.

2. “There are some points to cover in every interview: date and place of birth, what their parents’ and their own main jobs were. And whatever the topic, it usually helps to get the interviewee talking if you begin with their earlier life: family background, grandparents, parents and brothers and sisters (including topics such as discipline), then onto childhood home (housework, chores, mealtimes), leisure (street games, gangs, sport, clubs, books, weekends, holidays, festivals), politics and religion, schooling (key teachers, friends, favourite subjects), early relationships, working life (first job, a typical working day, promotion, pranks and initiation, trade unions and professional organisations), and finally later family life (marriage, divorce, children, homes, money, neighbours, social life, hopes). Most people find it easier to remember their life in chronological order, and it can sometimes take you two or three sessions to record a full life story.” (‘Preparing questions’ from http://www.oralhistory.org.uk/advice/).

3. The advantages and disadvantages of using oral history material for linguistic studies will be discussed in detail in section 3.

4. ‘How to do ORAL HISTORY – After the interview’ at http://www.oral-history.org.uk/advice/.

5. However, this procedure also raises the unfortunate problem of copyright. Especially for our older material, where many of the speakers have already passed away, it is near impossible to gain copyright clearance. Museums and archives are also often reluctant to provide copyright clearance, so that in the foreseeable future FRED will only be accessible to academic researchers, and cannot be published in its present form.

6. ‘How to do ORAL HISTORY – After the interview’ at http://www.oral-history.org.uk/advice/.

7. The tags used in FRED include pauses, different types of non-verbal elements, truncations, editorial corrections, dubious items, uncertain transcriptions, gaps in transcription.

Page 27: A Comparative Grammar of Brtiish English Dialects

The Freiburg English Dialect Project and Corpus 17

References

* Asterisks indicate publications or work-in-progress by members of the Freiburg research team on the grammar of British English dialects.

Anderwald, Lieselotte *2002a Negation in Non-Standard British English: Gaps, Regularizations,

Asymmetries. (Studies in Germanic Linguistics) London/New York: Routledge.

*2002b *I amn’t sure – Why is there no negative contracted form of first person singular be? In Anglistentag 2001 Vienna: Proceedings,Dieter Kastovsky, Gunther Kaltenböck, and Susanne Reichl (eds.), 7–17. Trier: Wissenschaftlicher Verlag.

*2003 Non-standard English and typological principles: The case of negation. In Determinants of Linguistic Variation, Günter Rohden-burg, and Britta Mondorf (eds.), 507–529. Berlin/New York: Mou-ton de Gruyter.

*2004 Local markedness as a heuristic tool in dialectology: The case of amn’t. In Dialectology Meets Typology, Bernd Kortmann (ed.), 47–67. Berlin/New York: Mouton de Gruyter.

*2005 Unexpected regional distributions: Multiple negation in FRED. In Aspects of Negation, Yoko Iyeiri (ed.). Tokyo: Yushodo Press.

*in progress Naturalness and dialect grammar: Evidence from non-standard past tense paradigms. Post-doctoral dissertation (Habilitationsschrift). English Department, University of Freiburg.

Anderwald, Lieselotte, and Bernd Kortmann *2002 Typology and dialectology: A programmatic sketch. In Present-day

Dialectology. Problems and Findings, Jaap van Marle, and Jan Berns (eds.), 159–171. Berlin/New York: Mouton de Gruyter.

Anderwald, Lieselotte, and Susanne Wagner *forthc. The Freiburg English Dialect Corpus (FRED) – Applying Corpus-

Linguistic Research Tools to the Analysis of Dialect Data. In Using Unconventional Digital Language Corpora. Vol. I: Synchronic Corpora, Joan Beal, Karen Corrigan and Herman Moisl (eds.). Basingstoke: Palgrave Macmillan.

Barbiers, Sjef, Leonie Cornips, and Susanne van der Kleij (eds.) 2002 Syntactic Microvariation. Amsterdam: SAND.

(http://www.meertens.knaw.nl/projecten/sand/ synmic) Black, James R., and Virginia Motapanyane (eds.) 1996 Microparametric Syntax and Dialect Variation. Amsterdam/Phila-

delphia: Benjamins.

Page 28: A Comparative Grammar of Brtiish English Dialects

18 Bernd Kortmann and Susanne Wagner

Bremann, Rolf 1984 Soziolinguistische Untersuchung zum Englisch von Cornwall.

Frankfurt: Lang. Bresnan, Joan, and Ashwini Deo 2001 Grammatical constraints on variation: ‘Be’ in the Survey of English

Dialects and (Stochastic) Optimality Theory. Unpublished manu-script. http://www-lfg.stanford.edu/bresnan/download.html

Chambers, Jack K. 2004 Dynamic typology and vernacular universals. In Dialectology Meets

Typology, Bernd Kortmann (ed.), 127–145. Berlin/New York: Mou-ton de Gruyter.

Chambers, Jack K., and Peter Trudgill 1998² Dialectology. Cambridge: Cambridge University Press. Feagin, Crawford 2002 Entering the community: Fieldwork. In The Handbook of Language

Variation and Change, Jack K. Chambers, Peter Trudgill, and Nata-lie Schilling-Estes (eds.), 20–39. Oxford/New York: Blackwell.

Fischer, Andreas 1976 Dialects in the South-West of England: A Lexical Investigation.

Bern: Francke. Hernández, Nuria *in progress Pronouns in dialects of English: A corpus-based study of non-

standard phenomena. Ph.D. thesis. English Department, University of Freiburg.

Hickey, Raymond 2003 A Source Book for Irish English. Amsterdam/Philadelphia: Benja-

mins. Hudson, Richard 1999 Subject-verb agreement in English. English Language and

Linguistics 3: 173–207. 2000 *I amn’t. Language 76: 297–323. Kastovsky, Dieter, Gunther Kaltenböck, and Susanne Reichl (eds.) 2002 Anglistentag 2001 Vienna: Proceedings. Trier: Wissenschaftlicher

Verlag.Klemola, Juhani 2002 Continuity and change in dialect morphosyntax. In Anglistentag

2001 Vienna: Proceedings, Dieter Kastovsky, Gunther Kaltenböck, and Susanne Reichl (eds.), 47–56. Trier: Wissenschaftlicher Verlag.

Kolbe, Daniela *in progress Complementation patterns in English dialects. Ph.D. thesis. English

Department, University of Freiburg.

Page 29: A Comparative Grammar of Brtiish English Dialects

The Freiburg English Dialect Project and Corpus 19

Kortmann, Bernd *1999 Typology and dialectology. In Proceedings of the 16th International

Congress of Linguists, Paris 1997, Bernard Caron (ed.). CD-ROM. Amsterdam: Elsevier Science.

*2002 New prospects for the study of dialect syntax: Impetus from syntactic theory and language typology. In Syntactic Microvariation, Sjef Barbiers, Leonie Cornips, and Susanne van der Kleij (eds.), 185–213. Amsterdam: SAND.

*2003 Comparative English dialect grammar: A typological approach. In Fifty Years of English Studies in Spain (1952:2002). A Commemorative Volume, Ignacio M. Palacios Martinez, María José López Couso, Patricia Fra, and Elena Seoane (eds.), 65–83.Santiago de Compostela: University of Santiago.

Kortmann, Bernd (ed.) 2004 Dialectology Meets Typology. Dialect Grammar from a Cross-

Linguistic Perspective. Berlin/New York: Mouton de Gruyter. Mair, Christian 2002 Three changing patterns of verb complementation in Late Modern

English: A real-time study based on matching text corpora. English Language and Linguistics 6: 105–131.

Mair, Christian, Marianne Hundt, Geoffrey Leech, and Nicholas Smith 2002 Short-term diachronic shifts in part-of-speech frequencies: a

comparison of the tagged LOB and F-LOB corpora. International Journal of Corpus Linguistics 7: 245–264.

Montgomery, Michael 1994 The evolution of verb concord in Scots. In Studies in Scots and

Gaelic: Proceedings of the Third International Conference on the Languages of Scotland, Alexander Fenton, and Donald A. Mac-Donald (eds.), 81–95. Edinburgh: Canongate Academic.

Orton, Harold 1962 Survey of English Dialects – Introduction. Leeds: Arnold. Orton, Harold, Michael V. Barry, Eugen Dieth, Wilfrid J. Halliday, Philip M.

Tilling, and Martyn F. Wakelin (eds.) 1962–71 Survey of English Dialects. Leeds: Arnold. Preston, Dennis R. 1985 The Li’l Abner syndrome: written representations of speech.

American Speech 60: 328–336. 2000 ‘Mowr and mowr bayud spellin’: confessions of a sociolinguist.

Journal of Sociolinguistics 4: 614–621. Siemund, Peter 2002 Animate pronouns for inanimate objects: pronominal gender in

English regional varieties. In Anglistentag 2001 Vienna: Procee-

Page 30: A Comparative Grammar of Brtiish English Dialects

20 Bernd Kortmann and Susanne Wagner

dings, Dieter Kastovsky, Gunther Kaltenböck, and Susanne Reichl (eds.), 19–34. Trier: Wissenschaftlicher Verlag.

Szmrecsanyi, Benedikt *2004 Persistence phenomena in the grammar of spoken English. Ph.D.

thesis. English Department, University of Freiburg. *2005 Creatures of habit: A corpus-linguistic analysis of persistence in

spoken English. Corpus Linguistics and Linguistic Theory 1: 113–149.

Tristram, Hildegard L. C. 1997 DO-periphrasis in contact? In Language in Time and Space,

Heinrich Ramisch, and Kenneth Wynne (eds.), 401–417. Stuttgart: Steiner.

Trudgill, Peter 1999² The Dialects of England. Oxford: Blackwell. Wakelin, Martyn F. 1975 Language and History in Cornwall. Leicester: Leicester University

Press.Weltens, Bert 1983 Non-standard periphrastic do in the dialects of South West Britain.

Love and Language 3: 56–74. Williams, Ann, and Paul Kerswill 1999 Dialect levelling: Change and continuity in Milton Keynes, Reading

and Hull. In Urban Voices: Accent Studies in the British Isles, Paul Foulkes, and Gerald Docherty (eds.), 141–162. London: Edward Arnold.

Page 31: A Comparative Grammar of Brtiish English Dialects

Relative clauses in English dialects of the British Isles

Tanja Herrmann

Abstract

This cross-dialectal analysis of adnominal relative clauses relates results from a variety of English dialects to typological hierarchies, particularly the Noun Phrase Accessibility Hierarchy (AH; Keenan and Comrie 1977). The AH – with a few reservations – is verified for all relative clause formation strategies found in the data, including the zero relative marker strategy. From a diachronic perspective, the AH also helps to reveal the pattern underlying the way individual relative markers enter or exit an existent relative marker system. The pronoun retention strategy, however, does not necessarily follow a reversed order of syntactic positions on the AH, as claimed by Keenan and Comrie. At least resumptive pronouns in nonrestrictive relative clauses, that type of relative clause in which the majority of resumptive pronouns in the dialect data are found, thwart AH-based expectations due to their prevalence of the subject position. Moreover, due to their explicating function, resumptives give rise to further embedded relative clauses and other nonstandard relative constructions (nonreduction; ‘connector’ which).

1. Introduction

This study is concerned with prototypical relative clauses, i.e. adnominal relative clauses. A relative clause is a subordinate clause that modifies an antecedent with which a relative marker in the relative clause is coreferential. An adnominal relative clause forms part of a constituent of the matrix clause, as in the examples in (1) below. In these examples, as throughout this chapter, the following notational conventions are used: The relative clause is put in square brackets and the antecedent is marked in boldface. The zero relative marker is indicated by the symbol . The initial three capital letters indicate the region from which the dialect data originate: CMI = Central Midlands, CNO = Central North, CSW = Central Southwest, EAN = East Anglia, NIR = Northern Ireland, SCO = Scotland (cf. Kortmann and Wagner, this volume: Trudgill’s [1999] 2003 Map 18: ‘Modern Dialect areas’). This information on the source region is followed

Page 32: A Comparative Grammar of Brtiish English Dialects

22 Tanja Herrmann

by the text code, the speaker’s identification code (where available), and (for the BNC texts) the sentence number (see section 2):

(1) a. ... they knew just where to stop and start, especially the last pony [ I had]. (EAN-K65<S: 0806>)

b. ... this Billy [that used to go round all the district] and, and [buy up all these old cast horses] and [bring them up there] .... (SCO-GYS<Person: PSSCO6><S: 136>)

c. ... they used to perhaps have competitions for the childrens[what used to want to go on]. (CMI-FYD<S: 303>)

Adnominal relative clauses are a central syntactic phenomenon in every dialect, taking however different forms in different dialects. Although adnominal relative clauses have been a fair center of interest in theoretical linguistics, including some works on individual dialects, a cross-dialectal study within a typological framework has never been undertaken. This framework apart, all formation strategies of adnominal relative clauses investigated in this chapter will be subjected to a cross-dialectal analysis with the aim of identifying salient properties that individual dialects have in common, and those properties in which they differ from one another and from the standard variety.

In a comparative study, it can be determined which standard features have made inroads into traditional dialects, and to what extent (e.g. whether there is a predominance of wh-pronouns). On the other hand, dialects also converge toward one another in a process of dialect levelling. General nonstandard features of informal speech have developed or are developing from traditional dialect features (e.g. the nonstandard relative marker what)and may in turn affect the future shape of Standard English. A comparative view can identify these supra-regional features of informal speech, thus allowing some prognosis as to their future development and whether they might find entry into Standard English.

2. Data

The material used for the present investigation constitutes largely a subcorpus of the Freiburg English Dialect Corpus (FRED; cf. Kortmann and Wagner, this volume). The data are taken from six areas (East Anglia,

Page 33: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 23

the Central Southwest, the Central Midlands, the Central North, Scotland, and Northern Ireland) and from several sources:

The East Anglian, the Central Midland, and the Scottish data are dialectal texts extracted from the British National Corpus (BNC),published in 1995. They were reformatted into easy-to-read texts, while selective tags were kept. The texts result from oral history projects and are based on recorded interviews with elderly, working-class people speaking naturally. The East Anglian data were recorded in Suffolk and Eastern Cambridgeshire, the Central Midlands data in Nottinghamshire, and the Scotland data in Midlothian, Ayrshire, Selkirkshire, Lanarkshire, and Invernesshire. A part of the Central Southwest data is Juhani Klemola’s Somerset Rural Life Museum data (texts SRLM 105, SRLM 107, SRLM 108, SRLM 109, SRLM 122, SRLM 123; SRLM 132).1The other part of the Central Southwest data and the Central North data are taken from the FRED Corpus. All Central Southwest data originate from Eastern Somerset. The Cumbrian data were recorded in the pre-1974 counties Westmorland, Cumberland, and Lancashire, in particular in the Ambleside area, formerly Westmorland. The Northern Ireland data are part of John Kirk’s Northern Ireland Transcribed Corpus of Speech (NITCS), in a re-edited form.2 Only texts with elderly informants (age group 3) were considered. The Northern Ireland data were recorded in Antrim, Londonderry, Tyrone, Fermanagh, Armagh, and Down.

The dialect corpus analyzed in this chapter totals approximately 480,000 words, divided into some 80,000 words for each of the six ‘dialect’ regions. The word ‘dialect’ here should be taken with a grain of salt, as it is rather a convenient means to partition the map than any claim to the existence and location of firm dialect boundaries. The reason why these regions were chosen was that they cover the British map areally, that means, they are areas which are sufficiently wide apart geographically and have an identity of their own. For the sake of convenience, the regional speech used in these areas is referred to as ‘dialect’. Each region is named after a larger geographical area, largely following Map 18 ‘Modern Dialect areas’ published in Trudgill (2003).

Page 34: A Comparative Grammar of Brtiish English Dialects

24 Tanja Herrmann

The six regional corpora are not equally dialectal. As a whole, the Central Southwest, the Central North, and the Northern Ireland data represent broader dialectal speech than do the data from the East Anglian, Central Midland, and Scottish subcorpora. In fact, the Central Southwest and the Northern Ireland corpora exclusively consist of data by broad speakers, while the composition of speakers of the other four corpora is more heterogeneous.

3. Overall distribution of relative clauses and relative markers

Table 1 presents the overall frequencies of adnominal relative markers in the six investigated regions (Central Southwest, East Anglia, Central Midlands, Central North, Scotland; Northern Ireland). Absolute numbers are typed in boldface; percentages are given in square brackets. A look at the totals of relative markers across the six regions reveals that the total of occurrences (1874) with relative particles (zero, that, what; as) outnumbers the total of occurrences (638) with relative pronouns (who, which, whom;whose) by almost 3:1. The most frequent relative marker in the corpus is that (39%), followed by zero (28.1%), which (15.1%), who (10.1%), nonstandard what (6.8%), and as (0.8%). The case-marked wh-pronounswhom (0.2%) and whose (0.1%) are very unusual in dialectal speech.

The regional differences in relative marker distribution are indicated by the order and the font size in Map 1, which is based on Trudgill’s (2003) Map 18 ‘Modern Dialect areas’, supplemented by a map of Ireland and Northern Scotland. Leaving wh-pronouns as a standard and supra-regional feature aside for the moment, Figure 1 presents the percentages of relative markers when moving from north to south.

Page 35: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 25

Table 1. Areal distribution of relative markers

CSW EAN CMI CNO SCO NIR TOTAL (EasternSomer-

set)

(Suffolk;EasternCam-

bridge-shire)

(Nott-ingham-

shire)

(Cum-bria:

Cumber-land, West-

morland, NorthernLanca-shire)

(Lothian, Borders, Strath-clyde,Inver-

nesshire)

N [%]3 N [%] N [%] N [%] N [%] N [%] N [%]

zero 84[28.9]

86[20.4]

80[17.7]

142[34]

123[23.6]

191[46.9]

706[28.1]

that 77[26.5]

93[22]

182[40.3]

182[43.5]

241[46.2]

204[50.1]

979[39]

what 65[22.3]

67[15.9]

26[5.8]

10[2.4]

2[0.4]

- 170[6.8]

as - - 11[2.4]

6[1.4]

- 2[0.5]

19[0.8]

who 26[8.9]

65[15.4]

57[12.6]

30[7.2]

70[13.4]

5[1.2]

253[10.1]

which 39[13.4]

111[26.3]

91[20.1]

48[11.5]

84[16.1]

5[1.2]

378[15.1]

whom - - 2[0.4]

- 2[0.4]

- 4[0.2]

whose - - 3[0.7]

- - - 3[0.1]

total 291

[100] 422

[100] 452

[100] 418

[100] 522

[100] 407

[100] 2512[100]

Page 36: A Comparative Grammar of Brtiish English Dialects

26 Tanja Herrmann

Map 1. Areal distribution of relative markers

that whatwhich who

0 50 100

that what

which who

that whatas which who whose whom

thatwhat as which who

that whatwhich who whom

thataswhich who

Page 37: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 27

zero that what as North Northern Ireland: 46.9% 50.1% - 0.5%

Scotland: 23.6% 46.2% 0.4% - Central North: 34% 43.5% 2.4% 1.4% Central Midlands: 17.7% 40.3% 5.8% 2.4%

East Anglia: 20.4% 22% 15.9% -

South Central Southwest: 28.9% 26.5% 22.3% -

Figure 1. Distribution of relative markers along the North-South axis in percentages

The Central Southwest data show rather even frequencies of zero (28.9%), that (26.5%), and what (22.3%). East Anglia presents a similar picture: that(22%), zero (20.4%), and what (15.9%) are quite evenly distributed. As we move northward, that steadily gains strength. In the Central Midlands, that(40.3%) is the predominant relative marker, at the expense of zero (17.7%) and what (5.8%), which both have less importance (in absolute numbers and percentages) when compared to the southern areas. What is more than twice as strong as the relative particle as (2.4%), which is not found in the south and is, percentage-wise, weak in the Central Midlands. In the Central North, that accounts for 43.5%, although zero (34%) is also prominent. What (2.4%) and as (1.4%) have about halved their percentages in comparison to the Central Midlands. Scotland is even more clearly dominated by that (46.2%). Zero (23.6%) is a much weaker second; what(0.4%) is almost nonexistent (one clear case in Glasgow and one dubious instance), and as is absent. Finally, in Northern Ireland, that (50.1%) is used in about half of all instances. The other half is almost taken up by zero (46.9%). While what is unknown, as (0.5%) is hovering around half a per cent.

With particular regard to the dialectal variants what and as, what is by far the stronger one, the more so the farther south we go. In the south (East Anglia; Central Southwest), what has a substantial number of instances, whereas in the north (Central North; Scotland), it plays a marginal role; in Northern Ireland what plays no role at all.

As has its stronghold in the Central Midlands. It is used, though not often, in the Central North and in one county (Tyrone) of Northern Ireland.

Page 38: A Comparative Grammar of Brtiish English Dialects

28 Tanja Herrmann

According to the present data, what has its strongest position in the Central Southwest.

All existing wh-pronouns have found their way into the investigated dialects. However, depending on the dialectal broadness of the individual subcorpus, their proportion varies from 2.4% in the very broad Northern Ireland subcorpus to 41.7% in the East Anglian subcorpus, which, as a whole, is closest to the standard variety. In other words, the composition of speakers and the dialectal quality of the data of the corpora are reflected in the frequency of wh-pronouns. In between, there are the broad subcorpora from the Central North (18.7% wh-pronouns) and the Central Southwest (22.3% wh-pronouns) and the less broad subcorpora from Scotland (29.9% wh-pronouns) and the Central Midlands (33.8% wh-pronouns). In addition to the overall percentage of wh-pronouns, the presence or absence of case-marked wh-forms is also indicative of how standardized or how dialectal (i.e. broad) a subcorpus is: whom (two instances in Scotland; two in the Central Midlands) and whose (three in the Central Midlands) only appear in two of the three less broad subcorpora. Thus, the frequency of wh-pronounsserves as a yardstick for the degree of standardization or traditionality of speech, respectively. While traditional dialect only comprises relative particles, (written) Standard English abounds in wh-pronouns, which are indeed a typical trait of Standard English. Thus, the more wh-pronouns,particularly of the case-marked variant, a corpus of dialect data contains, the closer this corpus is to the standard variety and the further away from traditional dialect.

4. Previous investigations of relative markers

4.1. Supra-regional studies and/or national surveys

Before giving a summary of earlier works on the six dialect regions in section 4.2, it is in this section that more detailed information will be given on the three supra-regional and national surveys conducted for the dialects of England in the late 19th and, roughly, first half of the 20th century.

Page 39: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 29

4.1.1. Wright’s English Dialect Grammar

In Joseph Wright’s English Dialect Grammar ([1905] 1961), the relative marker what is said to occur “in some of the north-midland counties and in nearly all the counties south of the north midlands” (Wright 1961: 77). Asis “occasionally used” (Wright 1961: 77) in Westmorland and generally used in Nottinghamshire, East Anglia, and East Somerset, while at is generally used in Scotland and Ireland. The zero relative marker is a recurrent phenomenon in dialect – also in subject position – whereas whomis never employed (cf. Wright 1961: 77).

Map 2. Map 207 31.1: (a man) that’s poor (Lowman Survey)

4.1.2. Lowman Survey

The Lowman Survey of Middle and South England was carried out in 1937/38 and supplemented by data from Henry E. Collins for the Southeast in 1950. For the Central Southwest, Map 207 and 208, which are published

Page 40: A Comparative Grammar of Brtiish English Dialects

30 Tanja Herrmann

in Viereck’s (1975) atlas, yield the following results: Map 207, featuring question 31.1: a man) that’s poor (i.e. linguistic environment: restrictive relative clause; subject position; personal, indefinite antecedent), displays as in all counties except for Dorset, where the relative marker is that. Map 208 (cf. next page) gives a similar picture. It reproduces question 31.2: he’sa boy) whose father, which asks for a genitive relative marker. It turns out that the periphrastic genitive as his father is used in all counties except for Dorset, where a paratactic continuation4 (his father) is used.

Map 3. Map 208 31.2: (he’s a boy) whose father (Lowman Survey)

In East Anglia the survey investigated four localities in Suffolk, three in Cambridgeshire, (and three localities in Norfolk and three in Essex). Map 207 shows nearly an even distribution of what (two in Suffolk; two in Norfolk; one in Northern Essex) and that (two in Suffolk; two in Cambridgeshire; one in Norfolk; two in Southern Essex) (cf. also Viereck 1980: 27). Map 208 yields similar findings: Periphrastic what his father(two in Suffolk) and that his father (one in Western Cambridgeshire; two in

Page 41: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 31

Norfolk), or its reduced variant that’s father (one in Suffolk), occur about equally often. A paratactic continuation5 (his father) was chosen as an alternative in all four counties (one in Suffolk; one in Eastern Cambridgeshire; one in Norfolk; three in Essex). This suggests that East Anglia, particularly Western Suffolk, is the heartland of relative what.6That was the major alternative, while as had moved further to the west.

4.1.3. Survey of English Dialects

The Survey of English Dialects (SED) was mainly conducted in the 1950s with non-mobile, old, rural males and (marginally) females in 313 localities in England.7 The informants were given a questionnaire which contained three questions asking for relative marker usage: Question III.3.7: ‘If I didn’t know what a cowman is, you would tell me: He is the man ... looks after the cows.’ (i.e. linguistic environment: restrictive relative clause; subject position; personal, nonspecific, definite antecedent; ‘that-frame’), Question IX.9.5: ‘The woman next door says: The work in this garden is getting me down. You say: Well, get some help in. I know a man ... will do it for you.’ (i.e. linguistic environment: restrictive relative clause; subject position; personal, specific, indefinite antecedent; ‘who-frame’), and Question IX.9.6: ‘That man’s uncle was drowned last week. In other words, you might say, that’s the chap ... (uncle was drowned).’ (i.e. linguistic environment: restrictive relative clause; genitive position; personal, specific, definite antecedent).8 Additional, non-elicited information on the questions was noted down as incidental material. Responses to Questions IX.9.5 and IX.9.6 are presented cartographically as S5 and M81 in Thelinguistic atlas of England (LAE), while Viereck’s The computer developed linguistic atlas of England 1 (1991) presents the responses (and incidental material) to all three SED questions as S8a and S8b, S9, and S10. All six maps are reproduced in Appendix 2.

In the Central Southwest, the SED investigated thirteen localities in Somerset, seven of which in Eastern Somerset, eight in Wiltshire, five in Dorset, four in West Berkshire, six in Oxfordshire, one in South Gloucestershire, and three in Western Hampshire.

In response to Question III.3.7, Dorset indeed seems to be an outsider in the Central Southwest in being a that and zero area, although what and asappear once in the incidental material. In the lower core counties of the Central Southwest, namely in Somerset, Wiltshire, and West Berkshire,

Page 42: A Comparative Grammar of Brtiish English Dialects

32 Tanja Herrmann

what, that (and its phonemic variant ‘at), and zero are about equally frequent. Who and as, on the other hand, are rather infrequent. When including the northern counties Oxfordshire and South Gloucestershire, ascatches up with what, that, and zero. Western Hampshire, on the periphery of the Central Southwest, already leans toward the Southeast in displaying who and zero. In the incidental material, however, what (above all in Eastern Somerset) and particularly as dominate. While both what and as do not transcend the Eastern Somerset/Western Somerset borderline, as has its stronghold in the more interior counties Wiltshire, (South) Gloucestershire, (West) Berkshire, and Oxfordshire to the north.

In response to Question IX.9.5, who is the predominant relative marker. As gains strength again, as one moves northward and further into the mainland, i.e. in (West) Berkshire, (South) Gloucestershire, and Oxfordshire. The standard genitive relative marker whose prevails in the entire Central Southwest, although in Eastern Somerset and Oxfordshire potential genitives are also promoted to other (‘higher’) syntactic positions, such as indirect object (dative) and subject (nominative) (cf. section 8 and 8.4.). Altogether, the wh-pronouns who and whose are prevalent (in Questions IX.9.5 and 9.6). What, that, zero, and as are all documented, yet as seems to be rather restricted.

In East Anglia, the SED data have been collected in five localities in Suffolk, one in Eastern Cambridgeshire, thirteen in Norfolk, as well as fifteen localities in neighboring Essex.

In response to SED Question III.3.7, the relative marker as dominates in Cambridgeshire and in the area to the west, whereas Norfolk, Suffolk, Eastern Cambridgeshire, and Essex are part of a what area which comprises the entire Southeast (cf. Viereck 1991: S8b). However, what seems to alternate with that in this area; the latter is not as frequent as what though (cf. Viereck 1991: S8a and S8b).

In response to Question IX.9.5, that occurs only once as a second choice in Norfolk. In all probability, at is no separate relative marker in East Anglia but a weakened form of that, whose initial th- was dropped, which is an occasional phonological feature there (cf. also Poussa 1996: 529; Peitsara 2002: 169). When looking at Map S9 in Appendix 2, the predominance of who in East Anglia (and surroundings, as well as the entire South) becomes evident. That is conspicuously rare in all England (although at is frequent in the north). As is absent in East Anglia, but starts to be very frequent in Bedfordshire, Northamptonshire, Buckinghamshire, and farther to the west (cf. Viereck 1991: S9). The synthetic genitive

Page 43: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 33

relative marker whose outnumbers the periphrastic genitive constructions by far.

In sum: While the dialectal relative markers what and as and the relative particles that and zero are preferred in SED Question III.3.7 ‘He is the man ... looks after the cows’ (subject position; nonspecific, definite antecedent), the standard wh-relative pronouns who and whose are preferred in IX.9.5 ‘Iknow a man ... will do it for you’ (subject position; specific, indefinite antecedent) and IX.9.6 ‘that’s the chap ... (uncle was drowned)’ (genitive case; specific, definite antecedent), respectively. Of the 46 instances of what accompanying Question III.3.7 in the incidental material, about twice as many instances of what were found in nonspecific (and definite) environments than in specific environments. Of the four instances of that,all seem to occur in nonspecific (and definite) environments. The eight instances of as do not seem to be affected by syntactic environment. That is to say that ‘(non)specificity’ seems to be the critical syntactic variable (cf. Poussa 1988: 447–448 and 465), in the sense that specific antecedents prefer who (such as in SED Questions IX.9.5 and IX.9.6), while nonspecific antecedents are left for the nonstandard relative particle what (instead of the standard relative particle that in the ‘that-frame’ Question III.3.7).9

In the Central Midlands, the SED investigated four localities in Nottinghamshire, one in Eastern Derbyshire, and four in North-Western Leicestershire. As occurs in the great majority of responses to all three SEDquestions, including the incidental material. The Central Midlands are part of a large as area, which extends over the entire Midlands and reaches up to Lancashire in the north, roughly Eastern Somerset in the south, and Cambridgeshire/Buckinghamshire in the east, excluding the Southeast (cf. Viereck 1991: S8a, S8b, S9, and S10). There are also sporadic instances of zero, that, what, and who, while whose is regularly used in genitive position.

In the Central North, the SED investigated six localities in Cumberland, four in Westmorland, and two in North Lancashire. The Central North belongs to an at area in the north of England, although the more dominant relative marker in North Lancashire appears to be as. (In the present study, at, ut, and t are considered as phonemic/phonetic variants of that.) There is a single occurrence of as in central Cumberland (i.e. in Cu4 = Threlkeld), and the single occurrence of what in the Central North occurs no farther north than North Lancashire (cf. Orton and Halliday 1962–1963: 243 and 1082–1085; cf. also Viereck 1991: S8a, S8b, S9, and S10).

Page 44: A Comparative Grammar of Brtiish English Dialects

34 Tanja Herrmann

4.2. Areal distribution of relative markers in previous investigations

With respect to the six dialect regions and, especially, the individual counties investigated here, the major findings of previous studies are summarized in Tables 15–20 in Appendix 1. Each table provides an overview of relative marker distribution in the respective region in chronological order. Below, only the main similarities and differences to the areal distribution of relative markers in the present data will be pointed out.

4.2.1. The Central Southwest – Eastern Somerset

When comparing the results of the present study on the Central Southwest to those of previous investigations, it becomes clear that in the past, asreached as far south as Gloucestershire, Berkshire, Wiltshire, and Eastern Somerset. In Western Somerset as did not occur, in Dorset just once in the incidental material of Question III.3.7 in combination with ‘all’ (cf. Orton and Wakelin 1967: 291). However, it has to be conceded that asoccasionally reappears in the Lower Southwest (Devon; Cornwall) in SEDmaterial in response to Question III.3.7 and in the incidental SED material (cf. Orton and Wakelin 1967: 291). According to the present data, asnowadays seems to have receded further north in relative clause formation. Apart from that, Ihalainen’s (1980) distribution of relative markers for Somerset is very similar to the present distribution. In contradistinction to that, van den Eynden’s (1992, 1993, 2002) frequencies on Dorset depart from it: While the percentage of the zero marker is still relatively similar to the percentage of the distribution found here, the percentage of that is remarkably higher, and that of what very much lower, attesting that Dorset is a distinct ‘dialect area’.

The nonstandard relative marker what has been part of the Central Southwestern dialect from the earliest citation onward, i.e. it dates back there at least to the middle of the second half of the nineteenth century, when Elworthy (1877) mentioned it for West Somerset. Some authors (Barth 1968 for Gloucestershire and SED Informant So[merset]1) hint at an antagonism between what and as, in which as is felt to be the older variant of the two, which was already superseded by the younger combatant at the time of the SED. Although who is the dominant relative marker in the SEDquestionnaire in Eastern Somerset, what is very frequent in the incidental material.

Page 45: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 35

4.2.2. East Anglia – Suffolk and Eastern Cambridgeshire

While as is described as ‘in general use’ for East Anglia in Wright’s English Dialect Grammar (cf. Wright 1961: 77), its occurrences became very sporadic after that time. In the Lowman Survey, as is not recorded for East Anglia any more, while the SED still records as as the dominant relative marker for Cambridgeshire in response to Question III.3.7 and in the incidental material of that question (cf. Orton and Tilling 1969: 301). Later on, as apparently retracted even further west than Cambridgeshire, where it is found only once by Ojanen (1982: 74) in Mid Cambridgeshire. In the present East Anglian data from Suffolk and Eastern Cambridgeshire, no instance of as was encountered.

On the other hand, what is almost unanimously recorded as the dominant marker in East Anglia, which can be regarded as the heartland of what. Even in Cambridgeshire, what is reported as the most frequent relative marker by Ojanen. Not surprisingly, the present data – mainly from Suffolk – are mostly in line with Kekäläinen/Peitsara’s (1985, 2002) findings for Suffolk: While Ojanen’s higher number of zeros may be partly attributable to the inclusion of non-relative clause constructions, the percentages for that are almost the same (ranging between 20.4% and 23%). Kekäläinen/Peitsara’s higher figures for what (29.1% and 21.2% versus 15.9% in the present data) are to be explained by the difference in broadness between their corpus and the one used here. The fact that the present corpus comprises less broad dialect speech is also reflected by the much higher figures for who (15.4% in the present data versus 2.9% and 7.4%) and, to a lesser extent, which (26.3% in the present data versus 21.4% and 16.9%), as well as by the total absence of archaic as.

Ojanen’s findings for (South) Cambridgeshire and Poussa’s (1994, 2001) and Francis’ (1999, 2001) for Norfolk are different from the present findings in two major respects: Their frequencies of the zero marker are considerably higher, although the high number of zero subject relative clauses should be taken with caution in Ojanen’s study. It contains instances which share a common surface structure with zero relative clauses yet are no relative clauses, such as nonfinite clauses, resultatives, and pro-drop constructions.10 In contrast to that, their figures for that are very low (1.6% for Cambridgeshire; 7.9% in Poussa’s and 3.7% Francis’ corpus for Norfolk versus 22% in the present data).

Page 46: A Comparative Grammar of Brtiish English Dialects

36 Tanja Herrmann

4.2.3. The Central Midlands – Nottinghamshire

All previous studies on the Central Midlands and even its neighboring counties report as either as the dominant or the only relative marker. Zero and what were also known. In the present data, by contrast, the relative marker as plays only a very minor role (2.4%). Partially this has to be seen as a consequence of the less broad dialect spoken in the present corpus. On the other hand, this is to be interpreted as a real time change. In the data investigated here, as has been driven back by other relative markers, particularly by that (40.3%), the wh-pronouns (32.7%), and zero (17.7%). The relative marker what is used, but not frequently (5.8%).

4.2.4. The Central North (Cumbria = Cumberland, Westmorland, and North Lancashire)

All previous authors agree on at as the regular relative marker in Cumbria, except for Wright ([1905] 1961), who is not very detailed on his recordings for the Central North. There is a disagreement among authors, however, whether at is a separate relative particle of Scandinavian origin or whether at – alongside ut or ‘t – is a phonemic variant of that with the initial th-being dropped. Although some scattered occurrences of at were found in the present Central North data, this dispute concerning its origin cannot be solved here. The suggestion offered on that point in Romaine (1982a), citing R. Girvan’s (1939) Ratis Raving and other early Scots poems on morals, sounds enticingly plausible: Even if at was a separate conjunction and later a relative particle in Northern England and Scotland in the past, it has become mentally merged with that over time, so that nowadays at no longer figures as a separate relative particle (cf. Romaine 1982a: 70).11

Interestingly, authors never argue about the current nature of at as a conjunction, but it is commonly assumed that the conjunction at is equivalent to the conjunction that, i.e. at = that. Since at and thatappear(ed) side by side, in the Central North and Scotland as well as in other areas of regular or occasional initial th-dropping (when unstressed), like East Anglia (cf. Poussa 1996: 529 and 531; cf. also Peitsara 2002: 169), the two were not formally distinguished, but the few transcribed instances of at were subsumed under that. Hence, that is the most frequent relative marker in the Central North data (43.5%), followed by zero (34%). Occasional or frequent occurrences of zero are also mentioned by all previous

Page 47: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 37

authors after 1905. While as seems to be a typical Lancashire feature (e.g. in the SED), it was also (occasionally) observed in Westmorland and Cumberland by Wright ([1905] 1961) and Reaney (1927) in the early twentieth century. In the present late twentieth century data, as only amounts to 1.4%, coming from the Ambleside area, formerly Westmorland. What is recorded as an occasional relative marker for West Cumberland by Brilioth (1913) and is mentioned also in the SED and by Wright (1979). What (2.4%) was found somewhat more frequently than as (1.4%).

4.2.5. Scotland – Lothian, Borders, Strathclyde, and Invernesshire

Like the present results, earlier findings document that/‘at (or its phonetic/phonemic variants ‘ut or ‘t) as the prevalent relative marker in Scotland (46.2% in the present data), followed by the zero relative marker (23.6% in the present data). Wherever a reduced form occurs, authors take it for a phonemically shortened that, with the exception of Dieth (1932: 153), who is uncertain on the matter, and Murison ([1977] 1980: 39). Grant (1931: 78) is the only author who cites two literary instances of asoccurring around 1900, which he notes as ‘rare’. Neither in the other studies, nor in the present investigation were there any instances of as,while what is very rare and seems to be restricted to Lower Scotland and Glasgow, in particular. Whilk, the Scottish equivalent to English whichcited by Murison (1980: 39), rather seems to pertain to an earlier stage of Scottish English.

4.2.6. Northern Ireland

Zero and that (or ‘at) dominate in (Northern) Ireland in previous studies, which conforms to the present findings (46.9% and 50.1%, respectively in the present data). The extreme conservativeness of the present Northern Ireland data can be read off the extremely low frequencies for who and which.

Page 48: A Comparative Grammar of Brtiish English Dialects

38 Tanja Herrmann

5. Restrictiveness/nonrestrictiveness

As is well known, adnominal relative clauses can be either restrictive or nonrestrictive modifiers of noun phrases. Restrictive relative clauses give determinative or essential information, typically narrowing down the range of possible referents of noun phrases (cf. Keenan and Comrie 1977: 63–64). Nonrestrictive relative clauses, by contrast, give supplementary, non-defining information about noun phrases (cf. Quirk et al. 1985: 1239; cf. also Huddleston 1984: 400). In contradistinction to restrictive relative clauses, nonrestrictive relative clauses are characteristically delimited by intonation breaks in speech and by commas (or dashes) in writing (cf. Quirk et al. 1985: 1258 and 366).

There is general agreement on the fact that, in the standard variety, nonrestrictive relative clauses are formed by means of relative pronouns of the wh-group. Nonrestrictive that may occur but is “very rare” (Quirk et al.1985: 1258), while nonrestrictive zero “cannot occur” (Quirk et al. 1985: 1258).

In the present study, the dichotomy between restrictiveness and nonrestrictiveness is determined on semantic grounds, since criteria based on prosody, punctuation, and syntax (whether prescriptive or described on the basis of written Standard English) are linguistic ideals that often fail to apply in informal speech (cf. Jacobsson 1994: 182–183 and 191; cf. also Newbrook 1997: 45–47; Biber et al. 1999: 602). Nonrestrictiveness itself is rather to be seen as a characteristic of (written) Standard English than of spoken dialect. Hence, the use (and frequency of use) of nonrestrictive relative clauses in a dialect or an idiolect is to be rated as an indicator of the relative move toward the standard variety of that dialect or idiolect. Table 2 gives an overview over the distribution of restrictive and nonrestrictive relative clauses across the six investigated regions.

Overall, restrictive relative clauses are four times as frequent as nonrestrictive relative clauses (80.1% vs. 19.9%) in dialectal speech. Generally speaking, restrictive relative clauses are predominantly formed with invariant relative particles, whereas nonrestrictive relative clauses are primarily constructed with wh-pronouns, especially with which. The nonstandard relative marker what falls out of line since it takes up a disproportionally high percentage (11.4%) of all nonrestrictives. As found in previous dialectal studies (e.g. Ihalainen 1980, Kekäläinen 1985, Miller 1993, Miller and Brown 1982, Peitsara 2002; van den Eynden 1992 and 1993), which is the a priori choice for (mainly nonpersonal; see section 6)

Page 49: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 39

nonrestrictives (61.9%), while personal antecedents of nonrestrictive relative clauses are primarily relativized by who (14.8%).

Table 2. Restrictiveness/nonrestrictiveness across regions

restrictive (r) nonrestrictive (nr) N % % r N % % nr

total

zero 687 34.2 97.3 19 3.8 2.7 706 that 942 46.8 96.2 37 7.4 3.8 979 what 113 5.6 66.5 57 11.4 33.5 170 as 19 1 100 0 0 0 19 who 179 9 70.8 74 14.8 29.2 253 which 68 3.4 18 310 61.9 82 378 whom 0 0 0 4 0.8 100 4 whose 3 0.2 100 0 0 0 3 total

% 2011 100.0

80.1501 100.0

19.92512

100.0

In the present data, the zero relative marker and that are not confined to restrictive relative clauses: 2.7% of all instances of zero and 3.8% of all instances of that appear in nonrestrictive relative clauses. Below, examples (2a–c) illustrate nonrestrictive zero relative clauses; examples (2d–f) illustrate nonrestrictive that relative clauses:

(2) a. ... there was Mr M. and B. W. from K. [ came round buying horses]. ... (CNO FRED Wes_015)

b. Yes she had her aunt [ was a widow there at the time [when she came to <gap cause=anonymization desc=address>]] and she just lived about a year. (SCO-K6N<Person: PSSCO28><S: 0114>)

c. ... because there’s the father [ was a cook], and, and my mother was a good cook, too, .... (NIR NITCS A19.3)

d. ... I seen E. A. [that lived there], he said it come one Sunday dinnertime, .... (CNO FRED Wes_014)

e. And my my aunt [that I] my grandmother [that I stayed with], their neighbour down the stair was quite indignant .... (SCO-K6L<Person: PSSCO24><S: 121>)

f. ... J. M., [that’s married to my niece there] {ahah}, his bog, we used to cut it. (NIR NITCS A51.3)

Page 50: A Comparative Grammar of Brtiish English Dialects

40 Tanja Herrmann

Nonrestrictive zero relative clauses are particularly recurrent in there- and have-existentials, which is the syntactic environment generally favored by zero relative clauses (compare 8.5.). Whereas nonrestrictive zero is found in four of the six regions, namely the Central Midlands, the Central North, Scotland, and Northern Ireland, nonrestrictive that occurs in all dialects. An older stage of English is preserved in the nonrestrictive use of that in dialectal speech: Nonrestrictive that was common in Middle English (cf. Mustanoja 1960: 190) and was still part of Early and Late Modern English (cf. Chevillet 1996: 26; cf. also Mustanoja 1960: 197). Nonrestrictive usages of the zero marker in Old and Middle English are arguable, since they may be interpreted as instances of parataxis (cf. Mustanoja 1960: 121 and 203–204).

Compared to the other relative particles, what shows a much stronger propensity toward nonrestrictive environments. Its distribution resembles that of the wh-pronoun who (70.8% restrictives vs. 29.2% nonrestrictives), and the percentage of nonrestrictive what (33.5%) exceeds the percentage of the overall nonrestrictives (19.9%).

Overall ratio of restrictives – nonrestrictives: 4 : 1 (80.1% 19.9%) vs. Ratio of what restrictives – what nonrestrictives: 2 : 1 (66.5% 33.5%)

Figure 2. Ratio of (non)restrictives versus ratio of what (non)restrictives

As will be demonstrated in section 6, what is not gender-marked and thus can relativize personal and nonpersonal antecedents alike (cf. Table 3 below). On closer examination, however, there are only nine instances (15.8%) of nonrestrictive personal what relative clauses in the data. In other words, what mainly competes with nonrestrictive which for the same syntactic environment (viz nonpersonal antecedents), since the vast majority (84.2%) of all nonrestrictive instances of what are nonpersonal. By and large, in those regional dialects that have what as a relative marker, broad idiolects employ nonrestrictive what, while idiolects that are closer to the standard opt for nonrestrictive which.

The few instances of the relative particle as (19 occurrences; a mere 0.8% of all relative markers) are all found in restrictive environments. This may be a result of the extremely low text frequency of as, i.e. as is numerically receding in all syntactic environments, including the confinement to exclusively restrictive environments (cf. also 8.3.2).

Page 51: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 41

Personal relative pronoun who prefers restrictive environments (70.8%), whereas its (mainly) nonpersonal counterpart which runs counter to all other relative markers by greatly favoring nonrestrictive environments (18% restrictives vs. 82% nonrestrictives). Finally, with regard to the highly marked use of whom and whose in our data (as few as four and three instances out of a total of 2512 relative clauses) it may be noted that whomseems to prefer nonrestrictive environments, while whose seems to follow the general trend in preferring restrictive environments.

6. Personality/nonpersonality

With regard to the personality/nonpersonality variable, the six regions show the following behavior:

Table 3. Personality/nonpersonality across regions

personal (+p) nonpersonal (-p) N % % +p N % % -p

total

zero 251 25.9 35.6 455 29.5 64.5 706 that 392 40.5 40.0 587 38.0 60.0 979

what 49 5.1 28.8 121 7.8 71.2 170 as 10 1.0 52.6 9 0.6 47.4 19

who 244 25.2 96.4 9 0.6 3.6 253 which 16 1.7 4.2 362 23.5 95.8 378 whom 4 0.4 100 - 0 0 4 whose 3 0.3 100 - 0 0 3

total %

96938.6

100.0 154361.4

100.0 2512 100.0

All six dialects conform to Standard English in restricting who and its case-marked forms whom and whose to personal antecedents. The nine cases of who relativizing nonpersonal antecedents belong to the sanctioned borderline cases, such as personalized animals and things (cf. Quirk et al.1985: 1245–1246 and 314–318): Three instances refer to domestic animals like cows and horses, as in (3), one to a car, and five to collective nouns.

(3) you had to have a cow the top side [who ’d make the other one stay her own side]. ... (CSW-SRLM 109<T 3320>)

Page 52: A Comparative Grammar of Brtiish English Dialects

42 Tanja Herrmann

However, the conventional gender contrast between which and who (and its case-marked forms) is overridden in dialectal English insofar as the gender concord constraint is lifted, or at least loosened, in the case of which: Apart from two instances denoting collective nouns and two referring to older (!) children, there are twelve instances of personal which outside the group of licensed borderline cases of personal which.12 Consider the examples in (4):

(4) a. ... And the boy [which I was at school with] G. O., .... (CNO FRED Wes_019)

b. Then, course then they used to <trunc> hav when the man stood in green, I knew a fella named <gapcause=anonymization desc=“last or full name”> [which we was talking about a little while ago] and .... (EAN-H5H<Person: PSEAN4><S: 501>)

c. ... And then there was C., [which caught his hand in the machinery up here] and he had his hand off, being severed (at the) wrist. ... (CSW FRED Som_009)

Personal which occurs in five of the six regions: in the Central Southwest, East Anglia, the Central Midlands, the Central North, and Scotland. (The data from Northern Ireland, which hardly has any wh-pronouns, only contains five instances of nonpersonal which.) In other words, the relative marker which is not confined to nonpersonal antecedents in dialects. In allowing nonpersonal and personal antecedents, which seems to have lost its status as a relative pronoun in dialect: Which relies on gender concord as the only kind of rigorously applied agreement between antecedent and relative marker for being considered a ‘relative pronoun’, as which shows neither case nor number agreement with the antecedent. Thus, in dialect which qualifies rather as a relative marker (‘relative particle’): It is open to any type of antecedent. In this respect, dialectal English has maintained a prior stage of English, since Middle English dialects permitted nonpersonal which (cf. Mossé [1952] 1991: 62; cf. also Mustanoja 1960: 195), while in the sixteenth century which started to be confined to nonpersonal referents (cf. Nevalainen and Raumolin-Brunberg 2002: 117 and 119). However, as illustrated in examples (5a,b), non-gender-marked relative marker whichcan still be governed by a preposition:

(5) a. ... and we had a meeting at headquarters, to which about forty or fifty people turned up, [of which two had probably blown a

Page 53: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 43

<pause> an instrument in the past], ... (EAN-K69<Person: PSEAN13><S: 130>)

b. But the yard one of the yard inspectors came to me and said, I wonder if you would make up a roster for the supervisors(UNCLEAR). (SCO-K6M<S: 304>) [Of which there were six, seven]. (<S: 305>) So I made a roster out for the (trunc) ro for the supervisors which meant that every week they got a rest day. (<S: 306>)

Being able to be governed by a preposition is a defining criterion of nouns and pronouns, in contrast to conjunctions/complementizers (relative particles), which cannot be governed by a preposition (e.g. *‘ofPREPOSITION

thatRELATIVE PARTICLE (CONJUNCTION)’). Hence, non-gender-marked relative marker which has to be attributed the status of relative pronoun.

The occurrence of nonpersonal which in combination with a partitive genitive appears to be triggered by two possible (and perhaps mutually reinforcing) factors: First, according to Christian Mair (personal communication), which is chosen in relativizing human antecedents to activate its ‘ascriptive’ property, i.e. when speakers want to express the ‘kind of person’, instead of the ‘identity’ (cf. Huddleston and Pullum 2002: 1048–1049; cf. also Quirk et al. 1985: 1246).13 For example:

(6) Remember that they have a house-keeper, [which we don’t have]. (Huddleston and Pullum 2002: 1049)

Mair (1998: 130–132) focuses on the ‘relaxation of a grammatical agreement rule’ concerning the who/which opposition, in analogy to the who/which distinction with collective nouns: “The use of which [in these cases] de-emphasises the status of members of a particular group as individuals and presents them as a collective” (1998: 130). This is illustrated in (7):

(7) It [the proposed law] is aimed at the external hacker [of which there are far fewer than press reports suggest] .... (Mair 1998: 130 [emphasis and bracketing mine])

Second, this combination of partitive genitive which with personal antecedents is rather due to an analogy mechanism: In interrogative clauses and nominal relative clauses, the relative pronoun which contrasts with the

Page 54: A Comparative Grammar of Brtiish English Dialects

44 Tanja Herrmann

relative pronouns who and what in denoting a (personal or nonpersonal) ‘limited set’, which is semantically close to the concept expressed by the partitive genitive of which. For example:

(8) a. indefinite interrogative pronoun who:Who is your favourite conductor?

b. definite interrogative pronoun which referring to a limited set: Which is your favourite conductor? (Von Karajan or Stokowsky?) (Quirk et al. 1985: 369)

c. indefinite nominal pronoun what:You can take [what you want].

d. definite nominal pronoun which referring to a limited set: You can take [which you want]. (X or Y)

By contrast, adnominal which is restricted to nonpersonal antecedents, even if it describes a definite, limited set of referents in a partitive genitive (cf. Quirk et al. 1985: 369–370; cf. also Huddleston 1984: 394). In this way, the semantic property ‘limited set’ is transferred from nominal relative or interrogative clauses to adnominal relative clauses by analogy, as in the examples in (5), repeated here as (5'a,b):

(5') a. ... and we had a meeting at headquarters, to which about forty or fifty people turned up, [of which two had probably blown a <pause> an instrument in the past], ... (EAN-K69<Person: PSEAN13><S: 130>)

b. But the yard one of the yard inspectors came to me and said, I wonder if you would make up a roster for the supervisors(UNCLEAR). (SCO-K6M<S: 304>) [Of which there were six, seven]. (<S: 305>) So I made a roster out for the (trunc) ro for the supervisors which meant that every week they got a rest day. (<S: 306>)

Where the dialectal relative marker as occurs (i.e. in the Central Midlands, the Central North, and Northern Ireland), it is not sensitive to the personality/nonpersonality variable, but combines with personal and nonpersonal antecedents almost in equal numbers. Below, example (9a) contains a personal antecedent, while example (9b) contains a nonpersonal antecedent:

Page 55: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 45

(9) a. ... I was on this er bottle washing stunt and (trunc) o one chap[as lived next door to us, back at er at <gap cause=anonymization desc=“last or full name”> Road] he got me his this job on the farm. (CMI-FYE<S: 166>)

b. ... there’s a lot of houses [ would be up that road] [as wasn’t there]. (NIR NITCS A41.3)

Unlike the nominal relative pronoun what in (10a), adnominal nonstandard what is not limited to nonpersonal antecedents in any region. About one third of all instances (28.8%) refer to a personal antecedent, as in (10b–d):

(10) a. nonpersonal relative pronoun what in nominal relative clause: preposition in is elided (cf. 10.1. below): I believe in wearing [what you feel comfortable [in]], you work better and ... (EAN-K65<S: 1059>)

b. ... And the lady [what was driving], she said, “Excuse me young man”, she said, “where ’s your dog? ” ... (CSW-SRLM 132<T 1280>)

c. See he was the man [what brought in decasualization during the war]. (EAN-H5H<Person: PSEAN4><S: 767>)

d. And er she used to sell corned beef and er pickled onions or anything like that for people [what was working round there] [what couldn’t get home for dinner]. (CMI-FYD<S: 201>)

7. Preposition placement

When a prepositional complement is relativized, the preposition can either be moved to initial position together with the wh-marker and govern the relative pronoun (preposition fronting or pied-piping), or the preposition is left behind in its normal clause position without its complement (preposition stranding). Both types are illustrated in turn:

(11) a. fronting: Now they were provided with a meal [for which the police, at that time paid sixpence], .... (EAN-K68<S: 030>)

b. stranding: ... our shovels [what they used to feed the boiler with] were all steel shovels. (EAN-H5G<S: 505>)

Page 56: A Comparative Grammar of Brtiish English Dialects

46 Tanja Herrmann

Table 4 presents the distribution of fronted versus stranded prepositions across regions.

Table 4. Preposition fronting/preposition stranding

fronted N stranded N tokens % tokens % total

zero 0 - 61 35.5 61 that 0 - 70 40.7 70 what 0 - 14 8.1 14 as 0 - 1 0.6 1 who 0 - 5 2.9 5 which 11 91.7 20 11.6 31 whom 1 8.3 1 0.6 2 whose - - - - - total

% 12

6.5100.0 172

93.5100.0 184

100.0

While pied-piping is typical of (written) Standard English, dialects vastly prefer preposition stranding (172 of 184 prepositions are stranded; 93.5%). To a large extent, this result is a corollary of the relative markers used: Only relative pronouns allow both preposition fronting (11a) and preposition stranding (11b). Relative particles demand preposition stranding, because in relative clauses neither conjunctions can be governed by prepositions (e.g. *‘of that’, *‘of what’; *‘of as’) nor gaps (e.g. *‘of

'’): recall the discussion concerning the word class status of relative markers in section 6. Accordingly, if adnominal what were found to be governed by a preposition, i.e. in an example of preposition fronting, this would attest to its status of a relative pronoun, parallel to its nominal counterpart what.

In all instances where the relative particles that, what, as and the zero marker figure as prepositional complements, the prepositions are invariably stranded (altogether 146 instances). Of the remaining 38 relative pronouns (i.e. the prepositional complements who, which, and whom taken together), twelve (31.6%) show preposition fronting while 26 (68.4%) show preposition stranding. In other words, even where preposition fronting is permitted, preposition stranding is preferred in dialectal speech. Regarding the relative pronoun who, the absence of fronted prepositions (and the

Page 57: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 47

presence of five stranded prepositions) is conspicuous. Apparently, the relevant speakers (two in East Anglia, one in the Central Midlands, and one in the Central North) neither wanted to apply case-marking (i.e. whom), nor did they want to have a non-case-marked relative pronoun governed by a preposition (e.g. with who). For example:

(12) a. Janet posted a letter for me last week to a friend [who I worked with at Ipswich], ... (EAN-HDK<S: 072>)

b. Well the agent, that be either <gap cause=anonymization desc=“last or full name”> or any agent who, [who the ship belonged to], ... (EAN-H5H<Person: PSEAN4><S: 372>)

There are two instances involving whom. In example (13a) the preposition for is fronted, whereas in example (13b) it is stranded:

(13) a. But anyway there was so many people and one chap who he he was, as a matter of fact, he was organizer with Communist Party [for whom I’ve got the very greatest respect], the very greatest respect. (CMI-FYH <S: 300>)

b. And er this bicycle well it would go out of fashion and was put in a, a loft in one of the, it must have changed hands from Mr<gap cause=anonymization desc=“last or full name”>[whom it was made for]. (SCO-GYW<S: 093>)

Out of the 31 relative clauses introduced by the relative pronoun which,eleven instances (35.5%) show preposition fronting, while the majority (64.5%) shows preposition stranding.

Of the eleven fronted prepositions governing which, six are partitive genitive ‘of’ (governing also personal which, as in example (14b) below). (Partitive) genitive ‘of’ should always be fronted, according to a prescriptive rule in (written) Standard English. Jespersen (1927: 188) phrases this rule as “there are certain cases in which it is unnatural to have of at the end, thus if it is the equivalent of a genitive” and “[p]artitive of is generally placed first” (Jespersen 1927: 188), while Huddleston and Pullum (2002: 1041) describe it as “[p]artitive of resists stranding”. For example:

(14) a. Most of the textbooks [ were handed to us] came from cupboards of storage, [of which they must have had about

Page 58: A Comparative Grammar of Brtiish English Dialects

48 Tanja Herrmann

sixty each], those classes were always, from then on, sixty boys in a class for one teacher. (CMI-FXU<S: 035>)

b. ... and we had a meeting at headquarters, to which about forty or fifty people turned up, [of which two had probably blown a <pause> an instrument in the past], ... (EAN-K69<Person: PSEAN13><S: 130>)

In dialectal speech, however, we do find examples of stranded of-genitives,in combination with which (15a) or even the zero relative marker (15b), which requires preposition stranding anyhow:

(15) a. resumptive personal pronoun them after stranded partitive genitive preposition of (preferred analysis): Now the first job he did was to get you well acquainted with the tools, [which we had quite a number of them]. (CMI-FY2<S: 003>)

b. Well you know they used, did you know there used to be atimber [ they used to make cog wheels of]? (CNO FRED Wes_004)

8. Accessibility Hierarchy

8.1. Preliminaries and overall Accessibility Hierarchy frequencies

The strong form of the ‘Noun Phrase Accessibility Hierarchy’ (in the following short ‘Accessibility Hierarchy’ or AH) and the ‘Hierarchy Constraints’ was put forward by Keenan and Comrie in 1977, stating universal principles in relative clause formation (cf. Keenan and Comrie 1977: 66–67):

Accessibility Hierarchy (AH): SUBJ > DO > IO > OBL > GEN > OCOMP

Spelled out as follows: SUBJECT > DIRECT OBJECT > INDIRECT OBJECT > OBLIQUE CASE > GENITIVE > OBJECT OF COMPARISON

Page 59: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 49

The Accessibility Hierarchy reflects the accessibility of various syntactic positions of noun phrases to relativization (in restrictive relative clauses with definite head noun phrases) in descending order from left to right. In psychological terms, the further a position is located to the left, the easier it is to relativize. The Hierarchy Constraints are phrased as follows:

Hierarchy Constraints (HCs): 1. A language must be able to relativize subjects. 2. Any R[elative]C[lause]-forming strategy must apply to a

continuous segment of the AH. 3. Strategies that apply at one point of the AH may in principle cease

to apply at any lower point.

Hierarchy Constraint 1 bars relativization on a lower position than subject without also having subject relative clauses in a language. Hierarchy Constraint 2 forbids leaving out a position. Hierarchy Constraint 3 maintains that any position is a potential ‘cut-off point’ for a relative clause formation strategy (cf. Keenan and Comrie 1977: 68).

Later in the article, Keenan and Comrie restricted the Constraints to primary strategies (in a language). A relative clause formation strategy is called primary if it can relativize the subject position. If a primary strategy can relativize a low position on the Accessibility Hierarchy (e.g. GEN position), this strategy will be able to relativize all higher positions (cf. Keenan and Comrie 1977: 68–69).

Over the years the formulation of the Hierarchy Constraints has undergone several revisions, while trying to come to grips with various (real and apparent) counterexamples to the Accessibility Hierarchy and its Constraints. However, the ranking of syntactic positions on the Accessibility Hierarchy – in terms of accessibility of individual relative clause formation strategies to these positions – has remained valid, provided that a language has this particular syntactic position in its grammatical make-up (cf. Keenan and Comrie 1977: 66).

For the present investigation of dialects of English, some minor modifications to the Accessibility Hierarchy were made on theoretical and also practical grounds:

The position OCOMP[ARISON] was eliminated for two reasons: First, than can function as a conjunction (particle) and as a preposition (cf. Maxwell 1979: 367; cf. also Comrie and Keenan 1979: 662), as in thanwhom (cf. Simpson and Weiner 1989, OED, 17: 861). Hence,

Page 60: A Comparative Grammar of Brtiish English Dialects

50 Tanja Herrmann

OCOMP[ARISON] could be subsumed under the position PREPOSITIONAL COMPLEMENT (cf. below). In addition, OCOMP is not sensed as “a grammatically identifiable position in the same way as the relative clauses that are higher than OCOMP” (Keenan and Hawkins 1987: 64) but rather conflated with OBL (cf. also Keenan and Hawkins 1987: 82–83). Second, even Keenan and Comrie (1977: 90 and 74) grant that examples like the man who John is taller than are of marginal acceptability. In the present dialect data, there was no instance of an object of comparison.

Although indirect objects are used in English (in dialects as well as in the standard variety), this position is commonly taken up by prepositional objects (i.e. objects of prepositions (OP)) in declarative clauses: He gave herIO the book – he gave the book to herOP. Resulting from that and from the fact that English is one of the languages which “[f]or purposes of relative clause formation, ... assimilate indirect objects to the other oblique cases” (Keenan and Comrie 1977: 72), indirect objects are very infrequently relativized (six instances in the entire data). Therefore, the syntactic position IO was subsumed under a bracketed group called OTHER, whose miscellaneous members are outside the scope of this version of the Accessibility Hierarchy.

The syntactic position OBL (oblique case) was renamed PCOMP (prepositional complement). In the initial set-up, there was a distinction between prepositional complements functioning as objects of prepositions (PCOMP(OP)) and prepositional complements functioning as adverbials (PCOMP(A)). In addition to that, a third subgroup of prepositional complements was distinguished, viz prepositional complements functioning as (‘of’)-genitives (PCOMP(GEN)). Since, in the course of the analysis, the three subgroups seemed to behave alike, all three were collapsed into one cover group PCOMP. At the same time, adverbial relative clauses were considered as a separate type of adnominal relative clauses, which do not enter the Accessibility Hierarchy as they are structurally and conceptually different. ‘Proper’ adnominal relative clauses can postmodify noun phrases with any kind of lexical meaning and their relative markers can serve all sorts of grammatical functions in the relative clause (i.e. subject, object, complement; adverbial). Adverbial relative clauses, by contrast, post-modify noun phrases with the lexical meaning of a spatial, temporal, causal, or modal adverbial; moreover, their relative markers (including relative adverbs such as where, when, or why) only serve adverbial function in the relative clause.

Page 61: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 51

Furthermore, infrequent positions like subject complements (SCOMP), as in (16a), and object complements (OCOMP), as in (16b), were grouped with indirect objects and so-called nonprepositional adverbials (A), as in (16c), to constitute the cover group OTHER:

(16) a. ... You can charcoal any timber [ there is] regarding different species. (CNO FRED Wes_004)

b. Quarterly meetings [which they call them] that was held in er the Odd Fellows Hall in Forest Road .... (SCO-K7G<S: 359>)

c. elided stranded preposition ‘with’ makes relative marker thatfunction as A (instrumental) instead of PCOMP(A): And never in my wildest dream did we imagine that Labour would get in with the resounding majority [that they did get in with]. (SCO-K6M<S: 010>)

In conclusion, the modified version of the Accessibility Hierarchy adopted here is outlined in Figure 3:

SUBJ > DO > PCOMP > GEN (OTHER)

PCOMP(OP) PCOMP(A) PCOMP(GEN) SCOMP OCOMP IO A

Figure 3. Modified version of Accessibility Hierarchy

The overall frequencies given in Table 5 confirm the Accessibility Hierarchy in all six regions (Central Southwest, East Anglia, Central Midlands, Central North, Scotland; Northern Ireland). (Frequencies of the group OTHER, which is not part of the Accessibility Hierarchy, are given for the sake of completeness). In all regions, subjects are by far most frequently relativized (mean 63.2%), followed by direct objects (mean 27.2%), prepositional complements (mean 7.3%), and genitives. The latter are found in only one region, where they account for less than 1% (mean 0.1%). The heterogeneous group OTHER varies between 1% and 3.6% across regions: Its 54 tokens are composed of 21 subject complements, eleven object complements, six indirect objects, and 16 nonprepositional adverbials.

Page 62: A Comparative Grammar of Brtiish English Dialects

52 Tanja Herrmann

Table 5. Overall frequencies

SUBJ > DO > PCOMP > GEN (OTHER) N N N N N % % % % %

total RCs

CSW 16556.7%

9432.3%

279.3%

00%

51.7%

291

EAN 24758.5%

13431.8%

348.1%

00%

71.7%

422

CMI 29865.9%

11725.9%

255.5%

30.7%

92

452

CNO 28568.2%

9723.2%

215%

00%

153.6%

418

SCO 33263.6%

12824.5%

489.2%

00%

142.7%

522

NIR 26063.9%

11428%

297.1%

00%

41%

407

total 1587 63.2%

68427.2%

1847.3%

30.1%

542.2%

2512

8.2. Individual relative clause formation strategies

While Keenan and Comrie (1977: 76) recognize only two strategies (+case/-case) for (Standard) English (cf. also Comrie and Keenan 1979: 656), more precise subdivisions were made for the present data. Each relative marker of the two major groups ‘declinable relative pronoun’ (wh-relative pronouns) and ‘indeclinable relative particles’ accounts for a distinct strategy (with the exception of whom and whose). In contrast to that, Keenan and Comrie’s ‘pronoun retention’ (in combination with a relative marker at the beginning of the relative clause) amounts to a separate relative clause formation strategy, which will be investigated in section 9. Wh-pronouns were split into the two gender-opposed components who and which, which surface in all six regions in the data. The case-marked wh-pronouns whom and whose only appear in Scotland and the Central Midlands (SCO: two whom; CMI: two whom; three whose). Whomand whose indeed have to be seen as the case-marked forms of who, since all seven instances refer to personal antecedents. In Tables 6 and 7 below, the occurrence of a relative marker in a syntactic position of the

Page 63: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 53

Accessibility Hierarchy is indicated by a plus sign [+]; nonoccurrence by a minus [-]. Absolute numbers of the total occurrences of who and its case-marked forms in all regions are given thereafter. The feature [+p] indicates that who, whom, and whose are wh-pronouns marked as referring to personal antecedents only. Featuring whose as [ p] would be permissible in principle but is not borne out by the present data:

Table 6. (Non)occurrence and frequencies for wh-pronoun [+p]

SUBJ > DO > PCOMP > GEN who whom whom whose

wh-pronoun [+p] + + + + total 244 1 3 3

On the other hand, who is not restricted to nominative case/subject position. In dialects, it can (even though only very rarely) relativize all syntactic positions (SUBJ, DO; PCOMP) except GEN (if the antecedent is personal).

Table 7. (Non)occurrence and frequencies for who

SUBJ > DO > PCOMP > GEN who + + + - total 244 4 5 -

(17a and b) show who in direct object position and examples (17c and d) in prepositional complement position:

(17) a. Mr H. went in to (gap ‘place name’), and he met an old manthere [whoDO he knows]: How today, Tommy? ... (CNO FRED Wes_019)

b. All those positions were going for want of an application, so I applied and because <gap cause=anonymization desc=“last or full name”> brought in people [whoDO he knew], certain of them were automatically filled .... (EAN-HDL<S: 335>)

c. ... The chap [whoPCOMP I got it off], he says, I ’ll come up on Monday .... (CNO FRED Wes_003)

d. Janet posted a letter for me last week to a friend [whoPCOMP I worked with at Ipswich], ... (EAN-HDK<S: 072>)

Page 64: A Comparative Grammar of Brtiish English Dialects

54 Tanja Herrmann

Which (predominantly but not exclusively nonpersonal) exhibits the following distribution on the Accessibility Hierarchy, according to regions:

Table 8. (Non)occurrence and frequencies for which

SUBJ > DO > PCOMP > GEN which + + + - CSW 17 15 4 - EAN 80 17 12 - CMI 64 18 5 - CNO 36 10 1 - SCO 58 11 7 - NIR 4 0 1 -

The numerical distribution of which supports the Accessibility Hierarchy very well (disregarding DO in Northern Ireland, whose overall number of which relative clauses is too low anyway). Since the occurrence and frequency of wh-pronouns serve as a yardstick of traditionality or degree of standardization of speech, respectively, the remarkable variance in subject position (ranging between 80 instances in East Anglia and four instances in Northern Ireland) is to be ascribed to the varied quality of the corpora used here in this respect. Whereas our corpora for East Anglian, the Central Midlands, Scotland, and the Central North also contain data from speakers whose dialect is less broad, the Central Southwest and Northern Ireland corpora only consist of data from broad dialect speakers.

A look at the frequencies of the relative particle that also corroborates the Accessibility Hierarchy. That can relativize all positions except GEN (see Table 9).

Although various dialectologists (e.g. Romaine 1980: 227, Seppänen and Kjellmer 1995, Aitken 1979: 105, Newbrook 1997: 41; Harris 1993: 150–151; Comrie 1999: 87) and surveys (SED, Lowman Survey; Survey of Anglo-Welsh Dialects) cite that’s as a dialectal GEN relative marker, the blank in the paradigm cannot be filled by the present data. That’sdeveloped from the invariant relative particle that + the reduced form of the possessive pronouns his or its into a general genitive relative pronoun that’s representing also feminine and plural antecedents (cf. Seppänen and Kjellmer 1995: 397–398). At least this seems to apply to that’s in Scottish English, as in The woman that’s sister mairriet the postie (from Grant and Murison 1974, The Scottish National Dictionary, IX: 265) or the people that’s houses were demolished (from Aitken 1979: 105).

Page 65: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 55

Table 9. (Non)occurrence and frequencies for that

SUBJ > DO > PCOMP > GEN that + + + - CSW 58 14 4 - EAN 58 24 10 - CMI 125 49 7 - CNO 145 32 4 - SCO 157 54 28 - NIR 137 49 17 -

The zero marker demands a more differentiated treatment, since it is ungrammatical as the subject of a relative clause in Standard English (cf. Quirk et al. 1985: 1250). In dialects, however, the zero marker can function as subject. At the same time, Table 10 shows that it is only in two of the northern regions, the Central North and Northern Ireland, that the subject position is easiest to relativize in all regions and thus corresponds to the Accessibility Hierarchy.

Table 10. (Non)occurrence and frequencies for zero

SUBJ > DO > PCOMP > GEN + + + -

CSW 29 43 12 - EAN 17 63 5 - CMI 27 46 7 - CNO 68 48 14 -SCO 45 62 12 - NIR 112 65 11 -

Occurrence and frequency of zero subject relative clauses can be tied to traditionality and/or colloquiality of speech. Belonging to the more traditional corpora, the Central North and the Northern Ireland data illustrate that traditional dialects can reestablish the proper order of positions along the Accessibility Hierarchy. Less traditional dialects cannot do that but they also remedy the situation on the Accessibility Hierarchy by filling the gap in SUBJ left by Standard English.

The relative particle what occurs in all regions except in Northern Ireland. In Scotland, there are only two occurrences of what, one of which is doubtful; the other is a clear case from Glasgow. The slight discontinuity

Page 66: A Comparative Grammar of Brtiish English Dialects

56 Tanja Herrmann

in the series between DO and PCOMP in CMI must be attributed to the very low figures. Analogous to that, what serves all positions except GEN in the data. What and what’s in genitive position were recorded in the SED(cf. Upton, Parry, and Widdowson 1994: 490), while what’s was also recorded in a survey conducted at schools in the 1980s (cf. Cheshire, Edwards, and Whittle 1993: 69). In the present data, however, there are no instances of what or what’s in genitive position.

Table 11. (Non)occurrence and frequencies for what

SUBJ > DO > PCOMP > GEN what + + + - CSW 36 21 7 - EAN 32 28 4 - CMI 19 1 2 - CNO 5 4 1 - SCO 2 0 0 - NIR 0 0 0 -

The traditional relative marker as is still to be found in three of the six regions (Central Midlands, Central North, and Northern Ireland). On account of its very low totals, the sheer (non)occurrence of as in a position via a +/- sign should be focussed on.

Table 12. (Non)occurrence and frequencies for as

SUBJ > DO > PCOMP > GEN as + + + - CSW - - - - EAN - - - - CMI + (7) + (3) + (1) - CNO + (3) + (2) - - SCO - - - - NIR + (2) - - -

Page 67: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 57

8.3. Interpretation

8.3.1. Who, which, that, and zero

In the present data, the wh-pronoun strategy who is the only strategy which can relativize all positions on the Accessibility Hierarchy including GEN. It either emerges in the shape of the case-marked forms who, whom, or whoseor in the shape of the non-case-marked form who in SUBJ, DO, and PCOMP. That is to say, GEN is the only position that requires an explicit, case-marked form in the very few cases when it appears (cf. also 8.4.). Both personal and nonpersonal which are well represented in all syntactic positions except GEN. Nevertheless, Romaine’s and Dekeyser’s results for Middle Scots and Early Modern English, respectively, are not borne out by the present dialect data: With the exception of GEN, wh-pronouns do not dominate the lower end of the Accessibility Hierarchy; likewise relative particles (that and zero) are not restricted to the higher positions (cf. Romaine 1980: 228; cf. also Dekeyser 1984: 76). Actually, the frequencies involving relative particles outnumber those involving wh-pronouns by far – from twice the number in the Central Midlands to almost 40 times as many in Northern Ireland. In particular, that and zero prevail over wh-pronouns in the less accessible positions DO and PCOMP. Whereas that is also prevalent in SUBJ, certain allowances have to be made for zero in subject position. Although zero relativizes subjects in all regions, it does not always have its stronghold there. In Keenan and Comrie’s framework, zero is rehabilitated as a primary relative clause formation strategy in dialects, because of its ability to relativize subjects. (The least broad East Anglian corpus, by contrast, is an exception to the rule in various ways: Its ratio of wh-pronouns to relative particles is just 174 : 243. Although that is also frequent in SUBJ and PCOMP, which is the most frequent relative marker in these two positions in East Anglia.)

8.3.2. Change in progress in terms of the Accessibility Hierarchy: whatand as

The nonstandard relative particles what and as seem to be following opposite developments (once again, East Anglia is a special case): What is on the rise (as a supra-regional nonstandard relative marker), while as is on the decline (as a regional dialect relative marker).

Page 68: A Comparative Grammar of Brtiish English Dialects

58 Tanja Herrmann

From its southeastern heartland East Anglia including Essex (cf. Poussa 1988: 448; cf. also Viereck 1975: Maps 207 and 208), what has been radiating out through the adjoining Midlands and the Home Counties, especially London, to the Southwest and, eventually, to the north (cf. Poussa 1988: 448 and 450; cf. also Cheshire, Edwards, and Whittle 1993: 64; Viereck 1991: S8b, S9, and S10; Poussa 1991: 311).

The frequencies in the present data reflect the process of dissemination. What is most frequent in the South, less frequent in the Midlands, and least frequent in the North. In our day, what spreads via the big cities (cf. Cheshire, Edwards, and Whittle 1993: 68), for which Glasgow is a case in point. In short, what has infiltrated all investigated areas, except for the countryside of Northern Ireland.

The Central Midlands data suggest that what enters the Accessibility Hierarchy by the subject position and gradually works its way down the hierarchy. Cheshire, Edwards, and Whittle (1993: 69–70) come to the same conclusion on the basis of results from questionnaires in a nation-wide survey at schools. They observe an implicational hierarchy, “such that all schools reporting the occurrence of what as a genitive pronoun also reported what as object pronoun, and all schools reporting what as object pronoun also reported what as subject pronoun” (Cheshire, Edwards, and Whittle 1993: 69). Being a hallmark of Present-Day Standard English, the wh-pronouns entered the Middle English relative marker system by the low positions on the Accessibility Hierarchy in formal and complex written language (cf. Romaine 1980: 234; cf. also Dekeyser 1984: 76). By contrast, the nonstandard relative marker what introduces itself into the Accessibility Hierarchy via the top end, because it is part of an informal straightforward spoken code, which has greater affinity to the simpler positions of the Accessibility Hierarchy (cf. Cheshire, Edwards, and Whittle 1993: 70).

What appears to have an even firmer grip on the Central Southwest (i.e. Eastern Somerset). Of all relative markers, what is recorded to be second strongest in the lower positions DO and PCOMP, as well as in SUBJ.

Although what originated in East Anglia, it is said not to be thriving in its place of origin and the prognoses of its future are not favorable. Unlike in urban centers, where it carries covert prestige among the younger generation, what has the stigma of an old and vulgar relative marker in East Anglia, which makes it unpopular among the present young and middle-aged generation, or even older people (cf. Poussa 1988: 443–444; cf. also Poussa 1996: 530; Poussa 1994: 419–422). However, among the traditional dialect speakers investigated here, what still proves to be the indigenous

Page 69: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 59

relative marker of East Anglia: Its overall number is fairly high and what can relativize lower positions on the Accessibility Hierarchy with equal ease.

In opposition to what, as has become a relic or completely disappeared, such as in the Central Southwest and East Anglia. Where it is still present, as seems to have retreated to the higher positions on the Accessibility Hierarchy, from which it will probably exit via the subject position. In addition, as appears to be confined to restrictive syntactic environments nowadays. The relative marker as once was almost the only relative marker existing in the Midlands (cf. LAE: Map S5). Hence, it is not surprising that the only instance of as in PCOMP in the data is found in the Central Midlands. In comparison to a total of 26 instances of what in the Central Midlands, there are just eleven instances of as. In the Central North, ten instances of what are faced with six instances of as. That is to say, whatseems to be driving as out.

8.4. Genitive avoidance

Only three instances (0.1%) of whose are recorded in the data, all of them in the Central Midlands. In other words, synthetic genitives are avoided in dialect. Only when prompted for a genitive relative marker – as in the questionnaire of the Survey of English Dialects – speakers tend to use the standard relative pronoun whose. The use of whose requires a coreference relationship between the antecedent and a determiner (possessor) of a noun phrase in the relative clause, which is more complex than a simple antecedent – relative noun phrase relationship (cf. Givón 1993: 133). According to Keenan and Comrie (1977: 69), “unrelativizable NPs can be systematically promoted to higher positions on the Accessibility Hierarchy, whence they can be relativized”. Thus, dialect speakers may circumvent synthetic genitives by promoting a potential genitive to a higher position on the Accessibility Hierarchy. The following possibilities are available: Possibility 1: In theory, speakers could promote a potential synthetic genitive to PCOMP (PCOMP(GEN) position) by using an analytical ‘of’-genitive. With possessive genitives, however, this was never done in the investigated dialect material. Johansson (1993) analyzed the spoken part of the Birmingham Corpus (1,3 million words of (nondialectal) British English; mainly from 1960–1981) with regard to whose and of which with nonpersonal antecedents (cf. examples [18a] and [18b], taken from Quirk et

Page 70: A Comparative Grammar of Brtiish English Dialects

60 Tanja Herrmann

al. 1985). Her findings showed that whose is unpopular but of which is even more unpopular (cf. Johansson 1993: 112).

(18) a. The house [whose roof was damaged] ... (Quirk et al. 1985:1249–1250 [emphasis and bracketing mine])

b. The house [the roof of which was damaged] ... (Quirk et al. 1985: 1249–1250 [emphasis and bracketing mine])

c. The house [of which the roof was damaged] ...

Possibility 2: In the literature on dialects, analytical genitives, which are formed by means of a resumptive pronoun (cf. section 9.1. below), figure very prominently (e.g. Edwards, Trudgill, and Weltens 1984: 27, Ihalainen 1985: 66, Miller 1993: 111 for Scottish English, Harris 1993: 150–151 for Irish English; Comrie 1999: 87; cf. Tables 15–20 in Appendix 1). In these cases, the relativized noun phrase is promoted from the genitive to the indirect object position. The possessor is expressed by a resumptive (i.e. possessive) pronoun. In the present data, an analytical genitive appeared just once:

(19) a. But (trunc) y you’d got to watch, there again, that er you didn’t exceed the width of er of your waggon, [whichDATIVEitsPOSSESSIVE PRONOUN maximum limit was er would be er eight foot three, or er eleven foot six, high]. (CMI-FY2<S: 026>)

The fact that the antecedent in this example is nonpersonal may additionally have favored the employment of nonpersonal relative pronoun which + neuter possessive pronoun its instead of whose, which is predominantly applied with personal antecedents (cf. Quirk et al. 1985: 366).

Possibility 3: (Re)constructing a proposition as a possessive have or getconstruction (example [20b]) or as a paratactic attributive with construction (example [20c]) are two well-described genitive avoidance strategies in the literature (e.g. Quirk et al. 1985: 1249, Keenan and Comrie 1977: 90–91, Johansson 1993: 112, Ihalainen 1985: 66, Elworthy 1877: 42; Kruisinga 1905: 38). In possessive have or get constructions, potential genitives are promoted to subject position while a resumptive pronoun expresses the possessor. Attributive with constructions avoid relative clause formation altogether.

Page 71: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 61

(20) a. The house [whose roof was damaged] ... (Quirk et al. 1985:1249–1250 [emphasis and bracketing mine])

b. The house [that had its roof damaged] .... c. The house with the damaged roof ....

Whereas no clear case of a possessive have or get construction avoiding GEN position was encountered in the present data, there was a handful of attributive with constructions. For example:

(21) instead of: Had a gun, you see, [whose cartridge was blank]:Had a gun, you see, with a blank, blank cartridge in. (EAN-K65<S: 0495>)

It can be concluded from the present investigation that dialect speakers usually resort to paratactic constructions to describe a genitive sense relation, rather than promote a relativized noun phrase up the Accessibility Hierarchy. Common parataxes, and-coordinations, and left dislocations are used in the place of a genitive relative clause (cf. 22a–c below). Of course, these paratactic constructions are not exclusively applied to evade genitive relative clauses, but subordinate clauses in general, including relative clauses, particularly of the nonrestrictive type.

(22) a. parataxis: And the <gap cause=anonymization desc=“last or full name”> used a little further past the <gap cause=anonymization desc=“last or full name”> and then there was another one further up near Cranfield well I can’t remember the name of that. (EAN-H5H<Person:PSEAN4><S: 479>) instead of: ... there was another one further up near Cranfield well [whose name I can’t remember].

b. and-coordination:... L. R., and her father was a farmer down at E. Hall, and he had rather an unusual Christian name. (CNO FRED Wes_017)instead of: ... L. R., [whose father was a farmer down at E. Hall], ....

Page 72: A Comparative Grammar of Brtiish English Dialects

62 Tanja Herrmann

c. left dislocation the grandfather:Now the boatmen they used to erm <UNCLEAR> the name, the family of a name of <gap cause=anonymization desc=“last or full name”>, and the old man, the grandfather his name was <gap cause=anonymization desc=“last or full name”> and .... (EAN-H5H<S: 345>) instead of: ... the grandfather [whose name was <gap cause=anonymization desc=“last or full name”>] ....

8.5. Interaction of zero subject relative clauses with topicalization hierarchies

Topicalization structures including clefts, pseudo-clefts, and all-pseudo-clefts (explained below) are not counted as involving relative clauses but form a distinct type of clause. In contradistinction to relative clauses, topicalization structures do not revolve around the modification of an antecedent, but the focussing of an antecedent (cf. Lehmann 1984: 363; cf. also Quirk et al. 1985: 1386–1387; Huddleston and Pullum 2002: 1035; Keenan 1985: 170; Stockwell, Schachter, and Partee 1973: 421–422). The subordinate clauses of these topicalization structures are related to relative clauses and can be seen as assuming positions along a continuum of relative clauses, with topicalization structures and ‘relative clauses proper’ at opposite ends (see Figure 4).

adnominal relative clauses

‘lexically empty’ antecedent cleft all-pseudo existential relative clause

topicalization ‘relativestructures clause

proper’ pseudo-cleft

nominal relative clauses

Figure 4. Continuum of relative clauses: from topicalization structures to ‘relative clause proper’

Page 73: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 63

The closer the types are located toward the right pole, the more similar to ‘relative clauses proper’ they are. In Figure 4, all clause types resembling adnominal relative clauses are arranged above the line, while those resembling nominal relative clauses are below the line. In contrast to clefts, pseudo-clefts, and all-pseudo-clefts, all of which are placed at or toward the topicalization end of the continuum, those subtypes in bold type, i.e. the so-called ‘lexically empty’14 antecedent relative clause and the existential, areincluded in the count of all relative clauses in the present analysis, as they sufficiently meet the defining criteria for relative clauses put forward in section 1.

Although resembling an adnominal relative clause, the cleft sentence is a topicalization structure that splits one proposition into two separate clauses for the sake of giving focus to an element (for the following discussion cf. Quirk et al. 1985: 1383–1387). The it-cleft follows the pattern: it + be + focussed NP + subordinate clause, which resembles a restrictive relative clause. However, there are various syntactic differences between an “‘annex’ clause” (Quirk et al. 1985: 1387) of a cleft and a relative clause, in addition to the prosodic difference arising from its focussing nature (cf. also Lehmann 1984: 362).15 One of the most striking differences between relative clauses and clefts is that the co-occurrence restriction between proper names (or other nouns or pronouns with specific or unique reference) and restrictive (!) zero or that clauses is lifted in cleft sentences, both in dialectal speech and in Standard English (cf. Quirk et al.1985: 1241 and 1387; cf. also Huddleston and Pullum 2002: 1056–1057). Consider the examples in (23):

(23) a. Isn’t it High House we ’re talking about? (CNO FRED Wes_011)

b. [...] I think it was Bristol Engineering that they did a lot of work for them during the war. (CNO FRED Wes_009)

c. [...] it’s you it’s up to. [...] (CSW-SRLM 123<T 1100>)

As shown in (24), clefts can also be introduced by the demonstrative pronoun that (or this):

(24) a. That’s the boss she’s speaking about. (SCO-G62<Person: PSSCO13><S: 1070>)

b. No me elder brother was, not me second brother, that was the Second World War he was in. (CMI-FXW<S: 037>)

Page 74: A Comparative Grammar of Brtiish English Dialects

64 Tanja Herrmann

c. And that was J. H. that they had from E. (CNO FRED Wes_015)

Like dialectal relative clauses, dialectal clefts also lend themselves to having a gap (zero marker) in subject position in the subordinate clause:

(25) a. It was J. H. lived in there. Her father raised in that house next, it was A. H. lived there. (CSW FRED Som_012)

b. [...] It was my grandmother owned this bit of land [...]. (NIR NITCS A13.3)

c. [...]’t weren’t everyone had a binder. (CSW-SRLM 108<T 1440>)

The subordinate clause of the all-pseudo-cleft sentence looks like a restrictive relative clause. The all-pseudo-cleft, as illustrated in (26), parallels the pseudo-cleft in structure, but differs from it by having the overt preceding noun phrase all in the matrix clause, followed by a subordinate clause.16 In the examples in (26) all and the subordinate clause are underlined:

(26) a. [...] All as we could get for this milk was four pence a gallon, [...]. (CSW-SRLM 105<T 1160>)

b. [...] Practically {PAUSE}, you practically kept the house on it, you know {I know} [insertion in curly brackets, T.H.], all was bought was bread. [...] (NIR NITCS A61.3)

c. all-pseudo-cleft focussing a verb: [...] I’d go the rounds with him and all I used to do was to er take the peoples [sic] things that they’d bought up the entry you see because they were all entries then. (CMI-FY5<S:415>)

d. all-pseudo-cleft focussing a verb: [...] so all as he had to do were go round in a circle all the time, [...]. (CSW FRED Som_001)

While the all-pseudo-cleft in example (26c) also occurs in Standard English, examples (26a,b and d) involve the dialectal particle as and the zero marker in subject position, which are typical dialect features. Topicalization structures incorporate particles (e.g. as) and/or maintain usages (e.g. the zero marker in subject position) which have dropped out of

Page 75: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 65

the relative marker system (in an area). In the Central Southwest, for example, the relative particle as has retracted from the relative marker system to semantically and syntactically related niches and exits the dialect via topicalization structures, such as it-clefts and all-pseudo-clefts.17

Historically, topicalization structures like all-pseudo-clefts served as as a stepping-stone from the comparative construction (such as) into the relative marker system in (Early) Middle English (cf. Smith 1984: 99–100 and 134). As seemed to have developed a stronghold in topicalization structures until the rising wh-pronouns also started infiltrating topicalization structures like it-clefts in Late Middle English (cf. Ball 1994: 183). In other words, these particles appear to enter a linguistic system via such topicalization structures and also appear to leave it through them. Mustanoja (1960: 191) reports all what as occurring in Old English, while in Middle English, what is said to mostly relativize so-called “antecedents of less definite character, like all and nothing” (Mustanoja 1960: 194). If a dialect (or an idiolect) shows a specific particle in a relative clause, such as what in the Central Southwest, it also shows this particle in topicalization structures like clefts and all-pseudo-clefts. For instance, in the Central Southwest, what as a relative marker occurs in each of the investigated texts. So it also occurs in all-pseudo-clefts, as in example (27):

(27) [...] so all what we had was oil lamp, oil lamp and a telephone up there for the police to keep ringing us up to see if we were all right. (CSW FRED Som_012)

Hence, the following implicational tendency emerges, which holds true for the vernacular particles what and as (in the Central Southwest) and the zero marker in subject function overall (cf. existentials and ‘lexically empty’ antecedent relative clauses):

particle in topicalization structures > particle in ‘relative clauses proper’18

A variety of syntactic structures that introduce new information or participants into the discourse, such as existential sentences and ‘lexicallyempty’ antecedent relative clauses (i.e. relative clauses that modify semantically bleached antecedents like the (only, best, first, last, next) thing, the (only, best, first, last, next) one, one thing, something, anybody,etc.), possess a focalizing aspect in addition to a weightier modifying function. Consequently, have- and be-existentials of the types ‘subject +

Page 76: A Comparative Grammar of Brtiish English Dialects

66 Tanja Herrmann

have + NP-[RC]’ and ‘there + be + NP-[RC]’, respectively, are regarded as involving relative clauses. Being a topicalization construction to some extent as well, however, existentials evince zero relative clauses after specific nouns and proper names (compare Lumsden 1988: 199–201). Hence, these clauses are nonrestrictive zero relative clauses syntactically, which, of course, are ungrammatical in Standard English (cf. Quirk et al.1985: 1258; cf. also Huddleston and Pullum 2002: 1056 and 1059). For example:

(28) a. there-existential:[...] there was Mr M. and B. W. from K. [ came round buying horses]. [...]. (CNO FRED Wes_015)

b. there-existential:(UNCLEAR) of course you there would just been my father and mother [ ’d be speaking Gaelic all the time in the house] you see. (SCO-K6N<Person: PSSCO28><S: 0473>)

c. there-existential:[...] there was only him {you know} [insertion in curly brackets, T.H.] [ used to preach], [...]. (CMI-FYH<S: 631>)

d. have-existential:Yes she had her aunt [ was a widow there at the time [when she came to <gap cause=anonymization desc=address>]] and she just lived about a year. (SCO-K6N<Person: PSSCO28><S: 0114>)

In addition, the zero relative marker is ungrammatical in subject position in (written) Standard English (cf. Quirk et al. 1985: 1250; cf. also Biber et. al.1999: 619). Supposedly, this restriction on the use of the zero marker in subject function is due to the difficult processing of constructions in which the zero relative clause modifies an initial noun phrase (subject of the matrix clause) (cf. Bever and Langendoen 1971: 444). The following example is taken from Bever and Langendoen (1972: 91):

(29) a. *Anyone owns a fleet of six cars deserves to be taxed at the highest rate.

b. sentence is bracketed as if there were two coordinate clauses at first blush: *[Anyone owns a fleet of six cars] deserves to be taxed at the highest rate.

Page 77: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 67

c. sentence is rebracketed as a matrix clause and a relative clause modifying the matrix subject, after coming across the second finite verb (deserves), which needs a subject: *[Anyone [ owns a fleet of six cars]RC deserves to be taxed at the highest rate]matrix

At first sight, the finite verb occurring first is, according to their analysis, bracketed together with the preceding noun phrase, in the treacherous assumption that the first clause is a main clause. Only after coming upon the second finite verb, the former misbracketing [NP-V-O]main clause-V-O is corrected to [NP-[NP -V-O]RC-V-O]matrix clause (cf. Bever and Langendoen 1972: 66). If an overt subordinator (relative marker) indicates the beginning of a subordinate clause, such erroneous analyses are ruled out from the start (cf. Huddleston and Pullum 2002: 1055). Bever and Langendoen (1972: 78 and 91) claim that this perceptual constraint was grammaticalized, i.e. extended to situations where perceptual problems would not arise (cf. also Huddleston and Pullum 2002: 1055).

Nevertheless, descriptive syntax makes some concessions to the appearance of zero in subject function: Zero subject relative clauses are either “clearly non-standard”, “between very informal and non-standard” (Huddleston and Pullum 2002: 1055), or “a marginally non-standard usage” that may occur “in some conversational varieties” (Biber et. al. 1999: 619) or “in very informal speech” (Quirk et al. 1985: 1250) “under certain conditions” (Huddleston and Pullum 2002: 1055).

In dialectal speech, the constraint on zero subject relative clauses is overridden the more a clause type moves away from the modification of an antecedental noun phrase (‘relative clause proper’) toward the topicalization of a noun phrase (topicalization structures). In other words, zero subject relative clauses pattern along the continuum of relative clauses in Figure 4 above. They occur with increasing ease (in a given dialect or idiolect), as one goes from existential and related presentational clauses like ‘lexically empty’ antecedent relative clauses to clear topicalization clauses like clefts and all-pseudo-clefts, while they are very scarce in pure modification structures (‘relative clauses proper’).

Looking at previous dialectal investigations, the overwhelming majority of zero relative clauses in subject function is reported to occur in the following syntactic environments (or those syntactic environments can be deduced from cited examples), irrespective of the region investigated: There- and have-existentials, existential-like constructions like copular ‘be’

Page 78: A Comparative Grammar of Brtiish English Dialects

68 Tanja Herrmann

sentences or equational sentences, ‘lexically empty’ antecedent relative clauses, and clefts (e.g. cf. Elworthy 1877: 41 and 58; cf. also Elworthy [1886] 1965: xxxiv, 660, and 688; Grant and Dixon 1921: 102; Harris 1993: 149–150; Ihalainen 1980: 189; Ihalainen 1987: 84 and 90–91; Kekäläinen 1985: 355; Miller and Weinert 1998: 107, 128, 301, and 347–348; Ojanen 1982: 75–78; Peitsara 2002: 171–172 and 174; Poussa 1999: 96; Tagliamonte 2002: 157–158 and 160–161; van den Eynden 1993: 170, 175, and 178–187; van den Eynden Morpeth 2002: 186–188). Moreover, a handful of examples occur in have-existential-like constructions involving the verb ‘to know’ or in other constructions which are not (or not easily) subsumed under topicalization structures (e.g. cf. Corrigan 1997: 335; cf. also Ojanen 1982: 77; Wilson 1915: 91). Shnukal (1981), who encountered all these various types of zero subject relative clauses in topicalization structures in a variety of Australian English, sketches the following implicational scale, which “parallels the frequencies of occurrence of the nonstandard variant as well as the history of obligatory S[ubject]R[elative]P[ronoun] presence in written Standard English” (Shnukal 1981: 324):19

Type IV Type III Type II Type I

According to Shnukal, speakers who have zero subject relative clauses in Type IV sentences may also show zero subject relative clauses in the other three syntactic environments but not vice versa. Type IV consists of a relative clause modifying anybody (here, ‘lexically empty’ antecedent relative clause type), Type III consists of the present ‘have-existential-likeconstruction involving the verb ‘to know’’, Type II comprises there- and have-existentials, and Type I clefts and equational sentences. Except that here, Shnukal’s Type III is grouped with (have-)existentials and that equational sentences are subsumed either under ‘lexically empty’ antecedent relative clauses or under existentials (depending on the type of antecedent), her implicational scale mirrors the continuum of relative clauses sketched above. All types of zero subject relative clauses are illustrated below:

(30) a. there-existential:[...] there’s nobody [ does that now]. (NIR NITCS A44.3)

b. multiple relative clause: first relative clause: have-existential-like construction involving the verb ‘to know’; second relative

Page 79: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 69

clause: have-existential; third and forth relative clause: there-existentials:No, no, no, no, eh, they went over, they would have been... I know two, or three [ went over], but they went till the building. They had, they had friends [ was in the building trade in Scotland], they went over to, there, to... But there was more [ went till the States], [ went till America], from this country {ahah}, in the young days. They had ones with uncles and aunts in America, and brought them out when they were young. (NIR NITCS A31.3)

c. existential-like construction involving the semantically weak verb ‘to see’:[...] I seen a chap at Broughton Moor, [ got his leg took off]. [...] (CNO FRED Wes_014)

(31) a. ‘lexically empty’ antecedent relative clause: [...] Yet, at them times, anybody [ wanted to learn] could learn, and I see anybody [ doesn’t want to learn yet], doesn’t learn, so I don’t know. [...]. (NIR NITCS A39.3)

b. ‘lexically empty’ antecedent relative clause: [...] The ones [ was here] was all going back to France. (NIR NITCS A49.3)

(32) a. ‘relative clause proper’: Aye, cogs, not cut up, a cog cut up, but cut up and sort of work them in together you know, they could make a cog wheel [would last quite a long time], of wood, you wouldn’t think that possible. [...] (CNO FRED Wes_004)

b. ‘relative clause proper’: [...] and he put in the paper about these sheep [ would live under t’ snow], [...]! [...] (CNO FRED Wes_015)

The total of 298 zero relative clauses in the present dialectal data corresponds to Shnukal’s findings: There are 228 there-existentials, 24 have-existentials, 9 equational sentences, 9 existential-like constructions involving the verbs ‘to see’, ‘to hear’, or ‘to know’, 18 ‘lexically empty’ antecedent relative clauses, and 10 ‘relative clauses proper’, which do not fit into any particular category. It can be observed that not only does the overall number of zero relative clauses correlate with the broadness of the

Page 80: A Comparative Grammar of Brtiish English Dialects

70 Tanja Herrmann

individual corpora (e.g. of the overall 228 there-existentials, 89 and 54 instances, respectively, originate from the broad NIR and CNO corpora, respectively) but broadness of speech also determines the occurrence or the frequency of zero relative clauses in particular syntactic environments: Broader corpora have (more) instances of zero in ‘relative clauses proper’ and ‘lexically empty’ antecedent relative clauses than less broad corpora, while existentials and existential-like constructions are open to all corpora. Thus, the implicational scale (supporting the continuum of relative clauses in Figure 4 above) looks as follows:

clefts > existentials > ‘lexically empty’ antecedent relative clause > ‘relative clause proper’

Total: 270 18 10

While resembling a nominal relative clause, the topicalization structure pseudo-cleft is a S-V-SCOMP sentence in which a subordinate clause, usually introduced by what, functions as subject (‘basic pseudo-cleft’). In ‘reversed pseudo-clefts’, the matrix clause, normally starting with demonstrative that (or this), precedes the subordinate clause, which functions as subject complement in the sentence (cf. Quirk et al. 1985: 1387–1388; cf. also Huddleston 1984: 462). Since pseudo-clefts give focusto the referent of the subordinate clause (cf. Huddleston 1984: 466) they are distinct from ‘nominal relative clauses’ (cf. Lehmann 1984: 361). For example:

(33) a. basic pseudo-cleft: Verb is highlighted: [...] what we used to do were, book a drop less milk and make a little tiny one, [...]. (CSW-SRLM 122<T 1320>)

b. reversed pseudo-cleft: [...] No doubt that was what did it. [...] (CNO FRED Wes_015)

9. Resumptive pronouns

Resumptive pronouns represent the antecedent within the relative clause – on top of a representation of the antecedent by a relative marker at the beginning of the relative clause. Thus, resumptive pronouns have to be distinguished from so-called ‘copies’, which represent the antecedent in the continuation of the matrix clause. Resumptive pronouns surface as

Page 81: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 71

pronouns (personal, possessive, demonstrative; existential there) or adverbs (e.g. demonstrative there), usually in the position they would have in a corresponding declarative clause. They point out or reinforce the grammatical function of the relativized noun phrase in the relative clause by case-marking and position, and they strengthen the coreference relation between the relativized noun phrase and the antecedent by agreement in gender and number.

9.1. Interpreting the pronoun retention strategy in terms of the Accessibility Hierarchy

The examples below illustrate resumptive pronouns in the syntactic positions subject, direct object, prepositional complement, genitive, and (nonprepositional) adverbial, respectively:

(34) a. resumptive personal pronoun it in subject position: Well, it ’s what they fed, you used to put it [i.e. treacle, T.H.]

on hay [that it was mouldy], you know, bad hay, and just sprinkle it on to give a better taste for t’ cow to eat, you see.(CNO FRED Wes_008)

b. resumptive personal pronoun it in direct object position; antecedent and resumptive pronoun are not strictly coreferential as number-markers do not match: And I have lots of letters which I discovered about five years ago, which I thought were probably letters [that my father had written it], Adam <gap cause=anonymization desc=“last or full name”>, because his name and my grandfather’s name were exactly the same. (SCO-GYW<S: 101>)

c. resumptive personal pronoun them in prepositional complement position:... it was, you know, looked upon then you were, were public transport and the public team [that you belonged to them].(EAN-HDL<S: 101>)

d. resumptive possessive pronoun its in genitive position: But (trunc) y you’d got to watch, there again, that er you didn’t exceed the width of er of your waggon, [which itsmaximum limit was er would be er eight foot three, or er eleven foot six, high]. (CMI-FY2<S: 026>)

Page 82: A Comparative Grammar of Brtiish English Dialects

72 Tanja Herrmann

e. resumptive demonstrative adverb there in (nonprepositional) adverbial position; preposition elision of at (cf. 10.1. below): they had the stallions down there then you see, they had stallions at the stud. (EAN-HYC<Person: PSEAN17><S: 0207>)[[at] Which we were there often]. (<S: 0208>)

Based on Keenan and Comrie’s findings, two predictions are made for the present data:

1. The distribution of resumptive pronouns in simple relative clauses mirrors the accessibility of syntactic positions (Accessibility Hierarchy) such that resumptive pronouns will be more likely to occur the lower a position ranks on the Accessibility Hierarchy (GEN > PCOMP > DO > SUBJ) (cf. Keenan and Comrie 1977: 92).

2. Resumptive pronouns combine more often with invariant relative particles than with relative pronouns because particles are less explicit in terms of structural markedness. In other words, in contrast to relative pronouns, relative particles lack case-marking, agreement in gender or animacy (i.e. with regard to the personality/nonpersonality opposition), and agreement in number with the antecedent (to some extent, relative pronouns show agreement in number with collective nouns). High structural markedness eases the retrievability of the relativized noun phrase (and thus the identification of the antecedent) and helps to identify the syntactic function of the relative marker in the relative clause.

Table 13. Resumptive pronouns in simple relative clauses across regions

SUBJ DO PCOMP GEN (Other:A) total zero - 1 - - - 1 that 1 2 1 - - 4 what 1 1 - - - 2 as - - - - - - who 1 - - - - 1 which 5 1 3 1 (3) 13 whom - - - - - - whose - - - - - - total 8 5 4 1 (3) 21

Page 83: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 73

Neither prediction is borne out by the data. In general, resumptive pronouns are extremely rare, as shown in Table 13. Indeed, the data seem to suggest the opposite of what was predicted. First, instead of a steadily increasing number of resumptive pronouns from subject to genitive position, the number of resumptive pronouns decreases with every position. Second, when all invariant relative particles and both wh-pronouns are taken together, twice as many relative pronouns co-occur with resumptive pronouns as relative particles do. In addition, the least explicit relative particle, i.e. the zero marker, combines only once with a resumptive pronoun.

How is this to be explained, given that the Accessibility Hierarchy should also apply to the pronoun retention strategy, which it clearly does in Keenan and Comrie’s (1977: 92–93) cross-linguistic investigation? Admittedly, Keenan and Comrie were primarily concerned with the occurrence/nonoccurrence of a resumptive pronoun in a position within a language, not with the number of occurrences in each position. Statements on frequencies of resumptive pronouns in individual positions they made only in passing (cf. Keenan 1985: 146–148; cf. also Comrie and Keenan 1979: 663). Second language acquisition, however, seems to provide evidence for an inverse relation between resumptive pronouns and the Accessibility Hierarchy (except for the genitive position) (cf. Hyltenstam 1984: 47–51; cf. also Gass and Selinker 1994: 113–114).

For one thing, the overall number of 21 resumptive pronouns is not particularly impressive. In fact, the frequency of resumptive pronouns in the investigated 480,000 words corpus is as low as 4.4 occurrences per 100,000 words, so the scarcity of the phenomenon might be responsible for a skewed distribution.

Secondly, new starts in favor of a paratactic construction might be mistaken for applications of a resumptive pronoun. The two phenomena are difficult to distinguish, especially in the face of missing information on prosody and speakers’ intentions. This is particularly true for resumptives in subject position:

(35) a. new start after ‘er’, starting with they, which introduces a new parataxis:

That was our marking, all our ships used to have the blue and er I think blue <UNCLEAR> and yellow in the square, cos they hired these <UNCLEAR> the people who do er you know suppose hire them off now would be the erm the Dutch people

Page 84: A Comparative Grammar of Brtiish English Dialects

74 Tanja Herrmann

cos they’re the people [what] er, they deal in all that type of thing, big dredging, that’s how Rotterdam was built <vocal desc=“clears throat”> (EAN-H5G<S: 846>)

b. new start after ‘er’, starting with it, which introduces a new parataxis:We then got the, the agency for Morris, which was a very popular car and erm we also got the agency for Austins,[which] er, it came about with father being interested in a hiring car, the Austin Twenty, he thought it was, it was the best that could be got. (SCO-GYW<S: 138>)

c. new start after truncated belong, starting with that, which introduces a new parataxis: And we had erm we, there was a few changes during that time, we used to get troops coming in and occupying the classroom, and we (UNCLEAR) maybe, at the very early days I can remember having to go to the <gap cause=anonymization desc=address> Street School and also the <gap cause=anonymization desc=address> Street Hall [which (trunc) belong] that was a church hall. (SCO-GYW<S: 169>)

All examples lacking an indication of a new start, such as punctuation, pause fillers, or self-correction, were counted as instantiating resumptive pronouns. This policy might be (partially) responsible for an inflated number of resumptives in subject position. Below, four instances of a resumptive pronoun in subject position will be cited and two instances of a resumptive pronoun in PCOMP position:

(36) a. resumptive personal pronoun they in subject position; antecedent and resumptive pronoun are not strictly coreferential: Semantics and number concord do not match; resumptive they refers to ‘the M. clan’, living at the ‘M. place’, which is represented by Mrs M.:Yeah. I was getting ready to go to school one morning, he said, “Where you going today?” I said, “Going to school, dad.” “No”, he said, “you gotta come with me.” And we walked from Barton St David down, round, well, ??? down round that corner and well on down to Catsham. Up through Southwood, up to Lottisham. And Mrs M., [what they live up here to, where the, (CSW-SRLM 132<T 1140>)

Page 85: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 75

Q: MUSEUM IS.... museum is]. Er, to her father that did, had the farm sale. And, er, course we went up there and stayed there till the sale was finished. Well then father took on some cattle then, to deliver, and we finished up out to Mr A.’s farm sale. We took some cattle and we finished up out to North Wootton, that night.

b. existential there is the resumptive pronoun in subject position (preferred analysis); alternatively, this might exemplify preposition elision of stranded partitive GEN preposition of;compare with example (36f) below: I can remember quite vividly the old tramcars running there er day and night, with the last service leaving the outskirts of Edinburgh around about er twelve er eleven thirty and you <trunc> g have about ten minutes or so to reach the depots[which there were many and varied <UNCLEAR> at this particular time]. (SCO-GYU<S: 224>)

c. resumptive personal pronoun it in subject position, pointing out singular number of relativized noun phrase in relative clause, in combination with relative particle that:Well, it ’s what they fed, you used to put it [i.e. treacle, T.H.] on hay [that it was mouldy], you know, bad hay, and just sprinkle it on to give a better taste for t’ cow to eat, you see. (CNO FRED Wes_008)

d. resumptive demonstrative pronoun that in subject position (preferred analysis); alternatively, that could also be analyzed as a vacuous conjunction in the Middle English relative marker relic which that:So I went and left I got a job at <gap cause=anonymization desc=“last or full name”> [which that was one of <gap cause=anonymization desc=“last or full name”>’s]. (CMI-FYE<S: 228>)

e. resumptive personal pronoun them after stranded partitive genitive preposition of (preferred analysis); alternative analysis: change to parataxis in the middle of the clause, i.e. to consider we had as a clause element which belongs to two adjacent clauses. It is the subject and verb in the preceding relative clause (which we had) and simultaneously the subject and verb of an ensuing main clause (we had quite a number of

Page 86: A Comparative Grammar of Brtiish English Dialects

76 Tanja Herrmann

them). Bracketing would look as follows: ... the tools, [which [we had] quite a number of them]:Now the first job he did was to get you well acquainted with the tools, [which we had quite a number of them]. (CMI-FY2<S: 003>)

f. resumptive personal pronoun us after stranded partitive genitive preposition of (preferred analysis); alternative analysis: Existential there is resumptive pronoun in subject position; compare to example (36b) above: So after a fortnight, thereabouts, the union had a meeting and decided that everybody in the industry like, [which of course there was only about a thousand of us any way], everybody would (trunc) s would stop until this man could start at work.(CMI-FYH<S: 556>)

The preponderance of resumptive pronouns involving which, in general, and in subject position, in particular, is conspicuous (compare Peitsara’s [2002: 175] examples in her Suffolk corpus and Ihalainen’s [1980: 190–191] examples in his Somerset corpus).

However, (36d) and another example involving which that in subject position are not excluded from an analysis as resumptive pronouns, although they could in principle be analyzed as instances of the relative marker relic which that found in Middle English (where that functioned as a vacuous conjunction). The occurrence of analogous combinations of a relative marker and a resumptive pronoun in subject position (like which it,who they, what they, or that it in the examples above) makes a special interpretation for which that unwarranted. In such cases an analysis of the second element as a conjunction is not possible, since the resumptive pronoun takes the shape of a personal pronoun. Nor is the occurrence of a demonstrative pronoun like that figuring as a resumptive pronoun, as in which that, unusual. Resumptive pronouns can, after all, take on the shape of personal, possessive, or demonstrative pronouns.

In contradistinction to what Jespersen (1927: 188) says about Standard English, namely that pied-piping is mandatory in partitive genitives (cf. also Huddleston and Pullum: 1041), examples (36e) and (36f) seem to document that stranding the preposition of in partitive genitives is permitted in dialect.

Page 87: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 77

9.2. (Non)Restrictiveness as a factor

Checking the restrictiveness/nonrestrictiveness of the which examples in subject position, and then of the which examples in all other positions, it turns out that, without a single exception, they are all nonrestrictive. Thus, the preponderance of which with a resumptive pronoun (contradicting Prediction 2) appears to correlate with the nonrestrictiveness of the relative clauses. In dialect, the relative particle what, which combines with personal and nonpersonal antecedents, is also available for syntactically nonrestrictive environments (compare [36a] above; the other what example in direct object position is also nonrestrictive). Which, however, is thedominant relative marker in nonpersonal nonrestrictives and a viable option in personal nonrestrictives (compare [36f] above), whereas the relative markers who, that, and zero are only minor options in nonrestrictive relative clauses (cf. Table 2 above). By contrast, the restrictive subject relative clause in example (36c) selects the relative particle that. The three restrictive examples in DO and the one in PCOMP equally opt for relative particles (three instances of that; one instance of zero).

Recalling that Keenan and Comrie set up the Accessibility Hierarchy on the basis of restrictive relative clauses with definite antecedents (cf. Keenan and Comrie 1977: 64), the prediction about the distribution of the pronoun retention strategy does not hold if, as in the present corpus, the majority of relative clauses involving the pronoun retention strategy are nonrestrictive. If all nonrestrictive examples of resumptives were discarded, we would be left with a mere handful of restrictive examples (one zero relative clause; four that relative clauses), which would neither save nor thwart Prediction 1. The question arises why resumptive pronouns in the present data mainly appear in nonrestrictive relative clauses. It seems that resumptives have a predilection for appearing in difficult and complex syntactic environments (cf. Keenan and Comrie 1977: 92). Since nonrestrictives are atypical of spoken language and uncommon or even alien to traditional dialect, their application demands the help of a resumptive pronoun, more than restrictives do. In sum, dialect speakers resort to resumptive pronouns in constructing nonrestrictive relative clauses in order to cope with a linguistic complexity with which they are rather unfamiliar.

With respect to Prediction 1, that the distribution of resumptive pronouns mirrors the accessibility of syntactic positions, the question remains why resumptives in nonrestrictive relative clauses, particularly those involving which, do not follow the hierarchy for pronoun retention in

Page 88: A Comparative Grammar of Brtiish English Dialects

78 Tanja Herrmann

restrictive relative clauses (GEN > PCOMP > DO > SUBJ). Or, to be more precise, what causes this abundance of nonrestrictive which resumptives in the subject position?

Dialect speakers who have less practice in constructing a nonrestrictive relative clause, since nonrestrictive relatives are rather a feature of written Standard English, fall back on the most basic type of (nonrestrictive) relative clause: the one which relativizes the subject position – in Keenan and Comrie’s (1977: 95) words, “subject relatives are psychologically simpler than nonsubject relatives”. That is to say, speakers start a nonrestrictive relative clause by employing the standard relative marker which in subject position and additionally reinforce the syntactic position of the relativized noun phrase in the relative clause via a resumptive pronoun. Hence, the predominance of the subject position within the pronoun retention strategy – or even a reverse order of syntactic positions from what was expected (compare Table 14 below) – arises from the dominance of the subject function in the general Accessibility Hierarchy, which also applies to the pronoun retention strategy. In this sense, the general Accessibility Hierarchy (SUBJ > DO > PCOMP > GEN) overrides the pronoun retention Accessibility Hierarchy (GEN > PCOMP > DO > SUBJ) in nonrestrictive environments.

Table 14. Resumptive pronouns in nonrestrictive relative clauses, with special emphasis on which

SUBJ DO PCOMP GEN (Other:A) total what 1 1 - - - 2which 5 1 3 1 (3) 13total 6 2 3 1 (3) 15

In sum, Keenan and Comrie’s Accessibility Hierarchy, which was set up for restrictive relative clauses with definite antecedents, is not invalidated by these unexpected result with regard to the pronoun retention strategy. It is the inclusion of nonrestrictive relative clauses which is responsible for Predictions 1 and 2 not being borne out by the present data. The pronoun retention strategy reflects the Accessibility Hierarchy in reverse order in restrictive contexts, yet it is not or not necessarily transferable to nonrestrictive relative clauses.

Page 89: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 79

9.3. Further embedded relative clauses via resumptive pronouns

According to Keenan and Comrie (1977: 92), pronoun retention enables or facilitates relativization in “‘difficult’ environments” such as “coordinate NPs, other relative clauses, indirect questions, and even sentence complements of NPs” (Keenan and Comrie 1977: 92; cf. also Comrie 1993: 140–141 and 147; Comrie 1999: 89; Keenan 1985: 155–156; Hawkins 1999: 265). In spelling out the relativized noun phrase in its normal clause position, resumptive pronouns maintain the coreference relation between the antecedent and the relative marker after intervening clauses or in otherwise deviant coordinate relative clauses.

(37a) and (37b) exemplify multiple relative clauses in which the relative marker serves different grammatical functions within the various coordinated relative clauses. Without a resumptive pronoun this would be prohibited by a coordinate construction constraint. With the aid of a resumptive pronoun, which explicates the changed grammatical function of the relative marker, the coordinate construction constraint is overcome so that these disparate juncts can be linked via and-coordination:

(37) a. The zero relative marker serves direct object function in the first coordinated relative clause, whereas it serves prepositional complement function in the second coordinated relative clause; the resumptive pronoun it indicates the change of grammatical function of the zero marker from the first to the second relative clause when it surfaces as prepositional complement after the stranded preposition with:

Er things like er <pause> crowbars and bull croppers and er rescue ropes and lines and things of that kind, which are very very simple, stuff [ you would buy in a hardware shop and probably be able to manage with it]. (EAN-K69<Person: PSEAN13><S: 065>)

b. Which serves prepositional complement function in the first two relative clauses (stranded preposition about is omitted in the second relative clause); which changes to direct object function in the third relative clause, where the resumptive pronoun it explicates the changed grammatical function of the relative marker: ... you know the decontamination side, the clearing up, the protecting of the environment from toxic chemicals er [which

Page 90: A Comparative Grammar of Brtiish English Dialects

80 Tanja Herrmann

we’ve all heard about in newspapers, and read reports and seen it on television], .... (EAN-K69<Person: PSEAN13><S: 038>)

In (38a) the presence of a resumptive pronoun in prepositional complement position keeps up the coreference relation between antecedent and relative marker after an intervening clause parenthesis (I’m now retired). Example (38b) exhibits a resumptive pronoun in a further embedded coordinate to-infinitive clause.

(38) a. ... so that that was quite an interesting area [which, even now [I’m now retired] I still have a little hand in that] .... (EAN-K69<Person: PSEAN13><S: 031>)

b. resumptive pronoun them in further embedded and-coordinated to-infinitive clauses indicates change of grammatical function from prepositional complement (PCOMP(OP)) to direct object; relative marker that functions as subject in the relative clause, which is reinforced by resumptive pronoun they after interrupting subordinate clauses (after a false start in which relative marker that would function as direct object: that they’d have done):... But, eh... {PAUSE} there’re quite a few old sayings, and one thing and another, in the country, [that [if you had time [to, to think about {mm} and study them]], they’d have, they’dhave done, gone by the board] {ahah}. ... (NIR NITCS A64.3)

In (38b), the resumptive pronoun them indicates the different grammatical function of the relativized noun phrase in the second to-infinitive clause: While in the first to-infinitive clause the relativized noun phrase serves prepositional complement function, in the second to-infinitive clause it serves direct object function (if you had time [to, to think aboutRELATIVIZED

NP: PCOMP {mm} and study themRELATIVIZED NP: DIRECT OBJECT]). After this intervening complex conditional clause, an additional resumptive pronoun they ensures the coreference relation between the antecedent and the distant relativized noun phrase and it indicates the subject function of the relativized noun phrase in the remainder of the relative clause (quite a few old sayings, and one thing and another, in the country that [if you had time [to, to think about {mm} and study them]], they’d have, they’d have done, gone by the board).

Page 91: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 81

Resumptive pronouns allow the construction of complex sentences by relativizing into otherwise inaccessible syntactic positions involving different levels of subordination. Thus, the presence of a mental prop in the shape of a resumptive pronoun gives rise to a type of complex relative clause in dialect which is not possible in Standard English (cf. Quirk et al.1985: 1299; cf. also Huddleston and Pullum 2002: 1091).

(39a) is an instance of self-correction: from a further embedded relative clause occurring in Standard English, in which the conjunction that is obligatorily omitted because the relative marker that assumes subject function in the relative clause (our own [thatSUBJECT we know THAT are])(cf. Quirk et al. 1985: 1298; cf. also Huddleston and Pullum 2002: 1083–1085; Comrie 1993: 16220; Comrie 1998: 65 and 80; van den Eynden 2002: 186 and 191), to a further embedded relative clause only occurring in nonstandard speech, namely one which shows relativization into an indirect question clause. The latter relative clause explicates the sense relation between the superordinate clause (we know) and the subordinate clause (how they are bred) by the conjunction how and spells out the subject of the ‘deep-embedded’ clause via a resumptive pronoun (they). While there is relativization across one clause boundary in (39a), there is relativization across two clause-boundaries in (39b) – first into an interrogative clause, then into a conditional clause with the resumptive pronoun them serving as prepositional complement:

(39) a. false start: Standard omission of conjunction that after weknow since relative marker that serves subject (passive) function (‘subject in partially object environment’ [cf. below]); however, this false start is abandoned for an interrogative clause after we know; new start (self-correction): starting with relative marker that; resumptive pronoun they is subject (passive) in a further embedded interrogative clause:

Yes, aye, or breed off our own [thatSUBJECT we know THATare] – [that we know [how they are bred]]. (CNO FRED Wes_002)

b. resumptive pronoun them in prepositional complement (object of preposition) function in further (first interrogative clause, second conditional clause) embedded conditional clause: ... When I went to school we had all sorts of games, [which the children of now wouldn’t know [what you were talking about [if you talked about them]]]. .... (CNO FRED Wes_019)

Page 92: A Comparative Grammar of Brtiish English Dialects

82 Tanja Herrmann

The examples in (40) to (42) exemplify resumptive pronouns in so-called ‘subject in partially object environments’ (cf. Quirk [1957] 1968: 101). That means, the relativized subject noun phrase in the ‘deep-embedded’ clause itself functions as the direct object of the superordinate clause and surfaces as a relative marker at the beginning of this complex structure:

(40) a. relative marker that in subject (passive) function is resumed as it (resumptive pronoun) after interjection you know; relative marker that introduces a nonrestrictive relative clause: ... He used to tell me about his country [that [you know] it was taken over by the Russians] and then it’d be taken over by the Germans and You know what I mean. (SCO-K6L<S: 064>)

b. further embedding of object clause after insertion you know of superordinate clause; conjunction that is obligatorily omitted due to subject function of relative marker that; instead, there is the resumptive pronoun it in subject (passive) function: ... He used to tell me about his country [that [you know][THAT it was taken over by the Russians]] and then it’d be taken over by the Germans and You know what I mean. (SCO-K6L<S: 064>)

c. relative marker that as direct object of you know; resumptive pronoun it has subject (passive) function in further embedded object clause: ... He used to tell me about his country [[that you know] [itwas taken over by the Russians]] and then it’d be taken over by the Germans and You know what I mean. (SCO-K6L<S: 064>)

Different ways of bracketing render different interpretations plausible: In (40a) you know is analyzed as an interjection after which the relative clause is continued; the relative marker that has subject function, as well as the resumptive pronoun it. This analysis of resumptive pronoun placement without further embedding would receive more support if there were some indication, like commas or dashes, that the speaker had paused before and after you know. In (40b) you know is taken to be an insertion in the relative clause on which the following object clause is dependent. The relative marker that and the resumptive pronoun it in the further embedded object clause assume subject function. In (40c) the relative marker that is felt to be the direct object of you know. The resumptive pronoun it in subject function is further embedded in the object clause.

Page 93: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 83

Syntax and/or semantics might lend support to the interpretations in variant (40b) or (40c). The analysis of example (40) advocated here is instantiated by variant (40b). In (41) below, an interpretation of the relative marker functioning as object is ruled out, since the superordinate clause hewas sure requires a prepositional complement instead of a direct object:

(41) relative marker which has subject function; he was sure is bracketed as an insertion; bracketing which as the direct object of he was sure is prohibited because he was sure demands a prepositional complement: he was sure of + PCOMP: ... And he come home one night, and he had two wee canaries,[which [he was sure] [was hens]]. ... (NIR NITCS A45.3)

Irrespective of the issue whether the relative marker functions as the subject of the relative clause or whether there is something ‘object-like’ about it, the “redundant” resumptive pronoun, which here is analyzed as an overt subject in the ‘deep-embedded’ object clause, makes these complex sentences more explicit, as in:

(42) a. resumptive pronoun it as overt subject in further embedded object clause: And they had a what they called the (UNCLEAR) twenty four girls dancing you know, and also they, they had the, a film in the show [which probably I think [it was a Bing Crosby film [that was on]]]. (SCO-GYW<S: 462>)

b. resumptive pronoun it as overt subject in further embedded object clause: I was the only one who stood. (SCO-GYW<S: 512>) In that second ward. (SCO-GYW<S: 513>)

Which was a ward [which I would say [it was built about the (trunc) ninet the eighteen seventies]]. (SCO-GYW<S: 514>)

In addition, nonstandard speech may disregard the prescriptive rule that the conjunction that has to be omitted if the relative marker serves subject function (cf. note 26), presumably to prevent that that be mistaken for the subject of the further embedded clause (cf. Quirk et al. 1985: 1050; cf. also Comrie 1999: 84–85). This is illustrated in example (43), taken from Quirket al. 1985:

Page 94: A Comparative Grammar of Brtiish English Dialects

84 Tanja Herrmann

(43) a. They pointed out the damage [which they supposed [had been done by last night’s storm]]. (Quirk et al. 1985: 1050 [emphasis and bracketing mine])

b. *They pointed out the damage [which they supposed [that had been done by last night’s storm]]. (Quirk et al. 1985: 1050 [emphasis and bracketing mine])

In example (44) below, the relative clause modifying the first one is constructed according to the rules of further embedded ‘subject in partially object environment’ relative clauses in Standard English. The second relative clause modifying the one appears to ignore two prescriptive rules at once: First, the relative clause seems to lack a subject, i.e. the zero relative marker is ungrammatical in subject function (cf. Quirk et al. 1985: 1250); second, the conjunction that is not omitted. While the occurrence of the zero relative marker as subject in such further embedded relative clauses is also noted for Standard English (cf. Quirk 1968: 102 for Standard British English; cf. also Huddleston and Pullum 2002: 1085; Graf 1996: 58–59 for Standard American English), it is regarded as an undesirable weak point in Standard English, which is “concealed by a push-down [i.e. further embedding] element” (Schmied 1993: 87). The presence of the conjunction that when the relative marker has subject function is ungrammatical in Standard English, but recorded in dialectal speech (e.g. van den Eynden 1993: 162 for Dorset), which has preserved a historically older, less restrictive way of using the conjunction that (cf. van den Eynden 1993: 224). Alternatively, if that were taken for a demonstrative pronoun in (44), the situation of a missing subject in the relative clause would be redeemed by the presence of a resumptive pronoun in subject position in the further embedded object clause.

(44) first relative clause: Relative marker that is counted as subject in ‘subject in partially object environment’; Herbie thought introduces a dependent, further embedded object clause in which a conjunction that is obligatorily omitted since the relative marker that functions as subject; second relative clause: The zero relative marker fills the subject position in ‘subject in partially object environment‘; when the relative marker has subject function, the presence of the conjunction that in an object clause is ungrammatical in Standard English, but not uncommon in dialect; if that were taken for a demonstrative pronoun instead of a vacuous

Page 95: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 85

conjunction, the relative clause would receive a subject in the shape of a resumptive pronoun: ... So I don’t know whether now I have two males instead of the, thefirst one [that Herbie thought [was a male]], we’re not so sure, but the one [ he thought [that wasn’t a male]], is a male. ... (NIR NITCS A45.3)

9.4. Nonreduction

A special variant of ‘resumptive’ is the nonreduction of the relativized noun phrase, that is, the antecedental noun phrase, or part of it, surfaces within the relative clause. Obviously, a nonreduced noun phrase is even more explicit than a resumptive pronoun or adverb:

(45) a. ... These are the points, [what we used to stand them on thepoints], yes. ... (CSW FRED Som_009)

b. And then they, the one where Johnny Weismuller was, it was alake in front of a great arena [which they did a lot of you know, swimming about in the lake and the different formations]. (SCO-GYW<S: 484>)

c. ... well then there used to be wire and used to have a big wheel in top, [which you couldn’t go over the top] and .... (EAN-H5G<S: 883>)

d. multiple relative clause: further embedded that relative clause consisting of two coordinated juncts linked via or, inside whichrelative clause: ... cos you got a radio tower on the dock now [which every ship [that come into port or leave the port] have to go through the radio tower] and .... (EAN-H5H<Person: PSEAN4><S: 425>)

e. stranded (or even fronted: at which there was sufficient room)preposition at was extracted from superordinate relative clause across one bracket into the further embedded to-infinitiveclause together with the nonreduced noun phrase: And of course the, with the boxes being all round the side, the dressing boxes, this was outmoded, we needed a building at theside, [which there was sufficient room at [to build something at

Page 96: A Comparative Grammar of Brtiish English Dialects

86 Tanja Herrmann

the side] [because we had a park at the side of it then]]. (SCO-GYW<S: 552>)

Nonreduced noun phrases should be subject to the same principles as resumptive pronouns. In other words, Prediction 1 and 2, set out for resumptive pronouns in 9.1., should hold for them, too: First, the frequency of occurrence of nonreduced noun phrases should reflect the Accessibility Hierarchy in the reverse order (GEN > PCOMP > DO > SUBJ). Second, nonreduced noun phrases should co-occur more often with relative particles than with relative pronouns. On the basis of only a handful of clear examples of nonreduced noun phrases, no valid interpretation can be made regarding these predictions. However, all examples relativize the prepositional complement position – a position located at the lower end of the Accessibility Hierarchy – which seems to lend support to Prediction 1. The nonreduced noun phrases fill the gaps created by relativizing on PCOMPs with stranded prepositions. Either the whole antecedental noun phrase re-emerges as a nonreduction or just part of the antecedent re-emerges: the head noun with its determiner in (45d), or the prepositional attribute in (45c). Examples (45d) and (45e) are complex relative clauses which involve another embedded relative clause, consisting of two coordinated relative clauses in (45d) and a further embedded to-infinitiveclause, followed by a causal clause in (45e). Thus, the distance between the relative marker and its governing (stranded) preposition is considerable, created by an intervening medial relative clause in (45d) and by an infinitival clause in (45e). The appearance of the redundant nonreduced noun phrase makes the laborious recovering process of the relativized noun phrase unnecessary and that of its grammatical function in the relative clause in such environments. In (45e), the stranded preposition at was extracted from its proper position at the end of the (superordinate) relative clause and moved into the further embedded infinitival clause together with its governed nonreduced noun phrase.

Four examples contain the relative pronoun which and one example the relative particle what, thus contradicting Prediction 2 that resumptive pronouns combine more often with invariant relative particles than with relative pronouns. This can, however, be attributed to the nonrestrictiveness of all examples. It is the cause of the application of the nonrestrictive relative markers which and what, on the one hand, and of the recourse to a resumptive in the form of a nonreduced noun phrase, on the other hand, as could be seen in 9.2. above.

Page 97: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 87

10. Which as ‘connector’?

In his investigation of Somerset English, Ihalainen (1980: 190) argues that which can function as a so-called ‘connector’ – not as a relative pronoun. In this function it “simply connects two statements” in cases in which there is no “obvious antecedent for which” (Ihalainen 1980: 190). For example:

(46) a. ... And, er, you had a great big chap up in between the hooves. Which the cow did go crippled. ... (CSW-SRLM 107<T 3004>)

b. Then we moved to the <gap cause=anonymization desc=address> and then I went to <gap cause=anonymization desc=address> Road School, which I finished my time there, fourteen. (CMI-H4B<S: 561>)

c. And of course the, with the boxes being all round the side, the dressing boxes, this was outmoded, we needed a building at the side, which there was sufficient room to build something at theside because we had a park at the side of it then. (SCO-GYW<S: 552>)

d. And they had addresses and I ran around, I went to get a job at the pit, which er I had no more sense so I’m glad now that I didn’t get on. (CMI-FXU<S:159>)

e. ... the one down Taunton left his pottery to his son, which all he wanted was drink. (CSW FRED Som_009)

In a similar vein, Miller and Weinert (1998: 110) remark in their study of spoken Lothian Scottish English that in these cases “[t]he function of whichis to signal a general link between the material that precedes it and the material following”. Instead of a relative pronoun, which is analyzed as a “general discourse linking element” or “complementizer” (Miller and Weinert 1998: 111). This analysis appears to receive support when the subsequent clause shows no grammatical gap, i.e. when which does not seem to have a grammatical function in the clause, as all syntactic positions are already taken by other noun phrases (cf. Miller and Weinert 1998: 110–111). Miller and Weinert observe parallel historical developments of thatand which, in the course of which the former relative pronoun which has caught up with that in having developed “a second function as general conjunction”.

Page 98: A Comparative Grammar of Brtiish English Dialects

88 Tanja Herrmann

Nevertheless, analyzing which as a general conjunction is unnecessary as well as unwarranted. The element of (post)modification is still strongly felt in these cases – even if the relative marker cannot readily be assigned a grammatical function in the clause at first sight. As will be shown below, in all alleged ‘connector’ instances which introduces either a nonrestrictive adnominal relative clause or a sentential relative clause, which is nonrestrictive by nature. The relative pronoun which is the nonrestrictive relative marker par excellence and the only available option in sentential relative clauses. Besides, all instances of alleged ‘connector’ which involve nonpersonal antecedents, except for one instance in text CSW- FRED Som_009: Example (46e) above contains a personal antecedent his son. In section 6 it was demonstrated that which does not exclusively modify personal antecedents, although this is its highly preferred environment. In the Central Southwest and in the idiolect of one CSW speaker in particular, which also combines with personal antecedents. These occurrences of personal which are clearly instances of adnominal relative clauses:

(47) a. ... And then there was C., [which caught his hand in the machinery up here] and he had his hand off, being severed (at the) wrist. ... (CSW FRED Som_009)

Accordingly, which does not seem to have parted with its properties as a relative pronoun but still seems to operate in its favorite syntactic environment, i.e. in nonrestrictive relative clauses modifying nonpersonal antecedents. A close look at all putative examples of ‘connector’ whichreveals that they can be analyzed more satisfactorily as instances of either preposition elision, resumptives, or new starts.

10.1. Preposition elision

The following instances can be explained as cases of preposition elision:

(48) a. formerly example (46e): stranded preposition ‘by’ elided (preposition ‘by’ could also have been elided in fronted position: [from/by] Which the cow did go crippled); relative pronoun which functions as adverbial (A) instead of prepositional complement in adverbial function (PCOMP(A)):

Page 99: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 89

... And, er, you had a great big chap up in between the hooves. [Which the cow did go crippled [from/by]]. ... (CSW-SRLM 107<T 3004>)

b. multiple relative clause: relative clause embedded in relative clause; preposition ‘with’ is elided; stranded preposition ‘with’is of questionable acceptability in the position above, but after the second coordinate noun phrase any haulage it would cause an obstruction between the antecedent any haulage and the following zero relative clause: Carriages ......... well, my father had two horses [which he used to do the local emptying of dustcarts and any haulage [with] [ there was to be done in the village]]; (preposition ‘with’ could also have been elided in fronted position: [with] which he used to do the local emptying of dustcarts and any haulage [ there was to be done in the village]); problematic positioning of preposition might have enforced preposition elision; relative pronoun which functions as A instead of PCOMP(A): Carriages ......... well, my father had two horses [which he used to do the local emptying of dustcarts [with] and any haulage [ there was to be done in the village]]. .... (CNO FRED Wes_018)

c. preposition elision of stranded ‘to’ in the face of a second preposition (off) which precedes ‘to’; relative pronoun whichfunctions as indirect object (IO) instead of prepositional complement functioning as object of preposition (PCOMP(OP)):So we were really answerable to the Ipswich Borough Council, rather than to private enterprise [which some people really wanted to sell us off [to]] as being a, you know, a weight round their necks because if we didn’t make a lot of money after the war, ... (EAN-HDL<S: 277>)

Examples like (48a–c) are the result of ‘preposition elision’. Which was supposed to function as a prepositional complement (PCOMP) with a stranded preposition. (Recall that preposition stranding is the rule in dialectal speech). However, stranded prepositions, which are separated from their governing relative markers, are often “forgotten”. Elision of (stranded) prepositions is a recurrent dialectal feature (also in nominal relative clauses; compare example (10a) in section 6) as well as a typical

Page 100: A Comparative Grammar of Brtiish English Dialects

90 Tanja Herrmann

feature of spontaneous spoken language. Preposition elision occurs with various relative markers and also outside relative clauses, for instance in to-infinitive clauses (cf. Miller and Weinert 1998: 106 and 108 for spoken Lothian Scottish English; cf. also Miller 1993: 112). Preposition elision is frequently found after massive and/or complex intervening material (as in example [48b]) and after another preposition (as in example [48c]). If the governing preposition is missing, which cannot function as a prepositional complement (PCOMP), of course. Instead which assumes the functions adverbial (A) or indirect object (IO), whereas with a preposition whichwould function as PCOMP(A) or PCOMP(OP), respectively.

Elision of prepositions is not limited to relative clauses introduced by which. It also occurs with other relative markers:

(49) elided stranded preposition ‘with’ after another preceding preposition in:And never in my wildest dream did we imagine that Labour would get in with the resounding majority [that they did get in with].(SCO-K6M<S: 010>)

10.2. Resumptives: Resumptive pronouns (or adverbs) and nonreduced noun phrases

The following examples of ‘connector’ which can be explained as instances of resumptives:

(50) a. formerly example (46b): For clarity’s sake, preposition ‘at’ is portrayed as having been elided in fronted position (preposition ‘at’ would rather have been elided in stranded position: which I finished my time [at]); instead of a preposition, resumptive demonstrative adverb thereunambiguously indicates the grammatical function of relative pronoun which; which functions as A instead of PCOMP(A):

Then we moved to the <gap cause=anonymization desc= address> and then I went to <gap cause=anonymization desc=address> Road School, [[at] which I finished my time there], fourteen. (CMI-H4B<S: 561>)

b. resumptive personal pronoun us after stranded partitive genitive preposition of; relative pronoun which and resumptive

Page 101: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 91

pronoun in PCOMP(GEN) function; (alternative analysis: Existential there is resumptive pronoun in subject position): So after a fortnight, thereabouts, the union had a meeting and decided that everybody in the industry like, [which of course there was only about a thousand of us any way], everybody would (trunc) s would stop until this man could start at work.(CMI-FYH<S: 556>)

c. sentential relative clause: relative marker which and resumptive pronoun that in subject function; (less preferable analysis: thatcould also be analyzed as a vacuous conjunction in the Middle English relative marker relic which that [cf. 9.1. above]): Because once the ship come up that’s still a certain amount of water in the hold [which that must be], cos then once you heave your door up then of course you load your ship again and ... (EAN-H5H<S: 684>)

d. conflation of sentential relative clause modifying bought thisand adnominal relative clause modifying this: slightly reconstructed version of: which to get I were lucky; resumptive pronoun it as direct object in further embedded to-infinitiveclause saves the sentence from ambiguity: ... They sold this and some at Cary and I jumped in and boughtthis, [which I were lucky in a way [to get it]], you know, had it, being a tenant you did get it cheaper, you see? (CSW-SRLM 105<T 1420>)

e. formerly example (46c): nonreduced noun phrase in PCOMP(A) function in further embedded to-infinitive clause: And of course the, with the boxes being all round the side, the dressing boxes, this was outmoded, we needed a building at the side, [which there was sufficient room [to build something at the side] [because we had a park at the side of it then]].(SCO-GYW<S: 552>)

Examples (50a–d) involve resumptive pronouns (or adverbs), while (50e) involves a nonreduced noun phrase. Due to the redundant nature of resumptives, the respective grammatical function is filled twice in a relative clause: once by the relative marker and a second time by the resumptive. Thus, the relative pronoun which is not deprived of its grammatical function in the relative clause by the presence of a resumptive, but the

Page 102: A Comparative Grammar of Brtiish English Dialects

92 Tanja Herrmann

resumptive steps in to explicate its grammatical function in some syntactic environments rather than in others (cf. 9.1. above). For instance:

1. In nonrestrictive adnominal and sentential relative clauses, which are difficult and rather unknown to dialect speakers (compare 9.2. above).

2. In cases of preposition elision, as in (50a): Instead of a preposition, the resumptive adverb there is inserted, which has taken over the task of unambiguously indicating the adverbial locative function in the normal clause position.

3. In inaccessible relativization environments, such as in (50b) and (50d): In (50b), relative pronoun which relativizes the PCOMP position, while in (50d), relative pronoun which relativizes a position in a further embedded clause (cf. 9.3. above).

In example (50e), inaccessibility of position (PCOMP) in terms of the Accessibility Hierarchy combines with further embedding (specifically, a further embedded to-infinitive clause).

Furthermore, resumptives are found with all kinds of relative markers (cf. Table 13 above). Yet the zero marker, that, what, and who are not analyzed as ‘connectors’ – instead of real relative markers – when they take a resumptive. For example:

(51) resumptive personal pronoun them in prepositional complement position, co-occurring with relative marker that:... it was, you know, looked upon then you were, were public transport and the public team [that you belonged to them]. (EAN-HDL<S: 101>)

10.3. New starts

The last group of examples of so-called ‘connector’ which can be explained as new starts:

(52) a. formerly example (46d): sentential relative clause: abandoned after relative marker which; pause filler er, followed by new parataxis with its own subject (I):

Page 103: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 93

And they had addresses and I ran around, I went to get a job at the pit, [which] er I had no more sense so I’m glad now that I didn’t get on. (CMI-FXU<S:159>)

b. sentential relative clause: abandoned after relative marker which; pause filler well, followed by new parataxis: I don’t really know, but er they managed to keep us looking nice and clean and tidy er because one respected teachers and elders [which], well, I don’t think I’ve ever changed anyway, but today they don’t]. (CMI-FXV<S: 090>)

c. formerly example (46e): adnominal relative clause (e.g. whichonly wanted (to) drink): abandoned after relative pronoun which in subject function, in favor of new start with all-pseudo-cleft:... the one down Taunton left his pottery to his son, [which all

he wanted was drink]. (CSW FRED Som_009)

Examples (52a–c) belong to a third category which comprises abandoned relative clauses after the relative marker which. A new start is made using a paratactic construction, often preceded by a filler element like ‘er’, ‘erm’,well, etc. and/or punctuation (like a comma in [52b]) to indicate a pause. Both elements are indicative of reorientation in sentence planning (cf. 9.1. above). Consequently, this type of ‘connector’ which is due to an involuntary midclause reconstruction, which proves the dominance of parataxis over hypotaxis, particularly in difficult and unfamiliar environments, such as nonrestrictive adnominal and (nonrestrictive) sentential relative clauses. Dialect speakers tend to start the nonrestrictive relative clause as a subject relative clause, which is the most basic type of relative clause (cf. Keenan and Comrie 1977: 95).

In contrast to object relative clauses, subject relative clauses demand no change of the normal SVO word order. According to the ‘relative marker fronting rule’ (cf. Downing 1977: 181), the relative marker in subject function is already in front position. Relativization of a prepositional complement with a stranded or elided preposition also avoids disrupting the normal (paratactic) SVO word order; the fronting of the relative marker in PCOMP position corresponds to the peripheral clause position of PCOMPs in declarative main clauses, where prepositional complements can precede or succeed the central SVO group. This may explain why most so-called ‘connector’ which examples either relativize nonrestrictive subjects or (primarily adverbial) prepositional complements. (Some) dialect speakers

Page 104: A Comparative Grammar of Brtiish English Dialects

94 Tanja Herrmann

plan to construct a nonrestrictive relative clause, but owing to online production constraints and unfamiliarity with nonrestrictives as well as its structural markers, i.e. wh-pronouns, these relative clauses are finished in an unconventional nonstandard way.

11. Conclusion

The relativization strategies found in the dialectal data are grammatically less constrained than those of Standard English. Often they represent earlier stages of English, containing grammatical and syntactic features which are disfavored or banned from Present-day Standard English. While traditional dialects use a pure relative particle, wh-pronouns are a hallmark of (written) Standard English. Hence, the infiltration of dialects by Standard English can be read off from the number of wh-pronouns in this dialect, and particularly from the number of case-marked wh-pronouns, since grammatical concord is generally absent from traditional dialects. In the present data, relative particles prevail over relative pronouns at a ratio of 3:1, whereas case-marked relative pronouns are very scarce. In contrast to Standard English, preposition stranding is generally preferred in the dialectal data, even with wh-pronouns. There are further supra-regional characteristics of traditional dialect: The relative marker which also refers to personal antecedents; nonrestrictive relative clauses may also be introduced by relative particles like what, that, and zero; zero subject relative clauses are legitimate; resumptive pronouns and nonreduced noun phrases are used for clarification, partially enabling complex constructions unknown in Standard English; and prescriptivisms like the omission of the conjunction that when the relative marker serves subject function in further-embedded relative clauses (push-down elements) do not apply.

However, as traditional dialects are declining in favor of a common nonstandard variety, the truly vernacular relative markers like as are receding: As has retracted from East Anglia and the Central Southwest. Nowadays as is still retreating toward the interior of the country, while its number has decreased dramatically since the days of the compilation of the Survey of English Dialects (in the 1950s). Even in the (Central) Midlands, where as used to be the dominant relative marker, as has diminished drastically in numbers. At the same time, as has backed out of several syntactic environments like nonrestrictive relative clauses and is moving toward relativizing only the subject (that is, the topmost) position on the

Page 105: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 95

Accessibility Hierarchy, while it has largely withdrawn from the lower positions. As is superseded by what, the novel general nonstandard relative marker, to the mutual exclusion of the two, as they both compete for the whole range of syntactic environments. What is spreading across Britain geographically, both numerically and in terms of syntactic environments. It has developed from a typical East Anglian relative marker toward a supra-regional nonstandard marker which may eventually find its way into spoken Standard English. Due to its affinity to informal non-posh spoken language and thus, to simpler syntactic positions, what enters a (new) region by the topmost position of the Accessibility Hierarchy (subject position) and percolates down the Hierarchy. If it were found governed by a fronted preposition (e.g. the man [for what Jack worked]), this would even signal the beginning of a change in word class, from conjunction (relative particle) to pronoun (relative pronoun). In addition, the relative particles that (or its phonemic variant ‘at) and zero proves to be prominent in each investigated dialect. While that increases its number as one moves farther north, zero appears to be independent of region.

The Accessibility Hierarchy was slightly modified for the present study: SUBJ > DO > PCOMP > GEN. This Accessibility Hierarchy is confirmed for all relative clause formation strategies in each of the investigated regions, with some reservations for the zero relative marker strategy. In contrast to Standard English, zero can relativize the subject position in all dialects, yet only in two of the broad corpora (Northern Ireland and Central North) does zero relativize most often on subjects. However, by its sheer presence in SUBJ in four regions and its supremacy in the remaining two, the zero marker strategy is rehabilitated as a primary relative clause formation strategy, as the gap in subject position on the Accessibility Hierarchy left by Standard English is remedied in dialectal English.

Relating the Accessibility Hierarchy (formulated on the basis of restrictive relative clauses) to the pronoun retention strategy yields unexpected results when including instances of resumptive pronouns in nonrestrictive relative clauses: First, instead of reversing the Accessibility Hierarchy (GEN > PCOMP > DO > SUBJ), resumptive pronouns appear most often in subject position, followed by direct object, prepositional complement, and genitive position. Regarding the ease of retrievability of the relativized noun phrase and the identification of the syntactic function of the relative marker, relative particles are less explicit than relative pronouns because they lack grammatical concord and case-marking. Nevertheless, instead of co-occurring with less explicit relative particles,

Page 106: A Comparative Grammar of Brtiish English Dialects

96 Tanja Herrmann

resumptives rather combine with more explicit relative pronouns, particularly which. The predominance of which is caused by the nonrestrictiveness of these examples, as which is the nonrestrictive relative marker par excellence. (The few restrictive relative clauses indeed select a relative particle.) Recalling that resumptives tend to occur in difficult syntactic environments and bearing in mind that nonrestrictive relative clauses are characteristic of written Standard English, but uncommon or even alien to traditional dialect speech, it becomes apparent that dialect speakers apply resumptives when moving on unfamiliar ground, such as nonrestrictive relative clauses. Dialect speakers use the principal nonrestrictive relative marker which, seeking the help of a resumptive to construct a nonrestrictive relative clause, which they construct in the most basic way possible, that is by relativizing the subject position. In other words, the reverse order of the Accessibility Hierarchy in the pronoun retention strategy is not necessarily transferable from restrictive relative clauses, for which the Accessibility Hierarchy was originally set up, to nonrestrictive relative clauses, in which resumptives mainly occur in the present data.

‘Connector which’ turns out to be an artifact to explain nonstandard constructions of which relative clauses. Such examples are the product of online production and analyzable as instances of preposition elision, resumptives, and new starts, which are recurrent in spontaneous speech. In all instances, the modifying character inherent in relative clauses is still pre-eminent.

Page 107: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 97

Appendix 1

Abbreviations: + = reported, - = reported as not occurring, GEN = genitive, poss pro = possessive pronoun

Table 1. Areal distribution of relative markers in previous investigations on the Central Southwest and Eastern Somerset in particular

Central Southwest author/study zero that what as who which whose, whom Elworthy (1877, [1886] 1965): West Somerset

+ + + only in East

Somer-set

- - –; GEN avoidance

Lowsley (1888):Berkshire

+ no whom;whosen

Dartnell and Goddard(1893):Wiltshire

+

Kruisinga(1905): West Somerset

+/ [hAt] + no whose;GEN

avoidance

Wright([1905] 1961): East Somerset

occurs gener-ally used

Wilson (1913):Burley, New Forest, Hampshire

+

Lowman (1937/38):EasternSomerset

+ as + poss pro

Page 108: A Comparative Grammar of Brtiish English Dialects

98 Tanja Herrmann

Table 1. cont.

Central Southwest author/study

zero that what as who which whose, whom

SED (1950s):EasternSomerset

occurs occurs/ at occurs occurs (older)

domi-nant

whose;as +posspro;

what’s;GEN

avoid-ance

Barth(1968):Naunton,Glou-cester-shire

+ (newer)

+(older)

Rogers(1979):Wessex

+ + + -; whathis + posspro)

Ihalainen(1980,1985,1987):Somerset

+[29.1%]

+[25.2%]

+[17.5%]

+[5.8%] in East

Somerset only

+[1.9%]

+

[20.4%]1-

Jones and Dillon(1987):Wiltshire

+ + + +

van den Eynden (1992,1993,2002):Dorset

+[24.9%]

+[37.1%]

+[7.1%]

+[13.9%]

+[11.6%]

+[0.6%]2

;zero + posspro;GEN

avoid-ance

Page 109: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 99

Table 2. Areal distribution of relative markers in previous investigations on East Anglia and Suffolk and Eastern Cambridgeshire in particular

East Anglia zero that what as who which whose,author/ study whom Forby ([1830] 1970)

predomi-nant

Wright ([1905] 1961)

occurs general-ly used

Lowman (1937/38):Suffolk, East Cambridgeshire

+ + what +poss pro;

that’s

SED (1950s): Suffolk, East Cambridgeshire

occurs occurs once as

at

dominant occurs only in Cam-

bridge-shire

occurs whose;as + poss

pro;GEN

avoidance

Francis SED Corpus [1956] in Poussa (1999, 2001): Northern Norfolk (Nf 1–9)

+[40%]

+[2.5%];

at [1.2%]

+[47%]

+[8%]

+ [1.7%] no whom

Claxton (1968) + hardly whom

Ojanen (1982): (South)Cambridgeshire

+[39.5%]3

+ [1.6%] + [53.3%]

+ [0.2%] + [3%]

+[2.3%]

-;GEN

avoidance

Kekäläinen/Peitsara (1985, 1988;20024): Suffolk

+[24.3%] [30.7%]5

+ / at[20.4%] [23%]

+[29.1%] [21.2%]

+ [1.9%][0.9%]

+[2.9%] [7.4%]

+[21.4%] [16.9%]

-;who +

poss pro

Poussa (1994): North-EastNorfolk

+[31%]

+ / at[7.9%]

+[25.6%]

+[19.4%]6

+[16.2%]

very rare

Poussa (2001): Docking, North-West Norfolk

very frequent

rare frequent occurs occurs -

Page 110: A Comparative Grammar of Brtiish English Dialects

100 Tanja Herrmann

Table 3. Areal distribution of relative markers in previous investigations on the Central Midlands and Nottinghamshire in particular

Central Midlands author/study zero that what as who which whose,

whomEvans and Evans (1881): Leicestershire

+

Pegge (1896): Derbyshire

+

Wright ([1905] 1961):Nottinghamshire

occurs generally used

SED (1950s): Nottinghamshire

occurs + whose; as +poss pro

Storr (1977): Selston, Erewash Valley, Nottinghamshire

+

Page 111: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 101

Table 4. Areal distribution of relative markers in previous investigations on the Central North (Cumbria = Cumberland, Westmorland, and North Lancashire)

Central North author/study zero that what as who which whose,

whomPeacock ([1862/63] 1968a, [1867] 1968b):Westmorland,CentralCumberland, North Lancashire

at

Ellwood ([1895] 1965): Cumberland, Westmorland,North Lancashire

at

Wright ([1905] 1961):Westmorland

occa-sional

Hirst (1906): Kendal,Westmorland

occurs at -

Brilioth (1913): Lorton, West Cumberland

occurs often

commonest:at

much less

com-mon

-

Reaney (1927): Penrith,Cumberland

frequent commonest: at

occa-sional

Round (1949): Broughton-in-Furness, Lancashire

often commonest: at

some-times

SED (1950s): Cumberland,Westmorland,North Lancashire

occurs +at is

dominant;ut; t

occurs once in NorthLanca-shire

occurs once in Cum-

berland;occurs in North Lanca-shire

oc-curs once

whose; at+ poss

pro, at’s;as his;GEN

avoidance

Wright (1979) some-times

usual:at

occa-sional

Page 112: A Comparative Grammar of Brtiish English Dialects

102 Tanja Herrmann

Table 5. Areal distribution of relative markers in previous investigations on Scotland and Lothian, Borders, Strathclyde, and Invernesshire in particular

Scotland author/study zero that what as who which whose, whom Wright ([1905] 1961)

at

Wilson (1915): LowerStrathearn District of Perthshire

+ ut + with‘all’

no whose but ut + poss pro; GEN avoidance

Grant and Dixon (1921)

+ + / ‘at / ‘t

‘at + poss pro; hardly whom

Wilson (1926): CentralScotland

+ + / ut - - -

Dieth (1932): Buchan,Aberdeenshire

at

Grant (1931); Grant and Murison(1974)

+ + / at rare at + poss pro / that’s

Murison([1977] 1980)

+ + / at - obsolete whilk

-; that + poss pro

Aitken (1979) no whose but that’s;GEN avoidance

Romaine(1982a, b, 1984)

+ + / ‘at rare + no whose but that’s;GEN avoidance

Page 113: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 103

Table 5. cont.

Scotland author/study

zero that what as who which whose, whom

Miller and Brown(1982);Miller(1983);Miller and Weinert (1998):Edinburgh,West and EastLothian

+ typical rare + no whosebut that +poss pro; no whom

Macafee (1983):Glasgow

+ most common

occa-sional in West of Scotland

Macaulay (1985,1989,1991): Ayr

secondfrequent

domi-nant

very rare

very rare no whosebut GEN avoidance

McClure (1994)

+ + / at that’s / at’s

Page 114: A Comparative Grammar of Brtiish English Dialects

104 Tanja Herrmann

Table 6. Areal distribution of relative markers in previous investigations on Northern Ireland

Northern Ireland

author/study

zero that what as who which whose, whom

Wright([1905] 1961):Ireland

at

Todd(1971):Tyrone

+ +

Policansky (1982):Belfast

most fre-

quent

fre-quent

occurs occurs no whose;zero +

poss pro; GEN

avoidanceHarris(1984,1993):IrishEnglish(North and South)

+ + rare rare that &zero +

poss pro; that’s

Henry (1995):Belfast

+

Corrigan(1997):SouthArmagh

most fre-

quentamongolder

people

+ compara-tively rare

compara-tively rare

GENavoidance

Page 115: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 105

Appendix 2

Map 1. S5 (LAE) (SED Question IX.9.5 who)

Page 116: A Comparative Grammar of Brtiish English Dialects

106 Tanja Herrmann

Map 2. M81 (LAE) (SED Question IX.9.6 whose)

Page 117: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 107

Map 3. S 8a: III.3.7 that (The computer developed linguistic atlas of England 1)

Page 118: A Comparative Grammar of Brtiish English Dialects

108 Tanja Herrmann

Map 4. S 8b III.3.7 that (The computer developed linguistic atlas of England 1)

Page 119: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 109

Map 5. S 9 IX.9.5 who (The computer developed linguistic atlas of England 1)

Page 120: A Comparative Grammar of Brtiish English Dialects

110 Tanja Herrmann

Map 6. S 10 IX.9.6 whose (The computer developed linguistic atlas of England 1)

Page 121: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 111

Notes

1. I would like to thank Juhani Klemola for kindly giving me permission to use his SRLM material.

2. I would like to thank John Kirk for kindly allowing the English Department at the University of Freiburg to work with the NITCS.

3. Percentages were rounded to one digit after the point, which might cause occasional incongruence with the total sum of percentages.

4. Alternatively, the continuation with his father could also be interpreted as an analytical genitive, i.e. as a combination of the zero relative marker + the possessive pronoun his.

5. Alternatively, the continuation with his father could also be interpreted as an analytical genitive, i.e. as a combination of the zero relative marker + the possessive pronoun his.

6. Based on Poussa’s (1988, 1991) investigations of the three SED questions on relative clauses, relative what seems to originate from Essex (cf. Poussa 1988: 448). In addition, Albrecht (1916: 136) in his treatise on the dialect poet Charles E. Benham of Colchester in north Essex, records what ( p) as the major relative marker, alongside as.

7. The results of the SED appeared in several publications: The original four Survey of English Dialects (B) Basic Material volumes, edited by Harold Orton et al., were published between 1962 and 1971. Dividing England into four large geographical areas, the volumes reproduce the SED questions and responses, including incidental material (i.e. additional, nonelicited information on the questions), according to counties. The linguistic atlas of England (LAE) (1978), edited by Orton, Sanderson, and Widdowson,presents selected SED questions as maps on which isoglosses are drawn as well as individual responses indicated. Wolfgang Viereck’s comprehensive The computer developed linguistic atlas of England (1991, 1997) illustrates SED questions and responses, including incidental material, in great detail. It lists individual responses and marks them via symbols on the maps. The Survey of English Dialects: The dictionary and grammar (1994), edited by Clive Upton et al., is an exhaustive reference work which arranges SEDquestions and responses according to feature occurrences in individual counties.

8. Occasionally, informants gave no answer or more than one answer. 9. (In)definiteness, however, did not prove to be a critical variable, neither in

investigations based on SED material nor in the present investigation, but tends to be a by-product of the variables (non)restrictiveness and (non)specificity, for which reason the variable (in)definiteness was dropped in the course of this analysis.

Page 122: A Comparative Grammar of Brtiish English Dialects

112 Tanja Herrmann

10. E.g., We used to have niggers go along in front of us. [bare infinitive clause] We had a combine come here ‘way from Holland. [causative resultative construction (We made a combine come here ‘way from Holland.)] (cf. Ojanen 1982: 77).

11. In A Middle English syntax (1960: 191), Mustanoja remarks on at: “At occurs by the side of that in the North, where it is common in the 14th and 15th centuries. It becomes rare after 1500, being supplanted by quhilk”.

12. For details on licensed borderline cases of personal which, please see Quirk etal. 1985: 1245–1246, 1260, and 367; cf. also Huddleston and Pullum 2002: 1048.

13. Mair found eight definitely human antecedents among some 90 downloaded examples when looking for the partitive genitive ‘of which’ in the British National Corpus (personal communication). For instance:

The hereditary sultans (of which there are only ...) ... (BNC-FEP: 426) ... the very special women, of which there are few, ... (BNC-GV8: 717).

14. ‘Lexically empty’ antecedents involve semantically bleached noun phrases like ‘anybody’, ‘somebody’, ‘the best’, ‘the only thing’, ‘the only ones’, ‘the one’, ‘the one thing’, ‘them’, etc., being relatively close to focus constructions semantically. Quirk et al. (1985: 1387) use the term "general antecedent" instead of ‘lexically empty’ antecedent.

15. For details, please see Quirk et al. 1985: 1387; cf. also Huddleston and Pullum 2002: 1046 and 1056–1057.

16. Peitsara (2002: 174) also calls them ‘cleft sentences’, but unfortunately includes them in her analysis of relative clauses nevertheless.

17. Compare also to Ihalainen’s (1980: 189 and 191; 1985: 68–69) findings. 18. The symbol > in X > Y means ‘Y implies X’. 19. In Middle English, the frequency of zero subject relative clauses was much

higher and they were not constrained in use (cf. Moessner 1984: 71). 20. “In English, non-subjects of subordinate clauses are freely relativizable, while

subjects can be relativized only if there is no conjunction: the girl [that you think (that) I love] the girl [that you think (*that) loves me]” (Comrie 1993: 162).

21. Total might include up to three sentential relative clauses. 22. Plus 4.8% sentential relative clauses and so-called ‘connector which clauses’

to add up to 100%. 23. It contains instances which share a common surface structure with zero

relative clauses, yet are not relative clauses. 24. Her 1985 study was based on the data of only nine of the 19 informants, while

her 2002 analysis exploited the whole of the Suffolk data. 25. The figures of her 2002 analysis are slightly re-worked here, such as to

exclude adverbial relative clauses, sentential relative clauses, and clefts;

Page 123: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 113

however, there might still be an unknown number of inappropriate constructions in the data.

26. The actual absolute number might differ slightly, since Poussa does not distinguish between who and its case-marked forms; on page 425 she records three whose and one whom, but her Figure 1 shows 2 who in genitive function.

References

Aitken, A. J. 1979 Scottish speech: a historical view, with special reference to the

Standard English of Scotland. In Languages of Scotland, A. J. Aitken and Tom McArthur (eds.), 85–119. Edinburgh: W & R Chambers.

Albrecht, Theodor 1916 Der Sprachgebrauch des Dialektdichters Charles E. Benham zu

Colchester in Essex. (Palaestra CXI.) Berlin: Mayer & Müller. Ball, Catherine N. 1994 Relative pronouns in it-clefts: The last seven centuries. Language

Variation and Change 6: 179–200. Barth, Ernst 1968 The Dialect of Naunton (Gloucestershire). Zürich: P. G. Keller. Bever, T. G. and D. T. Langendoen 1971 A dynamic model of the evolution of language. Linguistic Inquiry 2

(4): 433–463. 1972 The interaction of speech perception and grammatical structure in

the evolution of language. In Linguistic Change and Generative Theory. Essays from the UCLA Conference on Historical Linguistics in the Perspective of Transformational Theory. February 1969,Robert P. Stockwell and Ronald K. S. Macaulay (eds.), 32–95. Bloomington/London: Indiana University Press.

Biber, Douglas, Stig Johansson, Geoffrey Leech, Susan Conrad, and Edward Finegan

1999 Longman Grammar of Spoken and Written English. Harlow: Pearson Education.

Brilioth, Börje 1913 A Grammar of the Dialect of Lorton (Cumberland). Historical and

Descriptive. Inaugural diss., Uppsala: Philosophical Faculty of Uppsala.

Page 124: A Comparative Grammar of Brtiish English Dialects

114 Tanja Herrmann

Cheshire, Jenny, Viv Edwards, and Pamela Whittle 1993 Non-standard English and dialect levelling. In Real English. The

Grammar of English Dialects in the British Isles, James Milroy and Lesley Milroy (eds.), 53–97. London/New York: Longman.

Chevillet, François 1996 Relativization strategies in Early Middle English (and after). In

Speech Past and Present. Studies in English Dialectology in Memory of Ossi Ihalainen, Juhani Klemola, Merja Kytö, and Matti Rissanen (eds.), 18–32. Frankfurt am Main / Berlin / Bern / New York / Paris / Wien: Peter Lang.

Claxton, A. O. D. 1968 The Suffolk Dialect of the 20th Century. 3rd ed. Ipswich: Norman

Adlard.Comrie, Bernard 1993 Reprint. Language Universals and Linguistic Typology. Syntax and

Morphology. 2nd ed. Oxford/Cambridge: Blackwell. Original edition, 1989.

1998 Rethinking the typology of relative clauses. Language Design 1: 59–86.

1999 Relative clauses. Structure and typology on the periphery of standard English. In The Clause in English. In Honour of Rodney Huddleston,Peter Collins and David Lee (eds.), 81–91. Amsterdam/Philadelphia: John Benjamins.

Comrie, Bernard and Edward L. Keenan 1979 Noun phrase accessibility revisited. Language 55: 649–664. Corrigan, Karen Patrice 1997 The syntax of South Armagh English in its socio-historical

perspective. Ph.D. thesis, University College Dublin. Dartnell, George Edward and Edward Hungerford Goddard 1893 A Glossary of Words Used in the County of Wiltshire. (English

Dialect Society 69.) Dekeyser, Xavier 1984 Relativicers [sic] in early Modern English: A dynamic quantitative

study. In Historical Syntax, Jacek Fisiak (ed.), 61–87. Berlin/New York/Amsterdam: Mouton.

Dieth, Eugen 1932 A Grammar of the Buchan Dialect (Aberdeenshire). Descriptive and

Historical. Vol. I: Phonology – Accidence. Cambridge: W. Heffer & Sons.

Downing, Bruce T. 1977 Typological regularities in postnominal relative clauses. In Current

Themes in Linguistics: Bilingualism, Experimental Linguistics and

Page 125: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 115

Language Typologies, Fred R. Eckman (ed.), 163–194. Washington/London: Hemisphere.

Edwards, Viv K., Peter Trudgill, and Bert Weltens 1984 The Grammar of English Dialect. A Survey of Research. London:

Economic and Social Research Council. Ellwood, T. 1965 Reprint. Lakeland and Iceland Being. A Glossary of Words in the

Dialect of Cumberland, Westmorland and North Lancashire. Vaduz: Kraus. Original edition (English Dialect Society 77), 1895.

Elworthy, Frederic Thomas 1877 An Outline of the Grammar of the Dialect of West Somerset

Illustrated by Examples of the Common Phrases and Modes of Speech Now in Use among the People. (English Dialect Society 19.)

1965 Reprint. The West Somerset Word-Book. A Glossary of Dialectal and Archaic Words and Phrases Used in the West of Somerset and East Devon. Vaduz: Kraus. Original edition (English Dialect Society 50), 1886.

Evans, Arthur Benoni and Sebastian Evans 1881 Leicestershire Words, Phrases, and Proverbs. (English Dialect

Society 31.) Forby, Robert 1970 Reprint. The Vocabulary of East Anglia. Vol. 1. Newton Abbot,

Devon: David & Charles. Original edition, 1830. Fox, Barbara A. 1987 The noun phrase accessibility hierarchy reinterpreted: Subject

primacy or the absolutive hypothesis? Language 63: 856–870. Gass, Susan M. and Larry Selinker 1994 Second Language Acquisition. An Introductory Course. Hillsdale,

New Jersey/Hove/London: Lawrence Erlbaum. Givón, Talmy 1993 English Grammar. A Function-Based Introduction. Vol. II.

Amsterdam/Philadelphia: John Benjamins. Graf, Dorothee 1996 Relative clauses in their discourse context. A corpus-based study.

M.A. thesis, University of Freiburg. Grant, William (ed.) 1931 The Scottish National Dictionary. Vol. I. Edinburgh: The Scottish

National Dictionary Association. Grant, William and James Main Dixon 1921 Manual of Modern Scots. Cambridge: University Press.

Page 126: A Comparative Grammar of Brtiish English Dialects

116 Tanja Herrmann

Grant, William and David D. Murison (eds.) 1974 The Scottish National Dictionary. Vol. IX. Edinburgh: The Scottish

National Dictionary Association. Harris, John 1984 Syntactic variation and dialect divergence. Journal of Linguistics 20:

303–327. 1993 The grammar of Irish English. In Real English. The Grammar of

English Dialects in the British Isles, James Milroy and Lesley Milroy (eds.), 139–186. London/New York: Longman.

Hawkins, John A. 1999 Processing complexity and filler-gap dependencies across grammars.

Language 75: 244–285. Henry, Alison 1995 Belfast English and Standard English. Dialect Variation and

Parameter Setting. New York/Oxford: Oxford University Press. Hirst, T. O. 1906 A Grammar of the Dialect of Kendal (Westmoreland). Descriptive

and Historical. Heidelberg: Winter. Huddleston, Rodney 1984 Introduction to the Grammar of English. London: Cambridge

University Press. Huddleston, Rodney and Geoffrey K. Pullum 2002 The Cambridge Grammar of the English Language. Cambridge/New

York/Melbourne/Madrid/Cape Town: Cambridge University Press. Hyltenstam, Kenneth 1984 The use of typological markedness conditions as predictors in

second language acquisition: The case of pronominal copies in relative clauses. In Second Languages: A Crosslinguistic Per-spective, R. Andersen (ed.), 39–58. Rowley, MA: Newbury House.

Ihalainen, Ossi 1980 Relative clauses in the dialect of Somerset. Neuphilologische

Mitteilungen 81: 187–196. 1985 Synchronic variation and linguistic change: Evidence from British

English dialects. In Papers from the 4th International Conference on English Historical Linguistics, Amsterdam, 10–13 April, 1985,Roger Eaton, Olga Fischer, Willem Koopman, and Frederike van der Leek (eds.), 61-72. Amsterdam/Philadelphia: John Benjamins.

1987 Towards a grammar of the Somerset dialect. A case study of the language of J.M. In Neophilologica Fennica. Société Néophilologique 100 Ans, Leena Kahlas-Tarkka (ed.), 71–86. Helsinki: Société Néophilologique.

Page 127: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 117

Jacobsson, Bengt 1994 Nonrestrictive relative that-clauses revisited. Studia Neophilologica

66: 181–195. Jespersen, Otto 1927 A Modern English Grammar on Historical Principles. Part III:

Syntax, 2nd vol. London/Copenhagen: George Allen & Unwin/Ejnar Munksgaard.

Johansson, Christine 1993 Whose and of which with non-personal antecedents in written and

spoken English. In Corpus-Based Computational Linguistics, Clive Souter and Eric Atwell (eds.), 97–117. Amsterdam/Atlanta, GA: Rodopi.

Jones, Malcolm and Patrick Dillon 1987 Dialect in Wiltshire and its Historical, Topographical and Natural

Science Contexts. Trowbridge, Wiltshire: Wiltshire County Council. Keenan, Edward L. 1985 Relative clauses. In Language Typology and Syntactic Description.

Vol. II: Complex Constructions, Timothy Shopen (ed.), 141–170. Cambridge: Cambridge University Press.

Keenan, Edward L. and Bernard Comrie 1977 Noun phrase accessibility and universal grammar. Linguistic Inquiry

8 (1): 63–99. Keenan, Edward L. and Sarah Hawkins 1987 The psychological validity of the accessibility hierarchy. In

Universal Grammar: 15 Essays, Edward L. Keenan (ed.), 60–85. London/Sydney/Wolfeboro, New Hampshire: Croom Helm.

Kekäläinen, Kirsti 1985 Relative clauses in the dialect of Suffolk. Neuphilologische

Mitteilungen 86: 353–357. Kruisinga, E. 1905 A Grammar of the Dialect of West Somerset. Descriptive and

Historical. Bonn: P. Hanstein. Lehmann, Christian 1984 Der Relativsatz: Typologie seiner Strukturen, Theorie seiner

Funktionen, Kompendium seiner Grammatik. Tübingen: Narr. Lowsley, B. 1888 A Glossary of Berkshire Words and Phrases. (English Dialect

Society 56.) Lumsden, Michael 1988 Existential Sentences. Their Structure and Meaning. London/New

York/Sydney: Croom Helm.

Page 128: A Comparative Grammar of Brtiish English Dialects

118 Tanja Herrmann

Macafee, Caroline 1980 Characteristics of non-standard grammar in Scotland. Manuscript. 1983 Glasgow. Amsterdam/Philadelphia: John Benjamins. Macaulay, Ronald K. S. 1985 The narrative skills of a Scottish coal miner. In Focus on: Scotland,

Manfred Görlach (ed.), 101–124. Amsterdam/Philadelphia: John Benjamins.

1989 ‘He was some man him’: Emphatic pronouns in Scottish English. In Synchronic and Diachronic Approaches to Linguistic Variation and Change, Thomas J. Walsh (ed.), 179-187. Washington, D. C.: Georgetown University Press.

1991 Locating Dialect in Discourse. The Language of Honest Men and Bonnie Lasses in Ayr. New York/Oxford: Oxford University Press.

Mair, Christian 1998 Man/woman which ... – Last of the old, or first of the new? In

Explorations in Corpus Linguistics, Antoinette Renouf (ed.), 123–133. Amsterdam: Rodopi.

Maxwell, Daniel N. 1979 Strategies of relativization and NP accessibility. Language 55: 352–

371.McClure, J. Derrick 1994 English in Scotland. In The Cambridge History of the English

Language. Vol. V: English in Britain and Overseas: Origins and Development, Robert Burchfield (ed.), 23–93. Cambridge/New York/Melbourne: Cambridge University Press.

Miller, Jim 1993 The grammar of Scottish English. In Real English. The Grammar of

English Dialects in the British Isles, James Milroy and Lesley Milroy (eds.), 99–138. London/New York: Longman.

Miller, Jim and Keith Brown 1982 Aspects of Scottish English syntax. English World-Wide 3: 3–18. Miller, Jim and Regina Weinert 1998 Spontaneous Spoken Language. Syntax and Discourse. Oxford:

Clarendon Press. Moessner, Lilo 1984 Some English relative constructions. La Linguistique 20 (1): 57–79. Mossé, Fernand 1991 Reprint and translation. A Handbook of Middle English.

Baltimore/London: The Johns Hopkins University Press. Original edition, 1952.

Page 129: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 119

Murison, David 1980 Reprint. The Guid Scots Tongue. Edinburgh: William Blackwood.

Original edition, 1977. Mustanoja, Tauno F. 1960 A Middle English Syntax. Part I: Parts of Speech. Helsinki: Société

Néophilologique. Nevalainen, Terttu and Helena Raumolin-Brunberg 2002 The rise of the relative who in Early Modern English. In

Relativisation on the North Sea Littoral, Patricia Poussa (ed.), 109–121. München: Lincom Europa.

Newbrook, Mark 1997 Relative THAT, that isn’t always restrictive. I.T.L. Review of

Applied Linguistics 115–116: 39–60. Ojanen, Anna-Lisa 1982 A syntax of the Cambridgeshire dialect. Licentiate thesis, University

of Helsinki. Orton, Harold, Michael V. Barry, Eugen Dieth, Wilfrid J. Halliday, Philip M.

Tilling, and Martyn F. Wakelin (eds.) 1962–71 Survey of English Dialects. Leeds: Arnold. Orton, Harold, Stewart Sanderson, and John Widdowson 1978 The Linguistic Atlas of England. London: Croom Helm. Peacock, Robert Backhouse 1968a Reprint. A Glossary of the Dialect of the Hundred of Lonsdale,

North and South of the Sands, in the County of Lancaster, Together with an Essay on Some Leading Characteristics of the Dialects Spoken in the Six Northern Counties of England (Ancient Northumbria). (Transactions of the Philological Society.) Original edition, 1867.

1968b Reprint. On Some Leading Characteristics of Northumbrian; and on the Variations in its Grammar from that of Standard English, with their Probable Etymological Sources. (Transactions of the Philological Society: 232–264.) Original edition, 1862/63.

Pegge, Samuel 1896 Two Collections of Derbicisms. (English Dialect Society 78.) Peitsara, Kirsti 1988 On existential sentences in the dialect of Suffolk. Neuphilologische

Mitteilungen 89: 72–100. 2002 Relativizers in the Suffolk dialect. In Relativisation on the North Sea

Littoral, Patricia Poussa (ed.), 167–180. München: Lincom Europa. Policansky, Linda 1982 Grammatical variation in Belfast English. Belfast Working Papers in

Language and Linguistics 6: 37–66.

Page 130: A Comparative Grammar of Brtiish English Dialects

120 Tanja Herrmann

Poussa, Patricia 1988 The relative WHAT: two kinds of evidence. In Historical

Dialectology. Regional and Social, Jacek Fisiak (ed.), 443–474. Berlin/New York: Mouton de Gruyter.

1991 Origins of the non-standard relativizers what and as in English. In Language Contact in the British Isles. Proceedings of the Eighth International Symposium on Language Contact in Europe, Douglas, Isle of Man, 1988, P. Sture Ureland and George Broderick (eds.), 295–315. Tübingen: Max Niemeyer.

1994 Norfolk relatives (Broadland). In Verhandlungen des internationalen Dialektologenkongresses, Bamberg 29.7.–4.8. 1990. Regional-sprachliche Variation, Umgangs- und Standardsprachen. Vol. 3, Wolfgang Viereck (ed.), 418–426. Stuttgart: Franz Steiner.

1996 Gender-specific occupations and speech among North-East Norfolk working-class informants born 1899–1915. In IX International Oral History Conference, Göteborg 13–16th June, 1996. Vol. 3: People’s History; Political and Popular Movement, 524–533. From a pre-printed collected conference papers.

1999 The Flemings in Norman Norfolk: Their possible influence on relative pronoun development. In Thinking English Grammar. To Honour Xavier Dekeyser, Professor Emeritus, Guy A. J. Tops, Betty Devriendt, and Steven Geukens (eds.), 89–99. Leuven/Paris: Peeters.

2001 Syntactic change in north-west Norfolk. In East Anglian English,Jacek Fisiak and Peter Trudgill (eds.), 243–259. Cambridge: D. S. Brewer.

Quirk, Randolph 1968 Reprint. Essays on the English Language. Medieval and Modern.

London/Harlow: Longmans. Original edition, 1957. Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech, and Jan Svartvik (eds.) 1985 A Comprehensive Grammar of the English Language. London/New

York: Longman. Reaney, Percy H. 1927 A Grammar of the Dialect of Penrith (Cumberland). Descriptive and

Historical. Manchester: University Press and London/New York: Longmans, Green & Co.

Rogers, Norman 1979 Wessex Dialect. Bradford-on-Avon, Wiltshire: Moonraker Press.

Romaine, Suzanne 1980 The relative clause marker in Scots English: Diffusion, complexity,

and style as dimensions of syntactic change. Language in Society 9: 221–247.

Page 131: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 121

1982a Socio-Historical Linguistics. Its Status and Methodology.Cambridge: Cambridge University Press.

1982b The reconstruction of language in its social context. Methodology for a socio-historical linguistic theory. In Papers from the 5th International Conference on Historical Linguistics, Anders Ahlqvist (ed.), 293–303. Amsterdam/Philadelphia: John Benjamins.

1984 Some historical and social dimensions of syntactic change in Middle Scots relative clauses. In English Historical Linguistics. Studies in Development, N.F. Blake and Charles Jones (eds.), 101–122. Sheffield: The Centre for English Cultural Tradition and Language, University of Sheffield.

Round, Inabel G. 1949 The living dialect of Broughton-in-Furness in the county of

Lancashire. M.A. thesis, University of Leeds. Schmied, Josef 1993 Qualitative and quantitative research approaches to English relative

constructions. In Corpus-Based Computational Linguistics, Clive Souter and Eric Atwell (eds.), 85–96. Amsterdam/Atlanta, GA: Rodopi.

Seppänen, Aimo and Göran Kjellmer 1995 The dog that’s leg was run over. On the genitive of the relative

pronoun. English Studies 26: 389–400. Shnukal, Anna 1981 There’s a lot mightn’t believe this .... Variable subject relative

pronoun absence in Australian English. In Variation Omnibus, David Sankoff and H. Cedergren (eds.), 321–328. Carbondale, IL: Linguistic Research.

Simpson, J. A. and E. S. C. Weiner 1989 Oxford English Dictionary, R. W. Burchfield (ed.). Vols. 1–20. 2nd

ed. Oxford: Clarendon Press. Smith, Evan Shreeve 1984 Relative ‘that’ and ‘as’: A Study in Category Change. Ph.D. thesis,

Indiana University, 1982. University Microfilms International, Ann Arbor, MI.

Stockwell, Robert P., Paul Schachter, and Barbara Hall Partee 1973 The Major Syntactic Structures of English. New York: Holt,

Rinehart and Winston. Storr, Jeremy George 1977 Survey of the dialect of Selston in the Erewash Valley. M.A. thesis,

University of Sheffield.

Page 132: A Comparative Grammar of Brtiish English Dialects

122 Tanja Herrmann

Tagliamonte, Sali 2002 Variation and change in the British relative marker system. In

Relativisation on the North Sea Littoral, Patricia Poussa (ed.), 147–165. München: Lincom Europa.

Todd, Loreto 1971 Tyrone English. Transactions of the Yorkshire Dialect Society 13:

29–40.Trudgill, Peter 2003 Reprint. The Dialects of England. 2nd ed. Malden, MA/Oxford, UK:

Blackwell. Original edition, 1999. Upton, Clive, David Parry, and J. D. A. Widdowson (eds.) 1994 Survey of English Dialects: The Dictionary and Grammar.

London/New York: Routledge. Van den Eynden (Morpeth), Nadine 1992 Relativization in the Dorset dialect. In History of Englishes. New

Methods and Interpretations in Historical Linguistics, Matti Rissanen, Ossi Ihalainen, Terttu Nevalainen, and Irma Taavitsainen (eds.), 532–555. Berlin/New York: Mouton de Gruyter.

1993 Syntactic Variation and Unconscious Linguistic Change. A Study of Adjectival Relative Clauses in the Dialect of Dorset. Frankfurt am Main/Berlin/Bern/New York/Paris/Wien: Peter Lang.

2002 Relativisers in the Southwest of England – with special emphasis on subject contact clauses, preposition placement with particle and pronominal relative strategies and pushdown relatives. In Relativisation on the North Sea Littoral, Patricia Poussa (ed.), 181–194. München: Lincom Europa.

Viereck, Wolfgang 1975 Lexikalische und grammatische Ergebnisse des Lowman-Survey von

Mittel- und Südengland. 2 vols. München: Wilhelm Fink. 1980 The dialectal structure of British English: Lowman’s evidence.

English World-Wide 1: 25–44. Viereck, Wolfgang (ed.), in collaboration with Heinrich Ramisch 1991 The Computer Developed Linguistic Atlas of England. Vol. 1.

Tübingen: Niemeyer. Wilson, James 1913 The Dialect of the New Forest in Hampshire (as Spoken in the

Village of Burley). London: Oxford University Press/Amen Corner. 1915 Lowland Scotch as Spoken in the Lower Strathearn District of

Perthshire. London: Oxford University Press/Humphrey Milford. 1926 The Dialects of Central Scotland. London: Oxford University

Press/Humphrey Milford.

Page 133: A Comparative Grammar of Brtiish English Dialects

Relative clauses in dialects of English 123

Wright, Joseph 1961 The English Dialect Grammar. Oxford: Oxford University Press.

Original edition, 1905. Wright, Peter 1979 Cumbrian Dialect. Clapham: Dalesman Books.

Page 134: A Comparative Grammar of Brtiish English Dialects
Page 135: A Comparative Grammar of Brtiish English Dialects

“Some do and some doesn’t”: Verbal concord variation in the north of the British Isles

Lukas Pietsch

Abstract

Among the chief characteristics of the northern dialects since Middle English times has been the so-called Northern Subject Rule, a systemic split in the verbal concord system which allows for invariant verbal -s forms everywhere except when the verb is directly accompanied by a simple personal pronoun. This study provides a geographical and comparative survey of the reflexes of this pattern in the northern dialects, drawing attention to their variability and to their interaction with other related and/or competing patterns of concord variation. A corpus investigation reveals that over and above the ‘hard’ constraints that define the Northern Subject Rule as such, there exist a number of ‘soft’ probabilistic constraints governing its effects which are also near-universally shared between the varieties in question. I then go on to discuss the likely paths of historical development that have given rise to this grammatical pattern, and critically review some attempts that have been made to account for it in terms of formal syntactic theories. I show that existing formal models fail to account for the range of variability of this pattern, both in a comparative, diatopic perspective and on the level of individual speakers. I finally argue that variation phenomena of this kind can theoretically be better accounted for in a usage-based model in the vein of current functionalist and emergentist theories.

1. Introduction

The verbal concord system is an area of particularly rich regional variation in English. Several quite distinct regional types of non-standard agreement systems exist in the traditional dialects. Some of them are also well attested historically and can be traced as far back as the Middle English period. Today these patterns coexist, and sometimes compete, with other types of non-standard agreement forms, which have spread through the modern vernaculars. Some of these even appear to have the status of ‘vernacular universals’ (Chambers 2004).

Page 136: A Comparative Grammar of Brtiish English Dialects

126 Lukas Pietsch

Within this complex field of variation, one pattern stands out as particularly interesting, both from a historical and from a theoretical perspective: the so-called Northern Subject Rule. According to this rule, the Standard English contrast between verbal -s in the third person singular and zero forms elsewhere is observed only where the subject is one of the closed set of simple personal pronouns: he/she/it goes; I/you/we/they go.All other subjects can take an invariant -s form of the verb. Moreover, the agreement contrast in the pronominal subjects is found consistently only when the subject directly precedes or follows the verb; in other cases, even I, you, we and they may take the -s form. Agreement contrasts are thus organized not or not exclusively along the lines of the subject’s person-number features, but instead they are sensitive to the morphological type and syntactic position of the subject. This intriguing system has been among the chief grammatical characteristics of the dialects of northern Britain, including Scots, since the Middle English period. It is also found in Irish English, above all in the varieties of Ulster. Some of its reflexes, especially with regard to the usage of was and were, can be found in overseas vernacular varieties all over the English-speaking world (cf. for instance Montgomery 1988, 1997, Tagliamonte 2002). The latter varieties must remain outside the scope of the present paper, which will concentrate on those varieties where the pattern originated historically, namely those of northern Britain, as well as their immediate geographical neighbour and offspring, the dialects of Ulster.

Verbal concord variation in English nonstandard varieties has received some more attention in past research than many other phenomena of grammatical, especially syntactic, variation. This is certainly due to the fact that it is a high-frequency phenomenon, whose occurrences in texts can relatively easily be identified and counted. It has been given some treatment in traditional dialectology, at least occasionally (e.g. Murray 1873; Wright 1892, 1905; Orton et al., ed. 1962–1971). The Northern Subject Rule, in particular, is routinely mentioned in dialect descriptions, though often only in a summary fashion (Beal 1993, 1997, 2004; Miller and Brown 1982; Miller 1989, 1993, 2004; Macafee 1983, 1994; Harris 1993; Ihalainen 1994; Robinson 1997, Filppula 1999, Shorrocks 1999). Some of the major historical patterns of verbal concord have also been documented for older varieties (e.g. Macafee 1993, McIntosh, Samuels and Bensik 1986). Recently, verbal concord variation has become a stock-in-trade of corpus-based studies in variation and change. Some of them have been based on diachronic corpora (e.g. Meurman-Solin 1992, Kytö 1993,

Page 137: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 127

Montgomery 1994, Ogura-Wang 1996), while most studies used synchronic local corpora from individual speech communities, either historical (e.g. Bailey and Ross 1988, Bailey, Maynor and Cukor-Avila 1989, Montgomery, Fuller and DeMarse 1993; Schendl 1996, 2000, Wright 2002, McCafferty 2003) or contemporary (e.g. Tagliamonte 1999, Smith and Tagliamonte 1998, Godfrey and Tagliamonte 1999, Britain 2002, Schreier 2002, Peitsara 2002).

The present chapter reports on a new corpus-based study (Pietsch 2003, 2005) which attempts to complement this existing research in two ways. With respect to empirical description, a mostly diatopic-comparative perspective was chosen. Being based on dialectal speech recordings sampled over relatively large areas, the study traced the distribution of related variation patterns through space and also (in some parts) through the apparent-time dimension. The aim was to identify commonalities and differences in the linguistic conditioning of inherent, quantifiable variation across related varieties. The main result is that – over and above the defining constraints constituting the NSR itself – there exist a number of recurrent types of constraints which act as conditioning factors on concord variation. They can be detected statistically in the form of probabilistic effects, and they can best be described in terms of prototypical syntactic environments which idiomatically favour a certain morphological realization of the verb over the other.

The second contribution lies in an attempt to forge a closer link between this descriptive, empirical work on the one hand and recent grammatical theorizing on the other. Most of the existing studies of verbal concord have treated the phenomenon of quantifiable variation mainly under its societal (in Chomskyan terms: “E-language”) aspects. On the other side, those among the studies of dialect grammar that have been inspired by theoretical concerns about individual linguistic competence (“I-language”) have tended to concentrate on ‘variation’ in the sense of ‘differences between lects’, but have rarely addressed the issue of variability within lects (see papers in Black and Motapayane 1996 and Barbiers, Cornips and van der Kleij 2002; for a notable recent exception see Henry 2002). There are only a few studies offering specific analyses regarding a formal theoretical characterization of concord phenomena of the northern type (Henry 1995, Corrigan 1997, Börjars and Chapman 1998, Hudson 1999), and none of them addresses inherent variability. I will therefore propose a fresh attempt to relate the empirical observations regarding variable grammatical performance to the theoretical question of how variation is anchored in

Page 138: A Comparative Grammar of Brtiish English Dialects

128 Lukas Pietsch

grammatical competence. In doing so, I will seek explanations in a theoretical framework inspired by functionalist and cognitive research traditions (Bybee 1985, Bybee and Hopper 2001, Langacker 1987; Croft 1995, 2001).

This chapter is organized as follows. Section 2 gives a descriptive survey of the phenomena in question, and raises some of the problems involved in an adequate synchronic analysis. Section 3 reports on empirical corpus findings with respect to the patterns of quantitative variation and their geographical distribution as found in a range of twentieth-century dialects. Section 4 presents an excursus into the history of the northern concord pattern and the hypothetical diachronic processes of its first emergence in early Middle English. Section 5 presents a critical discussion of several theoretical proposals that have been made to account for the effects of the Northern Subject Rule in modern dialects. Finally, Section 6 presents a concluding discussion and an outlook on possible theoretical solutions to the problems presented in the earlier sections.

2. The Northern Subject Rule: Descriptive problems

The most concise descriptive definition of the phenomenon in question can be given as follows:

(1) The Northern Subject Rule (Version A): concord verbs1 take the -sform with all subjects, except with the personal pronouns I, we, youand they when they are directly adjacent to the verb.

However, this is only a somewhat idealized statement, describing a hypothetical, “pure” northern concord system. Indeed, varieties that come reasonably close to this have been identified, for instance in Older Scots and northern Middle English (Montgomery 1994; McIntosh, Samuels and Bensik 1986). In these older varieties, verbal -s in fact occurred with near-categorical regularity in all environments where the rule licensed it. In Modern English varieties, however, the system is always a variable one. It has therefore been customary in the literature since Montgomery (1994) to describe the variation patterns observed in terms of two separate constraints, the first of them most often called the Type-of-Subject Constraint, the second variously Position-of-Subject Constraint, Sequence Constraint, Proximity-to-Subject Constraint or similarly.

Page 139: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 129

(2) The Northern Subject Rule (Version B): a. All third singular subjects (and, where preserved, the old

second singular thou) always take verbal -s. b. The Type-of-Subject Constraint: All other subjects except

the personal pronouns I, we, you, they (and, where it exists, youse) take verbal -s variably.

c. The Position-of-Subject Constraint: Non-adjacency of subject and verb favours verbal -s.

Condition (2c) may apply, in principle, to all types of subjects. This means, on the one hand, that the prohibition of verbal -s with I, we, you and theymay be overridden if the verb and the pronoun are not adjacent. With respect to the other environments, on the other hand, it means that there is commonly a quantitative effect further increasing the likelihood of verbal -s. However, the nature of this constraint is difficult to define exactly. Some of the effects in question appear to be cross-dialectal universals that can be observed even in varieties not directly affected by the Northern Subject Rule. There seems to be a strong tendency in many dialects of English that clauses that diverge from the canonical structure, of a syntactically simple subject immediately followed by the verb, may display lack of agreement. For instance, it has often been observed that complex subjects consisting of two conjoined singular noun phrases may trigger an invariant third singular verb form, as in the following Early Modern English example (Visser 1963: 80):

(3) I and my company was arrested ij days at Dunckyrke.

Another environment that seems to stand out is subject-verb inversion, for instance in questions and under locative inversion. Non-agreement in the latter type of clause seems to have been common in English already during the Old English period (4). The same has been true for almost all forms of English in the case of existential there clauses, a special clause type that developed through grammaticalization out of the more general schema of the locative inversion. Non-agreement in this type of clause is found quite independently of the Northern Subject Rule (5).

(4) On þæm selfan hrægle wæs eac awriten þa naman ðara twelf heahfædra (‘On that same garment was also written the names of the twelve patriarchs’) [Ælfred, C.P. 6,15, quoted after Visser 1963: 73]

Page 140: A Comparative Grammar of Brtiish English Dialects

130 Lukas Pietsch

(5) There is two or three lords and ladies more married [Shakespeare, Mids. IV, II, 16, quoted after Visser 1963: 74]

Interestingly, it will be found that just among those modern dialects that otherwise follow the Northern Subject Rule in allowing many non-standard verbal -s forms, there are some that exhibit variation in exactly the opposite direction with respect to the existentials: here, they allow non-standard, seemingly plural verb forms even with singular subjects.

Whereas concord variation in existentials is thus largely independent of the Northern Subject Rule, the situation is different with subject-verb inversion in questions. This type of clause is neatly integrated with the general pattern of the Northern Subject Rule in the northern dialects. Thus, under interrogative inversion, full noun phrases are clearly distinguished from pronominal subjects just as in canonical subject-verb order. Whereas full noun phrase subjects in inversion may trigger a fairly strong effect in favour of verbal -s, even more so than full noun phrase subjects in canonical position, inverted pronoun subjects faithfully follow the Type-of-Subject Constraint, displaying agreement no less regularly than in other positions. The northern dialects thus regularly display forms such as havethey, not *has they. Indeed, it will be argued in section 4 that exactly these environments, of verbs with inverted adjacent pronouns, may in fact represent the historical core and point of origin of the Northern Subject Rule as a whole, and that the occurrence of the forms without -s in these environments has always been one of its central features.

Relative clauses, especially those following clefting it’s or existential there, are another environment where non-agreement is often observed, and these relative clause environments will be found to play a major role for the dialects affected by the Northern Subject Rule too.

A case that is problematic for the statement of the Northern Subject Rule is the one where subject and verb stand in the canonical order of the declarative clause but are separated from each other by intervening, clause-internal material. In Modern English, this may apply to either adverbs or so-called floating quantifiers, as in I often go or they all go. The Position-of-Subject Constraint as stated above predicts forms like I often goes or they all goes. Indeed, such forms occur, but only in the most ‘purely’ northern, older varieties such as Old Scots do they reach a high amount of regularity:

Page 141: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 131

(6) a. …that we lely heichtis and grantis… [Montgomery 1994: 89] b. we all hes mater to thank God… [Montgomery 1994: 89]

In present-day dialects such as Northern Irish English (cf. section 3.3.4 below), similar forms are also attested but much rarer:

(7) a. Oh never, they never was so strict, at that time, anyway[NITCS: L10.2]

b. And they, they both was yoked onto it. [NITCS: L19.3]

In other dialects such as Yorkshire and Lancashire English, similar forms with adverbs like often, never, always are fairly common, but it has been argued (Shorrocks 1999: 112, 116–117) that this usage is governed not so much by the syntactic environment but by the temporal semantics of the adverb, as verbal -s in these varieties also functions as a marker of habituality.

The last remaining type of environment where verbs typically occur in positions non-adjacent to their pronominal subjects is found in co-ordinated verb phrases. The second and any subsequent members of a series of verbs sharing the same subject will regularly take –s in typical northern dialects. Clauses of this type, they sing and dances, are often quoted as prototypical instances of the Northern Subject Rule. Indeed, of all the subtypes of Position-of-Subject effects, this appears to be the one that is most characteristic and specific to the northern dialects.

Summing up, it may be said that whereas the Type-of-Subject Constraint has been a stable and fairly unmistakeable feature of northern dialects, the Position-of-Subject Constraint may be regarded as the composite effect of several different patterns, of varying degrees of regularity, only some of which are specifically characteristic of the northern dialects whereas others are shared with many varieties elsewhere. Only in those older dialects which were least affected by standardizing influences from the south or other similar dialect contact or levelling effects, can it be said that the Position-of-Subject Constraint was a unified, tightly integrated feature of a consistent grammatical system.

Page 142: A Comparative Grammar of Brtiish English Dialects

132 Lukas Pietsch

3. Data from twentieth-century northern dialects

3.1. Data and methods

In the following sections, I will report on empirical findings regarding the quantitative variation patterns and the geographical distribution of reflexes of the northern concord system and some related phenomena in twentieth- century dialects. These data are based in part on the Survey of English Dialects (SED, Orton et al., ed. 1962–1971); partly on the Northern Ireland Transcribed Corpus of Speech (NITCS, Kirk 1991); and partly on a subset of a preliminary version of the Freiburg English Dialect Corpus (FRED;see Kortmann and Wagner, this volume). These data are of rather different kinds, and caution must be exercised in interpreting them in order to make results commensurable. Nevertheless, taken together they do give a fairly comprehensive picture of a range of grammatical varieties across a large geographic area.

The NITCS is a corpus of some 230,000 words, collected across a geographically regular grid of 38 mostly rural locations in Northern Ireland, sampling (ideally) one speaker from each of three age groups (9–12, 35–45, and 65–75) from each location. It is based on unscripted interviews conducted in the context of a dialect atlas project, the Tape-Recorded Survey of Hiberno-English Speech (TRS), during the 1970s (cf. Barry 1981b). In representing speech from different age groups, this project went an important step beyond the traditional design of dialect atlas surveys. Hence, it can be used for analyses not only along a diatopic but also along a diachronic, apparent-time dimension. For the purposes of this study, some 17,000 tokens of clauses with concord verbs were extracted and tagged. Obviously, many of these – actually, their great majority – are not of much interest for this study, as they exemplify environments that display no or only marginal variation. In general, this goes for all third person singular tokens, and for collocations with immediately adjacent personal pronoun subjects. The main focus of the analysis therefore were the approximately 1,000 tokens of third-person plural verbs with non-pronominal subjects. A separate study was made of existential there clauses(approx. 2,400 tokens). Tokens were classified for a range of different environment variables (such as syntactic constellation or morphological type of subject, in addition to the obvious variables of person and number). Multivariate analyses were then conducted, using the well-known Varbrulsystem, examining these intra-linguistic context variables as well as a

Page 143: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 133

number of social, extra-linguistic variables such as age, sex, and religious denomination.

These data, which provide for a fairly detailed view of the linguistic situation within the small geographical area of Northern Ireland, were complemented with data from FRED, illustrating a much wider geographical range of varieties. The subcorpus of FRED selected for the present study consisted of approximately 300,000 words of transcribed speech from 63 informants. It represented Scottish – predominantly Lowlands – speakers as well as different regions within the north of England. Most of the texts were conversations recorded for purposes of local ‘oral history’ projects. They were recorded between the 1970s and the 1990s, and typically consist of interviews between a fieldworker and an informant, centring about topics such as life in the old days, working conditions, war experiences and so forth. The informants were typically elderly people, of predominantly working-class background. Owing to the interview situation, speech styles tended to be relatively formal, but the speech nevertheless displayed a considerable range of local non-standard linguistic features. On the whole, these recordings are thus fairly similar in style to the interviews with the older age group in the NITCS. The Scottish recordings exemplify different speech styles along the continuum between broad Lowlands Scots and Standard Scottish English; similarly, the English recordings range from strongly local, containing conservative dialectal relic forms, to something fairly close to the standard. To these texts were added six additional recordings from the TRS, which for technical reasons could not be included in the NITCS. These are from Counties Donegal, Leitrim, and Louth. While these data, taken together, represent a much larger area than those of the NITCS, they evidently lack the systematicity of sampling characteristic of the latter, which means that they cannot be used with the same degree of reliability for investigations of social or fine-grained geographical variation. Nevertheless, multivariate analyses of this material revealed highly interesting patterns that matched or complemented findings from the NITCS in a number of ways.

To add a further dimension of real-time depth, as well as more fine-grained geographical information, the corpus data were supplemented with data from the Survey of English Dialects (SED). Obviously, these cannot be analysed with the same quantitative methods and in direct comparison with corpus data of the former kind. The SED data, collected during the 1950s, consist of isolated lists of elicited tokens, not exhaustive records of all forms produced in natural speech, and hence do not lend themselves to a

Page 144: A Comparative Grammar of Brtiish English Dialects

134 Lukas Pietsch

quantitative assessment of intra-speaker variation. Nevertheless, the SEDdata provides valuable evidence for the geographical distribution of some morphosyntactic variables in the traditional dialects, and, if interpreted with the necessary caution, also for some of their quantitative aspects. The SEDhas repeatedly been used for studies of subject-verb agreement (Orton, Sanderson and Widdowson 1978; Viereck 1991/1997; Trudgill 1990); Ihalainen 1991, 1994; Klemola 1996, 2000; Bresnan and Deo 2001; Wright 2002; and Britain 2002). However, none of these studies has fully exhausted its possibilities with regard to a study of the Northern Subject Rule and related phenomena.

Most of the relevant questionnaire items in the SED deal with the morphology of the primary verbs be, have and do. Some others deal with the agreement morphology of lexical verbs after various pronouns. Only two questions are specifically aimed at morphosyntactic environments for the Northern Subject Rule, namely, lexical verbs after full noun phrase subjects (III.10.7 ‘bulls bellow’, and VIII.7.5 ‘burglars steal them’). A number of other questions were primarily aimed not at morphosyntactic but at various lexical, phonological or idiomatic targets, but nevertheless provide material that contains tokens of present-tense subject-verb combinations which can be included in the analysis. As for most other questions eliciting verbs as answers, the material unfortunately does not regularly include subject forms and is therefore unusable.

Apart from these, the SED fieldworkers recorded a large number of additional utterances produced spontaneously by the informants during the interviews, whenever they felt these utterances illustrated interesting dialect features. These recorded tokens are known as “incidental material” in the SED. In cases where the features illustrated by these utterances were also the topic of one of the systematic questionnaire items, incidental material evidence was sometimes included in the published “Basic Material” of the SED (Orton et al. 1962–1971, henceforth SED-BM). However, this coverage of the incidental material is far from complete. For instance, the list of incidental tokens of plural verbal -s given under questionnaire item iii.10.7 (‘bulls bellow’) is mostly – though not quite consistently – restricted to tokens that match the elicited grammatical context in a rather narrow sense: subject NPs headed by lexical nouns in canonical, declarative SV clauses. Some of the most interesting grammatical environments with respect to the Northern Subject Rule are therefore missing: relative clauses, verbs after plural demonstrative and indefinite pronouns, to name but a few. Many incidental material tokens that illustrated verbal -s in these

Page 145: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 135

environments are either scattered across a large number of other headings in SED-BM, or have up to now not been accessible at all. Thus, a wealth of additional data is still enclosed in the original hand-written fieldworker notebooks and has never become available in publications. For this reason, the geographical picture that has so far been derivable from published SEDdata alone is in some respects distorted.

To remedy this situation, it was necessary to return to a study of the original fieldworker notebooks, held at Leeds University Library. This investigation was conducted for 139 of the 311 SED locations, covering the northern half of England roughly down to the Chester-Wash line and somewhat beyond. All incidental material tokens illustrating subject-verb agreement were excerpted from the notebooks, converted from the original phonetic to an orthographic transcription, and later collated with the data found under various headings in SED-BM. (The resulting token lists, which form the basis for the analyses in the following sections, can be found in an appendix to Pietsch 2005) They include approximately one thousand tokens of plural verbal -s related to the Northern Subject Rule, and another thousand tokens representing various other related phenomena.

In the following sections, I will first give a brief overview of the geographical distribution of several variation phenomena related to verbal concord in the north. I will then give a more detailed account of each of them in turn, starting with some phenomena of minor importance, proceeding through the special areas of was/were variation and of existential there clauses, and finally dealing with the reflexes of the Northern Subject Rule proper, found in the shape of plural verbal -s with non-pronominal subjects.

3.2. An overview

For the situation in the traditional dialects of England, a good overview can be gained from the data in the SED. It can be summed up as follows (cf. Map 1). There is a central northern area covering the three northern counties of Cumberland, Westmorland, Durham, and the southern half of Northumberland.2 They are characterized by a fairly consistent, homogeneous subject-verb agreement system, which conserves many features of northern Middle English. Its most important features are: preservation of the thou versus you distinction, with thou always taking verbal -s forms; generalized use of is as the singular present tense form of

Page 146: A Comparative Grammar of Brtiish English Dialects

136 Lukas Pietsch

be in all three persons; preservation of the was versus were number distinction as in the standard; hence parallel paradigms of be in both tenses (I/thou/he is, we/you/they are, I/thou/he was, we/you/they were); application of the Northern Subject Rule to all verbs, including be. This bundle of features can conveniently be called the Central Northern3

agreement system. In the northern half of Northumberland, we find more or less the same

features, but with less consistency, be paradigms more similar to the standard, and a lack of thou. This area can conveniently be labelled the “Upper North”.

Towards the northwest Midlands, more heterogeneous and variable agreement systems can be found. Verbal -s conforming to the Northern Subject Rule exists here too, but competes with inflectional forms in -n,another conservative relic form from Middle English. Unlike in the Central North there is a strong tendency to neutralize the was/were distinction in favour of generalized were forms. The second singular form of be is generally thou are or thou art. These three features together are characteristics of an area between Derbyshire and southern Lancashire, also reaching into the southwestern border areas of Yorkshire, and may be labelled the “Northwest Midlands” agreement system. To the north of it, one finds a gradual shift from the Northwest Midlands features towards the Central Northern features. It can be visualized as a bundle of successive isoglosses stretching through Yorkshire and northern Lancashire. These isoglosses are grouped roughly around a line from Morecambe Bay to the mouth of the Humber. This line, also called the Humber-Lune-Ribble line, has long been known as one of the most important and oldest boundaries in the traditional dialects of English (Trudgill 1990: 34; Ihalainen 1994: 219; Samuels 1988). For the present purposes, the area spanned by the whole isogloss bundle, i.e. most of Yorkshire and northern Lancashire, can conveniently be called the “Lower North”.

Further southwest into the Midlands, more variation in the paradigm of be is found, involving present indicative forms such as be, bist, and bin.Also, the area preserving inflectional -n overlaps in the south with an area that has generalized verbal -s independent of the Northern Subject Rule. The East Midlands differ from the West Midlands mainly in that present-tense paradigms of be are more similar to the standard, and second singular thou is generally absent. The two areas are divided by a line that runs between the counties of Derbyshire and Staffordshire in the west, and Nottinghamshire and Leicestershire in the east. However, one feature that

Page 147: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 137

both the west and the east Midlands share is the tendency to neutralize the was versus were number distinction, in the one or other direction.

Map 1. Important subject-verb agreement isoglosses in the SED

© L. Pietsch 2004

0 50 100km

Cheshire

Cumberland

Derbyshire

Durham

Lincolnshire

Lancashire

Leicestershire

Northumberland

Norfolk

Nottinghamsh.

North’sh.

Staffordsh.

Yorkshire

Westmorland

Shropshire

Cambridgesh.

Hunt’sh.

Herefordshire Worcestersh.

Warwicksh.

I amI is

you arethou is

he washe were (occ.)

I keepI keeps (occ.)

he were (freq.)

he were I isI am

thou isyou areburglars steals

burglars steals

burglars steal

burglars stealthou art

you are/thou bist

they keep

they keep(en)

they keeps

they keepen

they keep(s)

1

2

3

4

56

7 8

9

1

2

3 4

5

6

12

34 5

61

2 3

41

2

3

4

5 6

7 8 9

1011

1213

14

1 2

3

4

5

67 8 9

101112 13 14 15 16

17 18 1920

21 22 23 24 25

26 27 28

2930

31

32 33

34

1 2

34

56

12

34 5

6

7

1

2 3

4

12

3

4 5

6 7

8910

11

12

13 14

15

12

34

5

6 78

9 1011

12

34

5 67

8 9

10

11

1

2 345

6

78

9

10

1

2

1

2 3

45

7

12

34

56

7

1

2 3

45 6

7

1

23

4

5

1

2

1

2

1

4

7 8

Page 148: A Comparative Grammar of Brtiish English Dialects

138 Lukas Pietsch

Plural verbal -s with lexical noun subjects, as licensed by the Northern Subject Rule, is found everywhere north of a line running from Merseyside to the Wash. In Map 1 the isogloss is labelled burglars steals (named after a relevant item in the SED questionnaire). It cuts right through both the West and East Midlands with their otherwise very different systems.

For Scotland and the north of Ireland, data of similar geographical quality are not available. Data from the TRS/NITCS show a strong influence of the Northern Subject Rule in the northern parts of Ireland. Data from FRED and from other dialectological and sociolinguistic studies suggest that the dominant situation in Scotland for the most part resembles that of the Upper North. While modern Scottish varieties share with many English dialects a tendency for neutralization of the was/were contrast independently of the Northern Subject Rule (as in we was, you was), such tendencies were not discovered in the northern Irish data. On the other hand, one pattern that is found only in Ulster and in some parts of Scotland is the tendency of using neutralized singular there were, and possibly there are, in existential clauses only.

3.3. Minor patterns

3.3.1. Verbal -n

Verbal forms in -n are relic forms of the Middle English plural -enparadigms typical of the Midlands (Wright 1905: 296; LALME I: 467). In the SED, such forms are still well attested in one compact relic area in the northwest Midlands, covering southern Lancashire, Cheshire, Derbyshire, Shropshire, and Staffordshire, and reaching also into the southwestern corner of Yorkshire (Map 2; see also Pietsch 2005: ch. 4). They are also documented for some more recent dialects of that area (cf. Shorrocks 1999 on Bolton, Lancashire). The area overlaps with that affected by the Northern Subject Rule, which means that nonstandard forms in -n and in -scompete with each other in some areas.

Page 149: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 139

Map 2. Verbal -n forms in the SED

0 50 100km

© L. Pietsch 2004

Present-Tense bin

Generalized am/-m

Verbal -n(Other Verbs)

Cheshire

Cumberland

Derbyshire

Durham

Lincolnshire

Lancashire

Leicestershire

Northumberland

Norfolk

Nottinghamsh.

North’sh.

Staffordsh.

Yorkshire

Westmorland

Shropshire

Cambridgesh.

Hunt’sh.

Herefordshire Worcestersh.

Warwicksh.

1

2

3

4

56

78

9

1

2

3 4

5

6

1

2

34 5

61

2 3

41

2

3

4

5 6

7 89

1011

1213

14

1 2

34

5

67 8 9

1011

12 13 14 1516

17 18 1920

21 22 23 24 25

2627 28

2930

31

32 33

34

1 2

34

56

12

34 5

6

7

1

2 3

4

12

3

4 5

6 7

8910

11

12

13 14

15

12

34

5

6 78

9 1011

12

34

5 67

8 9

11

1

2 345

6

78

9

10

1

2

1

2 3

45

1

4

5

1

3

7

1

23

4

5

1

2

1

2

1

4

7 810

73

6 7

2

45 6

Page 150: A Comparative Grammar of Brtiish English Dialects

140 Lukas Pietsch

The examples in (8) show that verbal -n forms occur in all plural environments, with the pronouns we, you, and they, as well as occasionally with plural noun phrase subjects. However, the latter type appears to be rare. The example quoted under (8f) is the only one found among 335 tokens of -n forms in the SED, all the others having pronoun subjects. Shorrocks (1999: 114) states that -n occurs only “in the first, second and third persons plural, after pronominal subjects that do not trigger -s.” This type of distribution, where an -s versus -n alternation behaves according to the same pattern as -s versus - in more purely northern varieties, may have been typical of some parts of the border zone between the north and the Midlands as early as in late Middle English (Laing 1978: 244; McIntosh 1988: 117).

(8) a. We never sayen that. [SED: Y29] b. We callen it. [SED: Db1] c. You mowen. [SED: Db6] d. They taken more pulling than a cart. [SED: Db1] e. They think they knowen it. [SED: Ch1] f. Burglars thieven them. [SED: St2]

In the singular, verbal -n forms seem to occur only in the paradigm of be (in the form I/he bin), and of have (I/he han). In addition, he done is recorded, but only once (Sa5). The bin forms occur only in the southwestern half of the area in question, namely in Shropshire and some closely adjacent areas in the neighbouring counties. Areas yet further south, especially Herefordshire and Worcestershire, tend to have generalized beforms instead of bin (not charted in Map 2). Another phenomenon geographically related to the verbal -n forms seems to be the existence of am forms generalized to other persons and numbers than the first singular in the paradigm of be. Especially you’m and they’m are often recorded, and seem to be used as a contracted form both of you/they are and sometimes you/they have. These forms compete with both the bin and the beparadigms, in an area overlapping with the general -n area in Shropshire and southern Staffordshire.4

In the FRED data used for the present study, verbal -n was not recorded, but this may well be due to the area not being well represented in this part of the corpus.

Page 151: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 141

3.3.2. I is

As stated above, throughout the Central North – north of a line reaching from Morecambe Bay to the mouth of the Humber, but excluding the Upper North – the present tense of be generally has is in the first and second singular. In this area be has isomorphic paradigms in the present tense and past tense, with singular -s forms contrasting with plural -r forms. This means that the first singular present tense of be, just like its past tense in most other areas, falls by necessity outside the scope of the Northern Subject Rule.

While forms like I is (or I’s) are abundantly attested in the SED, they are almost completely absent in the more recent FRED data. Only two speakers in the corpus show residual I is as a relic form. Interestingly, one of these informants (Yks3:SL, a farmer from the Teesside area, born c.1910, and recorded during the mid-1980s) used I’s only in direct speech quotations in the context of narratives set in the old days, using standard I’m elsewhere. For this speaker I’s apparently functions as a signal of vernacular speech employed to give a certain stylistic touch to his narrative. The other speaker (Wes4:HL, a retired forest worker from the Lake District, Cumbria) used twice I’s and twice I’m in his recording.

(9) a. And t’ porter says, well, I’s about sick o’ this. [FRED:Yks3:SL]

b. And mi mother says, I’s not goin’ to have room to work in here. [FRED: Yks3:SL]

c. And now, so I got this one done and I thought, … This is mi last morning, I’s not going to bother. [FRED: Yks3:SL]

d. I used to dread to have to go down to give my boss a message when he was among all t’ boozey folks, ’cause they all knew me: “Fetch him a drink in. Fetch him a drink!” – “I’s comingback. I’ll just deliver this message!” [FRED: Yks3:SL]

e. I’s going on for eighty. [FRED: Wes4:HL] f. He says, I’s goin’ to have a real good go at it [FRED:

Wes4:HL]

Page 152: A Comparative Grammar of Brtiish English Dialects

142 Lukas Pietsch

3.3.3. Verbal -s with thou

The old second singular pronoun thou (in a variety of forms: thou, thee,[ð ], [t ] etc.) is found preserved in the SED in a large coherent area covering all of the north, with the exception of the East Midland counties. Only in northern Northumberland is it found less frequently, and ye/you is generally used instead (cf. Trudgill 1990: 86, Beal 2004). Where they exist, the thou forms almost invariably command agreement with -s forms (rarely also -st forms) of the verb, irrespective of the Position-of-Subject Constraint. Hence they do not pattern together with you under the Northern Subject Rule but rather align themselves with third person singular he/she/it. Variation in the verbal forms occurs only in the paradigms of be. As for the past tense, the variation between thou was and thou were will be dealt with together with that in the other persons and numbers in the next section below. As for the present tense, three fairly distinct regional types can be distinguished (Map 3). In the Central North one regularly finds thouis, alongside I is and he/she/it is. Further southwest, in southern Lancashire, southwestern Yorkshire, Derbyshire, Cheshire, and northern Staffordshire, the forms are thou are or thou art, in inversion sometimes contracted to art’. Going yet further south into the West Midlands, one finds forms such as thou bis(t), as well as other forms with the verbal stem be in which the plural forms have been generalized to the second singular (thou be, thou bin).

Like the I is forms, the thou forms are clearly obsolescent in the more recent data from FRED. Only four tokens of subject thou (thee/tha) with concord verbs are found in the data. One example from the Teesside data, and one from the Scottish Borders, again occur in narrated direct speech situated in the ‘old days’ (10). These two tokens have the old -s form of the verb and happen to be both in interrogatives. The other two tokens are in stereotypical collocations of the you see / you know type, and have been recorded in northeast England and southwest Scotland respectively (11). In these two tokens the pronoun thou has taken over the affixless verb form used also with you.

Page 153: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 143

Map 3. Verbal agreement forms with thou in the SED

© L. Pietsch 2004

0 50 100km

thou is

thou bist

thou are/art

thou be/been

thou -s, other verbs

Cheshire

Cumberland

Derbyshire

Durham

Lincolnshire

Lancashire

Leicestershire

Northumberland

Norfolk

Nottinghamsh.

North’sh.

Staffordsh.

Yorkshire

Westmorland

Shropshire

Cambridgesh.

Hunt’sh.

Herefordshire Worcestersh.

Warwicksh.

3

4

56

8

9

1

2

3 4

5

6

12

35

61

2 3

41

2

3

4

5 6

7 89

1011

1213

1 2

34

5

67 8 9

1011

12 13 14 1516

17 18 1920

21 22 23 24 25

26 27

2930

31

32 33

34

1 2

34

5

12

34 5

6

7

1

3

4

1 3

4 5

6 7

8910

11

12

13 14

15

2

3 5

78

9 1011

12

3

57

8

10

11

1

2 345

6

78

9

10

1

2

1

2 3

4

7

23

4

56

1

2 3

45 6

7

1

23

4

5

1

2

1

2

1

4

7 8

1

2

7

4

28

1414

2

2

6 46

9

1

7

1

4

6

Page 154: A Comparative Grammar of Brtiish English Dialects

144 Lukas Pietsch

(10) a. And uh, an old woman over North Gate said, where 's thoucome from then? Says, from Middlesbrough. She says, what a bloody muck-hole that is. [FRED: Yks1:WF]

b. And they used to go down and cook their meals, down in the, on the fire, hm-hm, and never anybody said, what doest thou,they never were put out or anything then. [FRED: Pee2:MT]

(11) a. Tha see it? See this here? [FRED: Dur1:ML] b. Oh, they, thee ken, there was no, ye see, there were, there was

a lot come in at what we call Huggins up there. [FRED:Dfs1:WH]

3.3.4. Verbal -s with I/we/you/they

Apart from the old use of I is, described above, and the widespread was/were variation which will be the subject of section 3.4 below, non-standard -s with the pronouns I, you, we and they occurs occasionally in many but not all northern varieties. We can distinguish several types of usage.

First, there are those instances which are covered by the Position-of-Subject Constraint, most typically in the coordinated structures of the theysing and dances type. This usage is characteristic of the more conservative varieties. In the relatively recent data of FRED it is not attested, but it appears with some regularity in data from the SED (12-13), even as far south as Nottinghamshire and Lincolnshire, and it is also found in a number of Ulster speakers in the NITCS (14–17).5

(12) a. They gang and never speaks. [SED: Du4] b. They peel ’em and boils ’em. [SED: La6] c. They break into houses and steals. [SED: La7] d. They go in and cuts ’em down. [SED: Y17] e. They throw that down and picks another sheaf up. [SED: Y17] f. They feel over ’em and weighs ’em, does butchers. [SED: Y22] g. They rope ’em and then pulls ’em in. [SED: Y29] h. They run at one another and brods one another. [SED: Y31] i. They cut ’em down low and lays ’em. [SED: Nt2] j. They lead bulls wi’ staffs, puts a band on their horns, puts it

undernien their hind legs. [SED: L4]

Page 155: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 145

(13) a. You stack it up and carts it up to where you want it. [SED:Y17]

b. You put middling of water into it and pulls it down. [SED:Y17]

c. You go fair down t’ middle and pulls one each way. [SED:Y22]

d. You take that there up and shakes it out. [SED: Y14]

(14) The women goes out and rickles, dear. I rickle my own turf. After I do my work in the daytime, I go out and rickles my turf. When I come home, I go away and rickles my turf. [NITCS: L17.3]

(15) a. You have, you shovel off the, makes the top of it smooth, you know. [NITCS: L17.3]

b. And then you just cut down, and makes the shape of the turf [NITCS: L22.1]

c. You pull a wee drawer there, hey, and puts the… maybe you have one of them? [NITCS: L2.2]

d. Like, you, if you go out to speak to anyone, know, ch-, has to challenge them about, if they’re doing any harm, or d-, your property or anything, they just answer you back. [NITCS:L4.2]

(16) a. But we lived, we lived about, about four mile out of the town here, and was taught at Ballyreagh school. [NITCS: L25.3]

b. We were all good neighbours, and is yet, I hope. [NITCS:L17.3]

c. Mm, we just t-, plays swinging on the ropes, and in the bars, you know, climbing bars, and swinging round on them.[NITCS: L17.1]

(17) a. And they season and gets lighter there then. [NITCS: L22.2] b. So they closed it and sends their milk to Manorhamilton.

[NITCS: L31.2] c. So they go on that way and takes about ten minutes for that

[NITCS: L8.1]

Note, incidentally, how in example (15b) the presence of an intervening adverb in the first conjoined phrase (you just go) fails to trigger a Position-

Page 156: A Comparative Grammar of Brtiish English Dialects

146 Lukas Pietsch

of-Subject effect according to the Northern Subject Rule, whereas the conjoining of verb phrases does trigger -s in the second verb.

A second, apparently quite unrelated pattern involving non-standard verbal -s with pronoun subjects is found in the so-called historic present. A tendency to mark narrative clauses in the present tense with generalized verbal -s forms is commonly reported for many modern varieties of English. Such verbal -s forms occur very commonly also with pronoun subjects where they are not licensed by the Northern Subject Rule.

The prototypical usage condition for such forms is in clauses introducing direct speech within the narrative. In the FRED data, 141 tokens from 21 informants in all parts of the survey area were recorded in this textual function; with one exception (I shouts) all using the verb says.In the NITCS data there were 20 tokens from 8 informants, all of them using says. One clear piece of evidence for the restricted, formulaic nature of the I says idiom is that for some of the older speakers it is associated with a special rule of subject-verb inversion (says I; also: says he etc.), which is likewise restricted lexically and pragmatically to exactly these formulae of introducing direct speech. 16 of the tokens mentioned above were of this type, and they were found in a few older speakers in Scotland and in Ulster.

Only 18 tokens were found of other verbal -s forms used in other textual functions within a narrative in the FRED data (and none at all in the NITCS), of which 13 came from a single informant and were produced during two longer narrative sequences. Apart from this, there is little evidence in the data of the present study that speakers have a productive stylistic rule of using verbal -s as a marker of the historic present as such, over and above the prototypical, idiomatic use of says (or its semantic equivalents).6 However, such wider usage of the narrative present has repeatedly been reported elsewhere, for instance by Harris (1993: 154–156), Robinson (1997: 127) and Henry (1995: 18) for Irish English. Henry – in a formal, generativist discussion of verbal concord – uses the existence of this usage as an argument for a formal analysis which involves a T (i.e. tense) node playing a central role in a feature checking mechanism that is responsible for licensing verbal -s. (1995: 27). The reference to the T node comes close to suggesting that the simple present and the narrative present must actually be two distinct morphological tenses, as that would presumably be the case by definition – within a formal theory such as Henry’s – if this node contained different features in each case. (For further discussion of Henry’s analysis, see section 5.1 below).

Page 157: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 147

The attestation of verbal -s with I/you/we/they in environments other than those described so far, i.e. where it violates the Northern Subject Rule, is quite marginal in both the NITCS and FRED. This is somewhat astonishing, since the existence of such forms is fairly well documented in some of the older dialects in the SED. There it is attested occasionally in a large area of Northern England, overlapping with the central northern I isarea in Yorkshire and Lancashire but also extending further south, and excluding most of Cumberland, Westmorland and Durham in the north (Map 4). It should be noted that the phenomenon of occasional verbal -susage in these areas must be distinguished geographically from the much more general usage of neutralized verbal -s in a different area towards the southwest (beginning in Map 4 in the counties of Herefordshire and Worcestershire; cf. Klemola 1996: 50–52). In this southwestern area, verbal -s is traditionally found quite regularly in all environments, irrespective of type and position of subject. Both areas are separated from each other by a broad belt in which non-standard verbal -s in the pronominal environments seems to have been largely absent in the traditional dialects, just as it was also absent further north. These observations confirm those by Wright (1905: 296), who likewise stated that exceptions to the Type-of-Subject Constraint occurred “occasionally in parts of Yks. Lanc. and Lin.” but not further north.

As for the function of this Yorkshire/Lancashire type of occasional verbal -s usage, nothing can be stated on the basis of the SED material, since the tokens are mostly documented without context. However, it seems likely that the phenomenon reflected in the SED attestations can be identified with that described for more recent Lancashire dialects, where verbal -s in such cases is reported to serve as a marker of habitual semantics (Shorrocks 1999: 112).

It should be noted in passing that many of the SED tokens that can be counted as potential examples of habitual -s occur with one of several typical adverbs between the pronoun and the verb: never, always, often, etc. According to the ‘ideal’ northern system, the presence of such an adverb would also in itself count as a condition for the Position-of-Subject Constraint to apply. Hence, it is sometimes difficult to decide whether verbal -s in this cases is triggered by the temporal semantics, or by the Northern Subject Rule.

Page 158: A Comparative Grammar of Brtiish English Dialects

148 Lukas Pietsch

Map 4. Verbal -s with I/you/we/they in the SED

© L. Pietsch 2004

0 50 100km

Verbal -s recorded with:

I

we

you

they

(large symbols: totalof 4 tokens or more)

1

2

3

4

56

78

9

1

2

3 4

5

6

12

34 5

61

2 3

41

2

3

4

5 6

7 899

1011

1213

14

1 2

34

5

67 8 9

1011

12 13 14 1516

17 18 1920

21 22 23 24 25

26 27 28

29 3130

32 33

34

1 2

34

56

12

34 5

6

7

1

2 3

4

12

3

4 5

6 7

8910

11

12

13 14

15

12

34

5

6 78

9 1011

12

34

5 67

8 9

10

11

1

2 345

6

78

9

10

1

2

1

2 3

45

7

12

34

5 76

1

2 3

45 6

7

1

23

4

5

1

2

1

2

1

4

7 8

Page 159: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 149

However, to the degree that the same dialects also exhibit occasional verbal -s in cases of no intervening adverb, with comparable frequency, the pattern is better described as independent of the Northern Subject Rule, at least in the case of Yorkshire and Lancashire. Matters are somewhat different in the northern Irish data of the NITCS: here, verbal -s with pronominal subjects is quite rare throughout; but it is, relatively speaking, a good deal less rare in cases not adjacent to the pronoun than in others. In this dialect, then, it is indeed the Northern Subject Rule that is responsible for the marginal option of verbal -s in these cases.

3.4. Was/were neutralization

Variation between was and were in the northern dialects is a highly complex field. This is due to the fact that it tends to follow only partly the pattern defined by the Northern Subject Rule, with was and were behaving like the -s and zero forms of other verbs. This pattern is often overlaid with other, complementary or competing, rules of variation specific to was and were alone. The high potential for irregularity and variation that characterizes this verb can be linked historically to two factors.

First, the was/were paradigm patterned differently from the present-tense verbs from the outset, and of course it still does so in Present-Day Standard English. The same form, was, is used for both the first and third person singular, whereas in present-tense verbs the -s form is unique to the third person. This is the last remnant of the Old English concord paradigms of the past tense, in which first and third singular regularly patterned together as against the rest. The breakdown of this system can best be seen when the old second person singular is taken into account too. Its original Old English form (ðu wære) resembled the plural form (wæron) more than the other singular forms (wæs), and it predictably fell together with the plural form in many dialects of Middle English (thou were). Elsewhere, however, analogy with other verb classes gave rise to a variety of other forms such as thou wert, thou was, thou wast, before the second person singular finally became obsolete in Modern English. It is not surprising that tendencies of analogical extension or levelling of forms have continued to operate on the was/were paradigms of different dialects, leading to a number of different outcomes.

Secondly, it must be noted that both was/were and is/are/am did not originally fall under the scope of the Northern Subject Rule at all. As will

Page 160: A Comparative Grammar of Brtiish English Dialects

150 Lukas Pietsch

be described in more detail in section 4, the diachronic development in these verbs went exactly in the opposite direction than with all others. In the lexical verbs, the (now non-standard) -s forms represent an older, conservative form both in the singular and in the plural, while the (now standard) zero forms in the plural are an innovation in Middle English. In contrast to this, the forms is and was are truly and exclusively singular in origin, and were only extended to plural use under the Northern Subject Rule by way of analogy with the other -s forms later. Apparently, the use of was in plural environments as licensed by the Northern Subject Rule never became quite as regular as the corresponding use of other verbal -s in the same environments, even in the ‘purest’ older dialects of the northern type (cf. Montgomery 1994 on Older Scots). But even though the forms of be were relatively slow in picking up the Northern Subject Rule pattern, they have also been slower than other verbs in giving it up again and replacing the northern with the standard pattern, under conditions of dialect levelling in more recent times. This results in a situation found in some modern varieties where be apparently tends to be the only verb to conserve reflexes of the northern pattern (see e.g. Tagliamonte 1999 on the English of York).

In the traditional dialects of northern England as reflected in the SED,the following situation obtains. In the Central North, the use of indicative was and were is, as a rule, parallel to that of is and are. The singular of all three persons (including thou) is invariably was, while the plural (including you) mostly has were but allows for was in accordance with the Northern Subject Rule. This is mainly true for the four northern counties of Cumberland, Northumberland, Westmorland and Durham. The SED data suggest a near-categorical validity of the number contrast with adjacent pronoun subjects in this area.

All the rest of the survey area shows a tendency of was/were levelling, either generalizing was to the plural, or were to the singular. Four regional clusters of locations can be distinguished in this respect. First, in a compact area centring around southern Lancashire, southwestern Yorkshire (i.e. southwest of a line from Morecambe Bay to the Humber) and Derbyshire, there is a strong preference for generalized singular were forms (cf. also Shorrocks 1999: 168). Second, in a broad transitional belt from this area into the Central North, covering northern Lancashire and the northeastern half of Yorkshire, singular were forms are also occasionally recorded but less frequent. Third, in a smaller area in the east Midlands, especially in northern Lincolnshire, the preference seems to be for neutralization in wasrather than were. In the Northwest Midlands there are also occasional

Page 161: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 151

attestations of plural weren. Finally, most other locations in the Midlands, further south, have highly variable or hybrid systems where both singular were and plural was may co-occur. A common tendency in many of these dialects seems to be that were is preferred in negated environments (cf. Anderwald 2002, Britain 2002). This effect is discernible in all parts of the SED data except the Central North and those parts of the NW Midlands and Lower North where were levelling is predominant in all environments. To the north of the central generalized were area, this negation constraint tends to affect only the singular forms (he was vs. he weren’t), while wereremains near-categorical in the plural, except where the Northern Subject Rule licenses was. The existence of systems of this kind is consistent with findings by Tagliamonte (1999), who reports the combined effects of singular were generalization, the negation constraint, and the Northern Subject Rule (though with no plural verbal -s usage preserved in the lexical verbs) in the local dialect of York city. In contrast to the SED findings, Beal (1993: 194) attests plural was levelling also for Tyneside and Northumberland.

In varieties where was/were levelling and the Northern Subject Rule compete with each other, it is also often found that the third person plural differs from the first and second person plural. Levelling typically results in we was and you was, whereas the third person plural tends to conserve near-categorical they were in conformity with the Northern Subject Rule (Tagliamonte 1999, Chambers 2004).

Page 162: A Comparative Grammar of Brtiish English Dialects

152 Lukas Pietsch

Map 5. Was/were neutralization in the SED

© L. Pietsch 2004

0 50 100km

Verbal -s recorded with:

I

we

you

they

(large symbols: totalof 4 tokens or more)

1

2

3

4

56

78

9

1

2

3 4

5

6

12

34 5

61

2 3

41

2

3

4

5 6

7 899

1011

1213

14

1 2

34

5

67 8 9

1011

12 13 14 1516

17 18 1920

21 22 23 24 25

26 27 28

29 3130

32 33

34

1 2

34

56

12

34 5

6

7

1

2 3

4

12

3

4 5

6 7

8910

11

12

13 14

15

12

34

5

6 78

9 1011

12

34

5 67

8 9

10

11

1

2 345

6

78

9

10

1

2

1

2 3

45

7

12

34

5 76

1

2 3

45 6

7

1

23

4

5

1

2

1

2

1

4

7 8

Page 163: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 153

Map 6. The was/weren’t negation constraint in the SED

© L. Pietsch 2004

0 50 100km

Verbal -s recorded with:

I

we

you

they

(large symbols: totalof 4 tokens or more)

1

2

3

4

56

78

9

1

2

3 4

5

6

12

34 5

61

2 3

41

2

3

4

5 6

7 899

1011

1213

14

1 2

34

5

67 8 9

1011

12 13 14 1516

17 18 1920

21 22 23 24 25

26 27 28

29 3130

32 33

34

1 2

34

56

12

34 5

6

7

1

2 3

4

12

3

4 5

6 7

8910

11

12

13 14

15

12

34

5

6 78

9 1011

12

34

5 67

8 9

10

11

1

2 345

6

78

9

10

1

2

1

2 3

45

7

12

34

5 76

1

2 3

45 6

7

1

23

4

5

1

2

1

2

1

4

7 8

Page 164: A Comparative Grammar of Brtiish English Dialects

154 Lukas Pietsch

Regional diversification in this respect, similar to what is evidenced in the SED, can also be seen in the FRED data. Table 1 displays the rates of non-standard was in the three pronominal standard were environments (we,you, they) across eight regional clusters of texts. For comparison, the table also shows the rates of standard I was as opposed to non-standard I were.

The regional clusters are defined as follows. ‘Ulster’ comprises the six TRSrecordings in the northern part of the Republic of Ireland, all in close neighbourhood to the six Northern Ireland counties covered by the NITCS(the label is used somewhat loosely, since not all of the locations are actually in the province of Ulster – four are in County Donegal, one in Leitrim and one in Louth). ‘Northern Scotland’ comprises one recording from Banffshire and six from the Scottish Highlands (Inverness, Eastern Ross and Sutherland). ‘Mid Scotland’ refers to a cluster of interviews from the east coast area, between Kincardineshire and West Lothian. ‘Southern Scotland’ comprises recordings from the Scottish Borders as well as one from Dumfriesshire. A group of Northumberland recordings represent the traditional English Upper North. Recordings from County Durham and northeast Yorkshire are grouped under ‘Teesside’ and represent part of the traditional Central Northern area. Another cluster, labelled as ‘Cumbria’, represents data from near Ambleside in the Lake District and also forms part of the traditional Central North. Finally, grouped under ‘Lower North’ are two recordings from southern Lancashire and southwest Yorkshire. The data in Table 1 suggest that speakers in Ulster, northern Scotland, as well as in the Upper and Central North of England, tend to have categorical was/were concord with pronoun subjects, identical to the standard. This is

Was with pronoun subjects in FRED

They We You IUlster 0/22 0% 0/5 0% 0/15 0% 11/11 100%N Sco 1/113 1% 0/23 0% 0/8 0% 66/66 100%M Sco 1/188 1% 8/68 12% 60/104 58% 115/116 99%S Sco 4/114 4% 4/15 27% 2/6 33% 48/48 100%Northumb. 0/39 0% 1/15 7% 0/14 0% 47/49 96%Cumbria 0/66 0% 1/27 4% 0/22 0% 76/77 99%Teesside 6/264 2% 3/128 2% 3/171 2% 331/344 96%Lower N 9/104 10% 15/61 25% 3/7 43% 99/115 86%Total 21/910 2% 32/342 9% 68/347 20% 793/826 96%

Table 1.

Page 165: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 155

also confirmed in the data from the NITCS, where non-standard was with adjacent pronoun subjects is extremely rare. In Middle and Southern Scotland, there is variable non-standard we was and you was, but hardly any they was. It must be stressed that these differences are not due simply to an uneven distribution of standard and dialectal speakers across the corpus. All clusters, including those that show no was/were variation at all, do contain informants who otherwise use strongly non-standard features, including heavy use of non-standard verbal -s according to the Northern Subject Rule. These data therefore fit in with the earlier findings in the SEDand the NITCS, which indicated that central Northern England and Ulster do not share in the pronominal was/were variation in the same way as so many other English vernaculars, and that this type of was/were variation is historically unrelated to the Northern Subject Rule.

The Lower North speakers, as can be expected on the basis of the earlier SED findings, have some use of non-standard were in standard wasenvironments, represented in the table by the figures in the I column (only 84% of standard was; the situation is similar in third person singular environments.) However, at the same time they also have variable plural was in standard were environments. These speakers even have they was,but they too use it less frequently than you was or we was. In all areas that allow plural was at all, you was seems to be more frequent than we was,and both are more frequent than they was. This fits in with observations made in other studies (e.g. Smith and Tagliamonte 1998, see also Chambers 2004). It is only in the speech of the Teesside informants that plural was is found with no marked differences between the three environments, but it is only of marginal frequency in all three. In a common Varbrul model for all areas that have plural was, factor weights of .26, .61, and .88 were calculated for they, we and you respectively.

Whereas there is no discernible Position-of-Subject effect in present-tense verbs with pronoun subjects in the FRED data, some effect of this type may be in evidence with was/were, although owing to small absolute token counts the evidence is hardly conclusive. While the overall rate of non-standard was with adjacent we/you/they is 115/1579 (7%), with non-adjacent pronoun subjects it is 6/20 (30%). Varbrul selects the factor group as significant at a 0.009 level, with factor weights of 0.49 versus 0.84 for adjacent and non-adjacent subjects respectively. The negation constraint had no discernible effect on plural was in the data.

Page 166: A Comparative Grammar of Brtiish English Dialects

156 Lukas Pietsch

3.5. Existential there clauses

In all data discussed in the previous sections, clauses with existential therehave been excluded because they require special consideration. Concord variation in existentials and concord variation in canonical clauses have followed quite different paths of historical development, they have different sociolinguistic status in the dialects in question (cf. also Wilson and Henry 1999: 12), and their distribution is governed by different sets of linguistic constraints.

As is well-known, existential clauses in formal Standard English require the verb to agree in number not with the syntactic dummy subject there, but with the so-called notional subject, the NP that follows it. This constitutes a systemic anomaly, as the morphosyntactic properties of subjects are divided between two constituents. The dummy subject there acts as the subject of the clause as far as word order is concerned, but the following NP acts as the subject in so far as it controls agreement. Many varieties of English share the tendency to level out this irregularity, by allowing invariant singular verb forms after there. Non-agreement in clauses beginning with there or similar adverbs have been a structural option in most forms of English even since the Old English period (Visser 1963: 62). In the modern dialects, the strong tendency for non-agreement means that the residual subject status of the following noun phrase is effectively lost. Historically this can be seen as part of a long-term trend of grammaticalization of the there construction, in which there has gradually changed its status from being originally a deictic adverb to being a subject (cf. Breivik and Swan 2000; pace Van Gelderen 1997: 88–109).

It is not surprising that evidence for non-agreement in existentials was also found everywhere in the data of the present study. However, there is a second, competing non-standard pattern in effect in some parts of Northern Ireland and Scotland, which works in just the opposite direction. In this pattern the distinction between singular and plural existentials is also neutralized, but the neutralization is in favour not of the normal singular verbal forms is/was, but of the verb forms that are otherwise used in the plural: are/were. These forms appear to be favoured in negated clauses, but are by no means restricted to them. The present-tense form of this pattern is often a contracted form in which the verb has lost all or most of its segmental phonological substance: there’ < there’re < there are, but full forms of are are also attested (18c). In the following discussion I will refer

Page 167: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 157

to these forms collectively as -r forms, as opposed to the -s forms is, ’s and was.

(18) a. Och, aye, there’ no blacksmithing work now, except odd wee bits. [NITCS: L7.3]

b. There’re no school. [NITCS: L7.3] c. The young would rather go away till Dungiven, or, or

Strabane, or some place where there are a big, eh, dance-hall.[NITCS: L10.2]

In the NITCS data, -r forms are found strongly concentrated in the speech of informants from the north and northwest of Ulster, whereas only a bit further towards the southeast (county Down and the southern half of county Antrim) such forms were not found or only very marginal. This finding is interesting from a dialect-geographical perspective, as this apparent division cuts right across what has long been recognized as an important dialect boundary within Ulster, that between ‘Core Ulster Scots’ and ‘Mid Ulster English’ (Gregg 1972; see Pietsch 2005: ch. 5.4.1 for details). In the FRED data, singular -r forms in the present tense are found only in the speech of six informants, two of them located in Ulster and four in southern Scotland. Some informants in the northwestern part of the ‘Core Ulster Scots’ area went on record using singular -r in about two thirds of all cases. However, the existence of the competing patterns of neutralization in -s and neutralization in -r means that for some speakers all eight of the options shown schematically in (19) and (20) seem to be possible.

(19) a. There’s a house now b. There’s houses now c. There was a house then d. There was houses then

(20) a. There’ a house now b. There’ houses now c. There were a house then d. There were houses then

The following authentic examples (21) give a first impression of the amount of variability observed in the corpus. They were all produced by one Ulster speaker within a single short stretch of discourse, in this order:

Page 168: A Comparative Grammar of Brtiish English Dialects

158 Lukas Pietsch

(21) a. There was fairs. b. Well, there was a fair in Kilrea. c. There were no pastime for you. d. There were no fences. e. There were better sheepdogs than there is now. [NITCS: L6.3]

Across the whole survey area, singular were in past tense existentials is more widespread than singular are in the present. All of the six FREDinformants who used singular bare there’ in the present tense also had singular there were in the past. Additionally, singular there were was also found in the speech of twenty other informants. Many of these informants are also found in Ulster and southern and middle Scotland. But, not surprisingly, singular there were is also found in the Lower North of England, that is, with those speakers who have singular were levelling in non-existential clauses as well. In the Central and Upper North of England and in northern Scotland, singular were in either type of environment is only marginally present or, in many speakers, not attested at all. The two phenomena can thus best be described as distinct in principle, and probably in origin: speakers either have singular -r forms in existentials (the ‘Ulster Scots’ type: there are a house, there were a house); or they have singular were in all clause types (the ‘Yorkshire’ type: he were going; there were a house); but no speakers have both. Structures of the type there were a house, representing a point of incidental overlap between the two patterns, seem to have spread across an area much larger than either of the focal areas of these two patterns, as a marginal option at least. This distribution can be seen in Table 2, cross-tabulating figures for the three usage types across the eight regional clusters within the FRED data.

Unlike the FRED data, the data of the NITCS allowed also for a study of the effects of sociolinguistic variables, such as age and sex (Pietsch 2005: ch. 5.4.2). Looking at the three age groups sampled in the NITCS, it was found, first, that the middle age group tended to observe standard-conforming number concord in existentials more often than either the children or the older informants in the corpus. The youngest age group showed a strong tendency to use generalized -s forms, approaching near-categoricity in some groups. In contrast to this, the use of generalized -rforms was strongly associated with the oldest age group. Moreover, across all age groups, male informants consistently used -r more often than women and girls.

Page 169: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 159

Table 2. Nonstandard -r forms in FRED

Pres. Sg. Ex. Past Sg. Ex. Past 3Sg. / 1Sg. Total there’re + Sg. there were + Sg I/he/she/it were

Ulster 9/43 21% 13/48 27% 2/110 02% 24/201 12% S Scot. 9/42 21% 26/92 28% 9/678 01% 44/812 05% M Scot. 1/45 02% 13/175 07% 9/1257 01% 23/1477 02% N Scot. 0/21 00% 1/68 01% 4/455 01% 5/544 01% Cumbria 0/22 00% 0/80 00% 4/428 01% 4/530 01% Northumb. 0/11 00% 1/58 02% 6/272 02% 7/341 02% Teesside 0/82 00% 8/362 02% 35/1850 02% 43/2294 02% Lower N 0/22 00% 22/73 30% 72/661 11% 94/756 12% Total 19/288 07% 84/956 09% 141/5711 02% 244/6955 04%

All of this clearly indicates that generalized existential -r has the status of a socially marked, conservative local variant in Northern Ireland and that there is a change in progress replacing this pattern with that of generalized -s. The gender differentiation can be seen as an expectable side effect of such a development, as it confirms the long-established rule in sociolinguistics (cf. Trudgill 2000: 73) that women tend to follow prestigious standard models of speech more than men do, and that men in rural societies tend to be more conservative in preserving local non-standard forms. In this case, women were found both to adapt more to the standard of formal English than men do, and simultaneously to be leading the trend, apparent in younger speakers in general, to move away from the more specifically local non-standard pattern of generalized -r forms, towards the supraregional colloquial norm of generalized -s forms. Table 3 displays the distribution of -s and -r tokens across the different social groups in two geographical sub-regions in the Ulster corpus.

In addition to this, there seem to be indications in the NITCS data that Protestant speakers in the north may have a stronger preference for the conservative -r forms than Catholic speakers. Although the findings are somewhat inconclusive in statistical terms (Pietsch 2005: ch. 5.4.2), such an effect would fit in plausibly with a hypothesis that linguistic features of traditional Ulster Scots serve as a marker of Protestant identity in Northern Ireland, whereas Catholics show more of an orientation towards varieties of English used in other parts of Ireland (Harris 1991: 46, cf. also McCafferty 1998a, 1998b, 1999).

Page 170: A Comparative Grammar of Brtiish English Dialects

160 Lukas Pietsch

Table 3. Existential -r in the NITCS, by area, sex and age group

Area Sex Age Group Singular Plural Total North m 65–75 82/188 44% 78/126 62% 160/314 51% 35–45 38/158 24% 51/93 55% 89/251 36% 9–12 13/38 34% 5/19 26% 18/57 32% f 65–75 8/30 27% 5/15 33% 13/45 29% 35–45 1/63 02% 15/28 54% 16/91 18% 9–12 2/56 04% 5/46 11% 7/102 07% South m 65–75 10/150 07% 15/92 16% 25/242 10% 35–45 4/129 03% 7/67 10% 11/196 06% 9–12 0/75 00% 1/40 03% 1/115 01% f 65–75 3/137 02% 2/69 03% 5/205 02% 35–45 0/71 00% 14/49 29% 14/120 12% 9–12 0/73 00% 3/44 07% 3/117 03%

Verbal concord in existentials is also influenced by linguistic environment factors. One of them is negation. With singular notional subjects, the use of -r is three times more frequent in negated than in non-negated clauses (27 per cent as opposed to 9 per cent); in plural clauses the difference is slightly less marked but the direction is the same (45 per cent as opposed to 28 per cent). This effect reflects the pattern already discussed in the previous section, of generalized weren’t being preferred over wasn’tin many dialects (cf. Anderwald 2002: 180–182). In Ulster, such an effect was found only in existentials, not in canonical clauses, and it seems to apply in parallel fashion both to was/were and is/are. The data suggest that for many speakers the negation constraint is a more powerful factor in determining the choice of form than the grammatical number of the notional subject. A Varbrul model shows a slightly larger span of factor weights, of 0.45 versus 0.71 for non-negated and negated clauses respectively, as opposed to a span of only 0.42 versus 0.64 for singular versus plural subjects. Another factor that seems to constrain the choice of verb forms in existentials is the syntactic environment in which it occurs.

Table 4 shows token counts tabulated across three types of structures: Type A, the bulk of the canonical existentials with the verb immediately preceding the notional subject; Type B, clauses where the verb and the notional subject are separated by intervening material such as adverbials; and Type C, all other clauses in which the element order between the verb,

Page 171: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 161

the notional subject and the existential marker there differs in some way. This group most notably comprises tag clauses, questions, and relatives. It can be seen that both of the special environment types have an interesting effect. In the cases with intervening adverbials, the proportion of -r and -sis almost exactly the same in the singular and in the plural. In these cases, the choice of verb form thus appears to be governed entirely by factors other than grammatical number. In other words, agreement between the verb and the notional subject – which, as seen earlier, is disregarded by many speakers much of the time anyway – is practically non-existent whenever other words intervene between the two. However, in the Type C cases, those with inverted or other non-canonical word orders, the picture is different: whereas for the plural the rate of -r usage is just about average, surprisingly few instances of non-standard -r were found in the singular. A definitive explanation for this effect cannot be given at the present moment, but it may be related to the fact that in these clause types (tags, questions and so on) the verb is usually stressed. Apparently, the use of singular wereand are in Ulster English tends to be restricted to unstressed environments.

Table 4. Existential -r in the NITCS, by number and clause type

Clause Structure Singular Plural Total Type A: Adjacent NP 152/1134 13% 196/612 32% 357/1755 20% Type B: Adv.

Intervening14/76 18% 10/51 20% 24/127 19%

Type C: Others 1/53 02% 14/44 30% 15/97 15% Total 167/1263 13% 229/716 32% 396/1979 20%

3.6. Verbal -s with plural NPs

Having so far discussed the distribution of phenomena that have no or only a marginal relationship to the Northern Subject Rule, we can now proceed to a more detailed discussion of the pattern that stands at its core and is most characteristic of the traditional dialects of the northern type: non-standard verbal -s with third person plural subjects other than the pronoun they.

The geographical distribution of this pattern in the traditional dialects of England can be traced well in the SED data. The SED questionnaire contained two items that were designed to elicit structures of the relevant type (III.10.7 ‘bulls bellow’, and VIII.7.5 ‘burglars steal them’), and atlas

Page 172: A Comparative Grammar of Brtiish English Dialects

162 Lukas Pietsch

charts based on these data have been printed in Viereck (1991/1997: II, M27 and I, M40); Klemola (2000: 334), and Viereck, Viereck and Ramisch (2002: 84); see also the isogloss labelled ‘burglars steals’ in Map 1 above. However, as was pointed out earlier, the published material these charts are based on is in some sense deficient, as it excludes a great number of incidental material tokens that did not match the elicited environment in a narrow sense, but still provide important evidence for the effects of the Northern Subject Rule.

When this additional material, recovered from the original fieldworker notebooks, is taken into account, two important observations can be made. First, the area affected by the Northern Subject Rule in the traditional dialects reaches a good deal further south into the East Midlands than shown in those maps based exclusively on the published SED material. The data suggest that there is a broad transitional zone in which plural verbal -s occurs but becomes progressively rarer the further one goes to the south. The outer limits of this transitional zone are still virtually identical to the limits of the northern system of six hundred years earlier (see section 4 below). This can be seen by comparing the attestations in the SED (Map 7) with those in the Linguistic Atlas of Late Middle English (LALME,McIntosh, Samuels and Bensik 1986), displayed in Map 8.

Second, the distribution of tokens in the border zone reveals an interesting concentration of some special environment types. Whereas in the core area of the northern dialects proper, plural verbal -s can be found in all types of clauses, with all types of subject noun phrases and all types of verbs, towards the transition zone in the south plural verbal -s is documented almost exclusively in a set of special environments. Among them are: relative clauses; clauses with demonstrative pronoun subjects such as them or those; and clauses with indefinite pronoun subjects such as some, some of them etc. Unfortunately, these are by and large just the types that were not documented systematically in the published SED material. These types of environments together make up for almost 80 per cent of all the (relatively sparse) tokens of plural verbal -s in the transitional zone. In the northern dialects proper, their predominance is less strong, but with roughly 50 percent of all recorded tokens they still appear over-represented. To these environment types that seem to be particularly favourable to verbal -s can be added cases of subject-verb inversion, as in questions and tag clauses.

Page 173: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 163

Map 7. Plural verbal -s in the SED fieldworker notebooks7 (all NP subjects)

© L. Pietsch 2003

0 50 100kmfieldworker

S. Ellis others

no tokens

1-2 tokens

3-5 tokens

6 or more

no tokens

1-4 tokens

5-10 tokens

11 or more

Cheshire

Cumberland

Derbyshire

Durham

Lincolnshire

Lancashire

Leicestershire

Northumberland

Norfolk

Nottinghamsh.

North’sh.

Staffordsh.

Yorkshire

Westmorland

Shropshire

Cambridgesh.

Hunt’sh.

Herefordshire Worcestersh.

Warwicksh.

1

2

3

4

56

78

9

1

2

3 4

5

6

12

34 5

61

2 3

41

2

3

4

5 6

7 89

1011

1213

14

1 2

34

5

67 8 9

1011

12 13 14 1516

17 18 1920

21 22 23 24 25

26 27 28

2930

31

32 33

34

1 2

34

56

12

34 5

6

7

1

2 3

4

12

3

4 5

6 7

8910

11

12

13 14

15

12

34

5

6 78

9 1011

12

34

5 67

8 9

10

11

1

2 345

6

78

9

10

1

2

1

21

2 3

45

7

12

34

56

7

1

2 3

45 6

7

1

23

4

5

1

2

1

4

7

Page 174: A Comparative Grammar of Brtiish English Dialects

164 Lukas Pietsch

Map 8. The NSR in Late Middle English (based on data in LALME I: 467, IV: 110–111)

Shown below are typical SED attestations of relative clauses, including presentational relative clauses after existential there, most typically with zero relativizer and with non-concord in the preceding existential construction (23), as well as after cleft it’s (24).

(22) a. You don’t see many has holes now [SED: La7] b. Them what’s got a few [SED: Lei5] c. Hedges that hasn’t been done [SED: Lei9] d. The ones that goes across was braces [SED: R2]

© L. Pietsch 20030 50 100km

Linguistic Profiles

Regular Verbal -s

Occasional Verbal -s

Rare Verbal -s

Page 175: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 165

e. It kills the thorns as grows round it [SED: Nth2] f. I know several signs as is pretty sure [SED: Nth4] g. People what was used to it [SED: Hu1]

(23) a. There’s a lot of the people now as doesn’t talk like they used to[SED: L9]

b. There’s any amount takes cattle on [SED: Nb4] c. There’s not so many fills a ten-quart tin [SED: Cu3] d. There’s a lot of people kills ’em [SED: L15] e. There’s two or three comes up at five o’clock in the morning

[SED: Lei7]

(24) We’re both right, aren’t we, it’s t’others ’at’s wrong [SED: Y16]

Typical attestations with indefinite pronouns are shown under (25), and examples with demonstratives are shown under (26). The indefinite items include some, most, many, a lot, etc., often with postmodifiers like of them,sometimes also in determiner function followed by nouns. As for the demonstratives, the most typical item is them, but standard these/those also occur; in the Ulster and Scottish data a demonstrative determiner they (also spelled thae in written Scots, and not to be mistaken for the simple pronoun) occurs and has a similar effect. Examples such as these are extremely common in all parts of the data.

(25) a. some uses a jug and gets it broke [SED: R1] b. some says lop but I’d say slat [SED: L13] c. some on ’em reckons it in’t [SED: Lei1] d. some on ’em’s red [SED: L13] e. most on ’em has one [SED: L14] f. a good many keeps hens [SED: Lei10] g. some folks steals ’em [SED: L9]

(26) a. them’s ourn [SED: Lei9] b. them’s gisters in the top field [SED: Lei9] c. these is the front of these [SED: Lei2] d. I wonder if them two’s married [SED: L11]

Plural verbal -s in inversion is exemplified in (27). Note that the list includes several instances where ’em, a weak form of demonstrative them,

Page 176: A Comparative Grammar of Brtiish English Dialects

166 Lukas Pietsch

is used as a tag subject just like a simple pronoun they, i.e. apparently without emphasizing or deictic force. It nevertheless takes verbal -s,retaining its status of a full NP, and not of a simple pronoun, in terms of sensitivity to the Type-of-Subject Constraint. This can be taken as a piece of evidence that the Type-of-Subject Constraint is defined in terms of particular items and the constructions they are part of, not in terms of some abstract semantic or formal features defining ‘pronoun-ness’.

(27) a. Is them two married? [SED: Nb2] b. Who’s them? [SED: Nb3] c. Is thy teeth warking? [SED: Cu5] d. Where’s my yorks at? [SED: We4] e. Doesn’t ’em? [SED: La4] f. Is ’em? [SED: La4] g. Has thi taties comed up yet [SED: Y7] h. Is both you women wed? [SED: Y7] i. Has them horses been served? [SED: Y27] j. Where’s them come fra? [SED: Y28]

A special idiomatic type of inversion environment which appears to be particularly favourable to plural verbal -s, found mostly in Yorkshire data, is in tag clauses of the kind exemplified under (28) and (29), used as a postponed expression of a sentence topic. Note that these are typically preceded by a clause with a co-referential pronoun they, and that this first clause invariably has the appropriate verbal form without -s.

(28) a. They’re real hard gossips, is them. [SED: Y2] b. They’re rough mutton, is tups. [SED: Y3] c. They’re rum things, is the pigs. [SED: Y11] d. They’re a bit queer, is pigs to manage. [SED: Y12] e. They’re very affectionate, is pigs. [SED: Y25] f. They’re laced boots, is these of mine. [SED: Y18] g. They’re not worth bringing up, isn’t little pigs. [SED: La6]

(29) a. They vary, does stee-steps. [SED: Y18] b. They’ve recently comed, has them. [SED: Y18] c. They feel over ’em and weighs ’em, does butchers. [SED: Y22] d. They always crawl upwards, does lice. [SED: Y25]

Page 177: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 167

Two tokens of this type, both from the same informant, are also recorded in the more recent FRED data, both using was (and both closely preceded by rule-conforming they were):

(30) a. They were proper slaves, was women, in them days.[FRED:Yks9]

b. They were wicked, was farmers, for playing nap.[FRED:Yks9]

As was mentioned above, the Northern Subject Rule also interacts in interesting ways with the concurrent pattern of was/were levelling. Tokens of plural was with plural NP subjects are found throughout the area affected by the Northern Subject Rule. To the degree that they was is rarer or absent in these dialects, these forms can be ascribed to the Type-of-Subject Constraint. Such tokens occur even in areas which otherwise have strong preferences for generalized were even in the singular, such as the core generalized were area in south Yorkshire and the northwest Midlands. Examples are shown under (31). Note that many of these tokens also exemplify the common occurrence of subjects involving demonstrative them and other demonstrative pronouns.

(31) a. Shops was open while ten. [SED:Y20] b. Them’s sideboards what was cut off. [SED:Y25] c. Them that was on that ship ’at went down. [SED:Y27] d. Half on’em was apt. [SED:Y28] e. Horses was baiting. [SED:Y28] f. These was like that. [SED:Y32] g. Them pigeons was there. [SED:Ch2]

These findings fit in with observations reported by Wright8 as early as in 1892: in the dialect of Windhill in the West Riding of Yorkshire, plural verbal -s had largely become restricted to relatives and to the forms of have and be, and it occurred in other environments only rarely as a relic form (1892: 156). Wright also includes several examples with the subject themamong lists of typical occurrences:

(32) a. Them’s the men that does their work best. b. Them men’s been very good to me. c. Us that’s done so much for him.

Page 178: A Comparative Grammar of Brtiish English Dialects

168 Lukas Pietsch

d. Me that’s so poorly. e. The coals isn’t done yet.

(33) a. I’ve done. b. They’re at it again. c. Them men do their work very well.

Wright’s location is situated well north of the burglars steals isogloss in the SED. Although the SED data does attest the continued presence of verbal -sin that same area even outside the special favouring environments, more than half a century after Wright, his observation suggests the existence of a strong favouring tendency, and as such it fits in nicely with the observation in the SED that verbal -s was preserved longer, and further in the south, in just the same set of favouring environments.

The impression that environments of this type are over-represented among the occurrences of verbal -s cannot be tested statistically in the SEDdata, because the SED does not fully record all utterances of any informant through a given stretch of discourse. However, very similar effects are also found in the corpus data of both the NITCS and FRED. In both corpora, the relative clause constraint is solidly in evidence, as relative clauses take verbal -s up to twice as often as other clauses. This goes especially for thatrelatives and for the non-standard zero relatives that are common in these dialects. Wh-relatives, on the other hand, tend to co-occur more often with standard concord behaviour, apparently due to their stylistic status of being more characteristic of formal Standard English. A similar stylistic differentiation can also be found with respect to the demonstratives. Whereas the non-standard demonstratives them and thae have a very strong favouring effect on verbal -s, the effect of the standard demonstratives these and those is rather the opposite.

Potential candidates for plural verbal -s in inverted environments are quite rare in the corpus (since the overwhelming majority of inverted clauses have pronoun subjects and/or modal verbs), but in the few cases in point, verbal -s is strongly over-represented (7 out of 9 cases in the NITCS,and 11 out of 12 in FRED). Inverted clauses must therefore be counted as one of the most strongly favouring environments.

As for the indefinite pronouns, effects are difficult to quantify, as the class of constructions that exemplify this type is a rather open one. Its typical ingredients (some, a lot, ones, of them, etc.) can occur in a wide variety of combinations, and they also often co-occur with some of the

Page 179: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 169

other factors mentioned above. For instance, many instances of a lot are followed by a relative clause; ones is often preceded by a demonstrative them ones, etc. It is therefore not easy to find the most appropriate classification when it comes to testing the effects of these and similar items statistically, as the statistic method always requires the counting of tokens across a set of distinct, clearly defined environment types. The classifications used for the statistics below are therefore somewhat arbitrary. However, intuitively it seems to be the case that the effect is linked not so much to a precisely circumscribed set of structural environments anyway, but rather to a set of prototypicality conditions. It will be discussed in more detail in section 6 how effects of this type can be integrated into a theoretical model of grammatical variation. Going through the attested examples of plural verbal -s, one is left with the impression that sentences such as there’s a lot of people kills ’em represent something close to a common prototype, and that a sentence’s likelihood of having verbal -sdepends on its relative degree of similarity or dissimilarity with this or a small number of other prototypes. Among the conditions that seem to play a role here is the subject’s role of being focussed, unlike in the most common constellation of a clause with a pronoun subject, where the subject is typically topical and unstressed.

However, apparent effects of favouring verbal -s may also be linked quite arbitrarily to specific items in the syntactic or semantic environment. For instance, it was found in the NITCS data that sentences with the subjects times and days displayed a far higher than average proportion of verbal -s, as in the examples below (34–35). This usage type must be characterized as a special stylistic idiom.

(34) a. The times is better in a way, like, as regards money. [NITCS:L4.3]

b. Oh, aye, times is a whole lot better now. [NITCS: L15.2] c. Oh, the times is a-changed very much. [NITCS: L36.3] d. The times was very bad. [NITCS: L8.3] e. If the people had sense, the times is perfect. That’s what the

times is. [NITCS: L8.3]

(35) After all, my young days was far better. [NITCS: L10.3]

Quite prominent among the factors favouring or disfavouring verbal -s is, not surprisingly, the type of verb involved. In the FRED data, a moderate

Page 180: A Comparative Grammar of Brtiish English Dialects

170 Lukas Pietsch

overall preference for plural was (used in 51 per cent of all cases of past-tense be) is in evidence in all parts of the corpus. Is is also used slightly more often (35 per cent) than the -s forms of other verbs (31 per cent). It is perhaps remarkable that this effect is not even stronger. Judging from some recent descriptions of northern varieties – e.g. Tagliamonte (1999) on York English, or Miller (2004) on the current spoken language in the Central Lowlands of Scotland – one might expect the trend towards a restriction of the Northern Subject Rule to was/were to have progressed much further. These studies attest effects of the Northern Subject Rule to be preserved only for be. As for modern Scots, Görlach (2002: 95) suggests that the Northern Subject Rule may have become a victim of social stigmatization in recent decades, despite its long tradition in Scots, because of its overlap with the non-standard verbal -s in other colloquial varieties of English. He reports that it is even avoided in some modern Scots writing. This stigmatization may explain the presence of some speakers in the FREDcorpus who consistently lack verbal -s but otherwise display many quite distinctively Scots grammatical features in their speech. The very solid presence of Northern Subject Rule effects both with be and with lexical verbs in many other speakers in the corpus confirms that FRED, on the whole, represents a range of predominantly very conservative types of speech.

In the NITCS data the situation is more complex, since internal geographic diversification was found within Ulster in this respect. Towards the northeast, in what has been known as the area of traditional ‘Core Ulster Scots’ dialects (Gregg 1972), the Northern Subject Rule seemed to apply without much distinction to all verbs, with relative frequencies of verbal -s at a uniform but only moderate level (around 21 per cent) for all verbs alike. Along the southern border of Ulster, plural was was found much more frequently (56 per cent) than other -s forms, including is (24 per cent). The verb have seemed to play another special role, showing exceptionally low levels of verbal -s in this area (5 per cent). Interestingly, the highest levels of verbal -s usage overall were found in neither of these two areas, but in the transitional zone between them, in central and southeast Ulster (see Pietsch 2005: ch. 5.3.1 for details).

Table 5 and Table 6 show the relevant statistics for the most important linguistic environment factors in terms of two (partial) Varbrul models calculated separately for the FRED and NITCS data. For technical purposes, the definition of the factors and factor groups is not exactly identical in both models, but the similarity of the results with respect to the

Page 181: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 171

principal effects discussed so far can easily be seen. (For details regarding the design of the models, interaction tests between the various factor groups, and the statistical integration of the social and geographic factor groups with the linguistic factor groups shown here, see the discussion in Pietsch 2005).

Table 5. Partial Varbrul model for NSR in FRED

Factor Group Factor Tokens -s/Total

% Factor Weight

Verb Was/Were 215/420 51% 0.58 Is/Are 32/92 35% 0.42

Other Verbs 52/167 31% 0.34 Clause Inverted 7/9 78% 0.91 Zero/That Relative 64/90 71% 0.78 Canonical SV (Adjacent) 198/475 42% 0.46

Wh- Relative 9/26 35% 0.37 Canonical SV(Non-Adjacent) 21/79 27% 0.35 Subject Them 19/29 66% 0.53 Indef. Pronouns 61/129 47% 0.56 Other Demonstratives 13/43 30% 0.31 Other NPs 206/478 43% 0.50 Total 299/679 44%

Summing up, there seems to be strong evidence that the use or non-use of verbal -s in the variable grammars of present-day dialect speakers is governed by a set of prototypicality conditions. Certain types of environments, defined in syntactic, lexical, or possibly also semantic terms, are associated with a relative preference for the use of the conservative dialectal option of verbal -s. The likelihood for any particular clause to display verbal -s seems to depend on its degree of closeness to one of these prototypes. The relevant conditions may range from very general, productive patterns (such as the relative clause or inversion constraints) to highly specific (such as the times constraint detected in the NITCS data). In diachronic terms, these effects may be described as a gradual loss of productivity of a once general, universal pattern, which may in the long run lead to its becoming fossilized in a highly restricted set of environments. In synchronic terms, the effects in the grammars of individual speakers may best be characterized as structural idioms: arbitrary properties associated

Page 182: A Comparative Grammar of Brtiish English Dialects

172 Lukas Pietsch

with specific construction types, which range somewhere in between fully general syntactic rules on the one end, and individual lexical properties on the other.

Table 6. Partial Varbrul model for NSR in the NITCS

Factor Group Factor Tokens-s/Total % Factor Weight

Clause Type Inverted 11/12 91% 0.97 Zero relatives 37/60 61% 0.79

That relatives 38/79 48% 0.58 Canonical SV 227/751 30% 0.46

Wh- Relatives 3/36 8% 0.23 Subject-VerbDistance

Non-adjacent 52/122 42% 0.62 Adjacent (and

zero subjects) 253/804 31% 0.48

Determiner of Subject NP Demonstrative

them/thae,interrogatives

30/38 78% 0.86

Quantifiers + ofthem

42/95 44% 0.69

Others 235/746 31% 0.47 Demonstrative

these/those9/59 15% 0.31

Nominal Head of Subject NP

Times/Days 18/21 85% 0.86 Ones 19/34 55% 0.69

Other nouns 192/643 29% 0.47 Total 316/938 34%

Page 183: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 173

4. The history of the Northern Subject Rule

So far I have dealt with the variable grammars of recent dialects, and centrally with the reflexes they preserve of the Northern Subject Rule, which was shown to be one of the main characteristic patterns of the north. In the present section I will explore what can be reconstructed of the historical development of this pattern. Unfortunately, much of its historical origins lie in the dark. Its first appearance in the dialects of northern England must be dated to the time of early Middle English, a period from which no written documents of northern provenance are extant. In the latest surviving Northumbrian Old English documents from before this gap (the Lindisfarne Gospels, the Rushworth Gloss, and the Durham Ritual, all mid-tenth century), there are signs of an ongoing change that can be understood as a corollary and prerequisite of the emergence of the later rule, namely the change from the verbal -eð/-að/-iað/-is affixes to neutralized -es. In the earliest reliable Middle English documents from after the gap, from c.1300 onwards, the Northern Subject Rule is already fully in place.

It has been suggested that the emergence of the Northern Subject Rule may have been a result of earlier contact of English with Brythonic Celtic (Klemola 2000). The idea is tempting because of certain typological parallels between the northern system and the concord systems of Welsh and related languages. Just like northern English, Welsh displays non-agreement with full NP subjects but agreement with accompanying personal pronouns. (For the different case of a rather more elusive parallel between the northern English concord system and that of Irish see the discussion in Corrigan 1997: 190–225 and Pietsch 2005: ch. 2.1.2.) However, I have argued elsewhere (Pietsch 2005: ch. 3.4) that a transfer explanation of the kind envisaged by Klemola is not tenable. This is mainly due to problems with the relative timing between the supposed period of contact and certain dateable linguistic changes such as the affix neutralization mentioned above, which are logical prerequisites of the emergence of the Northern Subject Rule and must therefore predate it. In what follows, I will instead sketch a possible alternative model of how the pattern could have evolved. It is based primarily on language-internal causal factors such as analogy and frequency-induced change.

It is, first of all, crucial to point out that the verbal -s forms in those environments where they do not match modern Standard English are not an innovation. They were not, as one might think, an intrusion into these positions from the third person singular. Rather (for most verbs, at least)

Page 184: A Comparative Grammar of Brtiish English Dialects

174 Lukas Pietsch

these -s forms are a conservative retention. Etymologically, verbal plural -s is no less genuine a reflex of an original agreement marker (northern OE plural -að>-as>-s) than third person singular -s (northern OE -eð>-es>-s).The main innovation thus lies not in a spread of the suffix but in the spread of the suffixless forms, in the environments with adjacent pronoun subjects. This innovation can plausibly be seen as part of the general drift of affix loss that affected all the Germanic languages. It may appear somewhat unexpected that it should have been the northern dialects that eventually preserved more of the traditional affixal system than others, in this particular domain of the grammar. After all, it is widely agreed that the trend towards affix loss was, on the whole, particularly vigorous in the north (possibly owing to a situation of intensive language contact with Scandinavian). But it must also not be overlooked that the apparent conservatism of the northern dialects in this particular respect only applies to a last remnant of the affixal system, and not to its actual function of agreement marking. Paradoxically, in keeping the affixed verb forms, northern English eventually preserved more of the phonological substance of the old agreement system than the standard did, but only at the price of having in effect less agreement.

This paradoxical outcome may be interpreted as the result of a ‘conspiracy’ of two independent developments originating in different dialects. One of them was the weakening and subsequent neutralization of a set of previously distinct but phonologically similar affixes (-eð/-að/-iað/-is> -s). This development originated in the north and was well advanced by late Old English. The other was the innovation of affixless, so-called syncopated forms, at first only in a certain restricted set of syntactic environments adjacent to pronouns. This development was apparently headed by the southern dialects and only began to reach the north at some time during late Old English. At this point, the previous neutralization process had already brought the older, fuller agreement system to the verge of breakdown in that dialect, having obliterated almost all distinguishing contrasts in the verbal paradigm. The new affixless forms that were the output of the second innovation were therefore apt to be reinterpreted and pressed into service as carriers of a new agreement contrast. A somewhat more detailed description of the whole process, as far as it can be reconstructed, is in order at this point (see Figure 1). The northern dialects of Old English originally shared the common Old English agreement system (Brunner and Sievers 1965: §352–378), which can be summed up as follows. The three persons in the plural had already been

Page 185: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 175

Syncopation

Midl. repl. -ð > -n

Affix

Spread of -s to

Northern Subject Rule

syncretized in pre-Old English and were -að or -iað in the present indicative of lexical verbs, -en or -un in other paradigms. The singular forms in the present indicative were originally the common Germanic forms -u, -is, -ið (in the case of the strong verbs; the endings in the weak classes were similar.) They underwent a process of vowel weakening to -e,-is, -eð during Old English. Despite this, the plurals remained distinct from the third singulars for some time (strong verbs -að versus -eð; weak verbs -iað versus -eð or -að respectively). In the course of the Old English period, the second singular added a -t affix after the -s, owing to a reanalysis of the cliticized subject pronoun ðu.

Figure 1. Development of agreement paradigms in Old and Middle English

During the transition to Middle English, the three principal dialect areas of England behaved differently. The Midlands dialects replaced the plural -aðendings with -en, apparently through analogy with the past and subjunctive paradigms. Through later phonological reduction of the endings, these -en

North

South

Spread of affixless forms

Middle English Modern English

sing-e sing-es(t)sing-eð sing-að sing-að

sing-e sing-est sing-eð sing-að sing-að sing-að

sing we sing ge

sing sing-est sing-eð sing-en sing-en sing-en

sing sing-est sing-ssing sing sing

sing-e sing-es sing-es sing-es sing-es sing-es

sing-s I sing sing-ssing-ssing-s we sing sing-s you sing sing-s they sing

sing sing-ssing-ssing-ssing-ssing-s

sing we sing ge sing ðai

we sing ge sing ðai sing

Page 186: A Comparative Grammar of Brtiish English Dialects

176 Lukas Pietsch

forms developed directly into the modern affixless forms of Standard English (cf. Figure 1, bottom). The south retained the -ð forms throughout, only to replace them with universal generalized -s much later in the modern dialects.

The north replaced the -ð forms, both in the plural and in the third singular, with -s, a change that occurred already during the Old English period. This change has repeatedly been linked to Scandinavian influence (Keller 1925, Samuels 1988). This idea has met with considerable skepticism (for a survey of the older literature see Brunner 1962: 177), but it seems to have gained ground again in recent decades in the light of current language contact theories (Bailey and Marold 1977: 45, Thomason and Kaufman 1988, Stein 1986; for a recent sceptical view see Ferguson 1996: 178). The older northern documents, such as Cædmon’s Hymn, the Leiden Riddle, and – with one notable exception – the runic inscriptions, all have the old -ð forms. The late Northumbrian documents from the mid-tenth century are witness to a stage where -ð and -s were used variably. The variation in these documents has repeatedly been the object of quantitative analyses (most thoroughly by Berndt 1956, but see also Holmqvist 1922, Ross 1934, Blakeley 1949 and most recently Stein 1986). These studies suggest that the variation was conditioned, among other things, by phonetic environment factors, but also already by a tendency to treat pronoun subjects differently from full NP subjects; hence in some way foreshadowing the later Type-of-Subject Constraint.

During the same time when -ð changed to -s in the north, the vowels in the two endings also lost their contrasts. The new -st ending in the second singular appears variably in the late Northumbrian documents, but seems to have been only a temporary intrusion from the south, and was then again replaced by -s. The -e ending in the first singular became mute. Taken together, these changes meant that by the time of early Middle English the present tense paradigms of lexical verbs must have contained only two distinct forms, - and -(e)s. In the present subjunctive, the neutralization had gone even further, through the loss of final -n, so that only one form -was left. In the present indicative, a hypothetical Late OE or Early ME stage must be posited in which all forms except the first singular had been completely neutralized in -s (cf. Figure 1, top left).

At about the same time, the second of the two innovation processes mentioned earlier seems to have set in: new affixless forms must have developed in the pronominal environments in the plural. The exact nature of this process is not known, but the most likely source is found in the

Page 187: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 177

southern dialects of some centuries earlier. In the West Saxon standard form of Old English, the original plural -að and -en affixes were sometimes deleted before the first and second person plural pronouns we and ge, when these were in an immediately post-verbal, probably clitic, position (cf. Figure 1, bottom left). Brunner and Sievers (1965: §360) suggest that these so-called syncopated forms first occurred in the subjunctive. Here they would be explained easily enough as an early application of the overall trend of deleting final -n. They would then have spread by analogy into the indicative. Why in southern Old English these syncopated forms remained restricted to the first and second person plural and did not also spread to the third person is another open question but need not concern us here. What is important is the fact that the syntactic positions these forms occurred in are a subset of those where the affixless forms occur in the later northern dialects.

Curiously, northern Old English, although more progressive in matters of affix reduction in most other respects, seems to have been slow to pick up these syncopated forms at first. In the surviving tenth century manuscripts they are attested, but marginal (Berndt 1956: 213–216). It can be no more than a matter of speculation what exactly triggered the sudden and much wider spread of these affixless forms during the following three centuries. In any case, once the affixless forms did start to appear in the north, they must have spread fairly quickly into all positions with pronominal subjects, irrespective of their pre- or postverbal position. Moreover, the affixless forms were also adopted in the third person plural, where they had not occurred in the older southern dialects, and where the northern dialects had in the meantime also innovated a new personal pronoun, the Scandinavian loan they. (cf. Figure 1, top centre). This process had approached completion by 1300 in the north, when written documentation of that area sets in again.

The functional motivation for the quick and thorough adoption of the affixless forms seems to have been that they were co-opted to serve as genuine agreement forms, re-introducing and stabilizing a formal contrast that was no longer signalled reliably by the older set of affixes. However, the newly introduced contrast remained restricted to pronominal environments, as it had been a subset of the pronominal environments where the affixless forms had originally emerged. Once they had assumed their new role as a carrier of agreement marking in these environments, they stopped spreading further across the system. Thus had emerged a relatively stable new system with an effective split of the once uniform

Page 188: A Comparative Grammar of Brtiish English Dialects

178 Lukas Pietsch

system of subject-verb agreement into two distinct paradigms valid for two complementary sets of environments: a contrast of -s vs. - with pronoun subjects, and neutralized -s everywhere else. In a certain sense, this paradigm split can be seen as a major structural innovation in the history of English, comparable perhaps to the common Germanic split of the adjective inflection into the strong and weak paradigms: “the occasional clear instance of an innovation that goes contrary to the main drift and persists along with it for long periods of time” (Ferguson 1996: 189).

Finally, the reorganization of the concord system was completed by two further developments. The first person singular – which had had no consonantal affixes from the outset – introduced new and unetymological -sforms in positions not adjacent to the subject, obviously by way of analogy with the other persons and numbers. Then, also by way of analogy, the usage of is/are and of was/were was partly aligned with that of the other verbs, as was and is gradually began to be used in plural environments in the same way as the -s forms of other verbs (Montgomery 1994). At this point, the ‘ideal’ Northern Subject Rule system as described in definition (1) above had been reached.

By the late Middle English period, the Northern Subject Rule had established itself as a more or less categorical pattern across all of northern England and Scotland (Mustanoja 1960: 481–482). There is some evidence that the pattern had been spreading from the north into the northeast Midlands, especially Lincolnshire, during the thirteenth century (Berndt 1982: 131). From then on, its southern boundary, roughly along a Chester-Wash line, was to remain stable up to the traditional dialects of the twentieth century, as was shown earlier in Section 3.6 (see Map 7 and Map 8 on pages 163–164).

Laing (1978: 244–247) and McIntosh (1988) draw attention to several types of intermediate systems that occurred along the boundaries of the northern area. In all of them, there was an alternation between two forms that was governed by similar constraints as in the northern system proper, but involved different morphological material (-s, -th, -en, or - ). It seems not entirely clear in all cases to what degree each of the affixes in these intermediate systems represents etymologically regular reflexes of older affixes in the respective dialects, or later analogical reintroductions triggered through contact with the northern system proper (cf. also Schendl 1996: 149).

Plural -s forms apparently conforming to the Northern Subject Rule are also widely attested in the emerging Early Modern English standard

Page 189: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 179

language, including the language of Shakespeare (Knecht 1911). This is usually interpreted as a dialect contact effect brought about by the massive amount of migration from the north into the London area (Schendl 1996, 2000; cf. also Bailey, Maynor and Cukor-Avila 1989; Bailey and Ross 1988; Montgomery, Fuller and DeMarse 1993; Wright 2002).

The Northern Subject Rule was brought to Ireland, especially to Ulster, by settlers both from Scotland and England (McCafferty 2003, 2004). Later, it was to be transported by Scottish, Irish and English settlers to many of the new overseas varieties of English. There has been some debate about just to what degree certain commonalities in verbal concord variation discernible across a wide range of overseas Englishes today can historically be ascribed directly to a common, specifically northern, dialectal source, or to what degree these similarities reflect universal, functionally explainable trends in English (see e.g. Montgomery 1988, Bailey and Ross 1988, Montgomery, Fuller and DeMarse 1993, Tagliamonte 2002, Chambers 2004). While strong northern input seems historically plausible for many of the varieties in question, there is also some evidence that variation patterns resembling the northern rule may have developed independently in some places, even within England. For instance, some studies of present-day southwestern varieties have found quantitative (though non-categorical) constraints similar to the Type-of-Subject Constraint to be active (Godfrey and Tagliamonte 1999, Peitsara 2002). The apparently independent rise of such structures strengthens the case for some dialect-universal mechanism (“vernacular primitives” in the terms of Chambers 2004).

5. Theoretical accounts of the Northern Subject Rule

There have been several attempts to characterize the northern concord system within formal theories of syntactic competence, using various models in the tradition of formal grammar. Besides some very brief discussion in Roberts (1993) and Van Gelderen (1997), and a somewhat inconclusive treatment in Corrigan (1997), I am aware of two more elaborate proposals: Henry (1995), and Börjars and Chapman (1998). A third and rather different approach, also within a rigidly formalized framework but with a radically different stance regarding linguistic universals, can be found in a brief sketch in Hudson (1999). However, each of the existing formal proposals is burdened with problems on the level of descriptive adequacy.

Page 190: A Comparative Grammar of Brtiish English Dialects

180 Lukas Pietsch

The difficulty which formal theories have had in dealing with the Northern Subject Rule is demonstrated by the fact that the existing proposals have mostly concentrated on either the Type-of-Subject Constraint or the Position-of-Subject Constraint, according to some descriptive definition or other, but none of them has succeeded in integrating both in a unified account. A part of the problem seems to be due to a somewhat puzzling feature of the Northern Subject Rule itself, which I have elsewhere called the “markedness paradox” (Pietsch 2005: ch. 1.2.4). All the existing formal analyses implicitly operate with the concept of marked and unmarked forms (even when they do not explicitly use that term). However, which of the two forms involved in the dialectal concord system (-s, - ) is the marked member of the paradigm and which is the unmarked one?

In Standard English, the -s form unambiguously encodes one particular person-number value, the third singular, whereas the - form acts as the default for all other persons and numbers. It is therefore fairly unproblematic to characterize the -s suffix as an agreement morpheme in the normal sense, and hence, the -s form as the marked member of the paradigm. This is a straightforward analysis even though the system may be a rather uncommon one in terms of a cross-linguistic comparison, as the third singular is generally the most likely candidate for zero marking within agreement paradigms (Bybee 1985). Matters are rather different in the northern system, even though in personal-pronoun environments it is identical to the standard, encoding the same grammatical information with the same means. In all non-pronominal environments it is the generalized -sform that leads to a complete neutralization of all person-number agreement contrasts. It can therefore be argued that the -s form in the NSR system, though formally carrying the overt agreement morpheme, acts as a functionally featureless form devoid of person-number information. The -smorpheme has repeatedly been characterized instead as a mere tense-mood marker. In contrast to this, the formally unmarked - form, where it occurs, has the effect of upholding agreement oppositions, particularly that between singular and plural in the third person. It is therefore usually regarded as the one that functionally does carry genuine person-number agreement features. This is a plausible synchronic analysis for the modern system, even though at the earliest stage, at the time when the zero forms first emerged, the system must have been just the reverse. As described in section 4, the zero forms were originally a product of the erosion of the agreement morphology. They were reanalysed as genuine plural agreement

Page 191: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 181

forms, taking on a new functional load as carriers of agreement information, only after the two formerly distinct endings -eð/-es and -að/-ashappened to fall together and were re-analysed as default singular forms. The resulting, somewhat paradoxical distribution of markedness properties will be seen as a recurrent problem that has affected several synchronic and diachronic accounts of the Northern Subject Rule.

5.1. Henry (1995)

Henry’s (1995) discussion of plural verbal -s (‘singular concord’, in her terminology) is part of an extensive analysis of the syntax of contemporary Belfast English, based on a late principles-and-parameters version of generative grammar. As such, it is an example of the type of variation studies that has come to be known as ‘microparametric syntax’ in recent generative work. Henry’s data are based on acceptability judgements elicited in interviews with Belfast informants. While Henry devotes extensive discussion to assumed parametrical links between the phenomenon of optional plural verbal -s and some other non-standard grammatical phenomena in this dialect, she makes no attempt at integrating her analysis of the narrowly local variety of Belfast with a historical or larger geographical account, with respect to the continuity of the phenomena in question across a wider range of varieties. As to her methodology, it must be noted as unfortunate that Henry fails to give any detailed account of the number and choice of informants, her methods of sampling, or the set of test sentences used. Neither does she give any detailed quantitative or qualitative account of how her informants reacted to each test sentence.

Henry’s account of the reflexes of the Northern Subject Rule in Belfast English differs descriptively in some respects from descriptions found elsewhere in the dialectological literature, and it is sometimes difficult to judge whether these apparent differences reflect genuine divergence between the dialects in question, or whether they are merely due to diverging research methodology. Most strikingly, Henry’s analysis does not deal with the Position-of-Subject Constraint in the form it is usually stated in the descriptive literature. Neither the relative clause structures of the type they that goes, nor the ‘classical’ northern pattern they sing and dances are mentioned in her analysis; nor does she say anything about the effect of intervening quantifiers or adverbs, as in they both goes, they often

Page 192: A Comparative Grammar of Brtiish English Dialects

182 Lukas Pietsch

goes (the latter structures admittedly being rare in Northern Irish English but by no means non-existent, at least in the traditional dialects surrounding Belfast). Interestingly, Henry does report (1995: 19, 26) on an apparent effect in exactly the opposite direction: with adverbs intervening between a full-NP subject and an operator verb (“raising verb” in her terminology, that is, a verb that raises to INFL before spellout), verbal -s is reported notto be licensed. This means that contrary to the common pattern observed elsewhere, non-adjacency of subject and verb in this case would favour the standard agreement pattern. Thus:

(36) a. The children really are late. b. The children is late. c. *The children really is late.

Henry explains (1995: 19) that “it seems to be the case that the adverb position between the subject and the topmost projection of INFL which exists in English is unavailable in singular concord”. She does not state explicitly whether the same kind of environment has any influence on agreement with personal-pronoun subjects (they really ?is/?are late), but from her statement it would seem that a fortiori the -s forms should be ruled out in that case too. A comparison of her observations with some data from the NITCS confirms the possibility that verbal -s in some of these environments may indeed be rare or avoided (Pietsch 2005: ch. 2.1.1). However, since the adverb position Henry is speaking of tends to be reserved for a rather restricted set of semantic-pragmatic environments in English (Quirk et al. 1991: §8.18), it does not become clear from Henry’s discussion whether her informants’ reluctance to accept verbal -s in these positions really was caused by some property of the phrase structure as such, or by other factors such as, for instance, stress conditions on the verb form, or possibly by the mere rareness of the construction type. It would seem that Henry’s set of test sentences, and her methodology in reporting informants’ actual reactions to them, are somewhat too limited to draw far-reaching structural conclusions on this particular issue.

A second, interesting constraint that Henry reports concerning the position of subjects is a prohibition of verbal -s under inversion (1995: 16–18). This would exclude sentences of the type does the children sing? in Belfast English. This constraint is not confirmed in the data of the present study: as pointed out in section 3.6 above, sentences of this type are regularly attested in conservative varieties of the northern type both in

Page 193: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 183

Northern Ireland and in Britain. They were even found to be among the environments most favourable to verbal -s. While the overall textual frequency of potential application environments of this type in the corpora is quite low – the bulk of all question clauses actually produced contain either pronoun subjects, or modal non-concord verbs, and thus fall outside the scope of the rule – the relative frequency of verbal -s in the few tokens produced is extremely high. Again, it seems unfortunate that Henry does not give any detailed account of her test procedure and results. The question whether these contrasting findings represent a genuine difference between present-day Belfast English and other dialects of the northern type, or just an artefact of diverging research methodologies, cannot be discussed in more detail here.

As far as the Type-of-Subject Constraint is concerned, Henry (1995: 23) finds that verbal -s outside the third person singular is possible in Belfast English whenever subjects are not openly marked as nominative. This formulation of the Type-of-Subject Constraint covers the prohibition of verbal -s with the simple personal pronouns I, we, and they. The pronoun you must be treated as an exception, as it fails to have an overt distinction between subject and object case but nevertheless patterns together with the other three pronouns with respect to the Northern Subject Rule. Henry takes as a strong piece of evidence for her claim that, according to her observations, the prohibition of verbal -s with these items does not apply only to the standalone pronouns but also to co-ordinated NPs involving any of them. The strong forms of the pronouns, which are identical to the accusative forms, are not subject to such a constraint. Co-ordinated NPs involving these can freely take ‘singular concord’:

(37) a. You and I are going. b. You and me are going. c. *You and I is going. d. You and me is going.

Henry reports that her informants found sentences of type (37c) “completely ungrammatical” to a degree which, as she argues, excludes the explanation that the ungrammaticality judgment could be due only to a “sociolinguistic mismatch – with singular concord being a non-standard feature, and thus sounding strange when placed together with the formal, prestige form involving co-ordinate pronouns in the nominative case” (1995: 24). Consequently, Henry makes the ‘nominative constraint’ the focus of her analysis.

Page 194: A Comparative Grammar of Brtiish English Dialects

184 Lukas Pietsch

In the principles-and-parameters framework adopted by Henry, agreement phenomena are assumed universally to involve ‘checking’ in a specifier-head configuration between the two elements. Therefore, the question of agreement or non-agreement logically leads to the question “whether the subject is in fact in SPEC/AGRSP in this structure” (1995: 21). Henry discusses, and rejects, an analysis according to which subjects of ‘singular concord’ clauses (those with non-standard verbal -s) would remain in VP, while those of standard agreement clauses would raise to SPEC/AGRSP. Her own proposal is similar but involves a different set of nodes in the tree: while subjects and verbs of standard agreement clauses raise to SPEC/AGRSP and AGRS respectively, just as in Standard English, subjects and verbs of ‘singular concord’ clauses raise only to one node lower in the tree, SPEC/TP and T (‘tense’) respectively. This reflects the intuition that the generalized verbal -s of the ‘singular-concord’ clauses is a mere tense marker, and that only tense features but no agreement features are involved in this position. Following common assumptions in generative work on English, the movement of the verb is supposed to happen overtly (at ‘spell-out’) only with the ‘raising verbs’ be and have, while all other verbs undergo the raising operation only at the level of LF. As for the subject, the T node is also assumed to be unable to assign nominative case to the subject in its specifier position, as this is an exclusive property of AGRS. However, Henry stipulates that T in this dialect has the property of being able to assign ‘non-nominative’ or ‘default’ case. Therefore, all subjects except those morphologically marked for nominative can appear in this position, and this explains the ‘nominative constraint’. The question of why and when the elements are forced to raise to either of the two target positions is handled, following the preferred mode of thought in recent generative work, by assuming that features are either ‘weak’ or ‘strong’. In this case, the NP features of AGRS can optionally be either weak or strong in the dialect, whereas in Standard English they are always strong.

One argument Henry adduces for her analysis (1995: 29) is the behaviour of negative-polarity items (NPIs) in this dialect, which she claims to be parametrically linked to the verbal concord behaviour. She finds that non-standard structures such as (38) are possible in Belfast English:

(38) a. Anybody wouldn’t be able to do it. b. I was surprised that anybody didn’t go.

Page 195: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 185

However, in the plural such negative-polarity subjects can only occur if there also is ‘singular agreement’:

(39) a. Any friends isn’t coming. b. *Any friends aren’t coming.

Following the generative tradition, Henry explains the licensing of negative-polarity items in terms of their being c-commanded by a raised negative operator on the level of LF. Her point is that for theory-internal reasons raising of the negative element above the negative-polarity subject is easily conceivable if the subject is in SPEC/TP, but not so easily if it is in SPEC/AGRSP. Again Henry seems to exclude the possibility that the perceived co-occurrence restriction between singular concord and negative-polarity subjects could be due simply to stylistic mismatches of elements from different registers:

That there is a link between singular concord and NPI [i.e. Negative Polarity Item] licensing in subject position seems clear from the the [sic] fact that the two phenomena seem to go together in speakers’ grammars; those speakers who allow singular concord also permit NPIs in subject position, and conversely non-users of singular concord find NPIs in that position strongly ungrammatical. This is a clear case where careful examination of dialects or sub-dialects can help to show whether proposed connections are real or not, and where it is important to check what co-occurrence constraints there are on dialect features. (1995: 29–30)

It is not my intention to comment on the technical, theory-internal merits of Henry’s proposal. While it is the most comprehensive and probably the empirically best argued formal analysis so far, it must be pointed out here that it fails to provide a unified analysis of the whole range of (contemporary or older) concord systems of the northern type. The presumed parametrical link between the Northern Subject Rule and the negative-polarity scope behaviour is just such a case where more “careful examination of dialects or sub-dialects” (to use Henry’s own words) would have been a welcome addition: Henry makes no attempt at establishing whether the two phenomena do indeed go together in other related varieties outside Belfast. A similar and even stronger argument holds for the ‘nominative constraint’. Even if that rule is descriptively adequate for those Type-of-Subject effects which Henry finds in her data, there does not seem to be a straightforward way how her analysis should carry over to varieties where the Position-of-Subject Constraint played a more prominent role.

Page 196: A Comparative Grammar of Brtiish English Dialects

186 Lukas Pietsch

Such varieties, after all, do allow overtly nominative pronouns to co-occur with verbal -s, in non-adjacent positions. It is difficult to see how Henry’s analysis could be adapted to account for varieties of this type, even by positing a couple of additional parameter settings. This, of course, does not logically invalidate Henry’s analysis as such. It may in principle be the case that present-day Belfast English differs fundamentally from northern systems in other (or older) dialects; in that case, Henry’s analysis of the former would stand but explanations along rather different lines would have to be sought for the latter. However, an argument in the other direction holds. If there was an explanation that could account for the ‘classic’ older systems with their stronger Position-of-Subject effects, such as Older Scots or late northern Middle English, then this explanation would very likely be able to account also for the more restricted range of nonstandard verbal -s in Belfast English in a unified way. Such a universal explanation would then certainly be preferable to the parochial one Henry proposes for Belfast English.

5.2. Börjars and Chapman (1998)

As we saw in the last section, at the heart of Henry’s proposal is the idea that the collocations of pronouns and affixless verb forms represent real agreement whereas the constructions with the -s form do not. This is entailed by the assumption that the former but not the latter involve checking of identically marked features of pronoun and verb in a Specifier-Head constellation under AGRSP. The proposal by Börjars and Chapman (1998) implies the exact opposite. They interpret the co-occurrence restriction on pronouns and affixed verbs in terms that might be described as ‘anti-agreement’ (in the terms of Corrigan 1997: 203, quoting Roberts 1997: 109), that is, the avoidance of double marking of some feature. In this perspective, then, it is the -s forms which represent the real agreement marking, and the - forms represent the lack of such. Obviously, the feasibility of these two diametrically opposed approaches is a consequence of the ‘markedness paradox’ as described earlier. Yet, none of the authors involved seems to be aware of the possibility of the opposite view.

Börjars and Chapman (1998) discuss the northern concord system from the perspective of Lexical-Functional Grammar (LFG). They propose a formal model based on the intuition, first developed in Chapman (1998), that pronoun subjects and adjacent verbs stand in a closer syntactic relation

Page 197: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 187

with each other than other subjects and verbs, and that together they form an integrated unit in syntactic processing – in the words of Chapman (1998: 39): “a syntactic unit which is interrupted if additional information is added in the form of a second pronoun, for example, or some sort of modifier between the pronoun and verb”. This proposal, then, differs from the preceding one in explicitly focussing on the effects of the Position-of-Subject Constraint, as illustrated in (40).

(40) a. They often talks. b. They talk.

When seeking a formal explanation for the contrast in (40), one must resort to either of two principal strategies: either posit that the two instances of they in (40a) and (40b) are in two different structural positions in the syntactic tree (which would not be the case in the corresponding sentences in Standard English), or posit that the two instances of they are somehow not the same thing. The analysis proposed by Börjars and Chapman combines both these options. As for the tree positions, they choose an analysis essentially opposite to Henry’s: whereas Henry assumes the subjects of the -s verbs to be structurally closer to the (original) position of the verb in the tree configuration, namely the T node instead of the AgrSnode, Börjars and Chapman ascribe a closer position to the subjects of the affixless verb forms. As for categorial status, they assume the existence of two partly homonymous sets of pronouns with different roles: those which co-occur with an -s ending on the verb: he, she, it, I1, you1, we1, they1; and those that occur without an -s ending: I2, you2, we2, they2. The central point of their proposal is that this latter set has the categorial status of verb inflections, more exactly: “pronominal inflections” – they are really not part of the syntax but part of the verb morphology (1998: 76, 83). In what follows I will adopt the authors’ convention of writing these constructions with hyphens to indicate this presumed affixal status: they=talk.

Börjars and Chapman go on to specify the necessary properties of the two sets of pronouns in terms of feature settings in the formal mechanism of Lexical-Functional Grammar. These properties entail the three most important aspects of their behaviour: that the second set must always be adjacent to the verb, that it cannot co-occur with another competing inflectional affix on the other side of the verb stem (namely, the -s affix) and that neither of the two sets can co-occur with each other or with another overt subject. In sentences with ‘pronominal inflection’, the affixal

Page 198: A Comparative Grammar of Brtiish English Dialects

188 Lukas Pietsch

pronoun is generated under the V node, and these sentences therefore lack a subject in the canonical subject position, the sister of the VP. They are therefore, technically speaking, pro-drop structures. This generates the following grammatical structures:

(41) a. [NP they1] [VP often [V talks]] b. [NP e ] [VP [V they2=talk]]

Structure (42), on the other hand, containing both the -s ending and a pronominal inflection they2, is correctly predicted to be ungrammatical:

(42) *[NP e ] [VP [V they2=talks]]

There are various problems with this analysis. First, as pointed out by Shorrocks (1999: 112), the assumption of affixal or clitic status intuitively conflicts with a central property usually connected with affixes (or clitics, for that matter): phonological dependence. While it may be true that preverbal pronouns in connected speech often tend to be realized as phonologically reduced forms (Börjars and Chapman 1998: 97), these pronouns can nevertheless also take contrastive stress, and will then still occur with the non-inflected verb form in NSR dialects. A second problem concerns the descriptive adequacy of the proposal with respect to the variability of the system. It seems somewhat unfortunate in this respect that the authors base their analysis – as they admit themselves – on rather sketchy and incomplete data, collected from some cursory descriptions of concord variation in the dialectological literature. This is particularly serious since the Position-of-Subject Constraint features so prominently in their discussion. As mentioned earlier, the effects of the Position-of-Subject Constraint in most modern dialects are rather marginal. Börjars and Chapman’s proposal predicts, incorrectly, that the presence of -s after intervening adverbs should be categorical, i.e. that they often talk shouldactually be ungrammatical – but it certainly is not, in any of the modern varieties I have looked at. Even if one concedes that the authors’ goal is only to describe an idealized system and that their generalizations “may hold only for a subset of subjects or a subset of verbs in each dialect” (1998: 75), this constitutes a weakness of the proposal that cannot easily be glossed over.

Moreover, problems may remain even within the proposed formal mechanism itself. To name but the two most important ones: first, I fail to

Page 199: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 189

see how the model, in and of itself, accounts for even as much as the ungrammaticality of *they talks. After all, instead of the ungrammatical structure in (42) above, repeated here as (43b), one could just as well construe this surface string by means of the structure in (43c), which does not seem to violate any of the principles posited by the authors.

(43) a. *they talks b. *[NP e ] [VP [V they2=talks]] c. (?) [NP they1] [VP [V talks]]

The second problem stems from the, by now well-known, markedness paradox inherent in the northern system. As stated earlier, the proposal is based on the assumption that the affixless verb forms are unmarked in terms of agreement features. This may work for lexical verbs, but it hardly works for the forms of be. Forms like am and are following the adjacent pronouns cannot easily be regarded as non-inflected in the same way as e.g. talk can. This is particularly obvious with I=am, where identical person-number information is clearly encoded twice, in just the way the LFG mechanism is designed to preclude.

5.3. Hudson (1999)

The most recent proposal for a formal analysis of the northern concord system is only a cursory, tentative sketch within a larger discussion of verbal concord in Standard English, found in Hudson (1999: 204). Hudson puts forward an analysis within his own brand of formal dependency grammar, called Word Grammar (cf. Hudson 2001). As his contribution is not centrally concerned with the northern dialects, the descriptive information on which his account of them is based is even sketchier than in the case of Börjars and Chapman – the only authority cited is a passage in Harris (1993). Nevertheless, it may be argued that his proposal constitutes a step forward over those discussed up to now, as it avoids the methodological imperative common to most branches of formal grammar, of having to deduce observed grammatical behaviour from a small set of postulated universal, innate elements. In Word Grammar, features and syntactic categories can be language-specific and can be freely assigned by ad-hoc rules. Words and grammatical entities form a taxonomic hierarchy (a network of so-called “isa relations”), in which all elements can either

Page 200: A Comparative Grammar of Brtiish English Dialects

190 Lukas Pietsch

inherit default properties from their superordinate nodes or override these by specific rules. To account for some special cases in the agreement behaviour of subject noun phrases in Standard English, Hudson introduces a novel ad-hoc feature of “agreement-number” that exists side by side with the traditional “number” feature. Both are linked through the default rule of identity, but may differ from each other in special cases (e.g., the noun family may have singular “number” but, by way of exception, plural “agreement-number”). A similar mechanism of exceptional feature assignments specified for individual words is used to account for the behaviour of the pronouns I and you, and of the verb be.

In this framework, the formal description of the Type-of-Subject Constraint becomes surprisingly simple: in the northern dialects, all nouns by default have either singular “agreement-number” features or no “agreement-number” at all, and only the pronouns I, you, we, they have an exceptional plural “agreement-number”. The main problem with this proposal seems to be that it does not provide for a way to deal with the Position-of-Subject Constraint, of which Hudson does not seem to be aware. As all the formal distinctions in this model are associated with individual items in the lexicon, there seems to be no easy way to account for differences in behaviour that are triggered only by accidental syntactic environment factors other than the choice of the verb or subject themselves.

6. Discussion: Variation and usage-based theories

As shown above, all the formal analyses that have been proposed to account for the northern concord systems so far have run into problems of descriptive adequacy. None of them is quite successful even with respect to their primary aim: descriptively integrating the conditioning factors of the Northern Subject Rule under some unified principle. The combinatory effect of the Type-of-Subject Constraint and the Position-of-Subject Constraint, this curious amalgamation of lexical and syntactical conditions, apparently resists such generalizations. Moreover, none of the proposals even attempts to deal with those conditioning factors that affect the choice of form in the modern dialects, over and above the core defining constraints of the Northern Subject Rule itself: the preference for relative clauses, the effects of indefinite and demonstrative pronouns, of is/are and was/were,and so on. None of the proposals addresses the question of the quantitative, probabilistic patterns of variation observed with respect to these latter

Page 201: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 191

factors. In addition, none of the proposals deals with the full range of diatopic variation and the various geographical sub-types of the northern system, especially with the cases of overlap or competition between the Northern Subject Rule and the other, peripheral patterns of concord variation, such as was/were levelling. And finally, none of the proposals can easily deal with the diachrony of the Northern Subject Rule, with its apparently gradual processes of spread and diffusion. In short, these formalist accounts cannot fully account for the northern concord system, either in its reconstructed nascent state of early Middle English, or in its ‘ideal’ form in the older northern dialects, or indeed in its highly variable forms affected by dialect levelling and dialect attrition, in the present-day varieties.

In this discussion, I will take for granted one thing that may sound like a truism in the context of variationist and sociolinguistic studies, but which in the context of formal grammatical theory requires some justification: that the morphosyntactic variation observed in the dialectal corpora is indeed of the type that has come to be called ‘inherent variation’. It cannot be reduced to variation between speakers with different grammars, nor to effects of code-switching between competing, distinct grammatical systems within the production of individuals. The choice between one form and the other within the performance of each individual speaker will vary unpredictably, at any point in discourse, often even within a single sentence – as in the example that forms the title of this chapter (44):

(44) Some do and some doesn’t. [SED: Y10]

Variation of this type can only be described with the help of stochastic models. Even though it may be intuitively evident that much of the observed patterns in the present case can be described in terms of a competition between two distinct rule systems – the rules of the ‘ideal’ northern system on the one hand, and those of Standard English on the other – such an account may in effect be no more than just a convenient descriptive abstraction. There is little evidence in the behaviour of speakers that these competing ‘grammars’ are represented as distinct, separate entities and that any wholesale ‘switching’ between them is involved in actual production. A description involving competing individual rules within one unified grammatical system seems more plausible.

The idea that quantitative variability may need to be taken into account as part of “core syntax” has recently been acknowledged by Henry (2002), a rare exception for an author coming from a generative research tradition.

Page 202: A Comparative Grammar of Brtiish English Dialects

192 Lukas Pietsch

Note, incidentally, that theoretical attempts at integrating statements of competing grammatical regularities with statements about the probability of their respective use – basically in a fashion similar to the Variable-Rules Theory of the Labovian school in the 1970s – have recently also been brought forward again in the shape of Stochastic Optimality Theory (Boersma 1998, 2000; Boersma and Hayes 2001; Bresnan and Deo 2001).

While probabilistic variation has often been regarded as an awkward problem in formalist theories of grammar, functionalist and emergentist theories have for a long time seen inherent variation as a natural and necessary part of linguistic knowledge. Quantitative variation is easy to accommodate for any theory that accepts the idea that grammars are not stable from the completion of language acquisition in childhood, but can continuously change, as a result of the language encountered throughout the lifetime of a speaker. In this process, a central role is commonly attributed to the relative frequencies with which different linguistic structures are encountered in discourse (Krug 2003). Different quantitative exposure to linguistic structures in discourse is directly mirrored by quantitatively different representation of these structures in linguistic knowledge. The principal cognitive mechanism invoked in modelling frequency effects on mental representation is “entrenchment”, strengthening of representations through their activation in use (Langacker 1987: 59).

I would like to suggest that frequency-based theories are highly relevant not only in accounting for the state of variation observed in the present-day dialects, but also in accounting for the historical process that led to the emergence of the Northern Subject Rule in the first place. At the very heart of the northern concord pattern are constructions whose principal distinguishing property is their high discourse frequency: combinations of personal pronouns and finite verbs. Pronouns have, by their very nature, a much higher discourse (token) frequency than any lexical noun has. The relatively much higher degree of entrenchment and routinization that must be involved in the processing of recurrent collocations of adjacent pronoun subjects and their verbs – as opposed to combinations involving lexical noun subjects – can plausibly be seen as causally involved in any diachronic tendency that leads to morphological reduction or irregularity associated with these constructions. As was explained in section 4, it was exactly such a process that seems to have provided one of the main triggers for the emergence of the Northern Subject Rule in early Middle English: the phonological reduction of the verbal inflexional suffixes in the high-frequency environment with adjacent clitic pronouns. In this perspective,

Page 203: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 193

the emergence of the Northern Subject Rule can be seen as a case of morphosyntactic irregularity associated with a high-frequency construction, induced by mental routinization.

We can finally turn again to the question of how the present-day northern concord systems can best be characterized descriptively, with their curious sets of multi-level conditioning factors (lexical, syntactic, idiomatic, semantic). I would like to suggest that a usage-based form of construction grammar (Croft 2001, Kemmer and Israel 1994, Langacker 1987) may offer the most natural account. A central tenet of all brands of construction grammar is the continuity between lexicon and syntax. Properties of grammatical constructions, on all levels of abstraction, are assumed to be represented in essentially the same way as idiosyncratic properties of lexical items. Constructions form a multiple taxonomy, ranging from highly schematic to highly specific, where the more specific constructions can inherit but also override the grammatical behaviour specified in the more general schemata of which they are instantiations. The behaviour of constructions at all levels is non-universal and arbitrary, which means that there is no need to search for “explanations” in the sense of the highly abstract generalizations common in the microparametric syntax of the Chomskyan tradition. Applied to subject-verb agreement in English, such an approach can easily describe, for instance, the combination they+verb as a construction that instantiates the more general schema subject+verb, and which, while sharing some or most of its properties with this more general construction by default, may also partly differ from it in some arbitrary way, for instance with respect to its concord behaviour. Such a construction-based approach is conceptually similar though not formally identical to that sketched out by Hudson (1999) in the framework of Word Grammar. Word Grammar too makes extensive use of hierarchies of more general and more specific rules providing defaults and overrides, but locates these in the lexical entries of specific words, not of constructions as such. I would like to suggest that the construction-based description can give a more natural account of those cases where identical items show different behaviour determined only by the word-order constellation they occur in, as it is the case with the Position-of-Subject Constraint.

Following Kemmer and Israel (1994: 165–171), I shall assume that different construction schemata can compete with each other during production of an utterance, and that variability in a speaker’s production can be explained by this competition. The probabilistic results of this

Page 204: A Comparative Grammar of Brtiish English Dialects

194 Lukas Pietsch

competition provide “the basis for a natural account of the sort of structured variability traditionally modelled by variable rules” (Kemmer and Israel 1994: 165). The more heavily entrenched a constructional schema is in memory, the higher is its probability of being selected as the relevant categorization unit for the production of a specific usage event. The following will give a short and tentative sketch of how this kind of competition might be modelled. Consider first a schematic representation of the agreement system of Standard English (see Figure 2). From the perspective of a maximally parsimonious grammar, a minimum of only three construction schemata, i.e. nodes in the taxonomic hierarchy of stored grammatical representations, need to be posited (ignoring, for the moment, the special cases of be and that of the non-concord verb forms). One is a maximally abstract schema specifying the relationship of subject and verb in general. It has two daughter nodes, one for third person singular subjects, specifying the use of the verbal -s morphology, and one for all other combinations, specifying the use of the base forms of the verbs. In a usage-based account, it may be further assumed that some additional, subordinate nodes might exist for more specific, frequent combinations, involving for instance pronoun subjects and/or high-frequency verb forms (shown in dotted lines in Figure 2). These may be stored and processed as separate units, having attained unit status through their entrenchment as high-frequency combinations, but they are not associated with any observably different formal behaviour and therefore formally redundant.

Compare this system with that of a northern dialect, considering first the case of a hypothetical, “pure” system in which the Northern Subject Rule is categorical throughout (Figure 3). Here, the structure of the inventory is somewhat different. There must be a set of nodes representing the exceptional behaviour of the pronouns I, we, you and they (taking the -forms), in opposition with a more abstract schema representing the use of the default verbal -s form for all the rest. Again, there might also be further, more specific units representing other high-frequency combinations, but these are formally redundant. In particular, it seems quite likely that those pronoun combinations that take the verbal -s forms (he/she/it + V-s) may also be stored and processed as independent units just like the other pronouns, owing to their high frequency of occurrence.

Page 205: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 195

Figure 2. A construction inventory for Standard English agreement

Figure 3. A construction inventory for NSR agreement

We can now turn to the representations needed in the case of a hybrid dialect involving variation between standard-conforming and agreement of the northern type. I assume that each of the formal options available to a speaker of such a dialect is represented by a construction schema of its own, and that these schemata can compete with each other in production.

Subj Verb

Subj+3Sg V-s

...it V-s

she V-s he V-s

...Subj+3Sg has

Subj+3Sg was

Subj+3Sg is

Subj–3Sg V-

...they V-

you V-we V-

I V-

...Subj–3Sg have

Subj–3Sg were

Subj–3Sg are

Subj Verb

Subj V-s

it V-s

she V-s

he V-s NP V-sthey V-

we V-you V-

I V-

Page 206: A Comparative Grammar of Brtiish English Dialects

196 Lukas Pietsch

Figure 4 shows a partial network structure of the construction inventory needed for such a system. The schemata NP V-s (from the northern system) and NPPl V- (from the standard system) will compete with each other. Any usage event of a particular subject and a particular verb will be processed as instantiating either of these two, whichever happens to be more salient when the construction is being processed. Depending on which of the two is more deeply entrenched, either the one or the other will have the higher likelihood of winning out.

Figure 4. A construction inventory for variable NSR agreement

The structured variation effects of the type captured by variable-rules models can be integrated in a system of this kind by again assuming further, subordinate nodes in the network. In a construction grammar approach, separate representation as a unit in the structured inventory of a speaker’s knowledge is commonly ascribed to “[a]ny construction with unique, idiosyncratic morphological, syntactic, lexical, semantic, pragmatic, or discourse-functional properties” (Croft 2001: 25; emphasis in original). In order to capture a system of structured, inherent variation, a further factor can be added to this list: any type of construction that is associated with distinct, quantifiable behaviour with respect to a linguistic variable must also have unit status and be separately represented. In this way, nodes in the construction network may correspond to what would be modelled as “factors” in a variable-rule model.

Figure 5 shows a partial network model of such a system. Both the competing nodes from Figure 4 (NP V-s and NPPl V- ) are here shown as having a number of daughter nodes. The subnodes shown represent schemata associated with individual verb forms. Yet others, not shown here, might represent different types of subjects, and so on. Each of the

Subj Verb

NP V-s

it V-s she V-s

he V-s

they V-we V-

you V-I V-

NPSg V-s NPPl V-

Page 207: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 197

schemata represents either a standard-conformant or an NSR-conformant use. Each of them may be thought of as characterized by its individual degree of strength of entrenchment (symbolized in the graph by boxes of different thickness). During production, there will be a pairwise competition between relevant nodes. In this way, a situation can be modelled in which, for instance, plural use of is as opposed to are is relatively more likely than plural use of goes as opposed to go, just as it was found in the corpus data from many northern dialects.

Figure 5. A partial construction inventory for structured variation in NSR agreement

In construction grammar, it is also commonly assumed that units in the taxonomic network of constructions can have multiple parents, and individual usage events can instantiate more than one construction schemata simultaneously. Each of the schemata activated for the production of any individual utterance will be responsible for a different aspect of its formal behaviour. Transferring this idea to the modelling of structured variation, it may be assumed that more than one set of competing nodes may be activated simultaneously in the production of an utterance, in such a way that each of them represents a different conditioning factor. For instance, it was shown in the previous sections that in many northern dialects the choice between the competing sentences them is going and them are going is influenced by two independent conditioning factors (represented as members of two “factor groups” in a variable-rules model): a relatively stronger preference for is over are as compared with other verbal -s forms; and a factor favouring the use of verbal -s with the specific

Subj Verb

NP V-sNPPl V-

NP is NP was NP has NP goes …NPPl are NPPl were NPPl have NPPl go …

Page 208: A Comparative Grammar of Brtiish English Dialects

198 Lukas Pietsch

item them as a subject. Such combinatorial effects of several independent conditioning factors can be modelled in a construction grammar approach by assuming that the production of a sentence such as them is going involves the simultaneous competition between a pair of schemata NP + isand NPPl + are, as shown in Figure 5; and between a second pair of schemata, them + V- and them + V-s.

A usage and schema based approach such as this can account in a natural way for the existence of the two types of constraints on subject-verb agreement that have been reported for many varieties of English, not only those that are directly related to the northern ones: Type-of-Subject and Position-of-Subject constraints. I hypothesize that a Type-of-Subject constraint may arise whenever combinations of verbs and personal pronouns, owing to their high discourse frequency, are entrenched separately and attain unit status. A Position-of-Subject constraint will be a corollary of the fact that both the more specific and the more general schema may compete with each other in the production of a usage event. I hypothesize that the more specific schema in memory – a gestalt consisting of a particular pronoun and a verb – will be more salient, and hence more likely to be activated as the relevant categorizing unit, if the utterance that is being formed involves a direct collocation of the pronoun and the verb. If both items are not adjacent, the construction less closely matches the gestalt prototype of the stored schema. It will then be less likely to be categorized as an instance of that specific schema, and by default the more abstract schema will be more likely to win out as the relevant categorizing unit.

I also hypothesize that the emergence of a Type-of-Subject Constraint is particularly likely in a situation where a verbal agreement paradigm has previously undergone heavy erosion and reduction. As was shown in section 4, such a state of affairs held in the historical situation in which the Northern Subject Rule first emerged in northern early Middle English. The present tense verbal system had been reduced, by a series of changes, to only one formal opposition between -(e) and -(e)s, so that only the first person singular was reliably distinguished from all the rest of the paradigm. If language learners are confronted with such a heavily eroded – typologically and functionally implausible – agreement system, then a distinction such as that between pronominal and non-pronominal subjects may become cognitively more salient in processing than the person-number distinction, and a re-structuring of the agreement system along these dimensions may be the long-term consequence. It may also be argued that a

Page 209: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 199

similar state of affairs has again been reached, in a different form, in Modern Standard English, where only the third singular is distinguished from all the rest of the paradigm (and where, in addition, the contrast between was and were patterns according to yet a different rule, contrasting first or third singular as against the rest). This may be a reason why variation phenomena resembling the northern system in some respects have emerged, independently as it appears, in some modern varieties outside the north.

With the foregoing discussion, I hope to have provided some basis for the claim that a usage-based, cognitive approach to grammatical theory can and should be fruitfully combined with variationist research using methods from the Labovian sociolinguistic tradition. Such a combination has repeatedly been envisaged in recent work. However, while sociolinguistic research has dealt descriptively with phenomena such as verbal concord variation in English quite thoroughly, much of the more theoretically oriented work has so far been restricted to more narrowly local phenomena of grammatical organization, such as inflectional paradigms, phonological reduction phenomena in specific word chunks, and so forth. It seems fair to say that this line of research has been somewhat reluctant to tackle questions of variation in the domain of syntax proper (cf. Bybee and Hopper 2001). The morphosyntax of agreement would appear to be a promising domain for an extension of such studies. Especially with respect to a language whose social and geographical variation is so well documented as English, there is still much fruitful ground for further research aiming to bridge the gap between theory and description.

Acknowledgments

I would like to thank all organisations and individuals who contributed the speech recordings which were used in FRED and form much of the empirical basis of the present article, especially Roger Leitch, Jane Petrie, the Ambleside Oral History Group, Pennine Heritage (Hebden Bridge) and the Oral History Group at Huyton (Merseyside). Thanks are also due to John Kirk for providing access to the NITCS, to him and Raymond Hickey for their efforts at recovering material of the TRS, and to Clive Upton for assistance with the unpublished material of the SED. I would also like to thank Kevin McCafferty, Sali Tagliamonte, Peter Auer, and Ingo Plag for valuable feedback and advice during the preparation of this work, and

Page 210: A Comparative Grammar of Brtiish English Dialects

200 Lukas Pietsch

Anthea Gupta, Clive Upton, Esther Asprey and Kate Wallace for special assistance with transcription problems.

Notes

1. For the purposes of this study, the term ‘concord verb’ refers to all verbal forms which have the morphological potential of displaying an agreement contrast. This includes all finite present tense verbs except for modals and subjunctives, plus the past tense forms was and were. The term verbal -s is used to include, where not otherwise specified, the irregular forms is and was.

2. Following the usage in the SED and much of the dialectological literature, I am using the historical (pre-1974) counties as a geographical reference frame.

3. Designations such as “Central North”, “Lower North” etc. mirror the well-known, partly homonymous labels introduced by Trudgill (1990) for the “modern dialect areas”, but are not intended as equivalent to them. As used throughout this chapter, these terms designate areas defined exclusively in terms of the subject-verb agreement data in the SED.

4. Three isolated tokens of ’m forms were also found in the incidental material in southwestern Yorkshire, but they can all be explained as products of a mere phonetic assimilation of han before labial consonants.

5. The examples from the NITCS include some tokens where the second verb is was or is. Since there is otherwise hardly any variation involving these verbs with these pronoun subjects in the Ulster data, these tokens can safely be attributed to the Northern Subject Rule together with those involving lexical verbs.

6. See, for instance, the common extension of the idiom to other verbs of saying, such as the modern vernacular I goes, attested for instance for Glasgow in Macafee (1983).

7. Owing to the different data collection habits of certain fieldworkers, it was necessary to use different weighting criteria for tokens recorded by one of them, S. Ellis, and those recorded by the rest. On average, Ellis tended to record only about half as many tokens per location than Wright did in the same geographical area. Locations whose fieldworker notebooks were not consulted are excluded in this map.

8. The examples from Wright (1892) are quoted after Chapman (1998: 38).

Page 211: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 201

References

Aissen, Judith, and Joan Bresnan 2002 Optimality and functionality: objections and refutations. Natural

Language and Linguistic Theory 20: 43–80. Anderwald, Lieselotte 2002 Negation in Non-Standard British English: Gaps, Regularizations

and Asymmetries. London: Routledge. Bailey, Charles, and Karl Maroldt 1977 The French lineage of English. In Langues en Contact – Pidgins –

Creoles – Languages in Contact, Jürgen M. Meisel (ed.), 21–53. Tübingen: Narr.

Bailey, Guy, and Garry Ross 1988 The shape of the superstrate: morphosyntactic features of Ship

English. English World-Wide 9: 193–212. Bailey, Guy, Natalie Maynor, and Patricia Cukor-Avila 1989 Variation in subject-verb concord in Early Modern English.

Language Variation and Change 1: 285–300. Barbiers, Sjef, Leonie Cornips, and Susanne van der Kleij (eds.) 2002 Syntactic Microvariation. http://www.meertens.nl/books/synmic/. Barry, Michael V. 1981a The southern boundaries of Northern Hiberno-English speech. In

Aspects of the English Dialects in Ireland, Michael V. Barry (ed.), 52–95. Belfast: Queen’s University, Institute of Irish Studies.

1981b The methodology of the tape-recorded survey of Hiberno-English speech. In Aspects of the English Dialects in Ireland, Michael V. Barry (ed.), 18–46. Belfast: Queen’s University, Institute of Irish Studies.

Beal, Joan 1993 The grammar of Tyneside and Northumbrian English. In Real

English: The Grammar of English Dialects in the British Isles,James Milroy, and Lesley Milroy (eds.), 187–213. London: Longman.

1997 Syntax and morphology. In The Edinburgh History of the Scots Language, Charles Jones (ed.), 335–377. Edinburgh: University Press.

2004 The morphology and syntax of English dialects in the North of England. In Varieties of English: The British Isles, Bernd Kortmann, and Clive Upton (eds.). Berlin: Mouton de Gruyter.

Berndt, Rolf 1956 Form und Funktion des Verbums im nördlichen Spätaltenglischen:

eine Untersuchung der grammatischen Formen und ihrer

Page 212: A Comparative Grammar of Brtiish English Dialects

202 Lukas Pietsch

syntaktischen Beziehungsbedeutungen in der großen sprachlichen Umbruchsperiode. Halle: Niemeyer.

1982 A History of the English Language. Leipzig: VEB Verlag Enzyklo-pädie.

Black, James R., and Virginia Motapanyane (eds.) 1996 Microparametric Syntax and Dialect Variation. Amsterdam: Ben-

jamins. Blakeley, Leslie 1949 The Lindisfarne s/ð problem. Studia Neophilologica 22: 15–47. Boersma, Paul 1998 Functional Phonology: Formalizing the Interactions between

Articulatory and Perceptual Drives. The Hague: Holland Academic Graphics.

Boersma, Paul, and Bruce Hayes 2001 Empirical Tests of the Gradual Learning Algorithm. Linguistic

Inquiry 32: 45–86. Börjars, Kersti, and Carol Chapman 1998 Agreement and pro-drop in some dialects of English. Linguistics 36:

71–98.Breivik, Leiv Egil, and Toril Swan 2000 The desemanticisation of existential there. In Words: Structure,

Meaning, Function, Christiane Dalton-Puffer, and Nikolaus Ritt (eds.), 19–34. Berlin: Mouton de Gruyter.

Bresnan, Joan, and Ashwin Deo 2001 Grammatical constraints on variation: ‘Be’ in the Survey of English

Dialects and (Stochastic) Optimality Theory. Unpubl. Ms., Stanford University.

Britain, David 2002 Diffusion, levelling, simplification, and reallocation in past tense be

in the English Fens. Journal of Sociolinguistics 6: 16–43. Brunner, Karl 1962 Die englische Sprache, Vol. 2. Tübingen: Niemeyer. Brunner, Karl, and Eduard Sievers 1965 Altenglische Grammatik. 3rd edition. Tübingen: Niemeyer. Bybee, Joan 1985 Morphology: A Study of the Relation between Form and Meaning.

Amsterdam: Benjamins. Bybee, Joan, and Paul Hopper 2001 Introduction. In Bybee and Hopper (eds.), 1–26. Bybee, Joan, and Paul Hopper (eds.) 2001 Frequency and the Emergence of Linguistic Structure. Amsterdam:

Benjamins.

Page 213: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 203

Chambers, Jack K. 2004 Dynamic typology and vernacular universals. In Dialectology Meets

Typology, Bernd Kortmann (ed.). Berlin: Mouton de Gruyter. Chapman, Carol 1998 A subject-verb agreement hierarchy: evidence from analogical

change in modern English dialects. In Current Issues in Linguistic Theory: Historical Linguistics 1995. Vol. 2, Richard Hogg, and L. van Bergen (eds.), 35–44. Amsterdam: Benjamins.

Clark, Brady Z. 2003 On stochastic grammar. Unpubl. ms., University of Stanford.

http://www.stanford.edu/~bzack/papers/stotgrammars.pdf. Corrigan, Karen P. 1997 The syntax of South Armagh English in its socio-historical

perspective. PhD thesis, University College Dublin. Croft, William 1995 Autonomy and functionalist linguistics. Language 71: 490–532. 2001 Radical Construction Grammar. Oxford: Oxford University Press. Ferguson, Charles A. 1996 Variation and drift: loss of agreement in Germanic. In Towards a

Social Science of Language: Papers in Honor of William Labov. Vol. I: Variation and Change in Language and Society, Gregory Guy (ed.), 173–198. Amsterdam: Benjamins.

Filppula, Markku 1999 The Grammar of Irish English: Language in Hibernian Style.

London: Routledge. Godfrey, Elizabeth, and Sali Tagliamonte 1999 Another piece for the verbal -s story: evidence from Devon in

southwest England. Language Variation and Change 11: 87–121. Görlach, Manfred 2002 A Textual History of Scots. Heidelberg: Winter. Gregg, Robert J. 1972 The Scotch-Irish dialect boundaries in Ulster. In Patterns in the Folk

Speech of the British Isles, Martin F. Wakelin (ed.), 109–139. London: Athlone Press.

Harris, John 1991 Ireland. In English around the World: Sociolinguistic Perspectives,

Cheshire, Jenny (ed.), 37–50. Cambridge: Cambridge University Press.

1993 The grammar of Irish English. In Real English: The Grammar of English Dialects in the British Isles, James Milroy, and Lesley Milroy (eds.), 139–186. London: Longman.

Page 214: A Comparative Grammar of Brtiish English Dialects

204 Lukas Pietsch

Henry, Alison 1995 Belfast English and Standard English: Dialectal Variation and

Parameter Setting. Oxford: Oxford University Press. 2002 Variation and syntactic theory. In Handbook of Language Variation

and Change, Jack K. Chambers, Peter Trudgill, and N. Schilling-Estes (eds.), 267–282. Oxford: Blackwell.

Holmqvist, Erik 1922 On the History of the English Present Inflections, Particulary -th

and -s. Heidelberg: Winter. Hudson, Richard 1999 Subject-verb agreement in English. English Language and

Linguistics 3: 173–207. 2001 Word Grammar. http://www.phon.ucl.ac.uk/home/dick/hcl.htm. Ihalainen, Ossi 1991 On grammatical diffusion in Somerset folk speech. In Dialects of

English: Studies in Grammatical Variation, Peter Trudgill, and Jack K. Chambers (eds.), 104–119. London: Longman.

1994 The dialects of England since 1776. In The Cambridge History of the English Language. Vol. V: English in Britain and Overseas: Origins and Developments, Robert Burchfield (ed.), 197–276. Cambridge: Cambridge University Press.

Keller, Wolfgang 1925 Skandinavischer Einfluss in der englischen Flexion. In Probleme der

englischen Sprache und Kultur: Festschrift Johannes Hoops, W. Keller (ed.), 80–87. Heidelberg: Winter.

Kemner, Suzanne, and Michael Israel 1994 Variation and the usage-based model. In Papers from the 30th

Regional Meeting of the Chicago Linguistics Society, Katherine Beals (ed.), 165–179. Chicago: Chicago Linguistics Society.

Kirk, John M. 1991 Northern Ireland Transcribed Corpus of Speech. Vol. 1 (with S.

West and S. Gibson): Textfile. Colchester: Economic and Social Research Council Data Archive, University of Essex, Colchester.

Klemola, Juhani 1996 Non-standard periphrastic do: a study in variation and change. PhD

thesis, University of Essex. 2000 The origins of the Northern Subject Rule: a case of early contact? In

The Celtic Englishes II, Hildegard Tristram (ed.), 329–346. Heidel-berg: Carl Winter Verlag.

Page 215: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 205

Knecht, Jakob 1911 Die Kongruenz zwischen Subjekt und Prädikat und die 3. Person

Pluralis Präsentis auf -s im Elisabethanischen Englisch. Heidelberg: Winter.

Krug, Manfred 2003 Frequency as a determinant in grammatical variation and change. In

Determinants of Grammatical Variation, Günter Rohdenburg, and Britta Mondord (eds.). Berlin: Mouton de Gruyter.

Kytö, Merja 1993 Third-person present singular verb inflection in early British and

American English. Language Variation and Change 5: 113–139. Laing, Margaret 1978 Studies in the Dialect Material of Medieval Lincolnshire. Unpub-

lished PhD dissertation, University of Edinburgh. Langacker, Ronald W. 1987 Foundations of Cognitive Grammar, Vol. I: Theoretical Prerequi-

sites. Stanford: Stanford University Press. Macafee, Caroline 1983 Glasgow. Amsterdam: Benjamins. 1993 A short grammar of Older Scots. Scottish Language 11/12: 10–36. 1994 Traditional Dialect in the Modern World: A Glasgow Case Study.

Frankfurt: Lang. McCafferty, Kevin 1998a Shared accents, divided speech community? Change in Northern

Ireland English. Language Variation and Change 10: 97–121. 1998b Barriers to change: ethnic division and phonological innovation in

Northern Hiberno-English. English World-Wide 19: 7–35. 1999 (London)Derry: between Ulster and local speech – class, ethnicity

and language change. In Urban Voices: Accent Studies in the British Isles, Paul Foulkes, and Gerard Docherty (eds.), 246–264. London: Arnold.

2003 The Northern Subject Rule in Ulster: how Scots, how English? Language Variation and Change 15: 105–139.

2004 ‘[T]hunder storms is verry dangese in this countrey they come in less than a minnits notice...’: The Northern Subject Rule in Southern Irish English. English World-Wide 25: 51–79.

McIntosh, Angus 1988 Present indicative plural forms in the later Middle English of the

North Midlands. In Middle English Dialectology: Essays on Some Principles and Problems, Angus McIntosh (ed.), 116–122. Aberdeen: Aberdeen University Press.

Page 216: A Comparative Grammar of Brtiish English Dialects

206 Lukas Pietsch

McIntosh, Angus, Michael L. Samuels, and Michael Bensik 1986 A Linguistic Atlas of Late Mediaeval English. 4 vols. Aberdeen:

Aberdeen University Press. Mather, James Y., and Hans-Henning Speitel 1975–86 The Linguistic Atlas of Scotland: Scots Section. 3 Vols. London:

Croom Helm. Meurman-Solin, Anneli 1992 On the morphology of verbs in Middle Scots. In History of

Englishes: New Methods and Interpretations in Historical Linguistics, Ossi I. Rissanen, Terttu Nevalainen, and Irma Taavitsainen (eds.), 611–623. Berlin, New York: Mouton de Gruyter.

Miller, Jim E. 1989 The grammar of Scottish English. In Regional Variation in British

English Syntax, James Milroy, and Lesley Milroy (eds.). Swindon: Economic and Social Research Council.

1993 The grammar of Scottish English. In Real English: The Grammar of English Dialects in the British Isles, James Milroy, and Lesley Milroy (eds.), 99–138. London: Longman.

2004 The morphology and syntax of Scottish English. In Varieties of English: The British Isles, Bernd Kortmann, and Clive Upton (eds.) .Berlin: Mouton de Gruyter.

Miller, Jim E., and Keith Brown 1982 Aspects of Scottish English syntax. English World-Wide 3: 3–17. Montgomery, Michael 1988 The Roots of Appalachian English. Clevedon: Multilingual Matters. 1994 The evolution of verb concord in Scots. In Studies in Scots and

Gaelic: Proceedings of the Third International Conferences on the Languages of Scotland, Alexander Fenton, and Donald McDonald (eds.), 81–95. Edinburgh: Canongate.

Montgomery, Michael, Janet M. Fuller, and Sharon DeMarse 1993 ‘The Black men has wives and sweet hearts (and third person plural

-s) jest like the white men’: evidence for verbal -s from written documents on 19th-century African American speech. LanguageVariation and Change 5: 335–357.

Murray, James A. H. 1873 The Dialect of the Southern Counties of Scotland. London: Philo-

logical Society. Mustanoja, Tauno F. 1960 A Middle English Syntax. Helsinki: Société Néophilologique.

Page 217: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 207

Newmeyer, Frederick J. 2000 Language Form and Language Function. Cambridge/Mass.: MIT

Press. 2002a Optimality and functionality: a critique of functionally-based

Optimality Syntax. Natural Language and Linguistic Theory 20: 1–42.

2002b A rejoinder to Bresnan and Aissen. Natural Language and Linguistic Theory 20: 81–95.

2003 Grammar is grammar and usage is usage. Language 79: 682–707. Ogura, Mieko, and William S. Wang 1996 Snowball effect in lexical diffusion: the development of -s in the

third person singular present indicative in English. In English Historical Linguistics 1994: Papers from the 8th International Conference on English Historical Linguistics, Edinburgh 1994,Derek Britton (ed.), 119–142. Amsterdam: Benjamins.

Orton, Harold, Michael V. Barry, Eugen Dieth, Wilfrid J. Halliday, Philip M. Tilling, and Martyn F. Wakelin (eds.)

1962–71 Survey of English Dialects. Leeds: Arnold. Orton, Harold, Steward Sanderson, and John Widdowson 1978 The Linguistic Atlas of England. London: Croom Helm. Parry, David 1977–79 Survey of Anglo-Welsh Dialects. Vol. I: The South-East; Vol. II: The

South-West. Swansea: University College. Peitsara, Kirsti 2002 Verbal -s in Devonshire: the Helsinki Dialect Corpus evidence. In

Variation Past and Present: VARIENG Studies on English for Terttu Nevalainen, Helena Raumolin-Brunberg (ed.), 211–230. Helsinki: Société Néophilologique.

Pietsch, Lukas 2003 Subject-verb agreement in northern dialects of English. PhD thesis,

University of Freiburg i. Br. 2005 Variable Grammars: Verbal Agreement in Northern Dialects of

English. Tübingen: Niemeyer. Poplack, Shana 2001 Variability, frequency, and productivity in the irrealis domain of

French. In Bybee and Hopper (eds.), 405–430. Quirk, Sidney G., Sidney Greenbaum, Geoffrey Leech, and Jan Svartvik 1991 A Comprehensive Grammar of the English Language. 9th revised

edition. London: Longman. Roberts, Ian 1993 Verbs and Diachronic Syntax: A Comparative History of English

and French. Dordrecht: Kluver.

Page 218: A Comparative Grammar of Brtiish English Dialects

208 Lukas Pietsch

1997 Comparative Syntax. London: Arnold. Robinson, Philip 1997 Ulster Scots: A Grammar of the Traditional Written and Spoken

Language. Belfast: Ullans Press. Ross, Alan S. 1934 The origin of the s-endings of the Present Indicative in English.

Journal of English and Germanic Philology 33: 68–73. Samuels, Michael L. 1988 The great Scandinavian belt. In Middle English Dialectology: Essays

on Some Principles and Problems, Angus McIntosh (ed.), 106–115. Aberdeen: Aberdeen University Press.

Sankoff, David, and William Labov 1979 On the uses of variable rules. Language in Society 8: 189–222. Schendl, Herbert 1996 The 3rd plural present indicative in Early Modern English: variation

and linguistic contact. In English Historical Linguistics 1994: Papers from the 8th International Conference on English Historical Linguistics, Edinburgh 1994, Derek Britton (ed.), 143–160. Amsterdam: Benjamins.

2000 The third person present plural in Shakespeare’s first folio: a case of interaction of morphology and syntax? In Words: Structure, Meaning, Function, Christiane Dalton-Puffer, and Nikolaus Ritt (eds.), 263–276. Berlin: Mouton de Gruyter.

Schreier, Daniel 2002 Past be in Tristan da Cunha: the rise and fall of categoricality in

language change. American Speech 77: 70–99. Shorrocks, Graham 1999 A Grammar of the Dialect of the Bolton Area. Vol. II: Morphology

and Syntax. Berlin: Lang. Smith, Jennifer, and Sali Tagliamonte 1998 ‘We was all thegither, I think we were all thegither’: was regulariza-

tion in Buckie English. World Englishes 17: 105–126. Stein, Dieter 1986 Old English Northumbrian verb inflection revisited. In Linguistics

across Historical and Geographical Boundaries. Vol. 1: Linguistic Theory and Historical Linguistics, Dieter Kastovsky, and Alexander Szwedek (eds.), 637–650. Berlin, New York, Amsterdam: Mouton de Gruyter.

Tagliamonte, Sali 1999 Was/were variation across the generations: view from the city of

York. Language Variation and Change 10: 153–191.

Page 219: A Comparative Grammar of Brtiish English Dialects

Verbal concord variation 209

2002 Comparative sociolinguistics. In Handbook of Language Variation and Change, Jack K. Chambers, Peter Trudgill, and N. Schilling-Estes (eds.), 729–763. Oxford: Blackwell.

Thomason, Sarah G., and Terrence Kaufman 1988 Language Contact, Creolization and Genetic Linguistics. Berkeley:

University of California Press. Trudgill, Peter 1990 The Dialects of England. Oxford: Blackwell. Van Gelderen, Elly 1997 Verbal Agreement and the Grammar behind its ‘Breakdown’:

Minimalist Feature Checking. Tübingen: Niemeyer. Viereck, Wolfgang 1991/97 The Computer-Developed Linguistic Atlas of England. 2 vols.

Tübingen: Niemeyer. Viereck, Wolfgang, Karin Viereck, and Heinrich Ramisch 2002 Dtv-Atlas englische Sprache. München: Deutscher Taschenbuch

Verlag.Visser, Frederic T. 1963 An Historical Syntax of the English Language. Vol. 1. Leiden: E. J.

Brill.Weinreich, Uriel, William Labov, and Marvin I. Herzog 1968 Empirical foundations for a theory of language change. In Directions

for Historical Linguistics: A Symposium, W. P. Lehmann, and Y. Malkiel (eds.), 95–188. Austin: University of Texas Press.

Wilson, John, and Alison Henry 1998 Parameter setting within a socially realistic linguistics. Language in

Society 27: 1–21. Wright, Joseph 1892 A Grammar of the Dialect of Windhill in the West Riding of

Yorkshire. London: Kegan Paul, Trench, Trübner and Co. 1905 The English Dialect Grammar. Oxford: Henry Frowde. Wright, Laura 2002 Third person plural present tense markers in London prisoners’

depositions, 1562–1623. American Speech 77: 242–263.

Page 220: A Comparative Grammar of Brtiish English Dialects
Page 221: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns: Southwest England

Susanne Wagner

Abstract

“In Dem, ees a er an’ ers a ee, all ’cept th’ aud Tom cat an’ even ees a er!” [In Devon, he is a her and she is a he, all except the old Tom cat and even he is a her!] (Marten 1973: 9)

This saying, attributed to a number of counties in Southwest England, nicely summarises one of the regional characteristics of the morphosyntax of Southwestern dialects. Traditionally, these dialects employ masculine pronominal forms (corresponding to StE he and him) to refer not only to humans and animals, but also to count nouns. It, on the other hand, is never used for anything except mass nouns. The feminine forms (she and her) are rare, generally referring only to women. This intricate system of gender assignment is made even more complex by the presence of the form [ ], which is equivalent to StE him, not her. Different from other features of non-standard morphosyntax, these gendered pronouns do not seem to be stigmatised. This may be one of the reasons why the feature could be successfully exported from the West Country to Newfoundland, and could survive in these varieties to the present day. Another reason for the relative strength of the feature today is the tendency of spoken varieties of English in general to employ non-neuter forms in contexts where emotional involvement or attitudes are expressed. The interaction of at least three different systems of gender assignment is investigated in traditional dialect data from Southwest England (Survey of English Dialects) and more modern corpus data for both Southwest England and Newfoundland. The factors responsible for gender assignment, it will be argued, differ with regard to their distributional preferences in the respective varieties, resulting in different systems of gender assignment in these varieties today.

1. Introduction

At a first glance, [...] it might seem that gender in modern English is a relatively straightforward category to discuss, in comparison with the phenomenon in many other languages; indeed, in many text-books for both

Page 222: A Comparative Grammar of Brtiish English Dialects

212 Susanne Wagner

native and non-native speakers of English it is barely mentioned, if at all. (Wales 1996: 134)

Katie Wales expresses here what almost everyone, be they laymen or linguists, native or non-native speakers, would agree to without a second thought – gender is an at most marginal category of Present-Day English (PrDE). For centuries, it has been largely equalled (and confused) with the biological category of sex, making it difficult to speak of “gender in English” at all. However, gender assignment in English is by no means as straightforward as grammarians describe it. Spoken and regional varieties of English in particular employ systems that are far more intricate than a superficial analysis would reveal.

This study originated from a long-standing interest in one particular feature found in the system(s) of gender assignment of the traditional English dialects of Southwest England and Newfoundland, which the author had first been encountered in her early student days. As with many interesting features of morphosyntax, literature on the subject was scarce to non-existent, and actual examples were hard to come by. However, over the years it was possible to gather what literature was available and remedy the lack of data by collecting a corpus of actual dialect speech (FRED; for details, see Kortmann and Wagner, this volume; also Anderwald and Wagner forthcoming). The following examples shall illustrate the core issue of this study, namely the seemingly confusing use of pronominal forms in basically all varieties of (spoken) English. All of the examples in (1) share one feature: the personal pronoun forms used (in bold print) are supposedly reserved for reference to human or at least animate entities.

(1) a. That is a dead teat with no milk into en [ n]. (38 Do 3, book III) b. We call en [ ] a peeth [well]. (36 Co 6, book IV) c. He used to say “Put un [candle] where ye can zee ’im [candle] and I

can zee ’im [candle] as well.” (FRED Wil_024) d. <u Int> Those flat irons interested me too because they had a handle

that would come off, removable handle, why was that? < u Inf> Da’s for you put one on da stove now and you put two on da

stove and when dey get warm you put da handle in you take and ah ah when he [iron] when he [iron] get cold off you put he [iron] on you put your handle in t’ udder one he [iron] be warm see, he’d [iron] be hot and you’d take en [iron] and you’d iron your clothes den and while one be, while one be warmin’ da t’ udder be coldin’, time be coldin’ off see. (MUNFLA 70-003: C0626)

Page 223: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 213

e. I joined in the first hall was down dere and den they build this one is down there now and they build he [hall/house] in 19-, in 1921 I believe they build it, I joined in the other one, I joined in 1917, in the ole lodge. (MUNFLA 71-131: C1034)

f. … press them like that and you’d see your thumb mark in them or any apple really when he’s [apple] ripe, wadn’t it, but when he’s [apple] not ripe he’s [apple] hard, isn’t he [apple], … (FRED Som_001)

g. Ok, crack ’er up! (From the movie Titanic [USA 1997]; the speaker is (presumably) an American male, talking about the safe being brought up from the ocean floor.)

h. Where is she? – If she will give us the pleasure … there she is! (From the movie The Red Violin; the speaker is [presumably] a Canadian male, talking as an auctioneer about the violin that is to be sold next.)

i. she’s up/down (in reference to a share [price]; Graham Shorrocks, personal collection); she’s up 30 today; she’s off $2 today (reference to the market [e.g. the Dow, Toronto, NASDAQ] and to an individual stock [price]; Newfoundland brokers; Graham Shorrocks, personal collection

j. This is another pot and saucer. A bit dusty! You see that one isn’t exactly glazed proper, burnt proper is she [pot]. (FRED Som_009)

At a first glance, the target nouns seem to have nothing in common: a cow’s teat, a well, a candle, an iron, an assembly hall, an apple, a safe, a violin, stock prices, a pot – this list could be continued, but suffice it to note for now that there seem to be more contrasts than shared properties among the referents of the relevant nouns.

As varied as the target nouns are the forms of the personal pronouns employed in the examples: en, un, ’im, he, she. While the last three forms can be readily identified as (reduced) object pronoun (masculine) and subject forms (masculine and feminine), the origin(s) of the first two forms may not be as obvious. Realized in numerous ways in speech, ranging from [ n] to [n], there is no uniform orthography for the item in question. Most commonly spelled either en or un, it is analyzed as a reduced form of Old English (OE) hine (acc. sing. masc.; cf. OED ‘un 1., being used as early as 1633), thus being equivalent to PrDE him. This en or un is a form typical of the Southwest. Its extension, based on data from the Survey of English Dialects (SED), is shown by Wakelin (1986: 35).

Scholars have not (yet) agreed on a suitable label for the phenomenon – terms that have been applied include “gender diffusion”, “animation”

Page 224: A Comparative Grammar of Brtiish English Dialects

214 Susanne Wagner

(Siemund 2001, 2002; forthcoming) and “individuation” (Pawley 2002). For reasons that will be discussed in detail in later sections, a more neutral term will be used here: “gendered” pronouns, referring to instances of personal pronouns which are marked for gender (i.e. either masculine or feminine) but which refer to inanimate (count) nouns.

Different from many other features typical of non-standard morphosyntax, gendered pronouns do not seem to be stigmatized, clearly operating below the level of consciousness. When asked about them, native speakers are generally aware of the “weird” use of pronouns, and readily offer all types of explanations. As can already be seen from the examples cited above, one should distinguish at least two systems of gendered pronouns: one where feminine forms are used and a second one that mainly employs masculine forms. As will be shown below, the two systems have different origins and different regional distributions. Where they interact, various factors have to be taken into account when discussing gender assignment in the relevant varieties.

In the following sections, gendered pronouns will be analyzed from a number of different perspectives. Section 2 will provide a brief overview of existing research on the subject. Also, when discussing gender assignment in non-standard varieties, the standard system deserves a closer look (Section 3). The corpora this study is based on will be presented in Section 4. Special pronominal referent classes such as animals are discussed in various studies on standard and non-standard varieties of English around the world and are thus also included here (Section 5). Studies of gendered pronouns in non-dialectal contexts are at the heart of Section 6, while Section 7 focuses on possible persistence phenomena (priming effects) in the use of gendered pronouns.

Sections 8 to 11 constitute the empirical part of the study. The actual use of gendered pronouns will first be analyzed in the material from the SED(Basic Material – Section 8 and fieldworker notebooks – Section 9), followed by detailed accounts of those gendered pronouns occurring in the oral history material from Southwest England (Section 10). For lack of space, the analyses of the Newfoundland material will only be briefly summarised here (Section 11), pointing out the most interesting differences between the English and the Canadian material.1 An overall summary concludes the study (Section 12).

Page 225: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 215

2. Gendered pronouns

The study of gendered pronouns has a long tradition in dialectology. In the late 19th and early 20th century, they are mentioned in most descriptions of dialects, both regionally restricted ones and general accounts (e.g. Wright’s English Dialect Grammar [1905]), as one of the few non-lexical features mentioned (but generally not explained) in descriptions largely dominated by lexical material. Interestingly, gendered pronouns never reached any considerable fame outside dialectology, and although native speakers readily offer examples when confronted with the issue, they do not seem to be aware of the extension of the phenomenon. Only few studies dealing with gendered pronouns are not restricted to dialect evidence, most extensively Morris’ doctoral thesis on Gender in English (1991). Because the evidence is scattered throughout the dialectological literature, mostly accumulating in the various publications of the English Dialect Society, the most important accounts will be summarized in the following sections.

2.1. Gendered pronouns in (traditional) English dialectology – 1879 to the present

Probably the earliest mentioning of gendered pronouns can be found in William Marshall’s “Provincialisms of the Vale of Glocester”, where the following is said about pronominal usage:

[T]his quarter of the island affords, among others, one striking deviation in GRAMMAR – in the use or abuse of the pronouns. The personal pronouns are seldom used in their accepted sense [...] sometimes he [is used] for she; as, ‘he was bulled’ – ‘he calved’; and almost invariably for it; all things inanimate being of the masculine gender. (1789: 56; emphasis original, boldface SW)

The excerpt shows the author’s clearly negative attitude towards the variety of English he encountered in Gloucestershire, which may explain why he offers neither concrete examples nor a more detailed description of the phenomenon.

To this day, Frederic Elworthy’s work on the traditional dialect of Somerset is unsurpassed in detail and richness of documentation. Although their authenticity is debatable2, Elworthy’s examples offer researchers the opportunity of catching at least a glimpse of what 19th-century pronominal usage may have been like in (West) Somerset (cf. Siemund forthcoming,

Page 226: A Comparative Grammar of Brtiish English Dialects

216 Susanne Wagner

who used Elworthy’s 1886 publication as the main corpus for his analysis) – an opportunity that is not available for any of the other varieties. In partly re-stating and summarizing his own earlier ([1875] 1965a) work, the author offers the most detailed description of gendered pronouns to be found anywhere in the literature in his 1877 Outline of the Grammar of the Dialect of West Somerset (Elworthy [1877] 1965a). It is this account that has influenced all later studies, and the distinctions and classifications made therein have been taken as reference points ever since.

Every class or definite noun, i.e. the name of a thing or object which has a shape of its own, whether alive or dead, is either masculine or feminine, but nearly always the former; indeed, the feminine pronouns may be taken as used only with respect to persons. (Elworthy 1965a: 32; emphasis SW)

Examples here include a pitcher as well as the nouns “tool, book, house, coat, cat, letter etc.,” that “are all spoken of as he” (Elworthy 1965a: 33). On the other side, “[i]t is simply an impersonal or ABSTRACT pronoun, used to express either an ACTION or a noun of the UNDEFINED SORT, as cloth in the quantity, water, snow, air, etc.” Weather, hay, and beer are used to exemplify this use (cf. Elworthy 1965a: 33; small capitals SW).

With this, Elworthy established a system that could be described in modern terms as a semantic gender system based on a mass-count distinction in nouns, with count nouns taking feminine or masculine pronouns, while mass nouns are neuter. Elworthy himself uses a similar wording in his 1886 West Somerset Wordbook (Elworthy [1886] 1965b), from which the following quotation (entry for he) is taken3:

The universal nominative pronoun to represent all things living or dead, to which the indefinite article can be prefixed. [...] He is used in speaking of a cow or a woman, but not of corn, water, wool, salt, coal, or such things as are not individual, but in the mass. (Elworthy 1965b: 328; boldface SW)

A contemporary of Elworthy’s, the Dorset poet William Barnes, whose poems contain numerous examples of dialect, also wrote a short treatise on the grammar of his home county. His Dissertation on the Dorset Dialect of the English Language (Barnes [1844] 1994) was originally published together with his first collection of poems, Poems of Rural Life in the Dorset Dialect. Although gendered pronouns are mentioned only very briefly in this sketch, Barnes’ poems show that he was aware of the phenomenon. More than 40 years later, his Glossary of the Dorset Dialect with a Grammar of its Word Shapening and Wording offers more detail on

Page 227: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 217

the matter. The heading “two classes of things” contains the following observations on pronominal usage:

Whereas Dorset men are laughed at for what is taken as their misuse of pronouns, yet the pronouns of true Dorset, are fitted to one of the finest outplannings of speech that I have found.

[There are two classes of things:]

1. Full shapen things, or things to which the Almighty or man has given a shape for an end; as a tree, or a tool: and such things may be called the Personal Class: as they have the pronouns that belong to man.

2. Unshapen quantities of stuff, or stuff not shapen up into a form fitted to an end: as water or dust: and the class of such things may be called the Impersonal Class, and have other pronouns and those of the personal class. (Barnes [1886²] 1970: 17; emphasis SW)

It is not difficult to equate Barnes’ class I (“personal class”) with Elworthy’s count nouns and Barnes’ class II (“impersonal class”) with Elworthy’s mass. According to Barnes (1970: 17), he is the pronoun of the personal class, with en serving as objective form. En is explained as deriving from the “Saxon-English accusative” (< he-ene > hine > hin). The impersonal class, on the other hand, uses it. The author illustrates the resulting contrast with the help of examples referring to a tree (personal, thus: he’s a-cut down, John vell’d en) and to water (impersonal, thus: it’s a-dried up). In addition, the same noun can be classified differently according to context. Thus, en has to be used when referring to a brick bat (take en up), but it for “a lot of brick-rubbish: take it up” (Barnes 1970: 18).

Work by Perry (1921) on North Somerset, Kruisinga (1905) on West Somerset, Jennings (1869) on the West of England (with special emphasis once more on Somerset), and Dartnell and Goddard (1893) on Wiltshire shows a mixture of implicit and explicit references to gendered pronouns, while explanations for the phenomenon are generally absent from these accounts. Very often, the given examples contain gendered pronouns, but their use is either only worth a side remark of the type that un stands for “him, or it” (Dartnell and Goddard 1893: 124), or not commented on at all (e.g. Jennings 1869).

Despite its title, Alexander Ellis’ On Early English Pronunciation provides a source for much more than phonological analysis. Ellis’ district 4, the “Southern division”, constitutes the focal area of the present investigation, consisting of – speaking in terms of modern dialect areas (see Trudgill 19992: 65) – the central Southwest as well as the southern part of

Page 228: A Comparative Grammar of Brtiish English Dialects

218 Susanne Wagner

the upper Southwest, with Somerset, parts of Dorset, Hampshire, and Gloucester forming the core area.

In his description of special “grammatical constructions” of this district, the author makes the following comments on gendered pronouns:

This ( ) is very widely spread in the S[outhern] div[ision], and is also used where it is said in received speech, on account of the general use of he applied to inanimate objects [...] (Ellis 1889: 43).

This statement seems rather insubstantial and unsatisfactory as an explanation. That Ellis was aware of the construction can be deduced from his specimens, one of which is full of gendered forms (see Ellis 1889: 151–152), including various masculine references to a tree (he 6x, un 7x, er 1x),a ravine, an oven, a roof (un), and a gate (en 2x).

Gendered pronouns are also encountered in the prose and poetry of West Country authors, most prominently, as already mentioned above, in the work of William Barnes and Thomas Hardy, who in particular has been subject of various studies focusing on his language use. However, most of these studies either emphasize quantity rather than quality (e.g. Hirooka 1980, which contains long but uncommented lists of examples), or comment only very insufficiently on the dialect features found in Hardy’s dialogues (e.g. Elliott 1984).

Unfortunately, then, Ossi Ihalainen’s summary of the 18th- and 19th-century dialect literature seems more than adequate in light of the material presented above: “Generally speaking, the picture that emerges from the early evidence is patchy, difficult to interpret and open to conjecture.” (1994: 197).

2.2. Gendered pronouns in modern dialectological investigations

Two scholars in particular, namely Ossi Ihalainen and Martyn Wakelin, contributed to a better understanding of dialects in the Southwest, publishing extensively on West Country English. Ihalainen’s focus is primarily on grammar, while Wakelin’s two monographs on Cornwall (1975) and the Southwest in general (1986) are mainly concerned with phonology and lexicography, but also include sections on morphology and syntax.

Ossi Ihalainen generally agrees with the accounts of his predecessors in evaluating gendered pronouns. However, since a century of language

Page 229: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 219

development must have had an impact on the traditional system of attributing gender, the author modifies Elworthy’s earlier rather strict system of “mass” versus “count” referents as follows:

On the whole, the evidence suggests that Elworthy’s description is basically correct. However, rather than say that it is used for “mass” referents and the personal forms for “thing” referents, the correct generalization today seems to be that it can be used for “thing” and “mass” referents, although it predominantly occurs with “mass” referents, whereas the personal forms do not occur with “mass” referents at all. (Ihalainen 1985a: 158)

This adaptation of the system is based to some extent on the observation that native speakers themselves seem to vary in their judgements of what is or is not a “thing”: “informants may disagree about thingness” (Ihalainen 1985a: 158).

Ihalainen’s main interest lies in the analysis of two details of pronominal distribution: i) what is the relationship between standard it andnon-standard he forms, and ii) what is the distribution of subject and object forms in this context? Ihalainen uses a partly typological approach to explain the results he observed in his corpus. When looking for gendered pronominal forms, he noticed that their distribution across grammatical cases, i.e. basically subject and object position, was by no means identical.

Based on his findings, the author suggests that standard forms establish themselves in the dialect system in less accessible positions of the Noun Phrase Accessibility Hierarchy (subject > direct object > non-direct/ indirect object > possessor/oblique; cf. e.g. Keenan and Comrie 1977; Comrie 1989²: 155–163) and spread from there. Thus, it is more frequent in object position than one might expect, at least in contexts where he is also possible in dialect (cf. Ihalainen 1985b: 69–70). Or, in Ihalainen’s own words:

[T]he most prominent linguistic contexts are also the most favourable to dialectal forms. Interpreted diachronically, this means that the changes concerned arose in non-prominent contexts and are spreading to more salient ones. (Ihalainen 1991: 105)

For the present investigation, this means that standard it forms first “invaded” the territory of personal forms in object position, later spreading to the more prominent subject contexts as well.4 This hypothesis is in fact borne out in the corpus data investigated in detail in Wagner (2004b).

Page 230: A Comparative Grammar of Brtiish English Dialects

220 Susanne Wagner

Contrasting with Ihalainen’s studies, which for the most part is based on authentic speech samples, Martyn Wakelin’s three major publications (1975; 1986; 1981²) are all based on the SED or at least on SED material, which is the main drawback of the respective monographs.

Largely following the traditional SED vein, Wakelin (partly) shifts the focus from lexicology to phonology and also includes general background information on the variety in question, from geographical to settlement information (e.g. paragraphs about the historical situation of English in the respective areas, including in the case of Cornwall some remarks on the situation of Cornish in relation to English, which is necessary for a full understanding of the language situation in present-day Cornwall).

The 1986 monograph contains a valuable compilation of texts from the core counties of the Southwest, including material from the 16th century to the 1970s. Unfortunately, not much information on morphological or syntactic peculiarities of the area is included. In the book on Cornwall, based on his Ph.D. thesis, Wakelin introduces the chapter headed “Morphological features” as follows:

The small collection of morphological features from SED assembled under several headings in this chapter is intended to support the phonological material in the preceding chapter and the lexical material in that following. (Wakelin 1975: 175)

Wakelin emphasizes that the respective SED material was selected to support his main hypothesis of the book, namely that Cornwall is divided into two parts: From a linguistic point of view, the East is close to its traditional Southwest neighbours, while the West is closer to StE, mainly because of the late introduction of English there (Cornish being the traditional first language until its extinction5). Probably due to this bias in the selection of the material, gendered pronouns are not even mentioned among the morphological features (four features are used, one from the system of personal pronouns, two from verbal morphology, and one from noun phrase morphology; cf. Wakelin 1975: 175–179). One cannot help but wonder why Wakelin chose not to include gendered pronouns, whose use should be very indicative of the assumed distribution of features. One would expect the traditional system for the East, with personal forms being used for count/“thing” nouns and it being used for “mass” referents, while in the West the StE system or at least something close(r) to it should dominate. Strangely enough, Wakelin chose the other peculiarity of Southwestern pronominal systems, pronoun exchange, to make his point.

Page 231: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 221

That the author is by no means unaware of the phenomenon can be seen from his description of gendered pronouns in his likewise SED-basedmonograph on English dialects:

En is used for it (object) as well as for him in the south-west of England, beside which the forms he, him and occasionally she, her may also be used to denote an inanimate object over a rather wider area in the west but in more scattered examples [...] The full implications of the use of he, she, him, her for inanimate objects have not yet been explored. (Wakelin 1981²: 113)

Although showing some phonological bias as well, The Southwest of England (1986) is much more balanced. An eight-page section on phonological features stands against a four-page overview of grammatical features. However, even though comments on the pronominal systems make up about half of the section, gendered pronouns are again only mentioned in passing. While the formal peculiarities (existence of old accusative ’n in a number of variant realizations/spellings) are listed, the slash giving him and it equal status in the example is not commented on (cf.1986: 34).6 The present author’s own investigations on pronoun exchange, a feature that does not avail itself easily to detailed study for a number of reasons (cf. Wagner 2001), show that it is far less frequent than gendered pronouns. Although Wakelin’s procedure is more than surprising, it is – unfortunately – not unique, as a look at some other publications shows: gendered pronouns do not figure prominently in any of the admittedly few modern publications on West Country dialect (cf. e.g. Attwell 1987; Downes 1986; Jones and Dillon 1987; Marten 1973).

3. Gender in English and elsewhere

This section gives a short overview of the history of gender in English, both in isolation and in comparison with other languages. Questions that need to be answered in this context include the following:

1. What is gender? – typological classification 2. How do gender systems develop and change? – diachronic

evidence, focus on English 3. Based on the definitions and classifications in 1. and 2., where

does gender fit in English today?

Page 232: A Comparative Grammar of Brtiish English Dialects

222 Susanne Wagner

These issues will be addressed from a functional-typological point of view, taking English as the focal point, but also taking a glimpse at both related and unrelated languages as dictated by the subject matter.

3.1. Gender in the world’s languages – basic terminology and classi-fications

Two major, sometimes competing systems for assigning gender exist in the world’s languages. On the one hand, there are semantic systems, “where semantic factors are sufficient on their own to account for assignment” (Corbett 1991: 8). Various features are used as the basis for gender assignment in such systems. Systems where masculine gender is attributed to males and feminine gender to females are often called “natural gender systems” (Corbett 1991: 9). Criteria for such systems are widespread; often, the general division is one between human and non-human, and humans are divided into male and female in turn (cf. Corbett 1991: 11). Sometimes the dividing line is animate – inanimate instead of human – non-human. English might be an example of this kind of semantic system, as animals (particularly domestic animals) are usually masculine or feminine according to sex; however, there are other factors that may influence pronoun choice (e.g. conventions of children’s stories; cf. Corbett 1991: 12). Cross-linguistically, “[t]he feature animate is particularly pervasive” in semantic gender systems (Corbett 1991: 31).

A more complex system can be found in Algonquian languages: Most of these have two genders, with a basic animate – inanimate contrast. An additional factor for gender assignment is power: powerful and/or dangerous things (although inanimate) usually belong to the animate gender, i.e. are grammatically animate (cf. Corbett 1991: 20–21). In Caucasian languages, a count – non-count distinction seems to play a role; for example, liquids and abstracts (non-count, non-rigid) belong to the same gender (cf. Corbett 1991: 24–30).

On the other hand, there are formal systems. Here, gender is determined to a large extent by formal criteria (usually phonological, e.g. in French, or morphological, e.g. in Russian; cf. Corbett 1991: 37–39). It is important to note that neither strict semantic nor strict formal systems seem to exist. Most of the world’s languages make use of mixed systems, but even in formal systems “gender always has a basis in semantics” (Corbett 1991: 63). Thus, when conflicting rules exist, semantic considerations normally

Page 233: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 223

take precedence (Corbett 1991: 66). This can be illustrated with the help of a German example.

German has a relatively formal morphological gender system (cf.Corbett 1991: 49–50) which usually assigns gender on the basis of derivations. Lexemes with the suffix -chen (indicating diminutives) are assigned to the neuter gender. With the lexeme Mädchen ‘girl’, there is a clash between semantics and morphology, as semantics clearly demands feminine gender rather than the neuter assigned on the basis of a formal rule. While it is ungrammatical to use a feminine article (i), feminine endings on attributive adjectives (ii), or a feminine relative pronoun (iii), it is possible to use sie ‘she’ as an anaphoric pronoun (iv), although neuter es‘it’ is also possible. The Agreement Hierarchy (AgrH; cf. v) nicely predicts this possibility (cf. Corbett 1991: ch. 8):

i. *die Mädchen ist jung “theNEUT girl is youngØ”ii. *eine junge Mädchen “aFEM youngFEM girl”

iii. das Mädchen, *die ich meine … “theNEUT girl who(m)FEM I mean” iv. das Mädchen ist jung; ich kenne sie/es

“theNEUT girl is youngØ; I know her/it” v. The Agreement Hierarchy

attributive < predicate < relative pronoun < personal pronoun

As we move rightwards along the hierarchy, the likelihood of semantic agreement will increase monotonically (that is, with no intervening decrease).7 (from Corbett 1991: 226)

It is interesting to see that the AgrH also works in the reverse scenario of a basically semantic system where nouns can exceptionally be assigned a category other than the expected gender. English “boat nouns” offer nice examples in this context (cf. Corbett 1991: 180–181; 236–238):

vi. the QEII is a beautiful ship vii. the QEII, on *whom I sailed recently, is a beautiful ship

viii. the QEII, on which I sailed recently, is a beautiful ship ix. I sailed on the QEII recently; she/it is beautiful

While it is impossible to use the relative pronoun who(m) with ship, the use of she as illustrated in (ix) is possible, thus supporting the AgrH.8 Corbett defines agreement as the “determining criterion of gender” (1991: 4) and adds that “[t]his is the generally accepted approach to gender” (1991: 4). Agreement can manifest itself to various degrees in numerous categories. Adjectives can agree with the nouns they classify, verbs with their subjects or even objects, etc.

Page 234: A Comparative Grammar of Brtiish English Dialects

224 Susanne Wagner

Control of anaphoric pronouns by referents is usually included as part of agreement; thus, a language like PrDE, where gender only becomes evident in personal pronouns, should be said to have gender as well (cf. Corbett1991: 5). Corbett calls such a system a “pronominal gender system” (1991: 5). In light of the AgrH, attributive modifiers and personal pronouns seem to be two opposite poles of a single hierarchy, i.e. “they should be treated as part of the same phenomenon” (Corbett 1991: 112). As illustrated in the above examples, personal pronouns (at the end of the AgrH), are the most likely stage at which semantic factors overrule formal ones. This could already be observed in Old English, where it was possible for a non-neuter noun to take a neuter anaphoric pronoun, especially when some spatial distance between pronoun and referent was involved (cf. x):

x. þæt þu þone wisdom þe þe God sealed, that you that wisdom which to you God gave,

þær þær þu hiene befæstan mæge, befæste. there where you itMASC implant may, implant.

Geþenc hwelc witu us þa becomon think what punishments to us then came

for þisse worulde, þa þa we hit nohwæþer for this world when we itNEUT neither

ne selfe ne lufodon, nee ac oþrum NEG ourselves NEG loved, NEG also each other

monnum ne lefdon … men NEG allowed …

(Dekeyser 1980: 101; Corbett 1991: 242–243)

In the example above, the masculine hiene is used in the immediate vicinity of wisdom (masculine in OE), but the next reference, spatially removed from its antecedent, is neuter hit. One can thus conclude that personal pronouns seem to be “the major initiator of changes in the balance between syntactic and semantic gender” (Corbett 1991: 242). English obviously took this direction in its history, developing from a language with a fairly

Page 235: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 225

formal gender system comparable to that of Modern German to a language with a semantic (or even “natural”) gender system.

3.2. Gender in English – the history

It is generally known that English inherited a formal gender system9 from its Germanic parent language which, roughly between the 10th and 14th century, was gradually replaced by the semantic “natural” or “logical” gender system that is known from PrDE, the change having been completed in early Middle English (ME) (cf. Kastovsky 2000: 709; Dekeyser 1980: 102). The dissolution and ultimate demise of the formal system is usually associated with phonetic changes in the syllable that eventually led to the loss of most inflectional endings on the noun. However, it has been noted that it is probably not entirely correct to classify the OE gender system as a purely formal category. Although some derivational endings showed a clear one-to-one match of ending to gender (e.g. -dom masculine, -nessfeminine), “many suffixes were affiliated to more than one gender” (Kastovsky 2000: 712). Also, for underived nouns, next to nothing could be predicted about their gender from their inflectional class, thus making it basically impossible to speak of a formal gender system in OE (cf.Kastovsky 2000: 712). Moreover, the distribution of nouns into the three genders masculine, feminine, and neuter had an almost semantic basis in OE already in that (in West Saxon) most male nouns were masculine, most female nouns feminine, and the majority of neuter nouns were “asexual” (Ross 1936: 321; cf. also Jones 1988: 35; Moore 1921: 89).10

Kastovsky (2000) illustrates how various stages and processes that took place in the inflectional morphology of the noun phrase were intimately linked to the dissolution of the formal category gender. It turns out that in OE already, very few combinations of an agreement-marking item (determiner, adjective) + noun were really unambiguously gender-specific (cf. Kastovsky 2000: 715–717). Thus, “wrong” gender assignment not only existed in OE times (cf. [x] above; Jones 1988: 10), but was presumably rather frequent.11 In addition, semantic and/or pragmatic factors can be expected to have overruled formal ones at that stage as well, particularly where formal gender and sex clashed (e.g. masculine wifman, neutermægden; cf. Kastovsky 2000: 711–712; Jones 1988: 36–38; Wales 1996: 137–139). Consequently, it seems appropriate to look for the beginning of the end of gender in English at a much earlier time, and it is likely that

Page 236: A Comparative Grammar of Brtiish English Dialects

226 Susanne Wagner

Moore (1921: 91) was correct in saying that “natural gender did not replacegrammatical gender in ME but survived it”.12

Anne Curzan (2000) adds another important factor to support this view, a factor which, judging from traditional accounts on gender in English, has largely been overlooked or at least neglected. She points to the status of English as a non-literary language when its first grammars were written, inheriting all systems and their classifications from Latin, a highly synthetic language with an elaborate grammatical gender system:

The early English grammarians were “beginners”, stumbling through English grammar with only Latin grammar and its terminology as their guide; despite the obvious inapplicability of many Latin categories for English, they often retained them in the English grammars either because they wished to adhere to tradition or because they could not conceive of other possibilities. (Curzan 2000: 563)

Judging from the inflectional paradigm of OE nouns, feminine noun endings were generally more distinctive than either masculine or neuter ones. This could support the view that feminine gender lingered longer than the other two, and that this continued association led to the persistence of feminine pronominal reference with some nouns to the present day (cf.Fennell 2001: 64). This is a fact worth remembering when the predominance of feminine referents is investigated in Section 6. Matters were not simplified by the gradual case syncretism of dative and accusative personal pronouns. Visser (1963: 427) notes that the reason for the change from hine to him “has as yet not been satisfactorily accounted for”. Possible explanations include the extension of the dative from rather frequent verbs taking dative forms as objects after speakers were no longer aware of the case distinction(s) (cf. Visser 1963: 427), or an economically motivated choice of the pronoun most distinct from the nominative (cf. Howe 1996: 114–115). Although dative-accusative levelling took place in most Germanic languages, some languages lost the dative forms, while others generalized the dative and lost the accusative forms. Howe (1996: 111) thus concludes “that no theory is at present able to account fully satisfactorily for both these directions.” Although the entries for it and him in the OED describe the concrete developments in detail, the only conclusion to be drawn from these observations is that there was a rather extended period of time in the history of the English language when the choice of a supposedly masculine personal pronoun (him) said nothing about the gender or sex of the referent. It could be masculine, male, neuter, or asexual – and every combination of those four.

Page 237: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 227

3.3. Gender in English today

Although rather marginal in PrDE, gender has always been a category (formal or functional) of English, and no one dares (yet) to let go of it completely. In the following, the view that grammarians have on gender in PrDE will be investigated more closely.

3.3.1. Gender in modern references grammars

As has already been indicated, modern scholars writing on PrDE are highly influenced by traditional accounts of gender in Indo-European languages. They adopt the traditional terminology which was developed for those languages, and no one seems to consider it inappropriate for describing a language like PrDE that is almost as far removed from a true “gender language” as it can possibly be. Quirk et al.’s Comprehensive Grammar of the English Language (1985) is no exception in this respect:

By gender is meant a grammatical classification of nouns, pronouns, or other words in the noun phrase, according to certain meaning-related distinctions, especially a distinction related to the sex of the referent. In English, unlike many other related languages, nouns, determiners and adjectives have no inflectionally-marked gender distinctions. Some 3rd person pronouns and wh-pronouns do, however, express natural gender distinctions:

it, which etc [nonpersonal] contrasts with the following:

who, whom etc [personal]

he, himself etc [masculine, chiefly personal]

she, herself etc [feminine, chiefly personal]

(Quirk et al. 1985: 314; original emphasis)

This definition of gender sounds suspiciously non-specific; yes, gender is a grammatical category, but is it really a grammatical category of PrDE? The authors choose the unfortunate path of equating gender with sex in talking about meaning-related distinctions, in particular those referring to the sex of the referent. The next paragraph tries to clarify matters, but one cannot help but feel slightly confused by the varied terminology:

Page 238: A Comparative Grammar of Brtiish English Dialects

228 Susanne Wagner

Gender in English nouns may be described as ‘notional’ or ‘covert’ in contrast to the ‘grammatical’ or ‘overt’ gender of nouns in languages such as French, German, and Russian; that is, nouns are classified not grammatically, but semantically, according to their coreferential relations with personal, reflexive, and wh-pronouns. We use the term male and female in reference to the ‘covert’ gender of nouns, as distinct from the ‘overt’ gender of pronouns. (Quirk et al. 1985: 314; original emphasis)

While the previous paragraph spoke of “natural gender distinctions”, now they are “notional” or “covert”. This in itself poses no real problem, but equating the existence of gender with the existence of gender-specific pronouns does. Claiming that English has gender because it has pronouns that show gender distinctions is a dangerous garden path – the traditional argumentation would rather be that the distinctions in the pronominal systems only exist because the nominal referent carries the feature gender, which in turn has to be mirrored in the pronominal system(s).

What is even more unfortunate in the terminology used in the above paragraphs is the use of male, female, masculine and feminine. Quirk et al.equate the gender terms masculine and feminine with the “overt”, “natural gender” of pronouns, while nouns with their “covert”, “notional gender” are male or female, which are biological categories referring to the sex of the referent. This seems to be a contradiction in terms. Why should different terms – which describe essentially the same phenomenon – be used for different word classes? A personal pronoun referring to a male human has either both features, [+ male] and [+ masculine], or, if one were to accept that English gender is a purely semantic category, only [+ male]. Also, it is not clear why the authors suggest that nominal gender is an overt category in German – grammatical, yes, but the categories are only rarely expressed overtly (although some derivational endings allow certain predictions about gender assignment, the rules for underived nouns are far too complex to call them overt). As Corbett (1991: 49) summarizes, gender assignment rules in German consist of “a complex interplay of overlapping semantic, morphological and phonological factors.” The gender system Quirk et al. derive from these observations must look confusing, if not wrong, to most speakers of English (cf. Fig. 5.104 in Quirk et al. 1985: 314).

According to this system, English has nine (!) gender classes (a term that is never explained nor defined) which reveal a high degree of overlap with each other. None of the sub-categories is sufficient for assigning gender to a noun. At first glance, “inanimate” seems to be sufficient, but if the classification were based on pronoun co-reference exclusively (as the

Page 239: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 229

authors claim it is), all categories that allow “which – it” (i.e. categories d, e, f, g, h, i) should be in the same sub-gender. Obviously, the authors applied some sort of hierarchy to arrive at the above classification in which humanity > sex > animation > pronoun choice, but the basis of such a system, if it exists, is never explained.

On the category of animals, the following can be found: “Male/female gender distinctions in animal nouns are maintained by people with a special concern (for example with pets)” (Quirk et al. 1985: 317). No reference is made to the fact that animals are generally he in spoken language (cf. section 5.2). The authors’ concern here is with “professional” language use that includes terminological differentiation, e.g. dog – bitch etc. Another exceptional noun class mentioned is that of country names, which can be used with neuter (emphasising the geographical unit) or feminine (political/economic units) pronouns (cf. Quirk et al. 1985: 318). Oddly enough, it is stated that in the latter case gender is either “class (b) or (g)” (ibid.) – (b) is acceptable, but it seems more than doubtful that a country could be classified as a “higher female animal” (class g) …

On the well-known use of feminine pronouns referring to ships, we are told that

[i]nanimate entities, such as ships, towards which we have an intense and close personal relationship, may be referred to by personal pronouns, eg:

That’s a lovely ship. What is she called?

In nonstandard and Australian English, there is extension of she references to include those of antipathy as well as affection, eg:

She’s an absolute bastard, this truck. (Quirk et al. 1985: 318, original emphasis)

With the generalized “we” in the first sentence the authors overdo things slightly. While it cannot be denied that a group of professionals (fishermen, ship crews, yacht owners, etc.) certainly have “intensive and close personal relationships” with the ships they are sailing on, it seems difficult to claim such a relationship for the average person. The following statement on Australian English sounds rather unprofessional, the wording and inclusion itself a bit unreasonable and rather unfounded. Although the co-ordination of “non-standard” and “Australian English” seems awkward (and is, as will be shown later, wrong), the authors at least mention that feminine forms are the pronouns of choice in speech (cf. Section 6). However, they offer neither comments nor explanations for this type of use.

Page 240: A Comparative Grammar of Brtiish English Dialects

230 Susanne Wagner

A more contemporary (or updated) approach to gender in English can be found in Longman’s Grammar of Spoken and Written English (Biber et al. 1999), incorporating, as the title suggests, considerable amounts of material from spoken language for the first time and thus complementing rather than substituting Quirk et al. (1985).

Gender is a less important category in English than in many other languages. It is closely tied to the sex of the referent and is chiefly reflected in cooccurrence patterns with respect to singular personal pronouns (and corresponding possessive and reflexive forms). The main gender classes are:

example noun pronoun personal/human: masculine Tom, a boy, the man he feminine Sue, a girl, the woman she dual a journalist, the doctor he, she non-personal/neuter: a house, a bird it

(Biber et al. 1999: 311–312)

The system presented here seems much closer to reality than the one proposed by Quirk et al. Biber and his colleagues are well aware of the fact that gender is a problematic category in PrDE. Thus their insightful caveat about oversimplifying matters: “However, gender is not a simple reflection of reality; rather it is to some extent a matter of convention and speakers’ choice and special strategies may be used to avoid gender-specific reference at all” (Biber et al. 1999: 312).

After this general introduction, Biber et al. largely discuss pragmatic motivations for pronoun choice, such as the use of specifically gender-marked forms on the one hand or avoidance on the other (e.g. chairman/-woman vs. chairperson). They note “a continuing sex-bias in English language use and society more generally” (1999: 313) towards masculine terminology, which is not only reflected in a much higher occurrence of masculine forms in pairs like the one mentioned above, but also in the distribution of third person singular personal pronouns in general: masculine forms are more frequent in all registers, occurring 1.5 to more than 3 times as frequently as feminine forms (Biber et al. 1999: 333–335).

Of major relevance for the present discussion is the section about personal vs. non-personal reference (Biber et al. 1999: 317–318). The authors state that “[p]ersonal reference expresses greater familiarity or involvement. Non-

Page 241: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 231

personal reference is more detached” (1999: 317). Items falling into the category that offers a three-way choice (personal he, she; non-personal it) are expressions for young children (infant, baby, child) and animals (pets in particular; cf. Biber et al. 1999: 318). An exceptional status is once again attributed to nouns denoting countries and ships, which offer a two-way choice (personal she, non-personal it). Although Biber et al.’s account comes much closer to the actual facts observed in the realization of gender English today, like Quirk et al. they fail to offer explanations for the exceptional cases, most of which are not even mentioned (e.g. why a ship can be she).

Huddleston and Pullum’s Cambridge Grammar of the English Language (2002), which contains the most recent comments on the subject, will be introduced by way of quoting their introductory remarks on “Gender as a grammatical category”.

[...] it is important to distinguish carefully between the grammatical terms masculine and feminine and the semantic or extralinguistic terms male and female. Until relatively recently it was usual to make a parallel distinction between gender (grammatical) and sex (extralinguistic) [...]. In the socialsciences, however, ‘sex’ came to be used to refer to biological attributes and ‘gender’ to the social construction of sex, and this usage has been incorporated into linguistics. A book on ‘language and gender’ will therefore not be primarily concerned with gender as a grammatical category, but will cover such matters as differences between the speech of men and women. Our concern in this section, however, is with gender in the old, strictly grammatical sense [...] (Huddleston and Pullum 2002: 484–485; italics SW)

For the first time, the change in terminology – and the resulting confusion – is topicalized in a grammar, which in itself points to a radical restructuring in the respective areas of research. The authors continue their introduction by justifying their decision to treat gender as a grammatical category, using an argumentation very much in Corbett’s vein (agreement as the defining criterion of gender English does show agreement, though in a very restricted sense English has gender, though it is not an inflectional category and not as strongly grammaticalized as in other languages; cf.Huddleston and Pullum 2002: 485–486). Although their use of the term “less grammaticalized” is debatable13, the authors take a very clear position in assessing the category, which is refreshing and helpful compared with the earlier descriptions.

Typical wording can be found in the actual distributional properties of masculine he, feminine she, and neuter it. He and she referring to males and

Page 242: A Comparative Grammar of Brtiish English Dialects

232 Susanne Wagner

females respectively, it referring to “entities which are neither male nor female”, are identified as the “core uses” of he, she, and it (cf. Huddlestonand Pullum 2002: 484). As this definition of it would exclude its use with both animal and human antecedents, which do exist, there is an extra section on these exceptional uses. In the case of animate nonhuman (i.e. animal14) antecedents, the authors state the following:

– It is generally used when the sex is unknown (cf. Huddleston and Pullum 2002: 489);

– he and she are “more likely with pets, domestic animals, and creatures ranked high in the kingdom of wild animals” (2002: 489; e.g. lions, tigers, ...);

– the use of he or she “indicates a somewhat greater degree of interest in or empathy with the referent than does it” (2002: 489).

It is the third factor that is remarkable, as this is what every native speaker would say in an impressionistic account and what has been part of socio-pragmatic gender studies for a long time, but what has not been taken up in grammars so far.15 As for the reverse scenario, the use of it with human antecedents, the authors again combine a traditional commonplace (it canbe used for babies) with modern specifications (used in such a manner, ittends to suggest resentment or antipathy; 2002: 489).

The other “special” section concerns the use of she with non-females. According to the authors, such usage is possible with two categories, namely with countries (when considered as political, not as geographical entities) and ships “and the like” (Huddleston and Pullum 2002: 484):

Ships represent the classical case of this extended use of she, but it is found with other kinds of inanimates, such as cars. There is considerable variation among speakers as to how widely they make use of this kind of personification. It is often found with non-anaphoric uses of she: Here she is at last (referring to a ship or bus, perhaps), Down she comes (with she referring, say, to a tree that is being felled). (2002: 484; boldface SW)

The (in)appropriateness of the label “personification” will be discussed in a later section (cf. section 5.1). However, it should be clear from the examples themselves that we cannot be dealing with personification if the pronoun is not used anaphorically – what is personified if the referent is not known? Perhaps an abstract idea or situation (as this is what most of the relevant instances of she seem to refer to; cf. section 5.3)? Although it is

Page 243: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 233

admirable that the authors mention such uses at all, which are more widespread than generally assumed, they fail to explain them.

All in all, the descriptions of gender in modern reference grammars are unsatisfactory in a number of respects. They either do not reflect actual language use, or they mix traditional with modern interpretation, which adds to the confusion rather than helps clarify it. If anything, the authors largely describe an idealized version of gender assignment in written StE.

3.3.2. Descriptions of gender in various works

Gender is not a category that figures prominently in descriptions of PrDE. Usually it is listed among those categories that have been weakened over time. Thus, Leisi and Mair (19998: 140–141) state that gender in English has lost much of its weight, primarily because it was a purely grammatical category without any solid basis in reality. Exceptional feminine and masculine nouns include names of countries and “machines men have a close emotional relationship with” (e.g. motorbike), classified as “adopted natural (psychological) gender” (Leisi and Mair 19998: 141, translation SW). This category is traditionally known as “metaphorical gender”.Additionally, there is the “allegorical gender” of abstract nouns, which, according to the author(s), is largely based on the gender associated with the noun in the original classical language. Thus, love can be masculine (< Ltn. amor), peace feminine (< Ltn. pax).

Figure 1. Gender categories in Brinton (2000)

More recently, Brinton (2000: 105–106) follows the dominating view that PrDE has “natural gender” as opposed to its earlier “grammatical gender”. She notes that “gender is generally a covert category” in nouns, while “a related category of animacy (animate/inanimate)” is not only expressed in personal, but also in interrogative and relative pronouns (what vs. who;which vs. who; 2000: 105). Interesting in her classification are the postulated animacy groupings: humans and higher animals on the one,

he, she; who

humans higher animals

it; which

lower animals inanimates

?

Page 244: A Comparative Grammar of Brtiish English Dialects

234 Susanne Wagner

lower animals and inanimates on the other hand. Animals thus appear on both sides of the scale. Although it is not mentioned explicitly by the author, it should be clear that the cut-off point can vary on all levels of lectal variation (dia-, socio-, idiolect), depending on the situation, context, addressee, etc. Figure 1 is an attempt to illustrate Brinton’s categories.

A description in an almost forgotten grammar of English is encountered time and again in publications on the status of gender in PrDE. George Curme’s English Grammar was first published in 1925, and to the present author’s knowledge, it served as a basic college grammar.16 Curme’s introductory remarks on gender read as follows: “Gender is a distinction in the form of words to indicate sex. There are two kinds of gender in English – natural gender and the gender of animation” (1962: 209). In the following paragraphs, the author illustrates natural gender, which “rests upon the conception of sex in nature” (1962: 209), with the help of the three strategies that English employs:

a) use of distinctive terms (generally relational opposites such as man– woman, husband – wife),

b) use of affixes (e.g. -ess vs. -or),c) use of pronouns (personal and possessive, i.e. he – him – his vs. she

– her – her)

In the context of the latter, Curme notes a tendency to regard animals as masculine, although they are generally treated as neuter (cf. Curme 1962: 211). What is really noteworthy about Curme’s description is his category of a “gender of animation”, which he explains as a sort of remnant from OE times (and its grammatical gender) that has developed a life of its own: “The old habit of associating lifeless things with sex continued and in our playful moods with their animated feeling still has strong sway” (1962: 212). Curme’s main point, the contrasting function of animated pronouns, will figure prominently in later parts of this study. The author distances himself strongly from identifying animation with personification:

This gender does not – as most grammars and rhetorics falsely suppose – rest upon vivid personification, but is merely an animated form to serve as a contrast to the scientific precision of our normal expression, which treats as neuter all living and lifeless things which lack personality. (Curme 1962: 213; emphasis SW)

David Crystal is an exemplary case of a scholar who uses personification as a way out of the dilemma that exceptional pronoun usage poses. He tackles a problem that will be discussed in detail in the main section of this study –

Page 245: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 235

namely why (at least in certain varieties) feminine pronouns are much more frequently found in “neuter” contexts than their masculine counterparts:

Many nouns are given variable gender, depending on whether they are thought of in an intimate way. Vehicles and countries are often called she as well as it (She can reach 60 in 5 seconds; France has increased her exports). Pets are often he or she. A crying baby may become it. It is not obvious why some entities are readily personified while others are not. Nor is it obvious why most entities are given female personifications. It is not simply a matter of feminine stereotypes, for she is used in aggressive and angry situations as well as in affectionate ones: guns, tanks and trucks which won’t go remain she. The only consistently male trend in personification which the author has heard in recent years is in computing, where word processors and other devices are widely given male pet names and pronouns. Why this should be so is beyond him [...] (Crystal 1995: 209)

Although it will be shown in the relevant section (see 5.1) that Crystal jumps too readily to the conclusion that all of these are cases of personification, his helplessness seems indicative of most scholars’ feelings towards the issue at hand. It is hoped that the discussion presented here will help shed some light on a largely uninvestigated (and underestimated) area of PrDE personal pronoun usage.

4. The corpora

The data used for this study stem from two very different sources, namely (1) the basic and incidental material collected by the SED fieldworkers and (2) various collections of an oral-history type from the Southwest of England and Newfoundland. While the previous material consists of unconnected utterances of sometimes just a couple of words, the latter is in interview form, so that it is obvious from the outset that different types of questions need to be addressed to these different sets of data. This section will provide some basic background information on the individual sources. In addition, problems that emerge in connection with the composition of the data (e.g. issues of comparability) will be discussed.

Page 246: A Comparative Grammar of Brtiish English Dialects

236 Susanne Wagner

4.1. The Survey of English Dialects (SED)

The fieldwork for the SED was undertaken between 1950 and 1961 (cf. Orton 1962: 14). 311 localities all over England were selected “according to their geographical position isolatively (sic) and relatively to each other” (Orton 1962: 15); agricultural communities with a population of approximately 500 were preferred. The informants selected are by and large of the type that in dialectal studies is known as NORMs, i.e. non-mobile, older, rural males (cf. Chambers and Trudgill 1998²: 29). As Orton states, informants were very rarely below the age of 60, which sets their birth dates at the end of the 19th century or earlier (cf. Orton 1962: 15). Speakers who had spent a considerable amount of time away from their home community “were constantly regarded with suspicion” (if only in terms of their linguistic authenticity; Orton 1962: 16). The method chosen for data elicitation is the questionnaire-based interview, which is one of the direct methods used in data collection (cf. Francis 1983: 78–80). The responses were taken down in phonetic script in the fieldworkers’ notebooks. An example of such a notebook page can be found in Francis (1983: 98).

The sheets [of the notebooks] were divided down the middle. The left side was reserved for the informant’s responses and for any remarks or explanations about them. [...] The right side of the page was intended for the fieldworker’s transcriptions of any significant expressions from the informant’s conversation that had relevance to problems under investigation in the Questionnaire. Relatively unconditioned by the somewhat artificial circumstances of the interview, this incidental material is particularly valuable for confirming, supplementing, amplifying or even contradicting the evidence of the responses themselves. All the fieldworkers made a point of collecting as much of this material as was feasible in the situation. (Orton 1962: 17–18)

The fieldworker notebooks provide an incredible wealth of material for a study concerned with personal pronouns, a fact that may come as a surprise considering the rationale behind the SED: of the 1322 “virtual questions” that constituted the questionnaire, 387 concern phonological issues, 730 are concerned strictly with lexical differences, and only 205 questions (128 + 77 or 15.5%) directly addressed morphological or syntactic phenomena (cf. Orton 1962: 15). It may thus seem inappropriate to use SED material for a study that is ultimately concerned with a morphosyntactic phenomenon. However, the make-up of the SED data (both the basic and incidental material) allows a number of analyses which, although not envisioned by

Page 247: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 237

its makers, are of a morphological and/or syntactic nature. A prerequisite for such a study is the salience of the feature under investigation.

Except for agreement phenomena, personal pronouns are probably the only class of items that can be found in the vast majority of responses. This is largely due to the fortunate circumstance that informants usually did not restrict themselves to one-word responses, but replied in complete sentences. Unfortunately, the basic material does not always reflect this. In the majority of cases, just the tokens relevant to the question are presented in the published material. Luckily though, the original responses are preserved in their entirety in the fieldworkers’ notebooks, which are accessible to researchers in situ at the University of Leeds.

In the basic material, care must be taken not to overlook possibly interesting questions/responses due to the overall make-up of the published data. Although there is a section in the questionnaire that deals with morphosyntactic items exclusively (book IX), this does not mean that all other questions are worthless for non-lexicological investigations. Just to give one example relevant to the present study, question I.11.6 has the form “How do you empty the cart the quickest way?”. The expected response is “to tip”. Most informants, however, did not simply reply “tip” or “to tip”, but used a whole sentence, generally in the form “(I/we/you) tip it (up)”, thus including a personal pronoun (object form) referring to “cart”.

For this particular question, a pronoun is present in 45 of 66 responses (68%) in the Northern Counties, while no pronoun was recorded in only 21 cases.17 For the West Midlands, out of a total of 86 localities, speakers at 54 localities included a pronoun in their response (63%; no pronoun at 34 localities). In the East Midlands and East Anglia, 58% of speakers used a pronominal form (55 out of 95 localities; no pronoun at 40 localities), and in the Southern Counties, the figure even climbs to 89% (67 of 75 localities with, eight without pronoun). The method introduced here will be taken up in the actual analysis of examples to support certain claims or to give details of a particular observation.

Used in such a manner, the basic material has more to offer in terms of morphosyntactic content than may first meet the eye. Still, however fruitful a search for particular forms in the basic material may be, it is nothing in comparison with the unsurpassed wealth of data that the fieldworkers’ notebooks provide. For the present investigation, a search through almost 80 notebooks from 11 counties yielded a total of roughly 700 pronominal forms of interest. Although critics claim that studies based on the incidental material fail to take the context of the utterance into account, such criticism

Page 248: A Comparative Grammar of Brtiish English Dialects

238 Susanne Wagner

is unfounded in this case.18 The following paragraphs will explain why this should be the case with the help of some examples from the notebooks.

Question IX.3.2 belongs to the “irregular verbs” section, in this particular case trying to elicit the present, simple past and past participle forms of find. It is a so-called completing question (cf. Orton 1962: 48) where the fieldworker is supposed to read out the stimulus sentence and pause before the key word (in bold print below) to allow the informant to complete the sentence. This and all further questions are reprinted here from the published version of the questionnaire (cf. Orton 1962); therespective contexts are:

He was looking for his knife but couldn’t ... find it.Next day he looked for it again and this time he ... found it.He came back looking pleased and told us that he had ... found it.

Irrespective of the provided context, informants very often chose to reformulate their response in a variety of manners. Such a reformulation could result in a more personal format (“I found it”) instead of keeping to the third-person singular context, or in a repetition of the trigger sentence, but once more in the first person (“I/we was/were looking for the knife but I/we could not find it”). No matter what the change, the fieldworkers usually put everything down that did not conform to the prescribed response 100%. Thus, the following responses to question IX.3.2 can be found in the notebooks (see Table 1a).

For the published material, only the token actually asked for was extracted from the response, while all additional information was omitted – i.e. it has been lost to everyone who does not have access to the notebooks. The policy of inclusion or non-inclusion of a personal pronoun (usually in the direct object position) in the basic material is unclear. For example, all of the following questions yielded a comparatively high output of pronominal forms when the notebooks were consulted, but none of them made it into the basic material, where only the required form was reprinted (see Table 1b; the underlined noun is the one that reappears in pronominal form in the responses).

Page 249: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 239

Table 1a. Responses to question IX.3.2 in SED fieldworker notebooks

county location infor-mant response19

Cornwall St. Ewe JJ I found it.

Devon Blackawton JW I #can’ #find en [ ]. I can’t find it.

Devon Kennford JW ’e #told us ’e’d a-#found en [ ].He told us (me) he’d found it.

Devon Swimbridge GY ’e #told I ’e’d a-#foun’ en [ ].He told me he’d found it.

Somerset Horsington TFW He gi’d en [ ] to en [ ]. He gave it to him.

Somerset Montacute EP We can never find it.

Somerset Montacute EP We found it where we put it down [to].

Somerset Pitminster CMM ’e was lookin’ for en [ ] but ’e couldn’ find en [ ]. He was looking for it but he couldn’t find it.

Somerset Wedmore WF (I) couldn’(t) see en [ ] nowhere.I couldn’t see it anywhere.

Wiltshire Burbage RH ’e couldn’ find en [ ]. He couldn’t find it.

Wiltshire Burbage RH I’ve found en [ ]. I’ve found it.

Table 1b. SED questions containing pronominal forms

VI.5.11 When I have an apple, I ... eat it.VIII.5.8 What’s a grave filled in with? Earth.VIII.7.3 A rubber ball that’s punctured won’t ... bounce.IX.1.3 A picture not hanging straight is hanging ... askew. IX.2.7 A door left like this, you say is ... ajar.IX.2.9 If a door has been made of unseasoned wood, before long it will be

sure to ... warp.

On the other hand, pronominal forms were included in the following questions, which also yielded numerous pronominal forms of interest:

Page 250: A Comparative Grammar of Brtiish English Dialects

240 Susanne Wagner

VIII.7.6 A dog buries a bone because he wants to ... hide it.IX.2.6 And now [stand sideways in front of it (door, SW)] ... in front of it.IX.2.8 If the door blew open on a cold day, you’d get up at once and ... shut it.

The decision to include or not include items with the same referent (here e.g. questions IX.2.5–9, which all concern a door) seems to be based on the presence or absence of a pronoun in the expected response. However, this principle is not always obeyed, as question VI.5.11 shows – it does includea pronoun that is not included in the basic material.

Most of the pronominal forms from the fieldworker notebooks that are relevant for the present study are unambiguous. Although the referent of the pronoun is not always obvious at a first glance, it can usually easily be deduced from the respective question or its context. The process of disambiguating unclear referents involved the following steps:

a) The formulation I always ate en skin and all was encountered at St. Ewe (Cornwall), in book VI;

b) the question that was recorded closest to this remark on the left side of the notebook page is VI.5.11, which is about an apple;

c) considering the content of the utterance, it seems impossible for the [ ] to refer to anything other than an apple;

d) as a precaution, the immediately preceding and following questions were also considered. In this particular case, question VI.5.10 reads “… this, where the roots of the teeth are”, enquiring about gums, and VI.5.12 “When, in eating, we crush apples or biscuits noisily with our teeth, we say we ... crunch them.” There is no possibility for the [ ] recorded to refer to either gums or to have anything to do with crunch them, making it a 99% certain example of a gendered pronoun referring to an inanimate noun, namely apple.

Although some ambiguous or problematic cases remain unclear even with such careful scrutiny, the results from the fieldworker notebooks are overall very transparent. Unclear cases will be pointed out in the discussion of examples, usually in connection with issues that have been debated in the respective literature. From the present author’s own experience with the notebooks, at most 10% of the examples are problematic or ambiguous. More often than not, referents are even included in the utterance, so that the possibility of misinterpretation does not even arise. Table 2 lists the notebooks that have been used for this study.

Page 251: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 241

Table 2. SED fieldworker notebooks used in this study

county ID in SED county locations 36 Co Cornwall 1. Kilkhampton

2. Altarnum 3. Egloshayle 4. St. Ewe 5. Gwinear 6. St. Buryan 7. 7. Mullion

37 D Devon 1. Parracombe 2. Swimbridge 3. Weare Giffard 4. Chawleigh 5. Gittisham 6. South Zeal 7. Kennford 8. Peter Tavy 9. Widdicombe 10. Cornwood 11. Blackawton

31 So Somerset 1. Weston 2. Blagdon 3. Wedmore 4. Coleford 5. Wootton Courtenay 6. Stogursey 7. Stogumber 8. Withypool 9. Brompton Regis 10. Stoke St. Gregory 11. Horsington 12. Pitminster 13. Merriott (14. Montacute20)

38 Do Dorset 1. Handley 2. Ansty 3. Whitchurch Canonicorum 4. Portesham 5. Kingston

Page 252: A Comparative Grammar of Brtiish English Dialects

242 Susanne Wagner

Table 2. cont.

32 W Wiltshire 1. Ashton Keynes 2. Sutton Benger 3. Avebury 4. Burbage 5. Steeple Ashton 6. Netheravon 7. Sutton Veny 8. Fovant 9. Whiteparish

In addition to the notebooks from the core Southwest, the notebooks for Hampshire (6 localities21), Gloucestershire (7 localities), Herefordshire (6 + 1 localities – Lyonshall, only four miles from the Welsh border, was already visited by Peter Wright in 1952, but was later excluded from the Basic Material, probably because the fieldworker noted that the Welsh influence was rather strong), Worcestershire (7 localities) and Oxfordshire (6 localities) were consulted, resulting in a total of 79 notebooks. Although the 5 notebooks for Berkshire were also consulted, it was decided not to use them as issues of legibility made it impossible to draw any clear findings from the material. A map detailing the exact locations of the SED localities can be found in Orton (1962: 30).

4.2. Collected material

The second major source of material that was tapped for this study are interviews from various oral history projects all over the Southwest of England. Advantages and disadvantages of using oral history material for linguistic purposes are discussed in detail in Anderwald and Wagner (forthcoming). For Newfoundland the major source was soon identified in the archive at Memorial University (Memorial University of Newfoundland Folklore and Language Archive – MUNFLA), where interviews with natives are stored that were conducted for a range of different studies, both individual (generally students’ papers and theses) and project-oriented (e.g. a collection of Folktales, which was also utilised in the present study).22

Page 253: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 243

4.2.1. Somerset material

The Oral Archive of the Somerset Rural Life Museum (SRLM) consists of some 350 interviews to date, recorded between approximately 1973 and the present, and is continuously expanded whenever someone’s interest in a certain subject is peaked. The material for this study was selected in reading through transcripts and listening to stretches of the tapes to find out if the speakers were “dialectal enough” for the project at hand. In the end, the Somerset corpus consisted of 30 interviews from the SRLM Oral Archive and one interview from the private collection of one of the Friends of the Museum. The statistics for the SRLM data can be found in Table 3.

Table 3. SRLM material

number of interviews 31number of speakers23 36number of words (total) 174,500number of words per interview (average) 5,629number of words per interview (maximum) 17,291number of words per interview (minimum) 1,671interviews recorded from – to 1968–1996recording dates (details)24 1960s (1x); 1970s (7x); 1980s

(20x); 1990s (3x)informants’ ages from – to 70–89informants’ ages (decades) 70s (10x); 80s (15x)informants born from – to 1884–1918informants born (decades) 1880s (3x); 1890s (7x); 1900s

(10x); 1910s (5x)

4.2.2. Wiltshire material

Although the Wiltshire Folklife Society had recently been dissolved when our interest in suitable dialect material started, we were able to get access to some of the material. Three of the interviews in very traditional “Wiltsdialect” are used here (details in Table 4). As copies of the tapes were not available, this material has to be taken with a grain of salt. Although speakers seem to be typical NORMs, born around the turn of the century, they show features that were rare even in SED times, such as the use of thicas demonstrative pronoun. Thus, these interviews are not taken to be as

Page 254: A Comparative Grammar of Brtiish English Dialects

244 Susanne Wagner

authentic as the rest of the material and will be treated accordingly in the analyses.

Table 4. WFLS material

number of interviews 3 number of speakers 4 number of words (total) 13,700

The second source for Wiltshire material was found at the Trowbridge Museum and its oral archive. The archive primarily holds a collection of interviews about the history of weaving, the primary industry in Trowbridge until recently. The interviewees are former workers in the weaving mills in and around Trowbridge, and this is one of only few locations for which at least some data for females (three speakers) is also available, as many of the jobs in the mills typically were women’s. The dialect is very strong in most speakers selected for this study; others who showed a more standard pattern of speech were disregarded. Most of the traditional features associated with Wiltshire or more general West Country speech are present in those speakers. Details for the Trowbridge material can be found in Table 5.

Table 5. Trowbridge material

number of interviews 6 number of speakers 10 number of words (total) 75,100 number of words per interview (average) 12,517 number of words per interview (maximum) 24,131 number of words per interview (minimum) 7,637 interviews recorded from – to 1968–1996 recording dates (details) 1987; 1992 (2x); 1994 (2x) informants’ ages from – to 66–85 informants’ ages (details) 66, 71, 81, 85 informants born from – to 1907–1926 informants born (details) 1907, 1909, 1913, 1923, 1926

4.2.3. Devon material

The material from Devon used in this study stems from the oral history material of the Totnes Community Archive at the Totnes Elizabethan

Page 255: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 245

Museum. The collection of tapes is large (ca. 300), but as no comprehensive catalogue exists and only about 10% of the interviews are transcribed, deciding what to select was difficult. Interviews with existing (standardized) transcripts were preferred in the initial stages of the selection process, replaced by time-consuming listening-in on the interviews at later stages. Details for the interviews, mostly recorded in and around Totnes, that have been selected for this study are provided in Table 6.

Table 6. Devon material

number of interviews 5 number of speakers 5 number of words (total) 51,100 number of words per interview (average) 10,020 number of words per interview (maximum) 11,320 number of words per interview (minimum) 9,550 interviews recorded from – to 1984–1985 recording dates (details) 1984; 1985 (2x) informants’ ages from – to 76–82 informants’ ages (details) 76, 79, 82 informants born from – to 1902–1913 informants born (details) 1902, 1906, 1909, 1910, 1913

4.2.4. Cornwall material

When first researching West Country speech, it appeared that Cornwall was a negligible area in terms of traditional dialect to be found there. No accounts from the 19th century comparable to Elworthy’s studies of Somerset speech exist, and secondary literature in general is scarce to non-existent. The successor of the Institute for Cornish Studies, the CornishAudio and Visual Archive, held a large amount of material suitable for the present study. Unfortunately, almost none of it was transcribed, again making the selection process very time-consuming. In addition, the transcription of the material turned out to be a particularly difficult enterprise, as those interviews prove that modern and traditional Cornish dialect is much better than its reputation – the material from Cornwall is much more dialectal than is to be expected from the literature. It has always been claimed that Cornwall is much closer to StE, as the result of the much shorter history of English in the area. Accordingly, dialects did not have time to develop, and the area was much more influenced by school English

Page 256: A Comparative Grammar of Brtiish English Dialects

246 Susanne Wagner

rather than traditional West Country speech – recall Wakelin’s position on this (Wakelin 1975: 100, quoted in full in note 5 on page 372).

In reality, nothing could be further from the truth. Being familiar with the relevant literature, one gets the distinct impression that researchers tended to over-generalize the situation they found in the phonological system and the lexicon, where not as many traditional dialect features / words were found as was expected – see, for example, Fischer’s lexicological description (1976: 298) or Bremann’s socio-phonological study (1984: 153; cf. Wagner 2004a). An overview of the interviews selected here, all from West Cornwall (Pendeen, Gurnard’s Head, Zennor and St. Ives) is presented in Table 7.

Table 7. Cornwall material

number of interviews 4 number of speakers 4 number of words (total) 18,900 number of words per interview (average) 4,725 number of words per interview (maximum) 7,506 number of words per interview (minimum) 2,835 interviews recorded from – to 1978–1980s recording dates (details) 1978 (2x); 1979; 1980s informants’ ages from – to 74–86 informants’ ages (details) 74, late 70s, 80s, 86 informants born from – to 1892–1904 informants born (details) 1892, 1895, 1901?, 1904

4.2.5. No material from Dorset?

Although every attempt was made to get hold of material from Dorset, none was successful. The Dorset Record Office holds a lot of material that seemed very promising on paper, judging from catalogue card copies. The reality, however, was rather disappointing – all transcripts that were consulted were StE, and none of the tapes listened to were more than slightly dialectal. Despite serious efforts to remedy the lack of data from Dorset (all libraries and local museums were contacted in search of relevant material, without any result), no good source for comparatively modern or even traditional dialect material could be found. At that time, the disappointment was great. However, it is relativized to quite some extent when considering the equally small – in comparison with the other

Page 257: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 247

Southwest counties almost non-existent – amount of really dialectal data that the SED fieldworkers were able to gather (cf. the relevant paragraphs in Sections 8 and 9).

4.3. Newfoundland material

4.3.1. Material from the Memorial University of Newfoundland Folklore and Language Archive (MUNFLA)

MUNFLA is listed as a major resource for research in Michael Linn’s collection of archives (1993: 444), and a first visit to assess the available material confirmed this. The sheer amount of material at MUNFLA is amazing: according to a list prepared by an archival assistant in 2000, there are more than 2,000 transcribed interviews, collected from the early 1960s to the present day. The interviews stem from various sources, but most of them are the results of students’ and staff’s research for papers, theses, or other publications.

The Folklore Department encourages students to deposit their tapes and papers at MUNFLA, and as it is the policy of MUN that all students, regardless of their major, should take at least one class in Folklore, the variety of material is unsurpassed. Topics range from gardening to ghost stories or a recording and description of traditional songs and music – this short list should suffice to give an impression of the wide array of tackled topics. Students usually went to their home communities to do the interviews, which is probably the greatest advantage of the MUNFLA material: The interviewers were almost always insiders, often talking to relatives or at least acquaintances. The informants were thus not as inhibited as is often the case when an outsider, least of all a non-native, intrudes on them.

Another major advantage of MUNFLA is that many of the interviews held in the collection are transcribed25, and although some are standardized, the transcription policies at MUNFLA have long been such that dialect features are to be presented as accurately as possible. Moreover, the transcripts were all done by professional transcribers, though not trained linguists. Even if an earlier version of the transcript existed (generally done by the student who had originally submitted the paper), the professional re-transcribed it. Thus, the standard of transcription is very high, and in the present author’s experience does not leave much to wish for.

Page 258: A Comparative Grammar of Brtiish English Dialects

248 Susanne Wagner

Originally, material from 17 collections had been requested, but as not all interviews were equally suitable for the present project, some were disregarded in the final corpus of Newfoundland English. The present corpus consists of 31 interviews with 34 speakers, totalling about 130,000 words, as detailed in Table 8.

Table 8. MUNFLA material

number of interviews 31 number of speakers 34 number of words (total) 132,000 number of words per interview (average) 4,265 number of words per interview (maximum) 10,770 number of words per interview (minimum) 1,261 interviews recorded from – to late 1960s to early 1980s informants’ ages from – to 27–88 informants’ ages (details) 27, 31, 49, 60s (5x), 70s (5x), 80s

(6x)informants born from – to 1885–1942 informants born (details) 1880s (4x), 1890s (6x), 1900s

(5x), 1920, 1938, 1942

4.3.2. Folktales of Newfoundland

The collection of Märchen published in 1996 in the two-volume Folktalesof Newfoundland is another valuable source for studies on Newfoundland English. It differs from the rest of the material used in this study in one important aspect: the telling of a tale constitutes a different discourse level than an oral-history-type interview. It will be interesting to see how far the genre influences the choice of language used when investigating gender diffusion in detail. The two authors/editors, Herbert Halpert and John Widdowson, are also the primary collectors of the more than 150 tales included in the books. While Halpert’s background is in folklore, Widdowson is a linguist with a long-standing interest in dialects. The authors became interested in the oral tradition of Newfoundland, which ultimately resulted in the publication of Folktales. For practical purposes, they ultimately used tales that had been collected between 1964 and 1979 (cf. Halpert and Widdowson 1996: xii). Although neither Halpert (born in NYC) nor Widdowson (an Oxford graduate) were natives, they were able to record the tellers without much inhibiting influence.

Page 259: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 249

The reasons for people’s openness lies primarily in the relative isolation of most of the communities that were visited, where easy access to mass transportation was still a couple of decades away. The major advantage of the tales is clearly their presentation in orthographic transcription. As the authors point out, and as should be obvious to anyone who has ever dealt with transcribing non-standard speech, the final results can only be a compromise between all possible extremes. For some issues, the folklorist’s point of view was prone to clash with the linguist’s, and vice versa. But however difficult the actual production was, the published versions of the tales are a dream for anyone interested in morphology or syntax.

Despite these and other necessary compromises in our editing procedure, we have constantly borne in mind our intention to present a text as close to the original speech as is both possible and practicable, always erring on the side of accurate representation rather than on the kind of editing which, in both obvious and more subtle ways, changes the original text radically in its insistence on presenting a text more acceptable from the literary viewpoint. We have retained the original lexis, grammar, and syntax because these are essential to the dialect, particularly since they also reflect the regional character of the tales. (Halpert and Widdowson 1996: lix–lx; emphasis SW)

An actual stretch of a tale is reprinted below:

[Int. B: Who were the best storytellers around?]

Uh … well I don’t know who the best one … I don’t know who the BEST storyteller is. Well ol’ S[mith] … now Eli Smith he’s up there to Port Anson. I think he was about so good a feller – as ever I … heard (could) tell stories – you know.

[Int. A: He used to tell these about Jack as well did he?]

Oh yes. He used to tell ’em … (right) ’bout Jack as well.

[Int. A: An’ you’ve picked them up when you heard people in the woods and … elsewhere have you?]

Oh yes. When I … whenever … whenever I hear a story told see that I pick un up – I’d knowed un. I could tell un then … right on – after he was finished. (Tale No. 32, p. 342–343)

The original tapes are held at MUNFLA today, and the authors are considering making them accessible for researchers. Detailed background

Page 260: A Comparative Grammar of Brtiish English Dialects

250 Susanne Wagner

information is generally available on the tellers, making it easy to single out informants with a West Country family background. The storytellers are traditionally males over 60 who work(ed) in the fishery or lumber industry, and thus could be termed the Newfoundland equivalent to NORMs. The procedure for selecting tales did not differ much from that of assessing other oral history material. Statistical details for the tales corpus are provided in Table 9.

Table 9. Folktales material

number of interviews 55 number of speakers 14 number of words (total) 146,000 number of words per interview (average) 2,655 number of words per interview (maximum) 7,353 number of words per interview (minimum) 401 interviews recorded from – to 1964–1975 informants’ ages from – to 44–89 informants born from – to 1877–1926

4.4. Summary

Some of the issues that may seem problematic due to differences in type, selection, or general make-up of the data need to be addressed. First, it may be argued that the SED material (both the basic and notebook material) is not comparable to the oral-history-type material, as the SED was questionnaire-based while the oral history data are usually one-on-one interviews. It should be mentioned initially that the type of analysis pursued here does not depend on long stretches of discourse. Often enough, a single phrase (and sometimes even an individual form) exemplifies the type of language use under scrutiny here. In addition, the differences are by no means as great as is generally thought. The atmosphere of responding to the SED questionnaire could not have been that different from a more general interview. In both cases, the interviewer(s) sat face-to-face with the informant(s). The only difference is that the contributions of the interviewer(s) are pre-determined in the case of the SED, but free (conversation) in case of oral history interviews. Also, the primary interest here is in the notebooks’ contents, which are usually side remarks that have nothing to do with the actual questionnaire answering. In style, these

Page 261: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 251

remarks come very close to an oral-history-type setting, which justifies a comparison of these data with those extracted from oral history material.26

Second, it could be argued that the time frames of the SED recordings and the oral history material differ too much for them to be comparable. As has been mentioned above, fieldwork for the SED took place between 1950 and 1961, and speakers were generally in their sixties or seventies then, setting their birth dates in the 1880s to 1900s. As will be recalled from the respective tables listing the birth dates of the oral history informants, these are identical or at least very close to those of the SED informants. Thus, we are essentially dealing with the same generation of speakers, an issue that will become important in later sections. As for the different oral history data, there is no obvious reason why they should not be comparable. Informants are generally NORMs, female speakers are scarce. Table 10 gives an overview of the corpus material that will be used in the analyses in this study.

Table 10. Corpora used in this study

area / source number of interviews number of words Newfoundland Folktales 55 156,000 MUNFLA 31 132,000 Southwest Wiltshire 9 88,800 Somerset 31 174,500 Devon 5 51,100 Cornwall 4 18,900 total 135 621,300

5. Special referent classes

In the course of this study, it has been and will be mentioned time and again that certain nouns which may trigger gendered pronouns deserve a special status. In addition to those nouns, a specific use of (particularly feminine) pronouns merits a closer investigation. The two major categories to be identified in this respect are instances of personification and references to animals. The specific use is subsumed under the label non-referential she. The reasons for excluding those two categories and assigning special status to the third use are manifold and complex. This section will provide background information on all three and explain why

Page 262: A Comparative Grammar of Brtiish English Dialects

252 Susanne Wagner

they are not treated here at all (personification and animals) or analyzed separately (non-referential she).

5.1. Personification

By definition, personification is classified as a figure of speech which attributes human qualities to non-humans and things (animals, plants, elements of nature, and abstract ideas). Although superficially easily mistaken for instances of personification, almost none of the examples in the present study really are. The entry for personification in The New Fowler’s Modern English Usage (Burchfield and Fowler 1998³; available online) is interesting insofar as it links the loss of grammatical gender to the rise of personification, giving examples from the OED:

Personification arises partly as a natural or rhetorical phenomenon and partly as a result of the loss of grammatical gender at the end of the Anglo-Saxon period. In Old English a pronoun used in place of a masculine noun was invariably he, in place of a feminine noun heo ( = she), and in place of a neuter noun hit ( = it). When the system broke up and the old grammatical cases disappeared, the obvious result was the narrowing down of he to refer only to a male person or animal, she to a female person or animal, and it to nearly all remaining nouns. At the point of loss of grammatical gender, however, he began to be applied ‘illogically’ to some things personified as masculine (mountains, rivers, oak-trees, etc., as the Oxford English Dictionary has it), and she to some things personified as feminine (ships, boats, carriages, utensils, etc.). For example, the Oxford English Dictionary cites examples of he used of the world (14c.), the philosopher’s stone (14c.), a fire (15c.), an argument (15c.), the sun (16c.), etc.; and examples of she used of a ship (14c.), a door (14c.), a fire (16c.), a cannon (17c.), a kettle (19c.), and so on. At the present time such personification is comparatively rare, but examples can still be found: e.g. Great Britain is renowned for her stiff upper lip approach to adversity; I bought that yacht last year: she rides the water beautifully; (in Australia and NZ) she’s right;she’s jake; she’s a big country, etc. (The New Fowler’s Modern English Usage, “personification”; boldface SW)

One has to distinguish between personification in its own right and personification as sub-component of metaphor (as in the mouth of the river). It should be obvious that, while the second use occurs frequently in everyday speech, particularly in idioms and proverbs, the discussion here only concerns the first use – which is, as mentioned above, rare.27

Page 263: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 253

In the corpus data used here, clear examples of personification are generally restricted to the telling of myths and legends, where they are typical of the genre. Personification is an extremely unlikely explanation when a cider maker uses he to refer to an apple, when a watchmaker refers to one of his watches, or a house mover to one of his houses. These are examples of true dialect use, based on a system that has nothing to do with personification. This can also be deduced from the provenance of masculine pronouns in these domains, while personification has been associated with feminine forms (recall Crystal 1995: 209) and the Spoken Standard system(s) described in Section 6.

According to this and other sources (e.g. The Oxford Companion to the English Language, online at www.xrefer.com; Wales 2002: 331; McArthur 1992, “personification”), the representation of ships as female is also an example of personification, probably based on the imagery of a ship as a womb-like container. However, one should hesitate to classify all of the literally hundreds of examples of feminine (though not female) ships in the Newfoundland data of this study as instances of personification. A Newfoundland fisherman would simply never use it to refer to his ship, and personification is more than unlikely when no standard forms are used. Other arguments are more plausible and convincing than personification, and it does definitely not suffice as an explanation accounting for gendered pronouns in general. Summarizing various researchers’ work on gender marking, Wales (2002: 333) argues similarly:

‘Personification’ is obviously too general a label to cover what seem to be quite complex analogical or metaphorical hierarchies of salience according to such value(s) as occupation, local environment and climate and general relevance to human needs, as well as subtle forms of gender symbolism.

5.2. Animal referents

At first glance, the major (and only) reason for excluding nouns referring to animals from the discussion of gendered pronouns is fairly simple, though maybe not obvious: Although most grammars of modern and earlier stages of English claim that it is the appropriate pronoun to use when referring to an animal, except for cases where the sex of the animal is known (in which case the pronoun corresponding with the sex of the animal may be used alternatively), actual language use could not be further removed from this

Page 264: A Comparative Grammar of Brtiish English Dialects

254 Susanne Wagner

prescriptive statement. When looking at speakers’ behaviour, it seems that – even at a very superficial level – forms of he and she by far outnumber instances of it in everyday casual speech.

In the corpora used here, there is essentially not a single example of itreferring to an animal, while literally hundreds of masculine and feminine pronouns can be found. In the West Country corpora, for example, there are about 500 references to animals with masculine or feminine pronouns, but only a handful with neuter forms. Though surprising at first, a more detailed investigation of additional corpus data and a number of studies dealing with the issue reveals that the observed pattern is the rule rather than the exception.

MacKay and Konishi (1980) investigated the use of what they call “human” pronouns (i.e. he, him, she, her) to refer to non-human antecedents. Though outlined as a study dealing with personification, it soon turned out that personification played only a minor or no role at all in those cases where non-standard pronouns were chosen.

The authors based their study on a database of approximately 35,000 pronouns collected from an anthology of children’s literature (cf. MacKay and Konishi 1980: 151). They distinguished three large classes of antecedents, namely “animals (including real, imaginary, and toy animals), fantasy creatures (including imaginary beings such as fairies, ghosts, giants, and trolls), and things (including abstractions such as thought and time)”.

The first major result from these counts was highly unexpected in light of prescriptive grammarians’ eyes: Of the approximately 450 pronominal references to animals, more than 80% were masculine or feminine – he(including object him, possessive his and reflexive himself) occurred in 62% of cases, she (her, hers) in 20%, and it (its) in only 18% of the examples (cf. MacKay and Konishi 1980: 151). Next the authors classified the pronouns according to whether or not the antecedent was personified, assuming that personification would play a significant role in triggering non-neuter pronouns. Although this was found to be true in general, the figures for the non-personified instances are striking (compare the percentage columns in Table 11).

Within the class of animals, personification could be held accountable for the use of a “human” pronoun in only about half of the cases (234 of 452). In the non-personified cases, a “human” pronoun was recorded in more than two thirds (69%) of the examples – a figure that clearly shows how rarely it is really used to refer to animals, here in only 31% (68 examples) of cases. The figures for the other two classes, on the other hand,

Page 265: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 255

are clearly within the expected norm. All examples of fantasy creatures being referred to by he or she are instances of personification, and in only six cases did speakers use a gendered pronoun to refer to things.

Table 11. The use of he and she vs. it for non-personified antecedents (from MacKay and Konishi 1980: 152)

Pronoun used Total he and she itNature of antecedent N N % N %

Animals 218 150 69 68 31Fantasy creatures 0 0 0 0 0Things 26 6 23 20 77

Total 246 156 64 88 36

In another study, Marcoux (1973) investigated students’ use of personal pronouns in tag questions. Tested noun classes included countries, ships, animals and humans. Like MacKay and Konishi, Marcoux found surprisingly high figures of “human” pronouns referring to animals of unknown sex. Some of the sentences that were used in this study are reprinted below (2a–d), together with the pronominal forms that were recorded in the tags.

(2) a. My dog will eat anything. he 88, it 5, she 3, aberrant 12 b. That cat looks hungry. it 46, he 43, she 9, he/she 2, aberrant 8 c. This canary sings beautifully. it 69, he 23, she 7, he/she/it 1, aberrant 8 d. Tweety, my parakeet, is sick. she 42, he 40, it 14, he/she 2, aberrant 10

Two conclusions can be drawn from these observations: First, “[t]he presence of a proper noun seems to encourage the use of either a masculine or a feminine pronoun rather than the neuter form” (Marcoux 1973: 104). And second, the masculine pronoun is highly favoured over the feminine one.28

Table 12. Pronouns for antecedent “dog” in a sample from the BNC (spoken)

N % masculine form 162 56.6 feminine form 23 8.0 neuter form 101 35.3

total 286 99.9

Page 266: A Comparative Grammar of Brtiish English Dialects

256 Susanne Wagner

Table 13. Pronouns for antecedent “cat” in a sample from the BNC (spoken)

N % masculine form 88 52.7 feminine form 38 22.8 neuter form 41 24.6

total 167 100.1

A cursory analysis of personal pronouns referring to animal antecedents in the spoken part of the BNC reveals the same pattern29: In the contexts of a search for “dog” and “cat”, all pronouns referring to the keyword were marked. Results can be found in Table 12 and Table 13 (percentages are rounded to first decimal so that figures may not add up to 100%)

Although slightly different from Marcoux’s results, the overall emerging pattern is identical: Masculine pronouns are the unmarked choice when referring to a pet such as a cat or dog. While it can be assumed that most of the instances of feminine pronouns referring to dogs are used by speakers who know the sex of the dog in question, cats are more likely to be shes generically, based on the biological-semantic pattern (dog = neuter or + male, vs. bitch; cat = neuter or + female, vs. tom-cat).

Pronoun switches are typical, and a number of emotive factors play a role in the choice of pronouns when referring to animals. For example, the a cat owner is very likely to refer to a dog chasing their cat as it rather than he or she, signifying their emotional attitude or intimacy towards their cat, but at the same time signalling distance towards the dog (see also Mathiot and Roberts 1979 and their idea of up- and downgrading in section 6.1.1). The reverse pattern naturally holds for dog owners. Some representative examples from the BNC can be found in (3a–e).

All of the examples in (3) highlight certain aspects of pronominal use connected with animal referents. In (3a), a police officer (PS1SF) is being questioned about dogs on the force. He himself has never owned such a dog, which, in addition to the rather formal situation of the discussion, should explain his four uses of it. Once he gets emotionally involved though, talking about a dog becoming a member of the family of the leading officer, he switches to he in the two final references.

In (3b), a farmer (PS2VX) is talking about hunting foxes. Although reporting a rather general procedure (“One dog would go in ...”), the speaker obviously has one specific dog in mind, which explains his use of she in all instances.

Page 267: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 257

Speaker (PS555) in (3c) has an obvious antipathy towards small dogs, such as a friend’s Chihuahua. Both the negative feelings and the animal’s size are responsible for the choice of pronoun – it in all but one instance, where the speaker uses a masculine pronoun, most likely referring to the true sex of the dog in question.

In (3d), the owner of the cat (PS1D1) uses masculine pronouns exclusively, while her friend (PS1CX) only uses neuter pronouns, a typical pattern signalling familiarity or ownership. Similarly, the speaker in (3e) only uses feminine pronouns, most likely in accordance with the sex of his cat.

(3) a. ... Alright? Next question. Yes young man. [PS000]: What was it like when you had your police dog?

[PS1SF]: I have never had a police dog. I’ve never had, never been on er the special course30. A lot of people like it ... because basically th er when you look after a police dog it becomes your pet as well, you take it home with you and you take it to work with you, and the u you’ll have a police dog for sort of like its working life of seven to eight years, so basically you’re gonna have him for seven to eight years and he becomes a fa like a family pet. I’ve never been on the course (i.e. force; SW) so I’ve never had a police dog. FM7 (257)

b. [PS2VX]: Aye. Aye. And erm say the fox had been in the ground, and the [...] and the the young cubs, for about three or four days. And we used to hear somebody saying there was a vixen there and some and some young ones. [...] we went up there with the dogs and let them in in to the burrow. Block everywhere, let them into the burrow. One dogwould go in, and she’d just shake her tail and come back, and you couldn’t get her in afterwards because she knew that they’d cleared off. HER (217)

c. [PS555]: I couldn’t stop laughing. The little dog’s going [yelping sound] [panting]

[...] This little dog was mad, man, did you see it? It was so ugly I would’ve

[PS55A]: Yeah. [PS555]: kicked it if I saw it. Same as Chris’s chihuahua. I’d, I’d love

to kick it. I’d love to kick her dog. He’s so tiny! I feel so sorry for ityou know, up at that house with all them big fat balls of, of fat. They’ve probably stepped on it enough times.

Page 268: A Comparative Grammar of Brtiish English Dialects

258 Susanne Wagner

[PS55A]: [...] [PS555]: And have you heard it crying at night? [PS55A]: Mhm. KPG (4043)

d. [PS1D1]: Come on puss, shh, shh, shh [PS1CX]: Where’s it gone Rebecca? Where’s pussy cat? [PS1D1]: puss, puss, puss, puss [PS1CX]: [laughing] Where’s it gone [...]? [PS1D1]: is he there? [PS1CX]: Can you see him? ... Can you see him? [PS1D1]: Where’s the cat? [PS1CX]: Go on out, out cat [shooing away] [PS1D1]: [laugh] where’s he gone? KB9 (1084)

e. [PS0H9]: I know, woke up this morning she was, she was obviously cold, cat was right under the covers, snuggled right up to me and got her, her chin on me arm like that, I was asleep ... KCY (438)

Clearly, pets are more likely to be hes than shes or its when their sex is unknown, he thus (still) serving as generic pronoun despite the arguments of recent feminist linguistics theories. Generally, researchers agree that personal involvement seems to be the most relevant factor in pronoun choice.

[T]he use of he and she seems to signal personal involvement or empathy for the referent in the case of [...] an owner of an animal, someone who is emotionally attached or values the referent, [...] or someone attached to a specific animal. By the way of contrast, the use of itseems to signal lack of involvement or empathy with the referent in the case of [...] [a speaker] who is not personally attached to the referent or wishes to devalue it, an entity which is acted upon, and finally a nonspecific animal or class of animals with which personal involvement is out of the question. (MacKay and Konishi 1980: 155–156; boldface SW)

The cut-off point within the class of animals differs from speaker to speaker, depending on profession, environment, or similar factors. For someone who grew up in the city and has never lived in the countryside, it is highly probable that only pets, or even just dogs and cats, can be he orshe, whereas a badger or fox, never having been encountered in their natural habitat, will be an it. On the other hand, it is extremely likely that a

Page 269: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 259

farmer will refer to the animals on his farm as he or she, that a hunter will refer to the hunted animal, the fisherman to the fish in his catch as he.

It should thus be concluded that the prescriptive rules in grammars concerning anaphoric pronouns to be selected to refer to animals are not reflected in everyday conversations. As some degree of personal involvement is usually present when speakers talk about animals, neuter pronouns are the least expected forms. Pets will be its only derogatorily or to signal detachment, while the status of wild animals depends to a large extent on the speaker’s background. Other factors that may influence pronoun choice are saliency of the animal in the discourse (“centrality” in MacKay and Konishi 1980: 155), size (the bigger the more likely he;MacKay and Konishi 1980: 153), and various (supposed or real) character attributes (brave, wise = male; weak, passive = female etc.; cf. MacKay and Konishi 1980: 154 and also section 6.1.1 on Mathiot and Roberts 1979).

5.3. Non-referential she

This category includes many examples that are mentioned elsewhere in this study. When investigating the use of gendered pronouns, examples like those in (4) occur with a regularity that warrants a closer analysis.

(4) a. “Okay!” Julia yelled. “Get ready. Here she goes!” “Timber!”

The post toppled slowly and as it landed on the grass with a thump, they both cheered. Woman (Julia) and her daughter are removing the posts that held a rope rail; reference seems to be to the situation as a whole rather than an individual post; Nicholas Evans, 2002 (Corgi ed.), The Smoke Jumper, p. 401.

b. Watch out! Here she comes! (speaker is sea-sick; Svartengren 1928, ex. 139)

c. Here she comes! (Paddock 1991: 30, referring to an approaching weather front)

d. She’s blowing hard out there. It almost blew over the tree. (Mathiot and Roberts 1979: 38)31

e. Well HE done it we’ll say, th’ (other) one had to do it. He climbed up you know an’ just as he got on the top part away she goes helter to skelter all over the ki ... all over the place. (Folktales 141; reference to situation of a pile of chairs falling)

Page 270: A Comparative Grammar of Brtiish English Dialects

260 Susanne Wagner

f. I only come down to help pull en [house]. But nevertheless I said, all right boys, straighten out, and away dey goes, my son, straightened out, we took dat house and here she come. (MUNFLA 72-089: C1187; cataphoric reference to what happens)

g. Stay the night with us? – Ah, she’s right. “all is in order” (R. Stow, 1963, cited in Ramson 1988: 531)

h. She’s fine; she’s cool; she’ll be joe. synonyms of ‘It doesn’t matter.’ (Orsman 1997: 717)

i. Well ... it rolled in at my feet and he’d pulled t’ pin out! I got out o’ that hole faster than I went in, and up she went! (FRED Yks_003; explosion caused by a grenade “it”)

All of these examples have one thing in common: The referent of the personal pronoun is either difficult to identify or cannot be named at all. Very often, she seems to refer to the (abstract) situation, circumstances, or side effects of the utterance rather than to a concrete thing. Items of this type can be found in all varieties of English, pointing to the fact that this use is restricted neither regionally nor socially. For that reason, non-referential she does not fall within the framework of this study. Nevertheless, some of the characteristics of these forms will briefly be outlined, since descriptions of this use are difficult to come by.

One of the major characteristics shared by most of these constructions is word order: More often than not, extraposition results in an output of the form X-S-V instead of standard S-V-X. “X” is usually a spatial or demonstrative adverb, most often here or there. Alternatively, the preposition of a prepositional verb is extraposed, resulting in patterns such as up she V or down she V.

Judging from the relevant literature, this type of fronting seems to be rare in English. Birner and Ward (1998), who analyzed pre- and postposed non-canonical word order patterns, do not mention this construction type. An analysis in terms of theme/rheme or given/new information is difficult in most cases. The fronted element, although usually containing new information, is generally not the topic of the respective utterance. Matters are further complicated by the fact that expressions such as here/there/PPshe V seem to assume an almost idiomatic meaning, making it impossible to attribute any type of information status such as theme/rheme or topic/comment to the individual elements at all.

Moreover, the verb is always in the present tense, although the action described would generally demand a progressive form. It seems that the need for fronting/extraposition overrides any aspect requirements. The

Page 271: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 261

origins of this type of use remain in the dark – flattering (or unflattering) though it would be, it is definitely not enough to claim that the situations in question show some feminine characteristics, as folk belief has it. It is probably true that most speakers who use non-referential shes are not aware of it, but that the construction seems to have established itself in everyday English conversations, and be it only because There she goes sounds much better than There it goes. From the present state of affairs, one can only conclude that further investigation is needed.

6. Non-dialectal studies of gender assignment

6.1. American English

6.1.1. Sociological perspective (M. Mathiot)

In their 1979 article, Madeleine Mathiot and her assistant Marjorie Roberts investigated the use of “referential gender” in American English. They assumed that patterns of speakers’ pronoun use reveal certain sex roles as manifested in language. Their approach is a sociological or even psychological rather than a purely linguistic one, and attitudes and mental representations are used to explain language use. viewpoint

Data were collected for a period of 10 years, resulting in two subsets, one from the Los Angeles (years 1-3.5) and one from the Buffalo area (years 3.5-10). The examples stem from informal face-to-face conversations (cf. Mathiot and Roberts 1979: 5). The authors do not specify whether they elicited their examples with the help of some priming sentences, or whether they occurred in natural discourse.

Mathiot and Roberts distinguish the standard pattern of referential gender (“normative pattern”) from the “intimate pattern”, which allows the use of heor she for an inanimate entity32 or of it for a person. While the normative pattern predicts constant use of one pronominal form, “in the intimate pattern, the same entity may be referred to with either one of the three pronominal forms by the same speaker” (Mathiot and Roberts 1979: 7).

At a very early stage, even before the actual analysis, the authors offer a generalization which they think explains the differing uses of he, she and itin the intimate pattern: “The choice of ‘he’, ‘she’ or ‘it’ depends on the speaker’s general attitude towards the entity referred to or his feelings of the moment towards that entity.” (Mathiot and Roberts 1979: 7; emphasis SW). Although details of Mathiot’s and Roberts’ analysis are debatable

Page 272: A Comparative Grammar of Brtiish English Dialects

262 Susanne Wagner

(some minor differences as well as some major contrasts in their assumptions compared to the present author’s analysis will be pointed out below), it should be mentioned here already that the general tenor of the article is very similar to findings from this study.

6.1.1.1. The intimate pattern

As in the standard variety, the intimate pattern manifests two basic oppositions in pronominal gender: he and she vs. it on the one and within that opposition he vs. she on the other hand. According to Mathiot and Roberts, the first contrast can be attributed to (semantic) upgrading (it he,she) or downgrading (he, she it). While the present author does not share the authors’ sentiment that upgrading in general corresponds to personification (cf. section 5.1), their association of “positive involvement on the part of the speaker” seems a good means of tackling the issue (cf.Mathiot and Roberts 1979: 11).33 Similarly, negative involvement is said to underlie instances of downgrading, which also extends to those cases of previously upgraded items (i.e. return to the standard pattern).

Mathiot and Roberts evidently did not expect to encounter instances of the intimate pattern at such a high frequency. As a consequence, they may over-generalize to some extent when they say that “[i]t seems that any nonhuman entity can be referred to as either ‘he’ or ‘she’, i.e. upgraded, without regard to its nature” (Mathiot and Roberts 1979: 11; emphasis SW). While the contrast between it on the one hand and he and she on the other is relatively straightforward, much more variation occurs within the intimate pattern, i.e. between he and she. The authors differentiate between men’s and women’s usage, as they assume that certain patterns of thought manifest themselves in the intimate pattern. A detailed description of men’s and women’s mental images of both themselves and the other sex can be found in the respective tables of the article (Mathiot and Roberts 1979: 14 and 16). A brief summary of their findings will follow.

In general, Mathiot and Roberts adopt a rather archaic picture of men and women and the values and ideas that are part of their heritage. Each of the sexes occupies one of the opposing ends of a continuum in each of the relevant categories, such as beautiful–ugly, incompetent–competent, challenge/reward–brave, prized possession–good-natured, and mature–infantile. While the first are traditionally associated with women, the second apply to men, at least according to the popular cliché.

Page 273: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 263

Although many examples are provided to illustrate all of the categories that Mathiot and Roberts mention, their work is rather impressionistic, and it seems as if theirs is not a clear-cut system of pronominal use, but rather an interpretation of more or less incidental facts. Under different circumstances, most of the examples could be interpreted differently. Also, the article discussed here was written and researched in the 1970s, and both authors are women, a fact one should not neglect. Although not (yet) established then, the sentiments expressed here are reminiscent of feminist linguistics concepts. Reading between the lines, a certain dissatisfaction with the current state of affairs can be recognized in statements such as the following:

Men define themselves both independently of their relationship to women and in terms of it. Women define themselves only in terms of their relationship with men. [...] Men regard themselves as intellectually superior to women. Women regard themselves as emotionally superior to men. (Mathiot and Roberts 1979: 27–28; italics original)

6.1.1.2. Summarizing the results

One of the major drawbacks of Mathiot’s and Roberts study is its lack of quantitative or statistical information. It is unclear from the article how many instances of gendered pronouns (or of the intimate pattern) the authors actually found in their data. The examples given in the analysis itself add up to approximately 130, with masculine and feminine forms distributed fairly evenly. That figure itself should cause suspicions as to the overall figures, as it is highly unlikely that gendered pronouns are distributed evenly between male and female speakers. Taking together the relevant forms of the appendices (excluding animals), the following picture emerges: Men use she or her about 40 times to refer to an inanimate entity, while not a single use of a masculine pronoun is mentioned. Women, on the other hand, use masculine pronouns about 60 times to refer to something inanimate, but there are also approximately 10 examples of women using feminine pronouns in the same context.

The most interesting conclusion to be drawn from these data is the following: If taking the examples presented here as representative of male and female use respectively, a clear picture emerges. Pronominal use in the intimate pattern is primarily dependent on the sex of the speaker – where males prefer feminine pronouns, females will generally use masculine ones.

Page 274: A Comparative Grammar of Brtiish English Dialects

264 Susanne Wagner

Only very rarely will a woman use a feminine pronoun that is not an instance of personification, nor are men very likely to use masculine pronouns for any inanimate entity. The use of he representing women’s inherent image of men is – obviously – “entirely limited to women” (Mathiot and Roberts 1979: 41).

The second conclusion to be drawn from Mathiot’s study is a rather sobering one: Although there might be a certain pattern in the use of non-neuter pronouns, this use is by no means systematic.34 This must be concluded from the numerous examples where speakers switch pronouns without any observable pattern (cf. examples in 5 from Mathiot and Roberts 1979: 34):

(5) a. Do you realize how many times I have picked him up? He keepsslipping off the shelf. Next time this happens I’m going to leave it onthe floor. See how he likes it! [towel]

b. This one has been around long enough. I say, get rid of it! He is A season [out of fashion], get rid of it! [bedspread]

c. What the hell is the matter with this thing? It just won’t work for me! He usually isn’t like this! [typewriter]

Mathiot and Roberts explain all of the above shifts as instances of attaching negative attributes to things that are usually upgraded (cf. Mathiot and Roberts 1979: 33). However, the explanative value of such an assumption is rather low. In (5a), for example, the speaker is clearly annoyed – why then not use it in all slots? The switch back to he is rather unexpected.

The pattern Mathiot and Roberts observed for everyday language use in Los Angeles and Buffalo in the 1970s is by no means exceptional. Examples of this type can be found aplenty in modern (American) fiction and movies and have been overheard in conversations of Newfoundlanders.

From those examples it can be concluded that the pattern of pronominal use as described by Mathiot and Roberts is rather prototypical of non-standard spoken English in general. It will be shown in the following sections that their study arrives at results that are very similar to those of Svartengren (1930s, fiction; section 6.1.2), Morris (1990s, Canadian English; section 6.2). On the other hand, a stark contrast exists between the results from these studies and those from the analyses of the West Country and Newfoundland dialect corpora (cf. Sections 10 and 11).

Page 275: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 265

6.1.2. Vernacular perspective (H. Svartengren)

6.1.2.1. Introduction

In three essays very similar in content (Svartengren 1927, 1928, 1954), Hilding Svartengren investigated the use of feminine pronouns used for inanimate referents. His study differs from most other studies mentioned here in that it is based on fiction, i.e. “non-natural” language use, for the most part from US writers. Premature as it may seem, the first “conclusion” from Svartengren’s studies can be drawn here already: Although obviously puzzled by the “weird” use of feminine forms for inanimate entities, Svartengren does not mention a similar phenomenon for the masculine counterparts.

Two explanations for this come to mind: a highly unlikely one would be that the author considered the use of masculine forms nothing unusual, thus thinking it unnecessary to mention it. The second (much more likely) explanation is that strange masculine forms were encountered at a far lower frequency than feminine ones, thus not deserving any comment.

The non-existence (or at least extreme rarity) of masculine pronouns referring to inanimate entities in American fiction supports one of the major claims of this study very strongly: For the average native speaker of English, the gendered pronoun of choice is feminine, while it is masculine for a speaker with a West Country dialect background. This could already be seen in the previous section(s) and sections, and will be supported by the analyses presented in the following sections as well.

Of course, this rule is not without exception. One factor which is apparently strong enough to overrule the general system is the speaker’s sex (cf. Mathiot and Roberts 1979), resulting in females using masculine pronouns in some situations. In addition, it should be clear that in our modern times pure dialect systems do not really exist any longer. In this particular case, the mixture of lects (true dialect vs. spoken “Standard”) results in outwardly conflicting uses of pronouns, as the dialect “predicts” masculine forms where the spoken variety would rather use a feminine form. This seems to be one of the rare instances in which neither of the most likely choices in casual style corresponds to the written standard language – which might be the explanation behind scholars’ puzzlement over the phenomenon. In the following paragraphs, a brief overview of Svartengren’s studies will be presented, largely based on his 1927 and 1928 papers, which are basically identical in terms of content. Because of the completely different outline and aims of Svartengren’s work in comparison

Page 276: A Comparative Grammar of Brtiish English Dialects

266 Susanne Wagner

with the present study, evaluation and adaptation of Svartengren’s observations are possible only to a very limited extent. Emphasis will thus be put on those points which either add new information to or support the assumptions made here.

6.1.2.2. Svartengren’s database

According to his 1928 article (Svartengren 1928: 7–9), Svartengren based most of his analysis on 79 texts of contemporary, mostly American authors (written ca. 1900–1925), among them such well-known names as Jack London (with 10 texts) and Mark Twain (one text). In another 37 books “written by Americans or describing American life nothing, or nothing worth quoting, has been found” (Svartengren 1928: 8), giving the impression that this particular use of feminine pronouns may be ideolectal (i.e. restricted to some authors) rather than universal.

Another noteworthy feature (which is probably at the heart of Svartengren’s explanation for the use of these feminine forms) is that almost all of the specimens stem from males, occurring either in direct speech or some sort of internal dialogue, or simply because the author is a man. In addition, Svartengren remarks that “[m]any novels dealing with upper and middle class life have contributed very little to our collection” (Svartengren 1927: 113). For him, the phenomenon is obviously a) not geographically restricted and b) vernacular and rural at heart, but he is aware of the bias of his database in this respect:

Examples show clearly that it is a distinct colloquialism at home chiefly among men familiar with the stern realities of life and whose speech is uninfluenced by literature – this practically all over the United States and Canada. Most of the material [...] hails from the fur, the timber, the mining, and the cow countries, which may, or may not, represent the actual state of things, for, we must add, works describing life in the industrial centers have been drawn upon only to a limited extent. (Svartengren 1927: 113; emphasis SW)

The high number of disregarded books (37 out of 116; roughly 32%) also relativizes the overall frequency of the feature in the first place. As the author himself states, “[t]he two hundred and odd ex[amples] are drawn from some 175 books, chiefly novels” (Svartengren 1928: 14; Siemund 2001 counted a total of 268 examples, based on Svartengren’s 1927 article). The resulting average is one to two examples per book, which is,

Page 277: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 267

based on the present author’s reading experience of the past years by no means exceptional (cf. examples from modern popular fiction elsewhere).

6.1.2.3. Classes of nouns

Not very surprisingly after the prologue above, the noun classes Svartengren identifies as being capable of triggering feminine pronouns are largely based on research by earlier authors. Due to the diversity of referent nouns, very often real classes cannot be identified at all, but rather represent a accumulation of nouns that often share no more than one semantic feature. Svartengren himself is well aware of this (Svartengren 1927: 110): “It will be seen, then, that every attempt to confine to certain categories of nouns the instances when the feminine is to be used, must be abortive.” The major classes Svartengren lists are (cf. Svartengren 1927, 1928):

1. Concrete things made or worked upon by man (a) Machinery, industrial plants (b) Hollow things, receptables

i. Rooms, houses, and their uses ii. Musical instruments

(c) Other things made, created, worked, or worked upon by man i. Various small objects not tools ii. Large scale undertakings iii. Picture, film, newspaper iv. Clothing, wooden leg v. Food and drink vi. Coins, money, amount of money, amount generally vii. Organized bodies viii. Districts ix. Road, trail, distance x. Natural resources exploited by man

2. Actions, abstract ideas (a) Actions

i. Expressions containing an imperative ii. Other expressions denoting actions

(b) Abstract ideas i. Pronoun referring to substantive mentioned ii. No substantival propword

3. Nature and natural objects not worked upon by man (a) Nature (b) Celestial bodies (c) Geographical appellations

Page 278: A Comparative Grammar of Brtiish English Dialects

268 Susanne Wagner

(d) Material nouns (e) Seasons, periods (f) Fire, temperature, weather conditions, ice, snow (g) Human body and its parts

The first category is very reminiscent of what Elworthy and Barnes described as man-made objects, hardly a coincidence. It is the second category in particular that commands our interest. Svartengren is, to my knowledge, the only author among those investigating gendered pronouns who proposes this category, which has been labelled “non-referential she”earlier in this study (cf. 5.3). “Non-referential” here stands for real instances (i.e. pronouns without antecedent) as well as abstract nouns referring to situations, events, etc. Svartengren found many examples illustrating this type of use, some of which are reprinted in (6a–e) below.

(6) a. Let her go! Let her went (“I am ready”) (Svartengren 1928, ex. 130) b. Start her off (ref. to making pancakes) (Svartengren 1928, ex. 134) c. Watch out! Here she comes! (speaker is sea-sick)

(Svartengren 1928, ex. 139) d. There she goes! (undertaking) (Svartengren 1928, ex. 59) e. “How do you like it, Tim?” – “She’s alright.”

(Svartengren 1928, ex. 161)

One major sub-category of the “action” category is connected with the use of an imperative, again a use that is encountered with high frequency in everyday conversations (such as Fill ’er up! referring to refuelling a vehicle). Although not explicitly mentioned, another syntactic peculiarity of this category is its frequent use of fronting or extraposition (examples of the type There she goes!).

Example (6e) is reminiscent of the use of she’s apples, she’s right, she’ll be right, she’s sweet, etc. to refer to the general situation or circumstances, supposedly restricted to Australian/New Zealand English (cf. Ramson 1988: 14, 31, 577, 656; Orsman 1997: 717), maybe hinting at the origins of that particular construction.

The comparatively high overall frequency of this category is another argument supporting the claim that we are essentially dealing with two completely different, largely unrelated systems. The dialectal system(s) described in Sections 10 and 11 do not “allow” the use of a gendered pronoun referring to something as abstract as a situation or, even more extreme, a gendered pronoun without any antecedent. We saw earlier that

Page 279: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 269

for the traditional dialect systems the theories put forward by 19th-century scholars still hold, at least to a certain extent. Most people will agree that it is difficult to attribute a high degree of individuality or human traits (which would justify an interpretation as personification) to a situation. In fact, this raises a problem in Svartengren’s own study: He himself claims that the items referred to with a feminine pronoun “must be capable of assuming at least some degree of individuality” (Svartengren 1927: 110). No comment is made as to how this is achieved with his category 2. While this statement supports the theory that Siemund (2001) proposes (roughly: only highly individuated nouns can be animated), the data clearly contradict it, with more than 20% of abstract referents pointing out that Svartengren’s material is of a completely different nature, illustrating a different system, than the rest of the data that Siemund analyzed.

Not much needs to be said about the rest of Svartengren’s categories, which include numerous items capable of triggering feminine pronouns even in the standard language (e.g. nature, celestial bodies, cities, …). As has been pointed out before, they are not systematic in any way – a fact that the author himself acknowledges in saying that none of the restrictions (e.g. artificial vs. natural objects; big vs. small, etc.) proposed by previous researchers hold for his data (cf. Svartengren 1928: 41).

6.1.2.4. Origins and explanations

As he had earlier advocated the vernacular status of gendered pronouns, Svartengren turns to the influence of other vernacular varieties as one of the conceivable origins of the phenomenon after having dismissed possible influence by foreign languages. Conveniently ignoring the fact that the Southwest of England is he-territory and that large portions of settlers, particularly working-class people, came from that area, he notes parallels between the use he observed for America and the one documented for Northern and Celtic English(es) (Svartengren 1927: 108). From the short and rather cursory statements in various reference works (the EDD among them) Svartengren concludes that, while it may have its origins in Great Britain, the phenomenon now is American at heart “and is, no doubt, rather slowly invading British English as well, aided possibly by northern dialectal influence” (Svartengren 1927: 113).

Even though thinking of gendered pronouns as a feature found only in the lower (working) classes, Svartengren does not automatically dismiss it as wrong or a result of imperfect learning. Rather, he assumes that the

Page 280: A Comparative Grammar of Brtiish English Dialects

270 Susanne Wagner

“emotional character is the distinguishing feature of the phenomenon” (Svartengren 1928: 51) and subsumes it under the more general label of personification, an error easily justified by recalling the sex-bias of his database and the general prejudices of that time.35 The importance of emotions in connection with the use of gendered pronouns is undeniably one of the deciding factors (possibly even the deciding factor) in triggering the phenomenon in the first place.

6.2. The Canadian perspective (L. Morris)

In her doctoral thesis (Morris 1991), Lori Morris investigated gender in modern Canadian English, drawing on both spoken and written data. However, hers is not a corpus study, but is rather based on impressionistic evidence and observations. The clear advantage of the study lies in its breadth of focus. Morris considers all possible types of referents, from humans to animals to inanimates, also including personification and other relevant sub-categories or factors that may influence pronominal usage. To my knowledge, this is the only work of this kind to date.

6.2.1. Animal denotata

Morris’ criteria for assigning gender are very similar to those already identified as crucial in other studies and our own analyses: Animals playing a (particular) role in discourse will be referred to by he or she rather than it.Table 14 shows Morris’ three categories (cf. Morris 1991: 112–139).

In her data, animals are much more frequently he than she, a pattern that is expected based on knowledge from other varieties. For animals, “examples involving variation between she and it are much more difficult to find” (Morris 1991: 124). Figure 2 shows the hierarchical system of assigning gender to animals according to Morris (1991: 125).

Table 14. Gender assignment for animal denotata according to Morris (1991)

it background, non-individual; generally “accepted” behaviour of species

he foreground, specific; individual; behaviour different from expected norm/peculiar

she behaviour typical of species

Page 281: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 271

Figure 2. Gender assignment for animals in Morris (1991)

The categorizations in Table 14 are typical; two traits that occur again and again when investigating animal references are highlighted:

– The major division is between neuter and non-neuter, or animate and inanimate (it vs. he/she).

– The factor that is mostly responsible for a change in the assignment pattern is pragmatic rather than grammatical: an animal that is foregrounded as the topic of a conversation will very likely be animated.

In addition, Morris attributes the choice between she and he to the behaviour of the animal in question, while stating at the same time that feminine pronouns referring to animals are rare in her data. The one example that puzzles the author fits in with our explanation of “non-referential she” in an earlier section (cf. 5.3): There she blows!, uttered about a spouting whale, would usually receive a masculine anaphoric form (cf. Morris 1991: 135). However, the utterance (or rather the pronoun) could also be interpreted as non-referential. Many of the non-referential examples show fronting or preposing of certain elements, which is otherwise rare – Here she comes! or There she goes! are similar in this respect. Only rarely is it possible to identify a referent; more often, the speaker seems to refer to the situation in general.

clearly animate inanimate

he it

+ female - female (neutral)

heshe

inanimate

Page 282: A Comparative Grammar of Brtiish English Dialects

272 Susanne Wagner

6.2.2. Biologically inanimate denotata

Contrasting with animal pronominalization, Morris found that for inanimate entities, she is favoured over he (cf. Morris 1991: 139). In her opinion, “speaker familiarity” is responsible for many of the she-pronominalizations in her data (Morris 1991: 146). Very often, the feminine pronoun is part of a (short) imperative; if it were used, it would feel like a simple order; she,on the other hand, has an inviting, “attenuating effect” (Morris 1991: 159–160; e.g. Let ’er rip!). Such an “attenuating effect” can easily be assumed as an explanatory factor for the occurrences of non-referential feminine forms in general. In addition, the author contrasts feminine and neuter pronominal forms with the help of a criterion that we will encounter again in Pawley’s analysis of Tasmanian Vernacular English (cf. Pawley 2004; Morris 1991: 163):

she particular denotatum, particular impressions of a given denotatum it concept/norm of that type of denotatum

What plays a particular role in choosing personal pronouns is the prototypicality of the referent in question: While an “average” denotatum will generally be it, the speaker is bound to shift to a feminine form as soon as anything peculiar or noteworthy about the referent is to be emphasized.

Contrasting with the use Morris observed for animate denotata, and also almost diametrically opposed to the situation described for West Country and Newfoundland dialects, masculine pronouns are basically non-existent for inanimate referents: “While masculine reference to any type of inanimate denotatum is extremely rare, no examples at all were found in which a native English speaker used he to represent an intangible, difficult-to-identify type of denotatum” (Morris 1991: 164; emphasis SW).

Based on the few examples of masculine pronouns referring to inanimate entities that the author was able to collect36, she establishes the following contrasts between she on the one hand and he on the other (cf.Morris 1991: 168):

she familiarity, well-known; predictable, foreseeable he maintains features of the unknown; less familiar, unpredictable,

more individualistic

Taking together the criteria Morris establishes for using she, he and it, the speaker has to make a number of choices when referring to any kind of

Page 283: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 273

noun. An attempt at showing all relevant relations is made in Figure 3. The data stem from Morris (1991: 175–177).

According to Morris, the “primary function of pronoun gender” is “to represent and express the manner in which a speaker has formed his mental image of the denotatum” (Morris 1991: 175). Overall, pronoun choice is thus largely based on discourse-pragmatic factors, and in Morris’ system, generalizations or predictions are difficult to impossible to make, as it is predominantly the speaker’s worldview that is responsible for the choice of a pronominal form. Although some patterns influencing this choice are obvious and well-known, this is by no means as systematic a procedure as those observed for Newfoundland and West Country dialects.

Figure 3. Overall system of gender assignment in Morris (1991)

6.3. Summary

The preceding sections have hopefully clarified a number of issues. Although the varieties (and methodologies) investigated could not have been more different, the results of the studies presented here are very similar. Taken together, all of the studies mentioned in this section support

yes

no

+ animate - animate IT

general use, no individualdenotatum

+ feminine

comparing againstother members ofcategory/ individual denotatum selected

corresponds to general image?

HE SHE

Page 284: A Comparative Grammar of Brtiish English Dialects

274 Susanne Wagner

one of the major arguments of this study very strongly: she is the gendered pronoun of choice for non-dialectal English.

Mathiot and Roberts (1979) investigated a “social” variety of American English, contrasting women’s and men’s language use with regard to their use not only of gendered pronouns (what they called “upgrading”), but also of the reverse phenomenon, “downgrading” (e.g. the use of it referring to a human being). Their major result was that the sex of the speaker has considerable influence on pronoun choice, with men using feminine pronouns in many situations where women would prefer masculine ones.

Svartengren (1927, 1928, 1954) analyzed more than 100 novels and other texts, primarily by American and Canadian authors, investigating the use of feminine pronouns referring to inanimate entities. As his main result, he identified a strong connection between the phenomenon and vernacular speech, and also proposed that this was an essentially male feature, as almost all of his examples stemmed from men. Svartengren could not confirm any of the restrictions on noun classes allowing gendered pronouns which had been proposed in earlier research.

The most important result of Morris’ research on Canadian English (Morris 1991) for this study was the rareness of non-standard masculine pronouns referring to inanimates in her database. While feminine forms were relatively frequent and have been explained as representing the mental image that the speaker has of the referent, masculine ones were almost non-existent.

In his analysis of Tasmanian Vernacular English, Pawley (2002, 2004) obtained very similar results: The Tasmanian (Vernacular English) universe is largely feminine. Based on oral-history-type interviews, Pawley’s data stem largely from men.

The conclusions to be drawn from these analyses are obvious: In everyday, casual spoken English, possibly world-wide, the pronoun of choice when referring to an inanimate noun and wishing to add extra information is (and, as Svartengren’s studies indicate, has been for some time) a form of she. Mostly, this “extra” has been identified as some sort of emotional information, either positive or negative. In contrast, the pronoun signifying non-involvement or simply disinterest is it.

The sex of the speaker may influence the pronominal form in so far that women are more likely to use masculine forms in a number of contexts where male speakers will use feminine ones, particularly in domains associated with gender or gender-biased behaviour (e.g. cars, tools, etc.). Although concrete nouns receive gendered reference more often than

Page 285: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 275

abstract ones, there seem to be no restrictions, semantic or otherwise, on the type of noun that can take a feminine form in anaphoric references.

An interesting category within the abstract domain is the use of she andher referring to a referent which is hard to identify, a situation or general circumstances, which is shared by male and female speakers alike. Although often identified as a peculiarity of Australian and New Zealand English, examples are reported from basically all major varieties of English.

The system outlined here stands in sharp contrast to the traditional dialect system(s) of Southwest England and Newfoundland, where he andthe corresponding object form(s) occur in a large number of slots that are occupied by she in the “spoken Standard” described in this section.

7. Persistence of gendered pronouns

7.1. Introduction and previous studies

The phenomenon under discussion here is known as “persistence” or “priming”, including various sub-forms. Very generally, theories behind these phenomena postulate that similar or identical forms cluster in speech. Clustering effects are a well-known feature and have been studied in variationists’ accounts for a long time.

Priming effects have been observed in a number of syntactic areas, with particular emphasis on subject-verb-agreement (cf. Poplack 1979, 1980; Poplack and Tagliamonte 1993, 1996; Scherre and Naro 1991). All of these studies found that in those dialects which allow a zero marker on nouns and/or verbs where StE would require a non-zero marker (usually third person or plural -s), sequences including only marked or only unmarked forms are much more likely to occur than a mixture of marked and unmarked forms:

[…] if a plural is going to be realized, the tendency will be for it to be realized on the first element; if it is not, subsequent developments will not tend to rectify this in a functional way. What follows might either be all markers or all zeroes, so that a case like øøs turns out to be virtually non-existent. (Poplack 1980: 64–65)

These results show quite clearly that a parallel marking process is occurring. Although isolated or first occurrences of verbs reflect no special

Page 286: A Comparative Grammar of Brtiish English Dialects

276 Susanne Wagner

influences, verbs that follow other verbs tend to mimic the marking of the previous occurrence. (Scherre and Naro 1991: 25)

[W]e observe a concord effect, whereby lack of marking on a preceding reference verb leads to a greater probability of zero marking on the current verb (at .68 for Samaná English and .66 for the Ex-Slave Recordings), while overt marking leads to more marking. (Poplack and Tagliamonte 1993: 190)

Other studies focus on the reasons for clustering, which are found in the way humans are assumed to process language. It is in this area of research, often situated within the field of psycholinguistics, that the idea of “priming” originated. In speaking, the use of one form activates that form, and a recently activated form will be easier to re-activate in subsequent discourse than an alternative (new) form. Also, less effort will have to be made to re-activate a form that has been used more recently than to re-activate one used much earlier in the conversation.

Potter and Lombardi (1998) used an immediate recall task to investigate double object constructions in English, assuming that the structure of the primes would influence the output of the recalled target, which was indeed the case (cf. Potter and Lombardi 1998: 270–271). Tree and Meijer (1999) obtained parallel results in a similar study, again focusing on double object constructions. In addition to an immediate recall task, they also compared structures differing in the complexity of the phrases investigated:

In two experiments we observed that sentences with the same major constituent structure shared syntactic routines. These syntactic routines are likely to be stored in a hierarchical structure and to be activated hierarchically. That is, major constituents are activated first after which subroutines are called to build the structures with these constituents [...] [S]yntactic priming occur[red] across conditions varying in complexity [and was] equally frequent across conditions. (Tree and Meijer 1999: 89–90)

Types of priming include lexical, morphological, form and syntactic priming (for definitions of these, see e.g. Szmrecsanyi in progress). The boundaries between the different types are not always clear-cut; for reasons detailed below, the phenomenon will simply be called “priming” here.

Page 287: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 277

7.2. Primed gendered pronouns?

7.2.1. Introductory background

The studies dealing with priming so far have usually analyzed syntactic phenomena that fell within clearly definable linguistic categories: Poplack (1979, 1980) investigated the use of plural -s in Puerto Rican Spanish (Vernacular), Scherre and Naro (1991) noticed a priming effect in the agreement behaviour of verbal strings in Brazilian Portuguese (Vernacular), and Poplack and Tagliamonte (1993, 1996) analyzed priming in past tense marking in varieties of AAVE and Nigerian Pidgin English.

All of these studies focus on variety-internal variation, usually those cases where both a standard and a non-standard marker can be used. Typically, no conflict arises between standard and non-standard markers, which are often realized as zero: The non-standard marker is not identical with another standard marker of the same paradigm, nor is the standard marker identical with another non-standard marker of the same paradigm. For gendered pronouns, the situation is slightly more complex.

In truly traditional dialect, be it the West Country or Newfoundland variety, variation between it (in all its forms) and he (in all its forms) referring to count nouns should not exist at all. In those varieties, only a masculine form can be used to refer to count and/or concrete nouns, while it is restricted to mass and/or abstract nouns. Standard Written English would use it for both of these noun classes, while masculine forms are restricted to animate entities.

In more “modern” dialects, such as those that form the basis of the present study, the traditional dialect paradigm and the standard paradigm overlap and interact. Thus, it is possible to use it with reference to a count noun, but he can also be used (cf. Figure 4).

Page 288: A Comparative Grammar of Brtiish English Dialects

278 Susanne Wagner

Standard English ____________ _ _ _ _ _

modern dialects ______________________ _ _ _ _ _ _ _ _ _

traditional dialects ________________________________________

humans animals count nouns mass nouns

Figure 4. Extension of “he” across noun classes in 3 varieties of English (simplified)

Priming can affect the choice between a neuter or a masculine form for these referents. What is problematic about this choice is that it may be switching registers or lects (namely dialect vs. standard) rather than an intra-lectal choice. Although none of the earlier studies involved such a switch, it seems appropriate nevertheless to use the concept of priming in this context. It has to be assumed that speakers are at least bi-lectal nowadays, i.e. have the capability to use the standard (or at least near-standard) variety of the language and at least one additional (regional, social, ethnic, …) variety. The speakers concerned here will use a regional lect in addition to a more standard one. The regional variety can use masculine pronouns for count nouns, while the standard variety cannot. Depending on the situation, speakers will switch between the codes available to them.

7.2.2. Problems and pitfalls

A number of issues have to be considered when analyzing possible priming effects in a corpus of spoken language:

– Each interview has to be looked at in isolation, as priming effects obviously cannot occur across different interviews.

– In oral-history-type interviews, the possible influence of the interviewer (if not a dialect speaker) must be checked and accounted for.

Page 289: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 279

– In interviews with more than one dialect speaker (dialogues, discussions between speakers), priming between speakers is possible and can be analyzed accordingly.

The current investigation faces one problem in particular: standard forms (i.e. its) are necessary for an analysis to be possible at all. Thus, only those interviews (or better: speakers) can be used that show a comparatively high influence of the standard variety, i.e. those where it rather than a masculine pronoun is used at least occasionally to refer to count nouns. In addition, the interviews have to contain a considerable number of tokens in order for statistical analyses to be possible at all. For example, an interview with 10 instances of masculine vs. two instances of neuter pronouns is of no use.

Another pitfall mentioned in connection with other studies is not applicable here, but will be mentioned for the sake of the argument. Certain constructions or forms do not occur in regular intervals throughout discourse. Tenses may be given as an example: Although in a narrative most stretches will be told in the simple past, the general narrative tense employed in English, there may also be passages that are told in past perfect or, for dramatic effect, simple present (historic present). When analyzing the clustering effect of simple past forms, there will thus be stretches of speech without a single simple past form – simply based on the fact that the narrative situation is different in that passage. Similarly, when analyzing future marking and possibly competing future markers (e.g. begoing to vs. will/shall), only those passages can be analyzed where future ismarked. Consequently, priming effects can only be studied in those stretches where the forms under scrutiny occur.

However, as it is basically impossible to carry on a conversation without using personal pronouns referring to both mass and count nouns, this issue is irrelevant for the present study. Although there will be certain passages where only count or only mass referents occur, the overall distribution of pronominal forms is fairly even within the texts that will be investigated below. Keeping these restrictions and problems in mind, we should be able to study possible priming effects on pronoun choice.

7.3. Priming in the corpus

In order to be able to investigate priming effects, a number of modifications needed to be made in the corpus material. First and most importantly, all occurrences of it had to be disambiguated according to their referents or

Page 290: A Comparative Grammar of Brtiish English Dialects

280 Susanne Wagner

rather the type of noun they referred to, i.e. mass or count nouns. Moreover, the three issues listed above had to be considered.

Each interview will be looked at individually, taking care of the first issue. Secondly, the interviewer influence is minimal to non-existent in the interviews selected. For the chosen Newfoundland material, the interviewers themselves are natives, so that only intra-speaker code-switching occurs. The Somerset speakers who were selected are of the very talkative kind, so that the interviewer only rarely intervenes to ask further questions, generally just making “supporting noises” to encourage the informant to continue. As only one-on-one conversations will be analyzed, the third issue is irrelevant.

Prerequisite for the selection of texts was a high overall frequency of gendered pronouns. After the disambiguation of all occurring its, the modified text was used as an input for the TACT text retrieval programme. Counting switches (i.e. sequence of either neuter–masculine or masculine–neuter forms) and non-switches (i.e. sequence of either neuter–neuter or masculine–masculine forms) produced a two-by-two table of the following kind:

it hepreceding it A B preceding he C D

If no priming effect were observed, as many switches (cells B and C) as non-switches (cells A and D) would be expected, the null hypothesis being that each scenario is equally likely (i.e. 25% probability for each cell).

7.3.1. Priming in the Somerset data

Five texts from the Somerset sub-corpus were chosen for an analysis of possible priming effects. This comparatively low output of interesting texts is based on the rather low overall frequency of gendered pronouns in the Somerset texts, making statistical analyses difficult or even impossible in many cases. The results of the analyses are presented in Tables 15 to 18.

The tables should be interpreted in the following manner: The likelihood of switches is much lower (between 14% and 30%) than the likelihood of non-switches. The bold figures show the likelihood of a masculine form following another masculine form, ranging between 64% and 82%. Although the chi-square values for the four tables above are not always as

Page 291: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 281

significant as hoped for, a tendency for priming or clustering of identical forms can be observed. In fact, it is highly unlikely that a speaker would switch pronouns in any given passage without outside influence. In comparison, Klemola (1996: 237ff) tested priming effects of unstressed periphrastic do in his thesis. His likelihood values for identical forms to occur subsequently only range between 10.5% and 46.1%, averaging 30%.

Table 15. Priming in Som_001

text code neuter form masculine form Som_001 n ratios n ratios preceding neuter form 32 0.86 5 0.14preceding masculine form 4 0.27 11 0.73

significant at 0.002%

Table 16. Priming in Som_009

text code neuter form masculine form Som_009 n ratios n ratiospreceding neuter form 22 0.71 9 0.29preceding masculine form 8 0.36 14 0.64

significant at 1.2%

Table 17. Priming in Som_021

text code neuter form masculine form Som_021 n ratios n ratiospreceding neuter form 31 0.70 13 0.30preceding masculine form 13 0.30 30 0.70

significant at 0.02%

Table 18. Priming in Som_022

text code neuter form masculine form Som_022 n ratios n ratiospreceding neuter form 6 0.75 2 0.25preceding masculine form 2 0.18 9 0.82

significant at 1.3%

7.3.2. Priming in the Newfoundland data

As the interviews from MUNLFA are more traditional in the sense that they contain more gendered pronouns, more texts could be investigated than in the Somerset sub-corpus. Also, the overall results should be more

Page 292: A Comparative Grammar of Brtiish English Dialects

282 Susanne Wagner

relevant because the speech of most of the informants is highly conservative. The results for texts that were investigated for priming effects can be found in Tables 19 to 23.

Table 19. Priming in C626

text code neuter form masculine form C626 n ratios n ratiospreceding neuter form 29 0.67 14 0.33preceding masculine form 13 0.34 25 0.66

significant at 0.28%

Table 20. Priming in C627

text code neuter form masculine form C627 n ratios n ratiospreceding neuter form 30 0.73 11 0.27preceding masculine form 12 0.39 19 0.61

significant at 0.33%

Table 21. Priming in C628

text code neuter form masculine form C628 n ratios n ratiospreceding neuter form 25 0.625 15 0.375preceding masculine form 16 0.29 40 0.71

significant at 0.09%

Table 22. Priming in C631

text code neuter form masculine form C631 n ratios n ratiospreceding neuter form 38 0.73 14 0.27preceding masculine form 14 0.32 30 0.69

significant at 0.005%

Table 23. Priming in C2914

text code neuter form masculine form C2914 n ratios n ratiospreceding neuter form 21 0.7 9 0.3preceding masculine form 9 0.14 57 0.86

significant at 0.001%

The generally higher frequency of gendered pronouns contributes to overall more significant statistics. The five texts tested here once again show high likelihoods of sequences of identical forms (61% to 86%). In fact, the

Page 293: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 283

overall distributions indicate that Newfoundland speakers are even more susceptible to priming than their Somerset counterparts.

7.3.3. Generalizations and summary

The scattergrams in Figure 5 and Figure 6 clearly show that priming is a factor in the tested interviews, thus nicely summarizing the results observed in the tables for the individual interviews (cf. Sankoff and Laberge 1978] for the procedure/methodology and background assumptions).37 Figure 5 plotsthe proportions of masculine to neuter switches against the overall proportions of neuter forms. Figure 6 plots the reverse option, i.e. the proportions of neuter to masculine switches against the overall proportions of masculine forms. If the null hypothesis held, as many switches as non-switches would be expected, i.e. all the dots should fall on the diagonal. The real distribution, however, shows all dots below the diagonal, which indicates that identical forms in sequence are highly unlikely to be switched.

Figure 5. Scattergram 1: MUNFLA and Somerset texts

Page 294: A Comparative Grammar of Brtiish English Dialects

284 Susanne Wagner

Figure 6. Scattergram 2: MUNFLA and Somerset texts

Note that even those interviews which were non-significant employing the chi-square test conform with the general tendency. The 13 interviews included in the scattergrams result in an overall picture that is fairly clear and rejects the null (non-priming) hypothesis. It can thus be concluded that priming is indeed a factor in pronoun choice. Speakers’ behaviour as observed for selected texts from both the Southwest of England as well as Newfoundland material confirms tendencies that have been reported for other grammatical categories: If confronted with a choice of two competing forms expressing the same category, speakers will generally prefer the already activated form over the alternative one, resulting in sequences of same forms rather than a mixture of forms A and B without a pattern. Although without further research and more detailed studies of priming it would certainly be premature to make any far-reaching predictions concerning pronominal choice, a tendency for “conservatism” (i.e. clusters of identical forms) has been observed for the investigated texts.

Page 295: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 285

8. SED – Basic Material

As mentioned in section 4.1, the Basic Material of the SED holds a lot of material that, although it may not be immediately obvious, contains information relevant for the present study. Based on the investigations of the fieldworker notebooks, it was possible to isolate certain SED questions that could evoke answers containing gendered pronouns. Those questions were checked in the Basic Material publications, not only for the Southwest but for all locations. The following paragraphs present the results of that investigation.

Some of the problems connected with such an endeavour were already mentioned in section 4.1. First of all, the policy for recording or not recording a pronoun is unclear. For certain questions, the pronoun was part of the expected response, and was thus written down by the fieldworkers. In those cases it is to be expected that the pronoun will also appear in the published Basic Material. For other questions, however, where a pronoun was not an essential part of the response, policies of reprinting pronouns in the Basic Material vary widely. Sometimes pronouns are included, sometimes they are not. Some of the entries in the Basic Material note that “where recorded”, the personal pronouns are included in the response.

However, it becomes clear that not all fieldworkers actually took notes of such uses, as additional pronouns are only rarely recorded for the North or East, while they are ubiquitous in the Southwest. The only alternative, but highly unlikely, explanation which would account for such a difference is that informants in the North simply use(d) fewer pronouns than those in the South. As a result of these fieldworker “preferences”, the Southwest shows by far the highest rate of “pronoun retention”, as will be seen from the data presented below. The codes that will be used in the following sections follow general SED conventions. The questions that were considered relevant can be found in Table 24; the formulations are taken from Orton (1962).

The different levels of pronoun inclusion are obvious in the third column (“expected response”). The part that was actually sought is in boldface, and while a(n additional) pronoun was obviously expected in some questions (e.g. IX.2.6), it was not in others (e.g. I.11.2). The following sections will discuss the responses to these questions as reprinted in the respective sections of the Basic Material, following its outline in the order of discussion. Questions will be identified by their number in the questionnaire, which can be found in the first column of Table 24. Not all of the 10 questions will be discussed for all regions, depending on the relevance of the recorded forms. If only standard

Page 296: A Comparative Grammar of Brtiish English Dialects

286 Susanne Wagner

responses were given or no pronouns were recorded at all, the respective question will not be discussed.

Table 24. SED questions likely to contain instances of gendered pronouns

book question expectedresponse

I.7.1 If you want to know how heavy a thing is, what do you do?

To weigh it

I.11.2 What do you use to prevent your cart going backwards when you stop on a hill?

Prop/chock

I.11.6 How do you empty a cart the quickest way? To tipVIII.7.6 A dog buries a bone because he wants to … hide itIX.2.6 And now [stand sideways in front of it = door] … in front of itIX.2.8 If the door blew open on a cold day, you’d get up at

once and … shut it

IX.3.1 If there’s a hole in the pocket where you keep your knife, you’re almost certain to …

lose it

IX.4.4 You can’t have my spade today because I want it, but you can have it tomorrow, because then (I) …

(I) shan’t want it

IX.8.2 Jack wants to have Tommy’s ball and says to him, not: Keep it!, but [gesticulate]: …

Give it me!

IX.9.3 You have something to give away and before deciding on the person to be given it, you might ask yourself: I wonder …

to whom I shall give it?

8.1. The Northern Counties

Of the 10 questions listed in Table 25, the Northern Counties show exceptional pronominal usage in only one question. Responses to question I.7.1, IX.8.2 and IX.9.3 are it in all 75 localities; no pronouns were recorded for question I.11.2, IX.3.1, and IX.4.4. For questions VIII.7.6, IX.2.6 and IX.2.8, the recorded response either contained a form of it (it, ’t)or no pronoun at all. Thus, the only question of interest is I.11.6.

At 40 of the total 75 locations, standard it was recorded; the seven exceptional uses are listed in Table 25; for the other locations, no pronouns were recorded. Although all non-standard variants appear in Yorkshire, no pattern can be detected. Locations vary from the far North (loc. 3, Skelton, is on the northern coast) to the West (loc. 14, Grassington), East (loc. 25,

Page 297: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 287

Newbald) and extreme South (loc. 32, Ecclesfield, and loc. 33, Tickhill). Also, although the preferred non-standard form is [ r], [ ] is recorded at loc. 32, thus making it impossible to say anything about the distribution of masculine or feminine forms. It is highly unlikely that that both pronouns can be attributed to a semantic contrast between the referent nouns (cart vs.load). While an explanation for those seven exceptional forms is thus difficult (at least for the time being), it should be noted that the preference for the feminine form is to be expected: Traditionally, feminine pronouns can be used to refer to vehicles, even in StE, at least metaphorically (cf. Section 3). Also, as will be recalled from earlier sections, feminine forms were identified as the general unmarked choice of gendered pronouns in spoken standard varieties, excluding Southwest England.

Table 25. Non-standard pronoun forms, question I.11.6, Northern Counties

county locality response Y 3 ,Y 4 Y 14 Y 25 Y 32 shoot her (=the load) up]Y 33

8.2. The East Midland Counties and East Anglia

Based on our knowledge about the distribution of gendered pronouns, we do not expect any real surprises (i.e. non-standard forms) from this area, which includes the region whose dialect has become Standard English. Interesting responses are given in Table 26.

Table 26. Non-standard pronoun forms, East Midlands and East Anglia

question county locality response I.11.6 Nt 1 L 9 (“tipe her”) Lei 6 also: (“tipe her up”) IX.2.6 Nth 2 (“in front on er”) IX.2.8 Nt 1 (“shut er”) Nth 2 (“shut er”)

Page 298: A Comparative Grammar of Brtiish English Dialects

288 Susanne Wagner

Loc. 1 in Nottinghamshire is geographically close to loc. 32 and 33 in Yorkshire, where almost identical pronominal forms ([ r]) were recorded (cf. responses to question I.11.6 above). Not much can be said about the other exceptional forms. It should be noted, however, that non-standard gendered pronouns seem to be completely absent from East Anglia.

8.3. The West Midland Counties

As most of the counties included in this region border on the traditional Southwest (Herefordshire, Worcestershire, Gloucestershire, Oxfordshire), a higher number of non-standard pronoun tokens is expected. Also, the frequency of masculine forms should be higher than that of feminine forms (which should approach zero). For better legibility, the questions will be presented in groups and individual distributions will be commented on; the detailed tables this discussion is based on can be found in the respective volumes of the SED Basic Material.

As was expected, the West Midlands present a picture that is radically different from either the Northern or the Eastern region. For question I.7.1, non-standard forms have reached a total of 10% (63 responses it, seven responses some form of he; no pronoun in four locations), all of them masculine. Although non-standard responses to question I.11.2 are too rare to include them generally, the two examples from Worcestershire and Gloucestershire stand for the general trend (increase in masculine forms) and are thus listed here.

Even more dramatic are the figures for question I.11.6. For the total 74 localities, no pronouns were recorded in 33, leaving a total of 41. In 23 of these, the pronoun is standard it, while we find 18 instances of masculine and two cases of feminine forms (i.e. 46.5% non-standard forms; occasionally, the total of standard and non-standard forms is higher than the grand total, as at some locations more than one response was recorded).Both feminine forms occur in comparatively northern locations (South Cheshire and Mid-Staffordshire). The distributions of the other forms show a very nice wave pattern depending on how far removed from the core Southwest they were recorded. The lone standard it in Worcestershire occurs in the northernmost location in the county, with the greatest possible distance to the Southwest. The same holds true for the only recorded pronoun in Warwickshire (also it).

Page 299: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 289

The two forms in Oxfordshire look like a textbook case: standard it wasrecorded in Islip on the eastern border of the county, while non-standard [ ] was chosen in Eynsham, close to the western border. In both traditional and modern maps of major dialect areas, the isogloss separating the Southwest from other dialect areas usually cuts vertically through the middle of Oxfordshire. Eynsham would thus be considered part of the Southwest, while Islip rather belongs to the Southeast. With five instances of masculine forms and no occurrence of it, Gloucestershire is, at least in the context of this question (referent = cart), truly Southwestern territory.

A complete picture of this wave structure will be presented below, also including the Southern Counties. The overall frequency of masculine forms for this question is 0% in Warwickshire (one it), 50% in Oxfordshire (one of two forms total), 66% in Worcestershire (4 of 6), 75% in Monmouthshire (3 of 4), 83% in Herefordshire (5 of 6), and 100% in Gloucestershire (5 of 5).

Although not as pronounced as for question I.11.6, the general tendency and patterning of forms is the same for question VIII.7.6: The only feminine form is found in the North, while standard it occurs frequently in border zones, be it to other dialect areas or the coast. Masculine forms, on the other hand, are the rule for all localities closest to core Southwest territory. Percentages of masculine forms range from 17% in Worcestershire (1 of 6) and 50% in Gloucestershire (3 of 6) to 66% in Herefordshire (4 of 6).

The two questions referring to door once again show the by now familiar pattern: The closer one moves towards the core Southwest, the more likely it is to encounter masculine forms. In both questions, no masculine forms were recorded in Warwickshire or Oxfordshire (nor Monmouthshire for IX.2.6). The ranking of the other counties goes from Monmouthshire via Worcestershire and Herefordshire (which change positions in the two questions) to Gloucestershire, which shows the highest ratios of masculine forms in both questions.38 Overall, it is recorded in 51 localities for question IX.2.6, while a masculine form appears in five localities (8.9%). Much higher ratios can be found for question IX.2.8, where it occurs in 30, a masculine form in 14 localities (31.8%). Even the comparatively high 1 : 1 ratio in Shropshire fits the general pattern: the masculine form is located in the South of the county, while standard itappears in the North.

The referent of question IX.3.1 is knife, which shows even higher overall masculine ratios than the previous questions. No pronouns were

Page 300: A Comparative Grammar of Brtiish English Dialects

290 Susanne Wagner

recorded in 42 out of 76 localities, attributing much more weight to the 10 masculine forms (one feminine form is found in Derbyshire) in relation to the 23 neuter forms, resulting in a 32% likelihood to encounter a non-standard form (11 of 34 forms).39 In question IX.8.2, which primarily investigates the order of constituents in a double object construction, the itrefers to a ball. Non-standard pronouns were only recorded at three localities in Gloucestershire, with speakers at all other localities using standard it. This low frequency of non-standard forms is striking in comparison with the figures for the same question recorded in the Southern counties, as will be seen below.

8.4. The Southern Counties

The Basic Material of the SED does not distinguish between Southwest and Southeast, which seems surprising considering the stark contrasts between those two dialect areas. On the other hand, this offers a nice opportunity to analyze transition zones between the Southwest and the Southeast with little effort. As in the previous section, the data will be presented in groups of questions, followed by brief discussions of the results.

Even when the Southeastern counties Sussex, Kent and Surrey are included, masculine forms outnumber neuter forms for question I.7.1 (38 of 73; 52%; for details for this section, see the respective tables of the Basic Material). Without those three counties, where only standard it wasrecorded, the percentage of masculine forms climbs to 69% (38 of 55 forms).40 With such high ratios of masculine forms even for a comparatively undefined referent (something), it comes as no surprise that a more concrete referent (cart in question I.11.6) should only increase the figures for masculine pronouns. Out of 63 pronominal forms, 47 (74.6%) are masculine. Excluding the three Southeastern counties, this figure climbs to 86.8% (46 of 53 forms).41 Similar results are obtained from question VIII.7.6 (ref.: bone), with 46.7% (28 of 60) masculine forms including the Southeast, 59.6% (28 of 47) without it.42

The door, referent in question IX.2.6 and IX.2.8, is [ ] or [ ] with some of the highest frequencies of all items. Overall ratios are between 46.9% (IX.2.6, incl. Southeast) and 81.4% (IX.2.8, excl. Southeast).43

Ratios of masculine pronouns as part of the responses to question IX.3.1 (ref.: knife) are never below 50%, with overall totals between 70.5% (31 of 44, incl. Southeast) and 79.5% (31 of 39, excl. Southeast).44 Noteworthy

Page 301: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 291

here is the sentiment of the informant at So 5 who considers the form [ ]“older” than standard it.

Question IX.8.2 is one of the few questions where it is an actual keyword. It is not impossible that fieldworkers checked informants’ responses by asking them to repeat their formulations, which in certain instances might have led to a highly stylized interview situation resulting in the use of more standard language. This might be an explanation for the low to non-existent frequency of masculine pronouns in the other regions in comparison with the Southern counties (recall that there were only three instances of masculine pronouns in the West Midlands, all in Gloucester-shire). For the Southern counties, however, overall ratios of masculine forms as part of the response are as expected, ranging between 46.8% (36 of 77, incl. Southeast) and 61% (36 of 59, excl. Southeast).45

Not surprisingly, a number of non-standard forms occur in the Southern counties in responses to questions where pronouns were generally not recorded in the other regions – probably because the response would have been standard it there anyway. This includes question I.11.2 (ref.: cart),IX.4.4 (ref.: spade) and IX.9.3 (ref.: something).

8.5. Summary and results

The results of this section are noteworthy in two respects: First, the clarity of the emerging pattern of the distribution of non-standard pronominal forms is striking. Figure 7 summarizes the results of all analyses in this section, showing a clear core Southwest area with adjoining transition zones. Only some localities fall out of the picture, which is generally textbook-coherent, starting from highest frequencies of masculine pronouns in the deep Southwest and slowly petering out the further east one moves.Interesting to note is also the “belt” of generally higher frequencies close to the eastern border of the West Country. Why this vertical strip exists is unclear at the moment. One might speculate that, from the equally high frequencies for the whole area, certain localities were more influenced by the standard than others. Those localities then exercised consecutive influence on the neighbouring ones. Somerset seems to be leading in this change towards the standard. Although this may not look surprising, this west-east cline contains the second noteworthy aspect: It has been accepted as common knowledge that West Cornwall is not really to be considered part of the West Country

Page 302: A Comparative Grammar of Brtiish English Dialects

292 Susanne Wagner

dialect region. The analyses presented here, however, prove that the opposite is true. All Cornish localities show frequencies of masculine pronouns of between 80%–100%, which is definitely not to be expected from the picture drawn in the literature.

Figure 7. Distribution of masculine pronominal forms (SED Basic Material)

9. The SED fieldworker notebooks data

The SED fieldworker notebooks provide a yet largely untapped data source for dialectological studies. Although by no means suitable for just any morphological or morphosyntactic investigation, the notebooks contain a wealth of examples in the domain of personal pronouns. Despite the obvious problems connected with using the fieldworker notebooks for a corpus study (or any study), it was felt that here the sheer quantity of data by far outweighs any problems of a qualitative nature. This does not mean

Page 303: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 293

that these problems will simply be ignored – where they are of particular interest and where it seems appropriate, they will be discussed, especially since unclear transcriptions can often actually provide the key to solving the theoretical issues. The following paragraphs will illustrate the system of pronominal gender assignment as found in traditional West Country dialects in the 1950s and 1960s.

9.1. Referent types

Based on the literature on gender assignment in the core Southwest, we expect to find a system that is (still) very similar to the one described by Elworthy and his colleagues in the 19th century. Speakers of such a traditional variety would use he and him, un, en, ’n etc. to refer to inanimate count nouns, while feminine pronouns are restricted to female animates (i.e. humans and animals) exclusively. Instances of masculine pronouns referring to mass nouns and instances of feminine pronouns referring to inanimate entities should not be encountered.

It should be clear from the beginning that the array of possible referents is heavily restricted by the design of the SED, most examples involving one of the words asked for in the questionnaire. As most of those items are concrete count nouns, the chances of encountering personal pronouns referring to highly abstract nouns are rare from the start. The overall predictive value of this SED-based section will therefore be rather low. First and foremost, the results obtained here are intended to serve as a reference frame for the investigations in the following sections, where more modern material will be at the core of the analysis.

9.1.1. Masculine referents

As has already been pointed out, the very nature of the SED fieldworker notebooks (and, indeed, the SED itself) is responsible for a very homogeneous picture regarding the kinds of nouns that occur with gendered pronouns. Although the term “homogeneous” may be misleading, it is appropriate when comparing the rather restricted number of referents from the fieldworker notebooks with the wide variety of possible referents that was found in the collected corpora from Southwest England and Newfoundland.

Page 304: A Comparative Grammar of Brtiish English Dialects

294 Susanne Wagner

The nouns to which the non-standard pronouns refer can be divided into three large classes (1 to 3 below); not very surprisingly (considering the nature of the SED), class 1, “man-made objects”, contains by far the largest number of entries. The following paragraphs will show typical examples of each class and its sub-categories. Also, overall frequencies of occurrence will be given for each class.

1. MAN-MADE OBJECTS(a) buildings, their parts and contents (e.g. chimney, furniture)(b) containers (e.g. box, cup)(c) tools and instruments (e.g. farming utensils) (d) vehicles (e) “creature comforts”

i. clothing ii. accessories of modern life (e.g. watch, pen, photograph, toys)iii. food and drink (excluding naturally occurring things)

(f) nature re-modelled (e.g. mine, lane, curb stone)

2. NATURE(a) trees and plants (b) other (e.g. fruit, vegetables; ground, hill, pond, river)

3. BODY PARTS (of animals and humans)

Although the class labels are slightly different from those proposed in earlier research, the reader will immediately discover similarities between Classes 1 to 3 and the traditional categories postulated by 19th-century writers such as Elworthy or Barnes and also Svartengren’s suggestions. The first class of man-made objects contrasts with the second one, which subsumes all “naturally” occurring things under it. The third class, on the other hand, could also be labelled “man and beast”, as its entries largely refer to animals or humans.

9.1.1.1. Man-made objects

BUILDINGSThis category comprises more than 40 relevant forms. Typical referents are door, house or chimney; representative examples can be found in (7).46

Those that occur with more than two examples (see Table 27) are generally either key words in questions or mentioned therein and taken up in the response by the informant.

Page 305: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 295

(7) a. (36 Co 5, book V) It had a top with a tassel on it. ref. = bed b. (31 So 14, book VIII) (Do you) hear it? ref. = bell (church bell) c. (32 W 6, book V) It was cleaned out only once a year, you see. ref. = chimney d. (36 Co 4, book IX) We generally say it’s half-way open. ref. = door e. (31 So 9, book V) Sweep the floor if you aren’t going to wash it. ref. = floor f. (36 Co 7, book IV) That’s the post it’s hung on. ref. = gate g. (24 Gl 7, book 9) I haven’t seen it; he hasn’t seen it. ref. = house

Table 27. Masculine pronouns referring to buildings, their parts and contents (Fieldworker notebooks)

referent # of examples # of speakersbed 2 1 bell 1 1 building, house 15 14 chimney 5 4 church 1 1 door 17 14 floor 2 2 gate 2 2 lock (hasp) 1 1 oven 2 2 roof 2 1 table 1 1 WC 1 1

total 51 38

The referents are usually typical count nouns which are expected to trigger masculine forms. WC may be a bit unusual, but according to the context, a sort of outhouse is discussed, which belongs in the same category as house.The status of floor may be debatable; one could argue that floor in the sense that it is used here (“layer on which people tread”) is a singular noun, as there is only one floor in a room. Nevertheless, it should be clear that the

Page 306: A Comparative Grammar of Brtiish English Dialects

296 Susanne Wagner

word is + count in general, which would justify the use of a masculine pronoun in any of its uses.

CONTAINERSWhen the different categories of noun classes were devised, a separate class was reserved for containers. The rationale behind this decision was based on the (folk) belief that containers have some features in common with women – “womb-like” is one of the terms one hears mentioned. On such a basis, at least some feminine pronouns that are used to refer to such a noun are expected. However, the data tell a different story: Not a single feminine pronoun referring to a container is found in the material. Containers, like all other man-made objects, are masculine pronoun territory, just as expected from the traditional literature. Some typical examples can be found in (8), detailed distributions of all container-like nouns in Table 28.

Table 28. Masculine pronouns referring to containers (FN)

referent # of examples # of speakers

bag 1 1 basket 2 2 bin 3 2 bucket, kettle, trough, tub 7 7 can 1 1 coffin 2 1 cup, bowl, glass 4 4 grave 7 7 parcel 3 2 pocket 7 7 pot 1 1 tumbler 10 9 vase 5 5

total 53 36

(8) a. (36 Co 4, book V) It was made by (with) withies. ref. = basket b. (37 D 9, book I) You can stand it on its head. ref. = bin (corn bin) c. (36 Co 1, book V) Put it over the bucket and empty it out. ref. = bucket

Page 307: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 297

d. (31 So 6, book IV) It is dented in. ref. = can (tin can) e. (31 So 14, book V) ... with a handle on it. ref. = bowl dish f. (36 Co 5, book IX) He shattered the glass and it broke to pieces. ref. = glass g. (37 D 2, book VII) There’s nothing in it. ref. = pocket h. (38 Do 5, book IX) I didn’t break it. ref. = vase

Referents occurring with some frequency are pocket and tumbler, which are either key words or mentioned in the respective question, and different articles of hollow- and flatware. The items all fall within the expected norm.

TOOLS AND INSTRUMENTSBased on the outline and purpose of the SED, we would expect this category to contain a high number of non-standard pronominal forms. Details for the actual referents found in the fieldworker notebooks are summarized in Table 29, and illustrative material is provided in (9). Also, as many tools are keywords in the questionnaire, the number of examples for a specific referent (e.g. spade) should be higher than in many of the other categories, where we often find just a single example per referent.

One such item is knife as part of questions IX.3.1 and IX.3.2, which are primarily concerned with the different forms of the verb find. These two questions include various sub-questions, which altogether yield no less than 22 references to knife with the help of a masculine pronoun. Plough (orparts thereof) is, not very surprisingly, also among the high-frequency items. Another favourite is spade, which once more figures prominently in one question of the grammar section (IX.4.4).

(9) a. (32 W 3, book V) We put it on the fire. ref. = fire grid b. (37 D 7, book I) It had four prongs to it. ref. = fork c. (32 W 9, book I) A beetle has got rings on it … ref. = beetle (hammer types)

Page 308: A Comparative Grammar of Brtiish English Dialects

298 Susanne Wagner

d. (36 Co 2, book IX) I’ve lost it; it fell through my pocket. ref. = knife e. (38 Do 4, book I) It keeps the linch from falling out. ref. = linch-pin f. (37 D 5, book I) Alter the wing there. Make it run differently. ref. = plough g. (37 D 5, book V) It has become old since I got it. ref. = poker h. (36 Co 4, book VIII) He borrows it very often. ref. = shovel

Table 29. Masculine pronouns referring to tools and instruments (FN)

referent # of examples # of speakers

bar 3 2 brake shoe 2 1 “The fireplace” 13 7 fork 6 4 “Horses & Co.” 12 9 knife 26 19 linch-pin 3 2 mallet, hammer 2 1 plough (or part of) 14 9 post 3 2 roller 3 2 scarecrow 1 1 screw 1 1 scythe 4 3 shovel 3 3 spade 21 20 stake 1 1 stretcher 1 1 string 1 1 various tools 25 18 whetstone 2 2

total 148 65

A sub-category is related to various items which can broadly be summarized as “having something to do with horses”; 11 examples belong here (belly band, brush, clog, harness, etc.). Another subgroup consists of items from the fireplace (book V.1) such as bellows, grate, or poker (eight

Page 309: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 299

examples). Moreover, there are numerous references to one tool or other, which will only be listed collectively (“various tools”) below; clearly identifiable items include coulter, flail, rake, or rope-twister.

In line with our expectations, the category “tools and instruments” is, with almost 150 examples (or almost 25% of all gendered pronouns in the SED notebooks), well-represented in the data. Many instances of non-standard masculine pronouns relate directly to questions in the questionnaire in that informants pick up a noun included in the stimulus question and use a personal pronoun referring to that noun anaphorically.

VEHICLESIf no neuter pronoun is used to refer to vehicles, the pronominal form of choice is generally feminine rather than masculine (cf. also Sections 3 and 6), which suffices as an argument in favour of treating vehicles as a separate category here. However, it turns out once more that the SED data are very traditional indeed – only one instance of a feminine pronoun referring to a vehicle could be found (10a). The reasons for such scarcity of feminine references are unclear. It is extremely unlikely that none of the informants used the occasional she or her to refer to a tractor etc. What is more likely is that such a use was not remarkable in the eyes of the fieldworker(s), as the use of feminine pronouns for vehicles is “sanctioned” by grammarians, if with reservations (cf. the ubiquitous fill ‘er up at petrol stations). One gets the distinct impression that only vehicles with a motor can be she – apart from sailing vessels, that is. At SED times, there were not that many motor vehicles around. Rather, horse-drawn cart and wagon were the norm, the advent of tractors is only occasionally mentioned. As a consequence, not a single she-tractor made it into the fieldworker notebooks. In addition, the one feminine reference to car does not stem from the core Southwest. It was recorded in Hatherden (Hampshire), which is in the North of the country, close to the Wiltshire and Berkshire borders.

The absence of feminine pronouns from the core Southwest counties does not mean that masculine forms do not occur either – in fact, the opposite is true. Table 30 shows the details for this category; examples are given in (10). The absolute number of denotata is relatively small, indicating that (motorized) vehicles did not yet play the central role in SEDtimes that they play today.

(10) a. (39 Ha 2, book VIII) We had to run it home. ref. = car

Page 310: A Comparative Grammar of Brtiish English Dialects

300 Susanne Wagner

b. (37 D 5, book IX) You can have my bike. I shan’t want it. ref. = bike c. (36 Co 5, book I) You’ve got to trig/scotch it up. ref. = cart d. (36 Co 5, book I) Call it drug or drag, which(ever) you want to. ref. = sledge e. (37 D 6, book IX) Drive it round so it won’t run back. ref. = vehicle

Table 30. Gendered pronouns referring to vehicles (FN)

referent # of masc. ex.s # of fem. ex.s # of speakers bike 2 2 bus 1 1 car 1 1 2 cart 34 23 sledge 1 1 vehicle 6 5

total 46 1 30

“CREATURE COMFORTS”With a label such as “creature comforts”, it will be clear from the beginning that this is not a homogeneous class. Rather, it is an accumulation of items belonging to three main sub-categories, namely clothing, accessories of modern life (e.g. watch, pen, photograph, toys), and food and drink (excluding naturally occurring items). To everyone familiar with the SED it will be obvious that these groups are inspired by frequently occurring referents in the questionnaire.

Articles of clothing Although few, a number of clear references to articles of clothing and a couple of questions that contain clothing items as possible referents justify the existence of this class. Detailed distributions and examples follow below (Table 31 and examples in 11). Once again, it is a question in book IX that yields the highest output of masculine forms. Although the context of IX.3.3 is plural (“... put them” ref. collar studs), the pronominal forms that are used by the informants are clearly singular (mostly [n]). Another question in book IX has suit as its referent, again with a comparatively high ratio of masculine forms.

Page 311: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 301

Table 31. Masculine pronouns referring to articles of clothing (Fieldworker notebooks)

referent # of examples # of speakers

bonnet, hat 6 4 boot 2 2 article of clothing 4 3 collar 1 1 sock 2 2 collar stud 12 10 suit, jacket 8 8

total 35 27

(11) a. (36 Co 6, book VI) A gook was a summer bonnet with lots of frills on it. ref. = bonnet b. (32 W 8, book VI)

It’s a brand-new one. ref. = boot c. (15 He 7, book VI) Pull it off. ref. = hat d. (37 D 1, book IX) I haven’t seen anything of your collar stud. Where have you put it? ref. = collar stud e. (37 D 10, book IX) Who made it? ref. = suit

Accessories of modern life Included here are all those items that might be classified as tools, but which have nothing to do with everyday work on a farm. Examples are watch or pen. Moreover, things that could be classified as “fun tools” (i.e. toys) such as ball, are also included in this category. Table 32 and the examples in (12) provide details.

Ball figures prominently as it is part of a number of questions, particularly VIII.7.3 “A rubber ball that’s punctured won’t ... bounce”(Orton 1962: 93). Informants usually form an independent clause and use an anaphoric pronoun to refer to the ball, resulting in formulations as in (12a).

Money, or rather coins and notes, are also frequently he or en, as in (12g). Part of question IX.1.3 (“A picture not hanging straight is hanging ... askew”; Orton 1962: 95), picture is another frequent denotatum (12h).

Page 312: A Comparative Grammar of Brtiish English Dialects

302 Susanne Wagner

Table 32. Masculine pronouns referring to accessories of modern life (Field-worker notebooks)

referent # of examples # of speakers

ball 20 13 book 1 1 clock, watch 9 5 coin, note 6 4 lamp 1 1 pen(cil) 3 2 photograph 1 1 picture 10 9 plate 1 1 see-saw 1 1 song 1 1

total 54 38

(12) a. (36 Co 1, book VIII) It won’t bounce. ref. = ball b. (31 So 14, book IX) I will lend it to you. ref. = ball c. (16 Wo 7, book VIII) I couldn’t read it. ref. = book d. (15 He 7, book VI) It’s about five minutes fast. ref. = clock e. (31 So 14, book VII) You can turn it and open it and shut it. ref. = watch

f. (15 He 5, book V) He called it that. ref. = lamp

g. (36 Co 6, book VII) You hold it against the light. ref. = note

h. (38 Do 3, book IX) It’s hanging out (askew), isn’t it? ref. = picture

Food and drink This category only includes countable items of food and drink – mass referents are treated in section 9.1.3. Also excluded are naturally grown things (such as apple), which are covered in section 9.1.1.3. As a result, this is a rather small class, with only 11 examples. Table 33 shows details; examples can be found in (13).

Page 313: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 303

In (13a), the speaker uses beer as a count noun, making masculine reference possible. In two other cases, speakers refer to bread or pastry as he,assuming an individual loaf of bread or pastry as referent (V.6.12 “When your bread or pastry has not risen, you say it is ... sad.” Orton 1962: 72).

Example (13e) may not really belong in this category, but oil can be subsumed under “food”. In addition, the semantic make-up of the example is very similar to others in this category: A mass noun is “converted” into a count noun by using a unit of measurement that is + count. Drop is used here, while the beer example is obviously meant to include glass or pint.

(13) a. (36 Co 6, book VI) He’s guzzling. He didn’t drink his beer. He guzzled it. ref. = beer

b. (36 Co 2, book VI) That’s clutchy cake. It’d be doughy, see. ref. = cake

c. (39 Ha 2, book V) The cake’s a bit sad. It hasn’t risen (up). ref. = cake d. (37 D 6, book VII) Onion and potato pasty. There might have been a bit of meat in it. ref. = pasty e. (36 Co 2, book V) Heave up a drop of oil and catch it afire. ref. = drop of oil

Table 33. Masculine pronouns referring to food and drink (Fieldworker notebooks)

referent # of examples # of speakers

(glass of) beer 1 1 bread/pastry, cake, pasty, pie 9 9 crumb, piece, loaf of bread 6 6 drop of oil 1 1

total 17 16

NATURE RE-MODELLEDThis category represents a sort of compromise or intermediate step between the class of man-made objects and the next bigger category, “nature”. Items included here are usually natural substances or similar things which man has changed to his liking. A hedge, for example, is usually left alone to grow once it has been planted, and often not maintained in any way – but it has to be planted in the first place. Similarly, a lane may consist of gravel

Page 314: A Comparative Grammar of Brtiish English Dialects

304 Susanne Wagner

or mud, i.e. natural substances, but someone had to build it. The same goes for a haystack, a rick, etc. Examples are given in (14); Table 34 shows details of distribution.

Table 34. Masculine pronouns referring to “nature re-modelled” (Fieldworker notebooks)

referent # of examples # of speakers

bundle of straw, stook 2 2 cesspit 1 1 dike/ditch 1 1 drain 1 1 field, meadow 11 8 furrow 3 3 ridge 1 1 garden 1 1 hedge 12 9 lane, path, road 4 4 rick, sheaf, stack 5 4 stick 3 3 tombstone 1 1 wall 4 3 yard 1 1

total 51 31

(14) a. (31 So 13, book II) Take out that bundle of straw and bring it along. ref. = bundle of straw b. (36 Co 6, book IV) It’s sixty acres that meadow. ref. = meadow

c. (36 Co 7, book II) They break it, plough it and till it again. ref. = field

d. (36 Co 2, book II) That’s pitched up well, that one is. ref. = ridge

e. (31 So 14, book IX) Water it by night. ref. = garden

f. (31 So 4, book IV) It is all grown over. ref. = hedge

g. (36 Co 4, book IV) It is very narrow here, Sir. ref. = path, road

Page 315: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 305

h. (36 Co 4, book II) You cut a young elm stick, it would bow down handsomely (very well). ref. = stick

i. (38 Do 1, book IV) A sheep wouldn’t go over it. ref. = wall

Field is included here as the OED states that the use of field meaning “open land as opposed to woodland; a stretch of open land” is obsolete (OED, “field” 1.a.). Moreover, SED informants are generally referring to their arable fields, clearly meaning the worked land, not some untouched stretch of land. Interestingly, the semantic change took just the opposite path with meadow, which originally only referred to the worked land, but now also stands for grass-covered land in general. There is just one clear masculine reference to a meadow in the SED fieldworker notebooks (14b), where the reference to its size shows that it is probably used for either hay production or as pasture.

Furrow and ridge are intrinsically connected in any field (see “ridge” 5. in the OED), and speakers refer to both with the help of masculine pronouns as in (14d), for example. Hedge is part of questions IV.2.3–4, and is generally taken up in the response with a masculine form, as in (14f). The walls included here (see 14i) are not parts of buildings, but are used to separate fields, fulfilling the same function as a hedge.

9.1.1.2. Man-made objects – summary

Summarizing the different sub-categories of man-made objects, we can conclude that the SED informants from the West Country still use a system of pronoun reference that is very close to the traditional system as described in 19th-century dialectological literature. Denotata are, almost without exception, concrete count nouns (cf. section 9.1.3 for non-standard referents). Table 35 summarizes the results from the different sub-categories (excluding the one feminine form referring to car).As has been pointed out at the beginning of this section, man-made objects constitute the overwhelming majority of the masculine pronoun referents in the fieldworker notebooks. With far over 400 instances, approximately two thirds of all examples belong to this category.

Page 316: A Comparative Grammar of Brtiish English Dialects

306 Susanne Wagner

Table 35. Masculine pronouns referring to man-made objects (Fieldworker notebooks)

referent subclass # of examples # of speakers

buildings 51 38 containers 53 36 tools & instruments 148 65 vehicles 46 29 “creature comforts” articles of clothing 35 27 accessories of modern

life54 31

food & drink 17 16 nature re-modelled 51 31

total 455 110

9.1.1.3. Nature

A category “nature” is postulated for the same reasons that were given for the classes of “containers” and “vehicles” respectively. Nature and naturally occurring things are often given feminine pronoun representations in StE, partly based on mythological explanations (e.g. Latin arbor “tree” – grammatically masculine, but feminine agreement based on the belief that nymphs inhabit trees), partly on patterns of metaphorical personification (“nature – she”).

Based on the make-up of the SED questionnaire and its general concern with things important in farming communities, a number of nouns one can subsume under the heading “nature” are part of the questions or even key words. In the following, “non-edible” denotata (basically trees and plants47)will be distinguished from others (“edible” ones such as fruit and vegetables, but also other features of nature such as river or moon). This distinction is based on categorizations in other studies (see, e.g., Siemund 2001 or Pawley 2002). Whether or not it is justified will be discussed at the end of this section.

TREES AND PLANTSTree or parts of a tree such as bough or branch are among the most frequent referent nouns, occurring 20 times in the context of six questions. In particular, trees are mentioned or at least implied in all questions in section IV.10 “Trees, bushes” and IV.12 “Parts of a tree”. In question VIII.7.4, it,

Page 317: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 307

referring to a tree, is part of the expected response (“If a boy wanted to get to the top of a tree, he’d have to ... climb it.” Orton 1962: 93). Examples of this category are given in (15) and summarized in Table 36.

(15) a. (16 Wo 7, book V) Put a big bough on and it would burn all week. ref. = bough b. (37 D 10, book II) Its face is wide open. ref. = flower

c. (32 W 8, book II) It’s broader, isn’t it? ref. = leaf d. (25 O 2, book II) It isn’t supposed to grow here. ref. = plant

e. (36 Co 4, book IV) If it breaks off, it’s bound to leave the stump behind. ref. = tree

Table 36. Masculine pronouns referring to trees and plants (Fieldworker notebooks)

referent # of examples # of speakers

bush 1 1 flower 1 1 grass 1 1 leaf 2 1 plant 6 6 tree (or part of) 21 17

total 32 25

OTHERReferents belonging to this category are rather heterogeneous: apples are represented, as are other naturally occurring things such as fire, hill orriver. Various references to the moon are interesting. Although traditionally associated with feminine rather than masculine pronouns (cf. OED, “moon” 1.b.), dialects use masculine pronouns almost exclusively in reference to the moon. Two potential explanations come to mind: for one, it seems possible that the traditional Germanic gender distinctions with a feminine sun and a masculine moon (as still found in Modern German) “survived” in the dialects and did not give way to the classic system of Latin/Romance languages, which has a masculine sun and a feminine moon (Lat. sol m.‘sun’ and luna f. ‘moon’).

Page 318: A Comparative Grammar of Brtiish English Dialects

308 Susanne Wagner

Alternatively or possibly reinforcing the first scenario, another folk belief took over in dialects, namely that of referring to the patches of darkness and light as a man’s face, thus “the man in the moon” (cf. OED, “moon” 1.a.). In the latter case it looks as if we would be well-advised to classify these examples as personifications. However, the OED definition explicitly states that no personification need be involved when using a feminine pronoun – so why should we necessarily assume it for masculine ones? Table 37 shows details of the referents found in this category, illustrative material is provided in (16).

Table 37. Masculine pronouns referring to other natural features (Fieldworker notebooks)

referent # of examples # of speakers

fruit, vegetables etc. apple 6 5 berry 2 2 nut 1 1 onion 3 2 potato 2 2 egg 2 2

forms of water pond 4 3 puddle 2 1 river, stream 8 8 well 1 1 ford 1 1

other fire 8 7 hill 2 1 stone 5 5 moon 7 5 thunder 1 1

total 56 42

(16) a. (36 Co 4, book VI) I always eat/ate? it skin and all. ref. = apple

b. (31 So 3, book V) Bank it. ref. = fire

Page 319: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 309

c. (38 Do 3, book VII) It has got a wheel around it. ref. = moon

d. (25 O 4, book V) That onion there, (if) you’d plant it you’d get chibols from it. ref. = onion

e. (36 Co 7, book IV) (There) isn’t any depth in it. It’s shallow. ref. = pond

f. (31 So 14, book VIII) Flinging it. ref. = stone

A by now familiar picture emerges when summarizing the results from the two sub-categories that constitute the superordinate category “nature”: Once more, not a single feminine pronoun is encountered. No matter if the denotata in question are big or small (e.g. tree vs. bush), dangerous as a fire or completely innocent, edible or inedible, moving or unmoving (e.g. rivervs. pond), the non-standard pronouns that are chosen by the SEDinformants are masculine without exception.

9.1.1.4. Body parts

Various questions in the SED refer to body parts, both of humans and animals. Occasionally, one could expect a metaphorical link in the form of a part-whole relationship that would justify the use of gendered pronouns even in StE. Individual cases will have to be put under scrutiny to see if such a condition applies. This is the only category dealing with animates (human beings and animals) – all others discussed so far strictly concerned inanimate entities. Relevant referents are divided into two groups (human – animal) in Table 38, with representative examples in (17).

(17) a. (36 Co 4, book VI) I’ve been (and) ricked it. ref. = ankle

b. (15 He 7, book VI) Drum of the ear, cleaning it out with a pin. ref. = (drum of the) ear

c. (31 So 14, book III) She tore it off. ref. = hoof

d. (38 Do 4, book III) That is a dead teat with no milk in it. ref. = teat

Page 320: A Comparative Grammar of Brtiish English Dialects

310 Susanne Wagner

Table 38. Masculine pronouns referring to body parts (Fieldworker notebooks)

referent # of examples # of speakers

human body parts ankle 4 4 belly 2 2 blister, boil, bump, swelling, wound 15 11 disease 1 1 ear (& parts) 4 4 finger, nail, part of hand 5 4 head 3 3 tongue 1 1 tooth 4 3

animal body parts bone 15 13 hoof 3 3 pelt 2 1 skin (of bacon) 1 1 teat 5 4

total 65 39

Bone is one of the most frequently recorded referents, owing to one question in particular in which it is part of the expected response (VIII.7.6 “A dog buries a bone because he wants to ... hide it.” Orton 1962: 93). Thedecision to include the one reference to disease under “human body parts” may be debatable, but seems justified as the question is part of the general section on the human body in the questionnaire and because diseases befall human (and animal) bodies. As far as one can tell from the short excerpts in the fieldworker notebooks, all of the 57 examples of masculine pronouns used for body parts are true instances of gendered pronouns, without any personification or metaphor involved. Referents are generally typical count nouns, and the use of a masculine pronoun would be expected according to the traditional system of gender assignment in West Country dialects.

9.1.1.5. “Over-generalized” cases – masculine female animals

It is assumed in the traditional literature that West Country speakers from time to time seem to over-generalize their masculine pronouns. As a result, a cow may become he, a reference that is usually too remarkable to go unnoticed by the SED fieldworker. It is unclear in how far such uses are

Page 321: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 311

really instances of over-generalization, but it seems as good an explanation as any. Table 39 lists the animals that receive masculine reference by SEDinformants. Unsurprisingly, cow is the most frequent denotatum (cf. 20a–c).As pronominal reference to animals is outside the scope of this study (see section 5.2), these cases are only mentioned here for their own sake and will not play any role henceforth.

Table 39. Masculine pronouns referring to female animals (Fieldworker notebooks)

referent # of examples # of speakers

cow 12 9 ewe-cat 1 1 hen 1 1

total 14 11

(18) a. (37 D 7, book III) Till it has been milked. ref. = cow

b. (37 D 8, book VII) That cow is looking pretty thin, isn’t it? ref. = cow

c. (32 W 5, book VI) That cow is wet; it’s got a chill. ref. = cow

9.1.2. Feminine referents

Not unexpectedly, feminine pronouns are only rarely used to refer to inanimate nouns in the fieldworker notebooks. In fact, “rarely” is much too tame an expression to describe the situation: Among the almost 700 interesting examples, there are only three, with two from the same speaker, which are clearly feminine. They can be found in (19).

(19) a. #Lot o’ #copper in #she. When #she went in was #bad #thing. Lot of copper in it. When it was liquidated ...” (36 Co 6, book IV) b. It takes some pulling. (21 NF 6, book VI)

The first reference is to a copper mine, the second to pulling out an eye tooth. The first example contains the only feminine references to an

Page 322: A Comparative Grammar of Brtiish English Dialects

312 Susanne Wagner

inanimate entity in the Southwestern notebooks. It is unclear why the speaker chooses this form here, although mines are traditionally among the entities which can take feminine pronouns. Also, it should be noted that St. Buryan, the locality in question, is at the westernmost tip of Cornwall, which is supposedly not as strongly influenced by West Country dialect as the more easterly parts. Any of these would suffice as an explanation.

The second example stems from Norfolk, which is location-wise of course neither part of the West Country nor example-wise really relevant to the discussion. However, it shows the basic west-east chasm, as three West Country speakers used a masculine pronoun in the same context in four instances (referring to an eye tooth; see Table 38).

9.1.3. Problematic cases

This section discusses those cases which “should not exist” in the traditional system, that is, masculine forms referring to abstract nouns, mass nouns, etc. It has already been mentioned that the fieldworkers’ transcriptions are not always consistent and thus not reliable. Most of the problematic cases are classified as such based on these inconsistencies.

The count-mass distinction is traditionally assumed to be at the heart of the principle of gender assignment in West Country dialects, with count nouns taking forms of he (and she), while all other nouns take itexclusively. The problem about this paradigm is its interaction and partial overlap with that of abstract vs. concrete nouns. At its most extreme, the system predicts that only concrete count nouns employ masculine forms, while all other nouns can only use neuter forms as anaphoric pronouns.48

The following examples illustrate the wide array of “unexpected” referents found in the SED fieldworker notebooks. It should be mentioned here already that exceptional referents are amazingly rare – totalling about 60 examples, they do not even amount to 10% of the grand total of almost 700 non-standard pronominal forms in the fieldworker notebooks.

The non-count referents encountered can be subdivided into four basic categories: a) liquids, food, b) weather phenomena, c) abstracta and d) (other) mass nouns. Referents of class a) include “nasty medicine”, “gravy”, “fat”, “bacon”, “milk” and “butter”, among others (examples in 20).

(20) a. (31 So 14, book V) That milk’s gone, it went to a crud (got sour). ref. = milk

Page 323: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 313

b. (36 Co 7, book V) Thicken it up a bit. ref. = gravy c. (32 W 5, book V) I’ve had to churn it. ref. = butter

The referents for class b) are more difficult to identify. “Fog” or “snow” can be deduced as in (21a) and (21b), but often, the referent seems to be “day” or “weather” in general, as in the example in (21c). It is interesting to note that we are almost exclusively confronted with bad weather – the negative feelings and implications might be responsible for gender choice. The importance of emotive factors in gender assignment were already mentioned in Section 6.

(21) a. (32 W 2, book VII) That’s a heavy fog. I got lost in it. ref. = fog

b. (36 Co 5, book VII) I can remember a time (when) we used to have it up to our knees. ref. = snow

c. (36 Co 5, book I) It’s damp weather today. ref. = weather

Another small subgroup of non-count referents is made up of a number of abstracta, mostly in response to two questions of the questionnaire which involve a “chance to go to college” (IX.3.7) and a “job” (IX.4.11; 22a–b).

(22) a. (36 Co 7, book IX) He didn’t take it. ref. = chance to go to college b. (37 D 8, book IX) You must get it done tomorrow. ref. = job

It is to be assumed that most of the “exceptions” to the traditional system of gender assignment are based on analogical extension: each of the three major categories (liquids/food, weather phenomena, abstracta) contains numerous other members which are count nouns, one can conclude that the masculine forms in the examples above were most likely used on that basis. Also, it should be obvious that we cannot really tell in many cases what the exact referent of the pronoun is. It might as well be a count noun expressing the same concept – just to give one example often used to illustrate this: bread is - count (i.e. it) while loaf (of bread) is + count, thus he.

Page 324: A Comparative Grammar of Brtiish English Dialects

314 Susanne Wagner

Table 40 summarizes the examples belonging to this section. Three speakers contribute more than two examples (three, four and eight respectively), while 14 speakers only add one example each. Two examples each are contributed by the remaining seven speakers. As the categories “noun class” and “number of speakers” overlap (one speaker can contribute to more than one class), the individual figures in the last column differ from the total.

Table 40. Summary of “problematic referents” (Fieldworker notebooks)

noun class # of examples # of speakers

liquids, food 21 11 weather phenomena 15 7 abstracta 11 8 (other) mass nouns 9 9

total 56 26

Cornwall figures rather prominently in this section. Eleven instances of “unexpected” masculine pronouns stem from St. Buryan and Mullion, and seven from Gwinear. Taking a look at the map of the individual SEDlocations, we see that all three localities are situated in West Cornwall. It will be recalled from earlier sections that it is generally assumed that West Cornwall is less dialectal than the more easterly regions. This assumption is based on the theory that the existence of Cornish and, following that, the relatively late introduction of English to the area are responsible for the comparatively standard dialect (of English) in West Cornwall.

However, the data presented here (and in the previous section on the SED Basic Material) make another scenario more likely: The Cornish people either seem to have done very well in adopting the West Country system of pronominal gender assignment in a rather short period of time – viewed in this light, the “misuses” we observe in the SED data are clear cases of over-generalization. Or, a more likely explanation, speakers were well aware of the system from the start and did not have to “learn” it at all, as it was part of their English competence from the their early childhood.

Page 325: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 315

9.2. Summary and results

9.2.1. Types of referents

At the beginning of this section we were interested in two issues in particular. First, we were curious to investigate the relationship between the fieldworker notebooks data and the Basic SED Material. Second, various claims made in the traditional dialectological literature as to the nature of the referents that allow gendered pronouns were to be tested.

In Section 8, we already learned that Cornwall is leading in the use of gendered pronouns. Devon, Wiltshire and Dorset follow, but Somerset, which is generally considered “core Southwest”, falls far behind – for unknown reasons, we have to admit. It is interesting to note in this respect that Montacute, the Somerset location not included in the Basic Material, shows figures and uses very close to those of Cornwall, resulting in a stark contrast between this location and the other 13 which constitute the Basic Material data for Somerset. It cannot be ruled out that Somerset speakers are less traditional than their Cornish counterparts, already having started the shift towards StE. This will have to be investigated more closely when comparing the modern corpus material for the two counties (cf. 10.2.1 and 10.2.3).

The dialect found in Cornwall in the SED period is of a very traditional Southwestern nature – this fact as such is probably the most important single result to be obtained from both the Basic Material and the fieldworker notebooks. Speakers from the two most westerly locations, St. Buryan and Mullion, use almost twice as many masculine pronouns referring to inanimate entities as their colleagues at the other Cornish locations. Based on experts’ views in the relevant literature, we would have expected to find the exact opposite in West Cornwall, namely a variety closer to StE.

When investigating which type(s) of noun(s) occur with a masculine (or feminine) anaphoric pronoun in the fieldworker notebooks, a surprisingly clear picture emerges. From a semantic point of view, almost 87% of denotata fall into three basic categories: (1) Man-made objects, (2) nature, and (3) body parts. Only 13% of the examples are referring to (in dialect terms) non-standard, difficult-to-classify or unclear referents. Non-standard feminine pronouns are so rare that the label “non-existent” seems justified. For the sake of convenience, Table 41 summarizes the figures from the individual tables given in this section.

Page 326: A Comparative Grammar of Brtiish English Dialects

316 Susanne Wagner

Table 41. Summary of masculine pronouns referring to inanimate entities (Fieldworker notebooks)

referent class subclass # of examples

MAN-MADE OBJECTS buildings 51 containers 53 tools & instruments 148 vehicles 46 “creature comforts” 106 nature re-modelled 51 NATURE trees & plants 32 other 56 BODY PARTS human body parts 39 animal body parts 26

subtotal 608 feminine referents 2 problematic referents 56 unclear referents 30

total 696

The SED fieldworker notebooks data show a variety of English that is (still) surprisingly close to the traditional West Country vernacular described by Barnes, Elworthy and other 19th-century authors. Critics may say that the material used here is atypical for a corpus study because only one of two possible values is used. Fieldworkers only noted non-standard languageuse, resulting in many exceptional masculine pronoun forms in the notebooks. We have no way of knowing how many standard forms speakers may have used. However, this is inconsequential here – we do not want to compare the system of gender assignment in SED times with that of other periods. Thus, if we were to play devil’s advocate, it is possible that SEDspeakers used it to refer to a door, one of the “standardly masculine” referents in dialect, in two or even three out of four cases, resulting in a frequency of non-standard forms as low as 25%–50%. As stated, this may well be possible – but it does not influence or change the results presented here.

Some possible tendencies towards a less rigid system of gender assignment were witnessed in the unclear and exceptional referents.

Page 327: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 317

Concluding ex negativo, these uses of masculine forms referring to “unacceptable” denotata (from a traditionalist’s point of view) may point to the diffusion and ultimate dissolution of the traditional system. Most of the problematic examples seem to be over-generalizations, where masculine pronouns are used to refer to a noun that may be close in meaning to another “acceptable” one, but whose semantics do not allow a masculine anaphoric pronoun in traditional West Country dialect.

9.2.2. Differences between counties

The distribution of noteworthy pronominal forms among the investigated counties is by no means equal. Although differences were expected, they turned out to be much more extreme than had been assumed. Table 42 summarizes the figures for the individual counties, including all localities of the core Southwest counties (i.e. also those not used in the Basic Material) and all clear examples of masculine pronouns as presented in Tables 27 to 38 (figures in fourth and sixth columns rounded to first decimal).

Table 42. Frequency of gendered pronouns per county and location (SEDfieldworker notebooks)

county # of examples

# of locations

examples per location

# of speakers

examples per speaker

Cornwall 163 7 23.3 20 8.2Dorset 40 5 8 8 5.0Devon 126 11 11.5 26 4.8Wiltshire 70 9 7.8 15 4.7Somerset 88 13+1 6.3 28 3.1total 487 46 10.6 97 5.0

Columns 4 and 6 are of particular interest in Table 42. Even on a very superficial level, the picture emerging could not be any clearer: Speakers from Cornwall produce most of the gendered pronouns by far and are responsible for almost exactly a third (163 out of 487; 33.5%) of all examples. Speakers from Dorset, Devon and Wiltshire are close to the average of five forms per speaker, while Somerset lags behind – once again, we have to say (recall Figure 7).

This overall picture does not change when looking at detailed distributions per location and per individual speaker. The order of counties

Page 328: A Comparative Grammar of Brtiish English Dialects

318 Susanne Wagner

is slightly different for examples per location – Devon and Dorset change places (see column 4). Examples per location range from two to 45. All but one of the Cornish locations are above the average of 10.6 examples per location, as are five out of nine in Wiltshire, six out of 11 in Devon, two out of five locations in Dorset, but only one out of 14 in Somerset (Montacute is not included in the Basic Material).

The order of counties stays the same when looking at who actually contributes masculine forms. In Cornwall and Devon, all informants do, while this is not so in the remaining counties. One of the nine Dorset informants (i.e. 11.1%) does not contribute, while this percentage climbs to 25% in Wiltshire (five of 20 speakers) and to 33.3% in Somerset (12 of 36 speakers, excluding Montacute). Although the order of counties in the detailed distribution list changes to a certain extent, individual informants who might distort these figures cannot be identified. Contributions range between one and 24 per speaker, with an average of five. 70 speakers are below that average or conform to it, while 27 contribute more than their share. Those 27 (or 27.8% of speakers) contribute 295 forms, i.e. 60.6% of the total of 487.

Thus, we are once more faced with a now familiar picture: (West) Cornwall is much more dialectal than has generally been assumed, at least when it comes to the use of non-standard personal pronouns. Why earlier researchers thought of Cornwall as a region where a rather standard variety of English is spoken remains an open question. Data from the SEDmaterial, both the Basic Material and the fieldworker notebooks, as well as more modern oral history material suggest that the opposite is the case.

Somerset in the 1950s, on the other hand, does not seem to have much in common with the Somerset of Elworthy’s times. While the gender system described in his studies can be considered the epitome of West Country dialect, the SED data show a system that is much closer to StE than to the 19th-century one. Although it is possible in theory that the variety changed so drastically in only one or two generations, we can immediately dismiss such a theory based on the data from the surrounding counties. Devon, Dorset, Cornwall and Wiltshire have only changed minutely in comparison with Somerset. It almost seems as if the Somerset informants’ language was not as dialectal as that of the rest of the West Country informants to start with – maybe better informants were simply not available at that time. It is very unlikely that Somerset was influenced by StE to a greater extent than its neighbouring counties. It will be interesting

Page 329: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 319

to see how the individual counties developed since SED times – an issue which will be investigated in the next section.

10. Southwest England oral history material

The material on which this section is based can be described very generally as oral history material. Although comparable in many respects, it differs from the SED fieldworker notebooks data in very important ways: The major difference between the two main sources for the Southwest certainly has to be seen in the fact that the notebooks only offer clauses, phrases or sometimes single words with minimal or even without any context at all. The oral history material, on the other hand, consists of complete interviews, generally in the form of a dialogue where the interviewer contributes only minimally in terms of quantity. While the former material does not allow quantitative or statistical analyses of any kind, the latter does.

In terms of content and types of referents, however, the SED and oral history material are very similar, maybe surprisingly so. These similarities are mainly caused by some of the oral history collectors’ interest in agricultural or related topics, such as cider making, cheese making, or farming in old times. Thus, it will be possible to compare the data from a qualitative (mainly semantic) point of view, and, where relevant, to trace changes in use.

The data from the different counties will be analyzed as a unit when discussing types of denotata. In a second step, however, counties will be looked at individually, partly in order to be able to point to differences between them, partly to illustrate peculiarities of a certain region better, and also to identify possible intra-county contrasts. The outline of this section will parallel that of the previous one wherever possible to make cross-referencing and comparisons easier. For easier legibility, there will be no tables detailing individual distributions; interesting figures are included in the text and in the summary in section 10.2.

10.1. Referent types

As in Section 9, semantically similar referents will be discussed together in groups, using the same three basic classes that were already postulated in that section. Accordingly, most denotata are expected to belong to class (1)

Page 330: A Comparative Grammar of Brtiish English Dialects

320 Susanne Wagner

of “man-made objects”. Class (2) is constituted by items referring to natural features, while class (3) referents are body parts; for sub-divisions and details, see page 294.

Problems of classification are bound to occur when a framework constructed for one data set is used on a different one. Such problems are expected here as well, since it is almost impossible that all referents found in the West Country material will neatly fit into one of the categories postulated for the SED fieldworker notebooks. Where they occur, these problems will be discussed and the decisions for putting an example into one or the other category will be justified wherever necessary and possible.

10.1.1. Man-made objects

10.1.1.1. Buildings

References to buildings, their parts and contents are distributed fairly evenly throughout the Southwest corpora. All counties are represented; seven examples stem from Cornwall, 10 from Devon, 10 from Somerset and two from Wiltshire. Most speakers contribute more than one example, as can be seen in (23) below; feminine examples could not be found.

(23) a. ’Twas a tall chimney, you see and I think they tried to bomb that chimney, thinking perhaps when he fell, you know, he’d do quite a bit of damage. (FRED Dev_009)

b. [talking about a roof] That hasn’t been on very many years, I said “He’s been on more than sixty years”, he said “How did you know?” I said “I can mind when he was (gap ‘indistinct’) it was all thatched then.” (FRED Som_022)

c. (gap ‘name’) Lovely church that is if you’ve never seen ’im. He’sworth looking at. Oh, father and mother’s buried there. (FRED Som_036)

10.1.1.2. Containers

This class is well-represented, partly as a result of the topics of the interviews. Tubs, barrels and similar items are used in cider and cheese making, as well as in mining ( coal tub). The informant of Som_009 is a potter, so it comes as no surprise that he should make numerous references

Page 331: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 321

to pots when relating the story of his life. Interestingly, there is one instance of she referring to a pot among those 21 masculine references, in a non-significant position (neither at the beginning nor at the end of the interview). What motivated this form is unclear.

Only ten of the total of 81 forms are not from Somerset; eight are used by speakers from Cornwall and two by Devonians. While the comparative ratio of number of words to number of examples is as expected for Cornwall (Somerset: ca. 175,000 words – ca. 70 examples; Cornwall: ca. 20,000 words – eight forms), it is too low for Devon. Also, it is unclear why there is not a single example of a masculine pronoun referring to a container in the Wiltshire data.

Example (24b) very nicely illustrates and supports Ihalainen’s theory that standard forms (i.e. the it here) are first used in the less accessible positions of the noun phrase accessibility hierarchy. The speaker has it inthe prepositional object slot, while he (still) uses dialectal he in the more prominent subject position (for details on and an investigation of this theory, see Wagner 2004b).

(24) a. But the best corn would come this side and when that chute did fill up his bag there so as you could tie un nicely, he would switch over, turn the chute over in that bag and while that bag was filling he would pull him back and tie him and put him back in a nice row and hang up another empty one there. (FRED Som_010)

b. So, therefore, if they had to use a stone jar they used to get the basket maker to put some wicker around it, so if he had a bump he wouldn’t break. (FRED Som_024)

c. Not fully, well he been fully, what I think he was glazed with a dark glaze, like that other one I think, only he hasn’t been fully burnt yes. He would have been back one side, you know and the fire didn’t actually get to him and made enough. So they gave him away I think… (FRED Som_009)

10.1.1.3. Tools and instruments

Based on the topics of most of the interviews, a relatively large number of examples for this class is expected. One of the Somerset interviews is concerned with sailing, boats and their parts in particular. For that reason, a sub-class “ship parts” is postulated, containing references to anchor, boom,

Page 332: A Comparative Grammar of Brtiish English Dialects

322 Susanne Wagner

flag, gaff, sail, rope, etc. Only those referents are included in this class that do not occur in other classes.

A second sub-class is concerned with items used in cloth making, the topic of many of the Wiltshire interviews. “Various tools” subsumes denotata such as fork, knife or shovel under its heading. Once more, we do not find a single feminine pronoun used anaphorically to refer to a tool or instrument. As was expected, examples are numerous for this class. Somerset speakers once again contribute more than their share with 78 relevant forms, while all other counties are underrepresented: 21 examples stem from Wiltshire, three from Devon, five from Cornwall.

(25) a. Then one of them let the bar fall. As he felled, he said, he hit both mi hands, for I had hold of a rope, and bounced off the ground, come on mi two feet. (FRED Som_028)

b. I’ve got a barley fork out there and I’ve had him there for years. I used to have him when I thatched the ricks - push him in the rick to keep the bottom reeds from slipping. (FRED Som_022)

c. I’ll tell you what I found - one of our old hay knives. I got him out and cleaned him up and put a new handle on him; he looks alright too. (FRED Som_027)

d. I drove into a lamp post. (v ‘giggling’) There’s the, the pavement, see and your lamp post was right here on the curb. This was a great big one and he got sort of battens all the way around en you know, this, one of the anchored posts this was. (FRED Dev_001)

10.1.1.4. Vehicles

It may be recalled that it was comparatively surprising that the SEDfieldworker notebooks data did not contain many references to vehicles at all. In the more modern material, on the other hand, we find numerous references to all types of farming machines, particularly to tractors (often referred to as engine in the corpora). This category interacts with the previous one because many farming machines are vehicles but also tools (e.g. thrashing machine). It was thus decided to establish a new category “farming machines” for these referents, which will be discussed after this class. As a consequence, only non-farming vehicles will be covered here. Ships, boats and other water vehicles are included here, and it is not surprising that reference to them is made with the help of feminine pronouns in an overwhelming majority of cases. There are some unclear

Page 333: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 323

masculine references to a tug boat and a steamer in Som_02849, which could be explained in reference to size and strength. As these references occur in the same interview where the speaker uses more than 100 feminine pronouns referring to various boats and ships, including tug and steamer,masculine reference seems not very likely. Thus, these masculine pronouns could also be interpreted as references to the captain(s) of the vessels rather than to the vessel itself. In the concrete examples as in (26a), however, contrasts between masculine and feminine pronouns seem to reflect emotional attitudes, supporting an interpretation in favour of semantic contrasts between he and she. He stands for large size or negative emotions, while she is smaller and/or positively connotated.50 In (26b) and (26c), we find two more examples supporting Ihalainen’s hypothesis. The speakers use he in subject position to refer to the lorry and the train, but a neuter pronoun in (prepositional) object position.

Masculine forms seem to be the pronouns of choice when referring to a “non-water” vehicle. From today’s perspective, we would have expected more feminine pronouns. However, there are not enough examples to make any far-reaching claims about the general situation in West Country dialects. It should be mentioned though that seven of the 14 masculine pronouns not referring to boats stem from women or very traditional dialect speakers. In other words, these speakers behave as expected. Included are all references to car, which is more frequently feminine these days, and three references to coach (two examples) and train (one example).

(26) a. That’s why you had, you know why you had a pilot, but there was no need to because you’m following the damn tug. Because he had to keep there, he had to keep behind un in any case. (FRED Som_028)

b. Oh I had a, I had a - I drove a lorry that was the same age as miself, 1926 Dennis and drove it for years and he never broke down on me all that time, and to start it you had to crank the handle … (FRED Wil_006)

b. … if they would change the time of the train so that the, we could travel on it and get to work at uh, eight o’clock, you see I think herunned a bit later, to go, I don’t think they in Totnes started until nine in the milk factory. (FRED Dev_010)

Page 334: A Comparative Grammar of Brtiish English Dialects

324 Susanne Wagner

10.1.1.5. Farming machines

This section will discuss all types of farming machines, regardless of whether they are vehicles such as tractor or rather tools such as swathturner. The examples are nicely distributed across the counties, obviously based on the farming bias in many of the interviews. At first glance, gender assignment seems to be highly variable for this category. While farming machines are consistently masculine for some speakers, they are exclusively feminine for others. Both “exclusive” uses are easily explained, as the traditional dialect requires masculine forms while the StE system allows feminine ones. Only one speaker uses both “non-neuter” genders for the same referent, probably as a result of the traditional system competing with the (spoken) standard one. Size, though a factor in the relevant examples, is not identified with either gender consistently enough to claim that it is responsible for the choice of pronouns. Based on frequency of use (three masculine, 22 feminine forms), feminine pronouns seem to be this speaker’s “first choice”, implying that his system is closer to StE than to the dialect for this category. Emotional involvement can be used as an explanation for shifts from standard it to dialectal masculine forms as in (27c). Although feminine pronouns do occur, only three speakers actually use them. With the one exception mentioned above, intra-speaker variation occurs only between feminine and neuter or masculine and neuter pronouns, but not between masculine and feminine forms. The traditional system thus seems to be fairly intact for this category – under closer scrutiny, the surface variability turned out to be quite systematic.

(27) a. Soon as the engine come over the main Bristol road, of course, down hill like that she dropped her nose and she begin to pick up speed and run. He couldn’t handle her, he couldn’t stop her. He jumped out of her, let her run away, and she crashed through the wall, and dropped down about sixty feet in this ravine, and smashed herself all to pieces. (FRED Som_014)

b. Ah, well now the swath turner was a marvelous machine because you see he’d turn it over on, turn it over on dry ground, he wouldn’t put it all together, he’d turn two swaths, one there and one there. (FRED Som_017)

c. … you can work a tractor and it don’t get tired like horses. When you finish you can put it in the garage or in the shed, ’n’ all you got to do is give ’m a drop of water ’n’ fill un up wi’ diesel … HDC (Devon)51

Page 335: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 325

10.1.1.6. “Creature comforts”

ARTICLES OF CLOTHINGConsidering that many of the Wiltshire conversations are concerned with work in cloth mills, it is strange that this class, with only nine masculine examples, is very underrepresented. It seems that here the traditional gender assignment system has already given way to the standard for many speakers, as can also be witnessed in (28b), where the neuter form seems to be in free variation with two instances of traditional he.

(28) a. … that were all splattered, she set he [riding habit] to the cleaners … (FRED Wil_001)

b. You know what the length of the cloth is before it goes in the machine, say 65 and the designer says, I want that 60 yards long, see, well you can calculate how much he was shrinking by measuring the yard up, see what I mean, he went in 61 inch, see what I mean, that’s, how many yards is that? (FRED Wil_001)

ACCESSORIES OF MODERN LIFEBased on the fact that the oral history material investigated here is more modern than the SED data, more references to this class are expected. Referents include a wide array of things, from coin to computer, from painting to paper. Example (29c) nicely illustrates the importance of emotions in gender assignment. The speaker talks about an oil lamp that is difficult to light, using it three times before shifting to he in the last reference, where he is clearly annoyed.

Although neuter references occur regularly for this class, not a single feminine example could be found. This is noteworthy in comparison with the observations that other researchers and also this author have made. In modern fiction and everyday conversations, modern technological appliances such as computers are shes rather than hes (if not its). We should thus conclude that the traditional system of gender assignment has been substituted by the StE system in the Southwest, at least for nouns referring to accessories of modern life.

(29) a. … you used to get these Litmus papers, you know what I mean to say. These what do change the colour, and if he did show green, see, he[cloth] were alright, see, but know up to you got to a certain point he

Page 336: A Comparative Grammar of Brtiish English Dialects

326 Susanne Wagner

might be red, purple, blue and all of a sudden he goes green. As soon as he goes to green you know he’s [cloth] alright … (FRED Wil_001)

b. Now, oh, you ain’t got en [tape recorder] on, have ye? (FRED Wil_010) c. Little oil-lamp, I had to get a match and light it and as I was coming

home it blew out, and there was I by there trying to light it and I, hewouldn’t light with nearly a whole box of matches. (FRED Wil_023)

FOOD AND DRINKThis section features a number of “irregular” referents belonging to the class of liquids, with cider as the expectedly most common one. It may be recalled that the SED fieldworker notebooks contained a number of references to non-count nouns of this class as well. In general, however, the number of entries for this class is, as expected, small. Most nouns referring to items of food and drink are non-count nouns, which should receive neuter reference, and obviously do so in most instances. “Non-dialectal” references in the Southwest corpora are few; denotata include butter, cider, milk and a type of cow feed.

In (30b), the informant describes the process of cheese making. While the first references are still to milk and consequently neuter, the change from milk to cheese is paralleled in the change from it to masculine forms. A period of indecisiveness as to the status of the milk-to-become-cheese mixture is obvious in the use of both forms in the middle of the passage. Cheese, both in its milk-based and apple-based variants, is a frequently occurring referent. The classification of cheese is problematic in some respects (cf. also 10.1.1.7). Cider and cheese are by far the most common and most evenly distributed referents in the Southwest material.

(30) a. … you just simply make the butter and put your hand in the salt and put so much on en, like. (FRED Con_005)

b. And tip it back in when you had it the right temperature. Well, she knowed she’d got electric steamers, boilers and the cheese stuff was all heating, you see. (gap ‘indistinct’) Like when you’d finished she used to have to lift it out to help it on, lift it out then cut it up in chunks and take it out and then keep it so long wrapped up because it was warm. Well then you had to grind it, salt it, put it in these vats, put it in the press. Next day you’d take un out, turn it over, put a fresh cloth around it, turn it over, press un again and then the day after that you would take un out and you’d sew, put a cloth over the top again and a

Page 337: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 327

canvas around the side. Sew it up and pin it through. And then you’d put un back in the press for the next morning and that press was tight. Well then you’d take un out, put un in the cheese room on a tray, and you would turn ’em every day. (FRED Som_013)

c. They used to put potatoes in the, the loaf of bread, you know, and when you cut un, you cut half, half a potato and half a bread. (FRED Som_028)

d. [Int.: … what did you do with the milk?] Oh, separated en. (FRED Con_005)

10.1.1.7. Nature re-modelled

Once more based on the agricultural bias of many of the interviews, we expect to find numerous examples illustrating this class. References to ricks and other forms of bundling cereal plants (sheaf, bundle) as in (31a+b) are particularly frequent. The distribution of referents across counties is fairly even, with 30 examples from Somerset, six from Devon and three from Cornwall. Wiltshire speakers only contribute one example, probably based on the fact that agricultural topics are not discussed as often as in the other counties. Feminine forms do not occur.

(31) a. And when you’d dig a ditch, if you hadn’t done him right for the water to run, that wasn’t right. (FRED Som_013)

b. Oh, I’ve thatched a rick, built a rick and thatched him. (FRED Som_22) c. ’Course if one sheaf was a little bit out or anything I’d patch him and

he’d go right up. (FRED Som_22)

10.1.1.8. Man-made objects – summary

The most noteworthy conclusion to be drawn from the data presented in this section is the following: Although examples of both standard pronouns (neuter it) and “spoken standard” forms (generally she) do occur in the modern oral history material, the traditional dialect system of gender assignment is alive and well. Masculine pronouns are frequently used anaphorically in reference to count nouns associated with man-made objects. The sheer quantity of examples is also an indicator for this. When comparing the SED fieldworker notebooks data with the modern Southwest material, we would have expected fewer examples for the latter, based simply on the amount of material and

Page 338: A Comparative Grammar of Brtiish English Dialects

328 Susanne Wagner

number of speakers. It seems that the questionnaire style of the SED did indeed inhibit speakers from using their local lect, resulting in a variety much closer to the standard than it would have been in free conversation.

Moreover, we were able to observe in many examples that Ihalainen’s “accessibility hypothesis” seems to be a very good explanation to account for the presence of both neuter and masculine forms referring to the same entity in a single utterance. While the subject form he still defends its territory, positions further down (to the right) in the accessibility hierarchy are losing ground. Thus, even though masculine forms are still very frequent in comparison with other traditional dialect phenomena, we have to assume that the dialect system is slowly giving way to the system of StE. The path it follows is very often the one outlined earlier: The StE forms “invade” the territory of gendered pronouns from the less accessible positions along the NP Accessibility Hierarchy. Thus, many speakers still use he in subject contexts, but have already shifted to it in oblique contexts. Table 43 summarizes the results from the tables presented above.

Table 43. Gendered pronouns referring to man-made objects (SW)

referent class subclass # masc. ex.s # fem. ex.s # of sp.

buildings 29 13 containers 81 1 17 tools & instruments 107 29 vehicles 32 142 12 farming machines 65 23 21 “creature comforts” articles of

clothing 9 4

accessories of modern life

37 14

food & drink 49 14 nature re-modelled 40 17

total 449 166 68

10.1.2. Nature

A considerable amount of the Somerset material stems from a project on cider making. It is thus not surprising that apple is one of the most frequently occurring referents. All examples stem from Somerset, two of

Page 339: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 329

which can be found in (32a) and (32b). Example (32a) shows yet again the variability of the system, with instances of both it and he referring to the apple. It should be noted that this example again conforms with Ihalainen’s hypothesis concerning the stages of change from dialect to standard.

Classification is difficult for cheese, which occurs as cider cheese and“regular” cheese made from milk. Both cheeses are ultimately man-made and serve as food (the cider cheese is often used as pig feed), and it was thus decided to include cheese in the food and drink section (see above).

(32) a. [talking about the right time to pick apples] Well, like if you picked one an’ cooked it early he isn’t same as when he’s been picked and kept, is he? (FRED Som_013)

b. … but the black withy is crips. Brittle. Well he’s crips, he ain’t no good, break him up like that. (FRED Som_022)

c. Oh, I remember when that were planted that tree, they cut it down just before they built the house. I went one day wi’ the car, they’d cut itdown, they had roots and all out, and I remember when it were a little one about as big as mi little finger, I think it were planted about the turn of the century there, but when they cut it down he were about that big round, yes. I didn’t know how big it was not ’til after I come down round there with the car, and I saw they were cutting the Walnut tree down, but when we were kids it was about as big as mi thumb and hewere up about six or seven foot probably, hadn’t been planted very long, when they cut it down then I should think he were about that big round, ’cause they had a timber carriage and that there to load ’im upon, and that’s how big it got in sixty or seventy years. (FRED Wil_003)

A closer look at example (32c) gives some interesting input for the discussion on the division of tasks between gendered and standard pronouns. The topic of the conversation is a walnut tree that the speaker has known for his whole life. He remembers its being planted and relates his emotions when he saw it being felled. It seems that two or even three systems interact in producing the personal pronouns of this passage.

The first five references are it, three in direct object and two in subject position. We can explain all its most easily by claiming that the speaker is using StE here. However, staying within the dialect framework, another explanation is possible. Ihalainen’s hypothesis can be held responsible for the three instances in object position. For the two examples in subject slots,

Page 340: A Comparative Grammar of Brtiish English Dialects

330 Susanne Wagner

we can come back to a factor that was thought relevant in much of the literature on this topic – size (cf. e.g. Mathiot and Roberts 1979). Although studies could not establish a direct link between size and gender assignment, folk belief uses size frequently as an explanatory factor, for example with animal referents – a tiger is much more likely a he or she thanan ant, which will probably be it.

The speaker recalls how small the tree was when it was planted. As soon as he refers to the grown tree that is being felled, he shifts to he in the sixth reference, either because of its size or because of the subject position (or both). The next two references are neuter once more, again based on small size or for the first example possibly also on the atypical syntactic context (subject complement). The next reference is to the young tree in subject position, resulting in a masculine form. The following it is again in object position, referring to the grown tree, as is the masculine pronoun in subject position after that. Following this is the only masculine form in object position, obviously a remnant of an earlier stage of the speaker’s dialect. Finally, the speaker uses another neuter pronoun in non-subject position. This excerpt nicely illustrates the factors that are “competing” for gender assignment. Although it may seem unsystematic at first glance, the gender assignment here is very systematic, if highly complex.

With 49 examples, Somerset speakers are once more the most prominent contributors to this class. Nine examples stem from interviews from Wiltshire, three from Devon and two from Cornwall. It is also obvious from the numerous examples that natural features are either neuter or masculine in the Southwest, but never feminine.

10.1.3. Body parts

Denotata of this class are rare, which seems surprising insofar as various aches and ailments are expected to feature prominently in conversations with elderly people. A closer investigation of the contrasts between neuter and masculine forms may reveal that West Country speakers abandoned the traditional system and have adopted the StE one for this category, although this does not seem very likely.

(33) a. I can’t clench that hand see for one thing, he won’t go back no more than that. (FRED Som_016)

Page 341: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 331

b. [what to do when you had a toothache in the “old days”] You did go into the doctor and he did get hold of un and drag un out with a pair of pinchers or something, you see. (FRED Som_031)

10.2. Individual counties

As laid out at the beginning of this section, we will now change our viewpoint from a macro- to a microperspective. Although the material constituting the Southwest corpus stems from comparable sources, we cannot assume unquestioningly that they are equal. Numerous factors can influence the type of variety that is found in an oral history interview situation. Without having access to all the relevant background information, however, we can only make certain assumptions about these factors based on the general degree of “non-standardness” that we observe in the interviews. As many of these factors are connected with the behaviour of the interviewer, and as many of the interviews in the Southwest corpus were conducted by only a handful of interviewers per county, a closer look at the individual counties might help to identify such “disturbing” factors.

Moreover, the comparison of the counties is interesting from a statistical point of view. Based on the total word number and total of gendered pronouns, combined with the overall frequency of pronominal forms, it should be possible to make claims about the degree of “genderedness” of certain counties (and/or speakers). Some speakers may use an over-proportional number of non-standard pronouns, while others rank far below the average. A comparison of the four main counties under investigation could also help to pinpoint certain changes that may have occurred since SED times. Table 44 summarizes the figures the following sections will be based on, showing only those examples that stem from one of the core Southwest corpus texts. One of the Somerset texts will be excluded from the calculations most of the time, as its unusually high frequency of gendered pronouns (179 in an interview of ca. 13,500 words, largely owing to more than 100 references to boats and 25 references to coal tubs) distorts the overall impression in this group of texts. Figures in brackets refer to the totals when this text is excluded.

Page 342: A Comparative Grammar of Brtiish English Dialects

332 Susanne Wagner

Table 44. Summary of gendered pronouns in the Southwest corpora

referent class # of masc. ex.s # of fem. ex.s # of speakers

buildings 22 (18) 0 10 (9) containers 78 (53) 1 14 (13) tools 93 (69) 0 24 (23) vehicles 32 (27) 109 (1) 9 (8) farming machines 36 23 14 “creature comforts” 72 (62) 0 22 (21) nature re-modelled 32 0 11 nature 59 (56) 0 16 (15) body parts 6 0 4

total 430 (359) 133 (25) 50 (49)

10.2.1. Cornwall

Speakers from the main Cornwall corpus contribute 42 examples out of the total of 384, i.e. 10.9%. This is almost exactly twice as many as we would have expected based on the percentage of total words for this county (18,900 of 319,800, i.e. 5.9%). Based on our observations in the previous section, we can thus say that English in Cornwall and West Cornwall in particular is much more West Country like than generally assumed.

One speaker is responsible for more than half of the examples (23), which are spread fairly evenly across the different semantic domains. This informant shows the most traditional lect of all Cornish speakers, using many typically Southwestern and also specifically Cornish dialect features.52 It is this speaker as well who uses a striking construction when talking about a situation or general circumstances: let’s put en this/that way.

A second speaker contributes eight examples, the third six and the fourth speaker five gendered pronouns. All speakers contribute to the domains containers and “creature comforts”, two speakers to those of buildings, tools and nature re-modelled respectively, and one speaker each to the categories vehicles and nature. No speaker uses a masculine or feminine form to refer to a body part in the Cornish interviews. An additional text that is not part of the main corpus provides the four remaining examples, adding up to a grand total of 46.

Page 343: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 333

10.2.2. Devon

Although they contribute 16% of the words constituting the Southwest corpus (51,100), speakers from Devon use only 16 non-standard (masculine) pronouns (or 4.2% of the total). The material from Devon is relatively standard in comparison with that from Cornwall or Somerset, which may explain this ratio. It should also be mentioned in this context that the Devon interviews are not primarily concerned with agricultural matters, thus contrasting thematically as well with the data for Cornwall and Somerset. Although universal non-standard features such as non-standard agreement, regularized irregular verbs and reflexive pronouns, non-standard use of relative and demonstrative pronouns, etc. are frequently found in the interviews, typically Southwest patterns are rare.

The distribution of non-standard forms between the five speakers is relatively even. One speaker contributes five, one four, two three and one speaker one example. The denotata can be categorized as buildings, tools, vehicles, farming machines or “creature comforts”. No examples belong to the domains of containers, nature re-modelled, nature, or body parts.

When comparing the modern Devon material with data from other dialect corpora (e.g. the Helsinki Dialect Corpus), it turns out that the material is by no means as exceptional as assumed. The Devon part of the Helsinki Corpus, recorded in the mid-1970s, contains even fewer gendered pronouns proportionally than the modern Devon material. Thus, judging from the Helsinki material, one should also consider the possibility that the Devonian dialect has already advanced further on the path towards StE than some of its neighbouring dialects since SED times.

10.2.3. Somerset

The 30 interviews from Somerset constitute almost exactly half of the total material from the Southwest in terms of word number (161,000 of 319,800, i.e. 50.3%). Based on the almost standard SED data for this county, a much lower percentage of non-standard pronouns was to be expected. Just the opposite is true, however.

With 258 examples, Somerset speakers provide more than two thirds (67.2%) of all examples – a result that comes as a surprise after the disappointing figures from the SED data. Given the numerous examples, a relatively even distribution across semantic domains is to be expected.

Page 344: A Comparative Grammar of Brtiish English Dialects

334 Susanne Wagner

Table 45 shows the absolute and relative frequencies of denotata for the Somerset material.

Table 45. Distribution of gendered pronouns across semantic domains (modern Somerset material)

referent class # of examples Somerset

# of examples total

% Somerset data of total

buildings 6 18 33.3 containers 46 54 85.2 tools 43 69 62.3 vehicles 10 28 35.7 farming machines 48 59 81.4 “creature comforts” 26 62 41.9 nature re-modelled 28 32 87.5 nature 46 56 82.1 body parts 5 6 83.3

total 258 384 67.2

Only in three domains, namely buildings, vehicles and “creature comforts”, do Somerset speakers contribute fewer than expected forms. In all other categories, they come either close to the overall ratio of 67.2% as for tools, or even surpass it with ratios of around 80% for the remaining categories. Contributions per speaker range from one to 45 examples, with eleven speakers above and 19 below the mean of 8.6 examples per Somerset speaker. Those eleven speakers use a total of 186 non-standard pronouns between them (72% of the total), but only 37.9% (61,000) of the Somerset word total.53 Only four speakers, among them the three with the most forms (41, 2x 25), use significantly more forms than their “colleagues”.54 Evenwhen we exclude these four texts from the calculations, Somerset speakers still use more gendered pronouns than expected (153 examples = 59.3% in 134,100 = 42% of words).

Based on these figures, any ‘distorting’ influence of individual speakers can be ruled out. Explanations for the unexpectedly high frequency of non-standard pronouns in the Somerset material thus have to be sought elsewhere. The informants of the modern corpus material are approximately one generation younger, aged 70-80 in the 1980s, than the SED informants, who had reached the same age in the 1950s. Although the modern material does not cover all of Somerset as evenly as the SED did, locations are more numerous and more widely spread than for any other area, and a certain

Page 345: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 335

representativeness should be guaranteed. We are thus forced to fall back on the same conclusion as in Section 8, namely that the Somerset speakers selected for the SED are not representative of the county’s dialect. The only alternative analysis would be that, judging from the modern corpus material, Somerset in the 1980s is more dialectal than in the 1950s – an extremely unlikely scenario given what we know about mobility, social and historical developments, and dialectology in general.

10.2.4. Wiltshire

As has already been emphasized earlier, the modern corpus material for Wiltshire stems from two different sources. While background information is available for the material from Trowbridge, next to nothing is known about the informants, recording details, etc. of the Wiltshire Folk Life Society data. It can be assumed, however, that the WFLS material was recorded in the 1970s or early 1980s, while most of the recording dates of the Trowbridge material are in the 1990s. In addition, the Trowbridge material is concerned with the history of cloth weaving, the primary industry of the region until recently, and not with farming, as is most of the Somerset and Cornwall material. In this respect, the Trowbridge and Devon material are comparable. Furthermore, many of the Trowbridge informants live in towns rather than villages.

Much of what has been said about the material from Devon also holds for the Trowbridge material: Although universal non-standard features are frequently encountered in the data, and some typically Southwestern features (unemphatic do in particular) are also ubiquitous, the dialect in most of the interviews is much closer to StE than that of the Cornwall and Somerset data. This is consequently also reflected in the figures concerning gendered pronouns: The Trowbridge material constitutes almost a quarter of the corpus material (75,100 words, i.e. 23.5%), but only 47 examples (or 12.2%) of non-standard pronouns are found in these data.

Contributions per speaker range from one to 13 (none of these differences is statistically significant), with a heavy bias towards tools as the favoured semantic domain, represented by 19 examples from four speakers. Other domains where Trowbridge speakers use masculine pronouns to refer to inanimate entities are buildings (two examples), vehicles (one example), “creature comforts” (18 examples), and nature (seven examples).

Page 346: A Comparative Grammar of Brtiish English Dialects

336 Susanne Wagner

Only in the data from the Wiltshire Folk Life Society do total word number and number of examples nearly match (13,700 words, i.e. 4.3% of the total, and 21 examples, i.e. 5.5%). Judging from certain features in the speakers’ dialect (e.g. demonstrative thick(y), relative particle as), the WFLS material is very traditional in nature. One speaker uses two, two three each, and one speaker 13 exceptional pronominal forms. With 10 examples, denotata stem most frequently from the domain of farming machines, to which all but one speaker contribute. Another four pronouns refer to car, leaving only one or two occurrences each for the domain of tools (one example), “creature comforts” (two), nature re-modelled (one), nature (two) and body parts (one example). When comparing the results from the two Wiltshire sub-corpora, we may be witnessing an apparent or real time change. WFLS informants use a variety of English that is (still) very close to that recorded by the SED fieldworkers. Trowbridge informants, on the other hand, still use many universal dialect features, but their inventory of Southwest features seems already reduced in comparison with traditional accounts and also the SED data. The development thus seems to be one towards StE, which we would expect based on the location of Wiltshire on the (present-day) border of the West Country.

10.3. Summary – developments and changes

The first result from the corpus analyses is striking in its clarity. When comparing ratios of word numbers and numbers of examples of non-standard pronouns, the following picture emerges: The ratio of pronouns to word numbers forms a West-East continuum, with Cornwall at the top, followed by Somerset. Both counties lie above the expected average ratio (> 1), while Wiltshire and Devon fall below it. Details can be found in Table 46. Even when the WFLS material is excluded, Wiltshire is still third on the list, with a ratio of 0.52. In other words: The further westwards we move, the more non-standard pronouns we encounter. The status of Devon does not seem to fit into this image, but when we consider a second continuum, namely North-South, new possibilities emerge. As the Devon data come from South Devon exclusively (Totnes and surroundings), it cannot be ruled out that data from the North would be closer to the Cornwall and Somerset data. Whether this North-South continuum hasdeveloped recently, i.e. over the past few decades, or has simply become stronger since SED times is impossible to say. But one can clearly observe

Page 347: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 337

the gravitational pull of StE or London, which is ultimately responsible for both continua.

Table 46. Ratio of word numbers to number of examples (modern corpora)

county % of examples % of words ratio ex.s to words

Cornwall 10.94 5.90 1.85 Somerset 67.19 50.34 1.33 Wiltshire 17.70 27.77 0.64 Devon 4.17 15.98 0.26

total 100 99.99 1.00

When consulting a map of the modern Southwest, it will be noticed immediately that almost all locations with a high percentage of gendered pronouns are located north of the modern access route to the Southwest, A30. In addition, Trowbridge is at the Wiltshire-Somerset border, as far west as one can move in the county. Totnes, on the other hand, is almost as far south as possible in Devon. These observations are supported by evidence from other areas of (linguistic) research: First, when looking at urbanized areas in England around 1950 (> 400 people/sqm), it becomes obvious that the region around and particularly South of Exeter was urbanized in the 1950s already, while North Devon, Cornwall and most of Somerset (except Bristol) remained essentially rural in character. Second, even when looking at analyses of completely different phenomena, (for instance the absence of third person singular -s; cf. Viereck, Viereck and Ramisch 2002: 98), the same picture emerges: South Devon behaves heterogeneously, shifting from complete absence to 100% presence and back to complete absence of -s forms, and in Cornwall, including West Cornwall, we find zero forms very homogeneously. When comparing this map with Figure 7, definite similarities will immediately be identified.

Moreover, a quantitative study of lexical contrasts based on SED data shows that bundles of heteroglosses separate the Central Southwest (more or less at the border of Dorset) from the Eastern Southwest more strongly than the Southeast is separated from the Southwest – a result that once more ties in with our observations (cf. Viereck, Viereck and Ramisch 2002: 94). The same analysis also shows that the contrasts between Devon and East Cornwall are more pronounced than those between East and West Cornwall, pointing towards the homogeneity rather than heterogeneity of English in Cornwall.55

Page 348: A Comparative Grammar of Brtiish English Dialects

338 Susanne Wagner

In comparison with the analyses in the previous section, the following similarities emerge: (West) Cornwall should be considered “core Southwest territory” in all respects, as the investigations of both the SED data and the modern oral history data have shown. Judging from the modern corpus material for Somerset, the suspicion we had in the preceding section is confirmed: The SED informants for Somerset are not representative of the dialect of the time and area. The situation for Devon and Wiltshire is not as straightforward. While most of Devon clearly belonged to the core Southwest at SED times, the modern situation is unclear. Our data indicate that the county is moving towards StE much faster than its neighbours, but we cannot dismiss the claim that our Devon material may not be representative of the county as a whole. As our data for Wiltshire largely stem from the western part of the county, it could be argued that the overall results pointing towards Southwest status of the county could be equally skewed. However, when comparing the modern data with the SED data for Wiltshire, it should by no means be considered impossible that the material used here describes the present situation fairly well.

Apart from these noteworthy facts, another fact must be acknowledged as remarkable, namely the stability of semantic domains to which gendered pronouns generally refer. Even though nothing can be said about general frequencies due to basic differences in the make-up of the two data pools, most denotata of non-standard pronouns were and are tools and instruments. It seems that Elworthy’s and Barnes’ class of man-made objects is still very relevant in gender assignment. The corpus analyses have also shown that feminine pronouns are (still) quite rare in West Country dialect. The only entities which are frequently feminine in anaphoric references are vehicles, a use that has most certainly been taken over from StE. In light of the facts from modern studies on spoken English in general and scholars’ general opinion as presented in grammars of PrDE, we would have expected a much higher frequency of feminine forms than we actually encountered. Based on all of these aspects, we can conclude that the traditional West Country system of gender assignment has remained largely intact to the present day, at least in elderly (rural) speakers. It will be interesting to see how far the “New World” and its dialects have influenced this system. An answer to this question will be attempted in the following sections, where corpora of Newfoundland English will be analyzed.

Page 349: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 339

11. Material from Newfoundland

11.1. MUNFLA material

For the category of man-made objects, a number of conclusions can be drawn from the analyses of the oral history material from MUNFLA (for details and figures, see Wagner 2003). First, the traditional (West Country) system of gender assignment seems to have survived fairly intact in Newfoundland. In comparison of total word numbers and total number of gendered pronouns for the Southwest and MUNFLA corpora, the Newfoundland data lead by far close to 900 forms in 132,000 words.56

Second, however, in comparison with the Southwest England material, it looks as though feminine forms are slowly invading the system. In some semantic domains (in particular accessories of modern life), these forms already constitute almost 30% of the total number of forms (the domain of “water vehicles” can be neglected here as it obviously always has been feminine). Table 47 summarizes the individual figures of forms referring to man-made objects in the MUNFLA material.

Table 47. Gendered pronouns referring to man-made objects (MUNFLA)

referent class subclass # masc. ex.s # fem. ex.s # of sp.

buildings 160 3 14 containers 33 1 12 tools & instruments 116 4 18 vehicles 1 6 3 water vehicles 1 322 24 “creature comforts” articles of clothing 14 5 accessories of

modern life 144 41 24

food & drink 13 5 nature re-modelled 26 4

total 508 377 38

There is one domain of referents that deserves special attention in the MUNFLA material, basically because it contains items of a class that is much more prevalent in the MUNFLA material and which was either not at all or only marginally encountered in the corpora analyzed so far: abstract referents.

Page 350: A Comparative Grammar of Brtiish English Dialects

340 Susanne Wagner

The label “abstract referents” is used here to refer to pronouns denoting abstract count nouns such as story or song. In addition to these denotata, we also find a handful of pronouns of the non-referential type described in section 5.3. Examples (34a) to (34c) illustrate the first, (34d) and (34e) the second of these uses. The speaker of (34d) describes how dynamite was used to make a passageway for a boat through ice. Although the feminine pronoun could refer to the boat or the stick of dynamite, an interpretation as non-referential is equally likely, particularly since the structure of the elements involved resembles the one identified as prototypical in such cases (X-S-V). The same structure is also used in (34e), where the feminine pronoun seems to be cataphoric, referring to the following events, rather than anaphoric.57

(34) a. Oh he’s the hell of a long story. (MUNFLA 70-003: C631) b. I learned, if I hear a song, we’ll say somebody sings all night and if I

mind to know en, (gap ‘indistinct’) I could sing en today. (MUNFLA 70-037: C751)

c. And he’s [song] all there but there’s different words in it you know. (MUNFLA 70-037: C751)

d. … he used to put the dynamite down by the side of the ship to loosen up the ice fer to get back and forth see, and they used dynamite at that time and they used to heat up the poker and heave out the red hot poker to put to the fuse and when they would touch the fuse it would be as long as that and the powder used to go right thru the cap and fling she’d go, and so (gap ‘name’) he had two men killed with the dynamite … (MUNFLA 73-046: C1958)

e. … I only come down to help pull en [house]. But nevertheless I said, all right boys, straighten out, and away dey goes, my son, straightened out, we took dat house and here she come. (MUNFLA 72-089: C1187)

Five speakers use 17 masculine forms to refer to abstract referents (poem,one example; song, 15 examples; story, one example), while five instances of non-referential she are used by three speakers.

Page 351: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 341

11.2. Traditional versus modern uses of pronouns – the past and future of gender assignment in Newfoundland English

The analysis of the MUNFLA material allows two basic conclusions: First, Newfoundland speakers adhere to the traditional West Country system of gender assignment more strongly than their fellow speakers in the mother country. This result is based on the comparison of the relative frequencies of gendered pronouns in the two corpora, with a higher frequency of non-standard pronominal forms the Newfoundland data. Southwest speakers, on the other hand, use proportionally fewer gendered pronouns. In other words, more Southwest English speakers have adopted the StE system.

Second, the Newfoundland data show a tendency not observed in the Southwest material. While British speakers seem to be shifting (or have already shifted) from the traditional dialect system ( masculine forms) towards the (written) standard system ( neuter forms), Newfoundland speakers seem to prefer the spoken standard system ( feminine forms) rather than the StE (neuter) system. This can be concluded from the higher frequency and more even distribution of feminine pronouns in the MUNFLA corpus. While feminine pronouns occurred in only three of the nine semantic categories in the Southwest material, the MUNFLA material contains feminine forms referring to all but two of the categories. Those categories, “nature re-modelled” and “body parts”, are represented with only very few examples, however, thus possibly precluding the occurrence of feminine forms simply though lack of examples. Feminine forms constitute only about 5% of all examples (25 forms out of over 500, excluding references to boats) in the Southwest corpora, but they make up approximately 10% of all gendered pronouns in the MUNFLA data (60 forms out of 621, excluding references to boats).

Based on what is known about Newfoundland settlement history, the present state of gender assignment in West Country Newfoundland dialect(s) can be explained as follows: Settlers brought with them the (strict) system that has been described in 19th-century literature on (English) West Country dialects. Although contact between them and other settler groups was relatively restricted, it must have existed to a certain extent. None of the varieties the West Country settlers came into contact with used masculine forms in the same manner as their mother lect. Where the pronominal system did offer a choice, it was between neuter and feminine rather than neuter and masculine forms.

Page 352: A Comparative Grammar of Brtiish English Dialects

342 Susanne Wagner

In some domains of usage, contact with speakers of other dialects as well as with standard speakers resulted in a (or possibly two) shift(s) in the traditional paradigm: Neuter forms became the unmarked choice in some domains, and only when a special need for emphasis was felt to play a role ( discourse-pragmatic factors), non-neuter forms were selected ( approaching StE). For some semantic domains, feminine forms as they were used in other dialects and were also possible in some instances in StE, began to compete with the traditional masculine forms. As a result, Newfoundlanders today use standard it alongside dialectal she and West Country he. All these factors come together to form a unique system that is conservative in some aspects and modern in others.

Where the change from the traditional to the StE system is in progress, it seems to follow the path outlined by Ossi Ihalainen: The standard form (it)is first used in non-subject positions. The form he, which may also be used in emphatic object position ( pronoun exchange), is the only remnant of the traditional system in many ideolects. This theory is supported by evidence from individual utterances as well as more general observations from whole interviews (cf. Wagner 2004b).

If one wanted to make any predictions about the future of gender assignment in Newfoundland English, it would be wise not to underestimate the strength of the traditional West Country system in claiming that a change towards StE is already dawning on the horizon. In comparison with Southwest England, Newfoundland English is much more conservative and successful in preserving and sometimes re-introducing or spreading features that have long since died out in the mother country.

11.3. Folktales material

Although not as well represented as in the corpora analyzed so far, non-standard pronominal references to man-made objects constitute the majority of gendered pronouns in the Folktales corpus as well. As was expected, the traditional nature of the variety of English used for the telling of tales is reflected in the data. Masculine pronouns are generally employed to refer to man-made objects. Only one sub-class contains more than impressionistic evidence of feminine forms ( accessories), the use of which conforms with StE rules (gun = she). As in the MUNFLA data, ships and boats are exclusively feminine for Newfoundland speakers, regardless of genre.

Page 353: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 343

Table 48. Gendered pronouns referring to man-made objects (Folktales)

referent class subclass # masc. ex.s # fem. ex.s # of sp.

buildings 30 13 containers 54 1 10 tools & instruments 49 7 ships and boats 77 8 “creature comforts” articles of clothing 8 6 accessories of

modern life 73 3 7

food & drink 22 4 nature re-modelled 5 3

total 241 81 13

As in the MUNFLA data, the class of abstract referents deserves special attention in the analysis of the Folktales. While gendered pronouns referring to man-made objects are very frequent in those varieties which employ them, they become less and less frequent the further one moves away from the concrete towards the abstract denotata. The traditional system of gender assignment as described in the 19th-century literature does not “allow” such uses at all. Some instances of speakers using masculine forms to refer to abstract denotata were already encountered in the MUNFLA corpus. The figures from the Folktales corpus, however, are even more striking.

Part of the “ritual” of story-telling is the fieldworker’s question about the origin of the tale (i.e. when and from whom the teller first heard it). When relating the tale’s history, the teller usually refers to the tale repeatedly. The pronoun that is used in those contexts is generally masculine (he or un), resulting in some 90 masculine references to story.Some illustrative examples can be found in (35).

(35) a. “Head Card Player o’ the World.” Uh? You [i.e., Int. B.] heard un.[recorder off] Now this uh ... I just as well tell un. I think I ... I think I can manage to tell un alright. (Tale 012)

b. [Int: Where did you get THAT one Mr. Snook?] Hey? That’s one I got from Uncle John Martin. Fella is dead now.[Int: Where was he? Here in...] He belonged Harbour Breton an’ I got he [ref. = man or story] up erm ... in Connaigre Bay. I only wants to hear un once an’ that settled -

Page 354: A Comparative Grammar of Brtiish English Dialects

344 Susanne Wagner

an’ I’ll get un. Now ... song I got to have un wrote off, I got to look unover he ’bout twenty-five times. (Tale 007)

c. [Int: How many times did you hear it?] Only heard it once.

[Int: An’ you ... an’ you ’ve been tellin it ... ] Yes. Only hear ... only hear it once, I only want to hear a story once an’ I knowed un all then. [Int: Mm-hm. When did you first start to tell it?] When did I first start [to] tell un? Oh uh ... oh a long spell ago uh .. (Istarted) ... first start to tell un. Long spell ago, must be thirty-five year ago I say ... start to tell un. Now I’ve aknowed that story ever since I was about ... about forty year old. An’ I can still remember un. (Tale 038)

d. [Teller’s wife]: (Now be) he’s goina have her all fooled up. [aside:] I got un mixed up. I ... yes I got un all fooled up now. I got un all fooled up now. [recorder off] (Tale 005)

In (35a), the teller uses masculine pronouns to refer to song as well as to story. (35c) contains neuter alongside masculine forms. The context makes it likely that the neuter (non-dialectal) forms are triggered by the fieldworker’s use of it. As soon as some distance to the neuter forms has been established, the teller switches (back) to his usual masculine forms.

(35d) is noteworthy as it supports one of the (by now) commonplaces of sociolinguistic research: Women tend to use a variety closer to the standard, while men will use the regional / vernacular variety (longer) (cf.e.g. Labov 1972: 243). The teller’s wife uses a feminine pronoun, that is the “spoken standard” form, to refer to the story, while her husband uses traditional (regional) masculine forms.

Another abstract denotatum found in the Folktales corpus is bet, around which one of the stories evolves. As already mentioned, one speaker refers to song with masculine pronouns. Yet another story involves mathematics: One of the protagonists is tested by having to solve a sum, which is frequently, but, as example (36a) shows, not always masculine.

In addition to these repeatedly occurring denotata, one speaker uses unto refer to a trip (36b), and in (36c) and (36d), we encounter a by now familiar use of she – non-referential. In (36c), she is used to refer to the event of the pile of chairs falling down. Although identical in wording, the interpretation of (36d) is less obvious, but a non-referential interpretation or an unspecified reference seems most likely.

Page 355: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 345

(36) a. “Oh” she said “I’ve got a sum here” she said “the teacher gi’ me to do” she said “an’ I can’t do it.” She said “I can’t do that sum” she said “’tis too hard for me.” He said “Let me have a look at un.”“Hah!” she said “What’s the good to show un to you” she said “Hard Head?” She said “No good to show un to you” /she/ said “you knows YOU can’t do un.” “Well” he said “there’s no harm for me to lo[ok] ... SEE un I don’t suppose.” (Tale 037)

b. “Well” he said “Jack you’re goina finish your trip?” “Yes” Jack said. “I’m goina finish un.” (Tale 010)

c. ... was in the kitchen preparin ... a meal for us. I was a devil-may-care an’ mi father didn’t care what I do. I piled all the chairs one an top o’ the other. Well they were right to the loft. An’ I said to the sailor “Now if you can’t climb up one side an’ come an the other you’re no kind of a sailor. Well up goes one of ’em you know an’ he come down alright an th’ other side. Well HE done it we’ll say, th’ (other) onehad to do it. He climbed up you know an’ just as he got an the top part away she goes helter to skelter all over the ki ... all over the place. (Tale 141)

d. “Now” he said “that’s all you’re gettin. An’ now don’t think you’re goin to serve me like you served the ... like ... [laughing tone] Bill” he said “because you’re not.” Uh ... uh ... th’ old man said “That’s what you thinks.” Huh! [laughs] An’ he jumped up. An’ when he did Jack jumped up too an’ he picked up the hot wa[ter] ... the boil water, he dumped un down over his head an’ away she goes. [laughs] Cap hair the whole lot down on the floor. [laughs] An’ th’ ol’ man took off an’ Jack after un. Very good. (Tale 003)

Summarizing the results from the Folktales, the distribution of gendered pronouns among the different semantic categories generally does not differ drastically from those observed for the other corpora. While the category of tools and instruments held the highest proportion of non-standard forms in the Southwest corpora, masculine (or feminine) forms referring to those denotata are comparatively rare in both Newfoundland corpora. When excluding vehicles from the calculations, tools and instruments reach 24.7% in the SEDfieldworker notebooks and 23% in the modern material of all gendered pronouns in the Southwest corpora, followed by non-standard forms referring to natural features (15.9% and 16.6% respectively).

In the Newfoundland corpora, on the other hand, the category of accessories contains most instances of non-standard pronouns (29.1% in the

Page 356: A Comparative Grammar of Brtiish English Dialects

346 Susanne Wagner

MUNFLA corpus, 24.9% in the Folktales data). The next four places on this frequency list are occupied by the same categories in both corpora, with tools ranking third, natural features fourth, and buildings and containers ranking second and fifth respectively. (The second place of buildings in the MUNFLA material is probably caused by the house mover’s frequent reference to houses, which slightly skews the picture.)

The close parallels between the corpora from the same area are surprising and rather unexpected in their clarity. The different “priorities” are very likely based on the contrasts in the major occupations of the informants. While most of the Southwest speakers are farmers, the Newfoundlanders are fishermen.

In one domain, however, the Folktales corpus deviates greatly from the other corpora: With over 140 examples, abstract denotata constitute more than a quarter of all non-standard pronominal references in the corpus (145 out of 527, i.e. 27.5%). Although it is possible that the Folktales simply contain more instances of anaphoric references to abstract nouns, another explanation is more likely. Storytelling is a very traditional genre, which includes the use of certain formulaic expressions as well as the use of a very traditional (conservative) language variety. Even though the use of masculine forms referring to abstract nouns is not really a traditional feature, the use of gendered pronouns as such certainly should be classified so. How and why this extension (from only count nouns to abstract nouns) occurred is unclear.

12. Overall summary

or: There’s never two people speaks the same …

(16 He, Lyonshall, book VI)

The analysis of gender assignment in a number of different varieties of (present-day) English proved to be enlightening beyond the expected aspects of the investigated features. The results from these analyses can be divided into different categories, depending on the type of variety, the region, and/or the genre being analyzed – or combinations of these factors.

The results we obtained from our analyses of gender assignment primarily concern the following varieties of English:

– traditional varieties of Southwest England

– traditional varieties of “West Country Newfoundland” (= those parts of Newfoundland that were primarily settled by people of

Page 357: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 347

West Country origin and that were only marginally influenced by other varieties of English or other languages)

– “spoken standard English” (= modern varieties of English in countries with English as a first language and only minimal input from other languages; primarily Great Britain, USA, Canada, Australia, New Zealand)

As this study only dealt with varieties of spoken English, not much will be said about Standard English. As far as the practise of assigning gender in StE is concerned, the myths still presented in grammars should be relativized from what we know about actual speech behaviour. What follows is a brief list of “dos and don’ts” in gender assignment, where the “don’ts” are generally not spelt out in grammars, but rather follow by deduction from the “dos”.

Based on the evidence presented in the preceding sections, it will be obvious to the reader that speakers either obey these rules only to a certain extent or not at all. Rather than the clear-cut division of tasks that StE postulates for the gendered personal pronoun forms, different varieties use different cut-off points along a continuum that ranges from (concrete) count nouns via mass nouns to abstract ideas and situations.

Dos (according to grammars) Don’ts (according to grammars

Use he and she when referring to human beings; it may only be used when the sex of a very young child is unclear

Use it when referring to animals; heand she may be used to refer to pets or animals the speaker has a close relationship with

Don’t use he and she to refer to animals in general

Refer to all things with it Don’t use he or she to refer to an inanimate object

Use it to refer to abstract denotata Don’t use he or she for purposes of “emotional emphasis”

Page 358: A Comparative Grammar of Brtiish English Dialects

348 Susanne Wagner

Figure 8 is an attempt at illustrating these cut-off points for the different varieties of English that were analyzed in this thesis. Letters only stand for the principal availability of the gender for that category in the variety in question, not for overall frequencies of occurrence.

humans animals count nouns (concrete) mass nouns abstract nouns

StE mfmfmfmfNfmNNNfmNNNNfNNNNNNfNNNNNNNfNNNNNspoken standard mfmfmfmfmmNfmmNfmNNmNNfNNNfNNNmNNNfNNfNNNtrad. West Country mfmfmfmmfmmmmmmmmmmmNNNNNNNNNNNNNNNNN mod. West Country mfmfmfmfmNfmfmNmNmfmNmNNNmNNfNmNNfNNmNNNtrad. West C. NF mfmfmfmmfmmmfmNmmNmNmfmNmNNmNNNmNNfNNmNmodern West C. NF mfmfmfmmfmfmNmNmfmmNmfmNmNNmNNNmNNmNfNN

m = masculine forms f = feminine forms N = neuter forms

Figure 8. Gender assignment in varieties of English (based on data from this study)

Certain difficulties emerge when it comes to categorizing the referent types. The major problem is posed by the interaction of the continua count – mass and concrete – abstract. Count nouns can be both concrete and abstract, resulting in a three-dimensional rather than two-dimensional model. However, for reasons of practicality and based on the observations made in the corpora, it was decided to reduce these categories to those which actually play a role in gender assignment. Thus, when count nouns are mentioned below, only concrete count nouns are meant, while abstract nouns are generally abstract count nouns.

Although necessarily very simplified, the most important contrasts between the varieties are nicely illustrated in Figure 8. Let me comment on the details and background assumptions implied:

(Written) Standard English (as presented in grammars) humans masculine or feminine (according to sex) animals supposedly neuter, but also masculine and feminine (not

always according to sex) count nouns boats, countries, etc. can also be feminine ( metaphorical

gender)mass nouns neuter; if anything else, then feminine abstract nouns

neuter; regionally also feminine ( AusE)

Page 359: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 349

Spoken Standard (based on studies presented here) humans masculine or feminine (according to sex) animals generally not neuter, but masculine and feminine (not

always according to sex) count nouns can be masculine and feminine, although feminine forms

seem to be more frequent (cf. studies by Svartengren, Mathiot/Roberts and also Pawley)

mass nouns can be feminine, masculine and neuter, but are predominantly neuter

abstract nouns although generally neuter, often feminine in special constructions ( non-referential she etc.)

Traditional West Country dialects (19th century) humans masculine or feminine (according to sex) animals generally masculine (often even when female cows in

SED)count nouns masculine mass nouns neuter abstract nouns neuter

Modern West Country dialects (oral history) generally a mixture of the traditional West Country and the Spoken Standard system humans masculine or feminine (according to sex) animals generally masculine, but also feminine and occasionally

neuter count nouns still predominantly masculine, but also neuter and

occasionally feminine mass nouns diffusion of traditional system into mass nouns can be,

though rarely, masculine (and feminine SpokenStandard) as well as neuter

abstract nouns see mass nouns, although to an even lesser extent

Page 360: A Comparative Grammar of Brtiish English Dialects

350 Susanne Wagner

Traditional West Country Newfoundland dialect (Folktales) in general, the system is almost identical to that of the traditional West Country; all cut-off points diffuse to the right humans masculine or feminine (according to sex) animals generally masculine, but also feminine and occasionally

neuter count nouns can also be feminine (rare) in addition to masculine and

neuter mass nouns can also be masculine (in addition to neuter) abstract nouns can also be masculine (probably also by extension from

concrete (count) nouns)

Modern West Country Newfoundland dialect (oral history) generally close to traditional Newfoundland dialect, with neuter forms diffusing lefthumans masculine or feminine (according to sex) animals generally masculine, but also feminine and occasionally

neuter count nouns masculine and neuter, but also feminine mass nouns abstract nouns

generally neuter, but also masculine and occasionally feminine abstract nouns

In light of the findings presented here and in the preceding sections, it seems appropriate to return to the term “gender diffusion”, which was used as a label for what was later called gendered pronouns, “non-standard pronominal forms”, etc.

The comparisons of the modern data with the more traditional material revealed certain tendencies both in Southwest England and Newfoundland: The traditional system is slowly dying out, witnessed by fewer and fewer masculine forms in those domains which at one time were their exclusive territory. The formerly obligatory system has developed into an optional system. But on the other hand it seems as if gendered pronouns spread to domains they were formerly not used in ( abstract referents, particularly in the Folktales corpus). Although such uses will probably not be continued, the current stage seems to be a mixture of one, two or all three of the following systems:

Page 361: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 351

1. traditional West Country system – masculine forms for all count nouns; – mass and abstract nouns = neuter; – females (humans, rarely also animals) = feminine

2. Spontaneous Spoken (Standard) system (and elsewhere?) – feminine forms for everything that deserves closer attention, no

matter what its (semantic/biological) status; – animals generally masculine; in addition, – non-referential she to refer to situations; – masculine forms in reference to inanimate objects are rare to

nonexistent

3. (Written) Standard English – gendered references to nouns also acceptable in StE (e.g. feminine forms for boats, guns, cars) are more easily/readily accepted in spoken language than those non-standard pronouns that are “out” according to prescriptivists’ views.

System (1) is the most endangered one, as it is under pressure from both systems (2) and (3). The co-existence of both feminine and masculine forms denoting the same (inanimate) referent(s) leads to mixed – diffused – systems. The Spontaneous Spoken Standard and the StE system slowly invade the lects of those speakers who only would have known their regional dialect some decades ago, but who now cannot help but be influenced by the supraregional standard(s).

Those features found in the non-regional (weaker) but not in the regional (stronger) system may also lead to the extension of the stronger forms to cover all areas of use (e.g. employing masculine forms to refer to abstract referents). Alternatively, speakers may simply employ forms from different systems to cover the whole semantic spectrum (e.g. masculine forms for count nouns, feminine forms for non-referential “situation-type” uses).

Not very surprisingly, the Spoken Standard system in Southwest England seems much closer to StE than the English spoken in Newfoundland. As a result, West Country speakers are losing their traditional system of gender assignment more rapidly than their Newfoundland counterparts. The West Country system is being substituted by the StE one (this can be deduced from the differences between the SED material and the modern West Country oral history data), without the intermediate “Spoken Standard” step that is

Page 362: A Comparative Grammar of Brtiish English Dialects

352 Susanne Wagner

obviously employed in Canada, which leads to an overall slower demise of the traditional system there (and a higher frequency of gendered pronouns).

The one feature that has accompanied us throughout this study is that native speakers of English are obviously not very happy about the ubiquitous status that it has assumed in their mother tongue. Although we can only make educated guesses about the exact reasons behind the individual phenomena, all speakers choose to employ personal pronouns other than “neuter”, non-distinct, semantically empty it if they want to add “feeling” to an utterance, no matter if the emotion to be conveyed is positive or negative. The natural gender system of StE thus seems to be on its way to becoming a pragmatic gender system in the Spoken Standard, where forms marked for gender are used according to the different requirements, emotionality and general circumstances of the situation.

In certain domains of use, the gender chosen by speakers of modern Spoken Standard varieties depends on extralinguistic factors such as the sex of the speaker (particularly relevant for cars, etc.), professional background (gendered pronouns = non-professional), emotional attitude (generally, neuter forms = disinterest, negative attitude), etc. The traditional systems, on the other hand, are based on intralinguistic gender assignment rules (semantic domains of nouns). As both systems can be employed at the same time, a diffusion is not only likely, but expected. As a result of the preference of traditional dialects for masculine and that of modern dialects for feminine forms, the mixture of systems looks like free variation to an outsider. The factors that determine pronoun choice in modern dialects of Southwest England and Newfoundland, and probably in native varieties of English worldwide, are manifold and complex.

Acknowledgments

None of the research presented in this article would have been possible without the help of numerous individuals and organisations. Special thanks are due to Treve Crago, Ann Heeley, Clare Lyall and Norman Rogers for providing access to archives and museums, to Oliver Pickering and Clive Upton for help with and access to the SED material, to Phil Hiscock, Graham Shorrocks, Patricia Fulton and all the MUNFLA staff for their help during my stays in Newfoundland, and of course to everyone in Freiburg for providing the most essential ingredient of all – a stimulating environment where such work can be carried out.

Page 363: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 353

Appendix

(Additional) corpus material

– The British National Corpus (BNC)

– The Freiburg English Dialect Corpus (FRED)

– The Helsinki English Dialect Corpus (pre-final version)

– Parts of (*) and complete transcripts of MUNFLA Tapes by the following collectors (see References for details; partial passages were noted down in situ at the archive)

– Burton 1982 – Churchill, Kean and

Stratton 1978 – Du Pree 1969 – Feener 1983 – Fudge 1969 – Hynes 1977 – Ivany 1974* – Lake 1977 – Lehr 1977 – Mifflin 1972*

– Powell 1976 – Quinton 1972* – Ryan 1985 – Ryan 1978 – Simms 1975 – Squires 1978 – Welsh 1971* – Widdowson 1963* – Wiseman 1976 – Wood 1971 – Yother 1971

Notes

1. The reader interested in details of the Canadian material is referred to Wagner (2003), which includes the full analyses.

2. Although Elworthy claims that all examples given in his books are authentic utterances by native speakers (cf. 1886: viii), the modern reader has obviously no possibility to verify this. Thus, modern narratives in dialect seem not that far removed from Elworthy’s data, particularly since all of Elworthy’s examples are out-of-context utterances, generally not more than one sentence, with no detailed information on speaker background etc. Any statistical analysis of such material cannot be taken seriously, as the comparability of data sets is in no way guaranteed.

Page 364: A Comparative Grammar of Brtiish English Dialects

354 Susanne Wagner

3. Elworthy’s own view on the phenomenon seems much more biased in his later work, where he states that “it is unknown to us in W. Somerset as a neuter pronoun” (1886: xxi), a gross over-generalization and/or over-simplification.

4. See also Ihalainen 1991: 114–115. The link between “more accessible” and “more frequent” is established in, e.g., Keenan (1987: 49): “The frequency with which people relativise in discourse conforms to the [AH], subjects being the most frequent, then direct objects, etc.”

5. “[S]peakers of Cornish in the Modern Cornish period would learn not the ancient Wessex dialect of east Cornwall, Devon and Somerset [...], but a version of English taught them in schools and by the upper classes and better-educated (note that it was the gentry who gave up Cornish and spoke English first), an English deliberately acquired, as distinct from a regional dialect passed on from generation to generation.” (Wakelin 1975: 100)

6. Wakelin’s highly eclectic methods in data selection have been previously criticized; cf. e.g. Klemola (1996: 28) on Wakelin’s treatment of periphrastic do.

7. Note that German does not mark predicative agreement; thus der Mann ist nett “the man is niceØ” and die Frau ist nett “the woman is niceØ”.

8. It is debatable whether or not she provides better semantic agreement than it.Thus, it would probably be more appropriate to speak of a modified Agreement Hierarchy that can be based on numerous factors in addition to the original “likelihood of semantic agreement increases” criterion, e.g. on pragmatic or stylistic factors. Another problem is posed by the question of what exactly connects he and she with who(m), while it is linked with which.Animation seems the most likely factor.

9. As this is the viewpoint adopted by most scholars, it will be assumed here, too. However, as will become clear from the following paragraphs, this view is at least questionable, if not outright wrong.

10. “Asexual” is used here and in the remainder of this study as a label for the third sexual (i.e. biological) category, i.e. male, female, asexual = sex; masculine, feminine, neuter = gender.

11. See, for example, Baron’s (1971) summary of Moore’s (1921) work on the topic. Baron (1971: 119) states that Moore “statistically demonstrates the prevalence of natural gender in OE”. It should also be kept in mind that there were obviously huge regional differences in the gradual process of gender loss, with the North leading the way (cf. Kastovsky 2000: 715; Baron 1971: 127; Dekeyser 1980: 99).

12. Original emphasis. Scholars like Mitchell (1985: 29–31) argue that Moore’s analysis is over-generalized, as his choice of texts is not representative of the respective periods. Still, it should be recalled that in those cases where a

Page 365: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 355

conflict between sex and gender is potentially possible, sex seems to overrule gender more often than not.

13. As English used to have grammatical gender, its loss would have to be classified as one of the rare cases of de-grammaticalization, as previously highly grammatical items lost their grammatical status. On the issue of unidirectionality in grammaticalization, cf. e.g. Hopper and Traugott 1993, ch. 5.

14. This is debatable from a biological point of view, as plants and trees are also animate and non-human.

15. Biber et al.’s description contained a similar remark, but did not follow this path as consequently as Payne and Huddleston. Note that specialist studies such as that by Morris (1991) or Mathiot and Roberts (1979), which will figure prominently in a later section, have long since argued in such a manner.

16. This is the conclusion drawn from the fact that the umpteenth reprint (1962) was published as trade-market paperback as part of Barnes and Noble’s College Outline Series.

17. The figure for “no pronoun recorded” includes irrelevant responses or unanswered questions, i.e. the sum of the total of responses without a pronoun plus the total of responses with a pronoun gives the grand total.

18. One major point of criticism concerns the fact that any use of the fieldworker notebooks depends on the researcher’s interpretation of the – usually exclusively – phonetic material. It should be pointed out in this context that the whole SED (i.e. the published material as well) is ultimately based on these notes, i.e. the interpretations of the individuals who were responsible for the publication. Anyone who has worked with the notebooks in quite some detail will agree that the conventions of transcription and orthographic interpretation varied greatly between different fieldworkers, making it almost impossible to claim that one standard was used for the whole SED.

19. The first line, if in phonetics, is transcribed letter for letter, as found in the notebooks (as far as the typescript was available). Italic script is the author’s ‘translation’ of what was found in phonetic script in the notebooks. The regular script version is the author’s “standard” translation.“#” stands for word stress as indicated by the fieldworker. Some fieldworkers provided orthographic translations while others did not.

20. For unknown reasons, Montacute was excluded from the Basic Material. The dialect may have seemed too conservative even for SED purposes, retaining such “ancient” features as utch for I.

21. Excluding the Isle of Wight, where the fieldwork was done by Michael Barry, whose note-taking style proofed to be highly unsuitable for this investigation – most information taken from his notebooks would have been wishful thinking rather than actual data, as his handwriting is close to illegible.

Page 366: A Comparative Grammar of Brtiish English Dialects

356 Susanne Wagner

22. For lack of space, a detailed description of the dialects under investigation will not be included here. Information on these two regions and their dialects can be found e.g. in Clarke (2004) on Newfoundland and Wagner (2004c) on the Southwest.

23. In some of the interviews, a spouse or child was present during the recording and gave the occasional comment. However, there is only one main speaker in most of the interviews. The same is true for the material from the other counties.

24. The discrepancies between the totals in the age and recording date column are due to the fact that this information was not always available.

25. As transcription is a time-consuming task, and the amount of new material coming in is much greater than the production of transcripts, MUNFLA has to deal with an approximate 10-year backlog. Thus, the most recent interviews used here stem from the late 1980s.

26. Cf. also Klemola (1996: 39), who argues in a similar vein: “The formal nature of the questionnaire interview situation probably did not encourage the use of [non-standard] features [...], whereas the incidental material contains utterances that the fieldworkers picked up from their more informal conversations with the informants.”

27. It is absolutely unclear how the Australian and New Zealand uses she’s rightetc. can be classified as personification in the quoted dictionary entry – the personal pronoun is usually non-referential, excluding personification as an explanatory factor.

28. Morris (1991) came to similar results; see section 6.2.1. It is unclear why the results for the two birds (canary and parakeet) differ to such a large extent. We can assume that a parakeet is more pet-like than a canary, which would explain the comparatively high output of it for the latter. Why she and heoccur with almost identical frequencies in example (2d) remains unclear, though.

29. A search for “dog” and “cat” was run, restricted to the spoken sub-corpus. The total figures are incidental, based on the printed output of eight pages after marking those cases where pronouns occurred in the “maximum scope” setting.

30. The transcriber probably misunderstood; “force” is more likely in this context, as the police units with dogs are usually part of a special force.

31. The first pronoun (she) most likely refers to the weather in general, while the second one (it) refers to the wind specifically, which the authors identified as referent in both cases.

32. The primary distinction made by Mathiot and Roberts is between human and non-human, not animate vs. inanimate as assumed in the earlier sections. However, this difference is of no consequence here. As explained in section 5.2, animals should be treated separately anyway.

Page 367: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 357

33. This concept will underlie all future uses of the term involvement in this thesis. It should be treated separately from the more pragmatic meaning (ensuring collaboration between speakers; thus e.g. in Cheshire 1997). While the latter involves inferencing, negotiating of meaning, etc., the former is purely referring to the individual speaker’s sentiments (e.g. sympathy or antipathy) without involving the addressee.

34. It is unclear how Siemund (2001) can detect a system in the pattern. It is not surprising that he concludes (Siemund 2001: 104) that the men’s pattern is similar to the one found for Tasmanian English – the Tasmanian informants are almost exclusively males (cf. Pawley 2004). However, this nicely supports the present author’s view that we are essentially dealing with a pattern of spoken English in general rather than a (regionally restricted) variety-specific pattern.

35. “[E]motional interest that is mirrored by the feminine gender” (Svartengren 1927: 110); “familiarity and the feeling of companionship between an artisan and his tools” (Svartengren 1927: 110), etc. What Svartengren means seems to be some form of “personal involvement” as used by Mathiot and Roberts rather than personification in its strict sense.

36. Morris’ database for this category is very small. Of the approximately 1,500 examples which constitute her overall database, only 80 instances of masculine pronominalization of biological inanimates could be found. These include 15 instances of personification and about 30 examples from other authors’ studies (cf. Morris 1991: 166).

37. As mentioned above, the interviewer’s influence is minimal in all of the chosen texts; accordingly, the scattergrams only include the informant’s switches. Included here are C 626, C 627, C 628, C 631, C 751, C 1187/8, and C 2914 from the MUNFLA material, and Som_001, Som_009, Som_014, Som_021, Som_022 and Som_028 from the Somerset material. The additional figures not included in the tables above on which the scattergrams are based are:

text it/total switches he-it he/total switches it-he C 626 0.53 0.34 0.47 0.36 C 627 0.57 0.39 0.43 0.27 C 628 0.42 0.29 0.58 0.375 C 631 0.54 0.32 0.46 0.27 C 751 0.38 0.33 0.62 0.45

C 1187/8 0.07 0.06 0.93 0.63 C 2914 0.31 0.14 0.69 0.3

Som_001 0.71 0.27 0.29 0.14 Som_009 0.58 0.36 0.42 0.29 Som_014 0.43 0.31 0.57 0.41 Som_021 0.51 0.30 0.49 0.30 Som_022 0.42 0.18 0.58 0.25 Som_028 0.14 0.10 0.86 0.63

Page 368: A Comparative Grammar of Brtiish English Dialects

358 Susanne Wagner

38. Individual rations: Monmouthshire 0-25% (1 of 4), Worcestershire 20%-40% (1 of 5; 2 of 4), Herefordshire 17%-57% (1 of 6; 4 of 7), Gloucestershire 50%-83% (3 of 6; 6 of 7).

39. Individual county ratios: 50% for Monmouthshire and Worcestershire (1 of 2; 2 of 4), 60% in Herefordshire (3 of 5), 100% in Gloucestershire (4 of 4).

40. Individual county ratios: 20% in Berkshire (1 of 5), 25% in Somerset (3 of 12), 75% in Wiltshire (6 of 8), 80% in Dorset (4 of 5), 85.7% in Hampshire (6 of 7), 100% in Cornwall and Devon (7 of 7; 11 of 11).

41. Individual county ratios: Berkshire 25% (1 of 4), Sussex 50%(!) (1 of 2), Somerset and Hampshire 83.3% (10 of 12; 5 of 6), Wiltshire 87.5% (7 of 8), Cornwall, Devon and Dorset 100% (7 of 7, 11 of 11, 5 of 5).

42. Individual county ratios: 27.3% Somerset (3 of 11), 28.6% Wiltshire (2 of 7), 77.8% Devon (7 of 9), 85.7% Cornwall (6 of 7), 100% Dorset and Hampshire (5 of 5 each).

43. Individual ratios (first figures for question IX.2.6, second for IX.2.8): total incl. Southeast 46.9% (15 of 32), excl. Southeast 55.6% (15 of 27); Devon 37.5% (3 of 8), Wiltshire 40% (2 of 5), Hampshire 50% (3 of 6), Somerset 83.3% (5 of 6), Cornwall and Dorset 100% (1 of 1 each); total incl. Southeast 63.2% (36 of 57), excl. Southeast 81.4% (35 of 43); Sussex 20% (1 of 5), Berkshire 25% (1 of 4), Hampshire 50% (2 of 4), Wiltshire 71.4% (5 of 7), Cornwall 83.3% (5 of 6), Somerset, Devon, Dorset 100% (8 of 8, 10 of 10, 4 of 4).

44. Individual county ratios: Berkshire 50% (1 of 2), Dorset and Hampshire 66.7% (2 of 3, 4 of 6), Somerset 70% (7 of 10),Wiltshire 75% (3 of 4), Cornwall and Devon 100% (6 of 6, 8 of 8).

45. Individual ratios: Somerset 46.2% (6 of 13), Dorset 50% (3 of 6), Wiltshire 60% (6 of 10), Hampshire 71.4% (5 of 7), Devon 81.8% (9 of 11), Cornwall 100% (7 of 7).

46. Although great care has been taken in reproducing the examples from the hand-written notes in the fieldworker notebooks, misinterpretations and mistakes cannot be ruled out completely. However, it is hoped that those only affect the parts that are irrelevant to the present analysis. Where certain phonetic symbols where either not available to me or could not be interpreted in positions relevant for the analysis, this is generally mentioned. As the notebooks where consulted three times in as many years, certain inconsistencies and “developments” on my part cannot be excluded. Interesting (dialectal) lexical items will not be commented on in the examples.

47. I am aware of the fact that in other cultures this distinction may be non-existent or that the cut-off points may differ. “Plant” is taken in its popular rather than scientific sense (which would include vegetables as a sub-category).

Page 369: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 359

48. Recall Elworthy’s criterion for gendered pronouns: the noun in question has to be a “definite noun, i.e. the name of a thing or object which has a shape of its own”, which would exclude abstracta (Elworthy 1877: 32).

49. “’Twas an American steamer in camouflage. He went by her he turned, come back by her again, didn’t know ’twas a trick on.” The steamer (“he”) is passing by a small boat with survivors from a sunken ship (“she”).

50. This theory is supported by data from other regions. 51. This is an example from the pre-final version of the Helsinki Dialect Corpus.

I would like to thank Kirsti Peitsara for granting us access to the material in situ at Helsinki University.

52. Among the features that occur in the interview are the Cornish habitual belong (see Hancock 1994), periphrastic do, purposive for to, pronoun exchange, non-standard use of prepositions (on for of, “directional” prepositions), non-standard use of relative pronouns, agreement features, regularized verb paradigms, present tense used in the extended-now sense usually (in StE) expressed by the present perfect (he’s dead years).

53. Detailed distribution: Two speakers contribute one example each, four speakers two, three three, four four, two five, two six, one seven, one eight, one nine, three ten, one 12, one 14, two 15, two 25, one 41.

54. Text AH: 15.9% of examples, 6.3% of words; significant at < 0.001%-level (chi-square); SRLM 44: 5.4% of examples, 1.6% of words; significant at < 0.001%-level; SRLM 224: 9.7% of examples, 4.1% of words; significant at 0.002%; SRLM 108: 9.7% of examples, 4.7% of words; significant at < 0.05%-level.

55. 615 forms in ca. 320,000 words by 68 speakers in the Southwest England corpus, 885 forms in ca. 132,000 words by 34 speakers in the MUNFLA corpus. Even when excluding the 300 or so boat references, the MUNFLA material contains proportionally more gendered pronouns than the Southwest corpus. Table 48 does not include the 20 or so examples that stem from interviews which do not belong to the core corpus as described in Section 4.

56. Though house seems to be the most likely referent when analyzing this example superficially, it may be recalled that this is the speaker who uses more than 100 masculine forms referring to houses in the remainder of the interview. It is thus extremely unlikely that he should here use a lone feminine.

57. Due to the nature of the data chosen for the analysis, questions of a sociological nature such as influences of age, sex, education, urban vs. rural, etc. could not be addressed at all. Moreover, sophisticated statistical analyses are often not possible due to small sample size. To remedy this situation, future studies would have to be based on material collected to fit these requirements. As basically no other studies on the phenomenon were available at the time of writing, the present author chose to tackle the basic

Page 370: A Comparative Grammar of Brtiish English Dialects

360 Susanne Wagner

issues first. A follow-up study could investigate those areas that turned out to be a promising field for future work.

References

Anderwald, Lieselotte and Susanne Wagner forthc. The Freiburg English Dialect Corpus (FRED) - Applying corpus-

linguistic research tools to the analysis of dialect data. In Using Unconventional Digital Language Corpora. Vol. I: Synchronic Cor-pora, Joan Beal, Karen Corrigan and Herman Moisl (eds.). Basing-stoke: Palgrave MacMillan.

Attwell, James 1987 Dorset Dialect Days. Sherborne: Dorset Publishing Company. Barnes, William 1886² A Glossary of the Dorset Dialect with a Grammar. Guernsey/St.

Peter Port: Toucan Press. 1844 A dissertation on the Dorset dialect of the English language. In

William Barnes - Selected Poems, Andrew Motion (ed.), 117–138. Hardmondsworth: Penguin.

Baron, Naomi S. 1971 A reanalysis of English grammatical gender. Lingua 27: 113–140. Biber, Douglas, Stig Johansson, Geoffrey Leech, Susan Conrad, and Edward

Finegan 1999 Longman Grammar of Spoken and Written English. London:

Longman. Birner, Betty J. and Gregory Ward 1998 Information Status and Noncanonical Word Order in English.

Amsterdam/Philadelphia: Benjamins. Bremann, Rolf 1984 Soziolinguistische Untersuchung zum Englisch von Cornwall. Frank-

furt: Lang.Brinton, Laurel J. 2000 The Structure of Modern English. Amsterdam/Philadelphia: Ben-

jamins. Burchfield, R.W. and H.W. Fowler (eds.) 1998³ The New Fowler’s Modern English Usage. Oxford: Oxford

University Press. Burton, Terry 1982 MUNFLA ms 82-319/C10260, 89pp. Chambers, Jack K. and Peter Trudgill 1998² Dialectology. Cambridge: Cambridge University Press.

1970

1994

Page 371: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 361

Cheshire, Jenny 1997 Involvement in ‘standard’ and ‘nonstandard’ English. In Taming the

Vernacular. From dialect to written standard language, Jenny Cheshire and Dieter Stein (eds.), 68–82. London/New York: Longman.

Churchill, Shirley, Betty Kean, and Nita Stratton 1978 MUNFLA transcripts of Tapes 79-037/C3831-32. Clarke, Sandra 2004 Newfoundland English: morphology and syntax. In A Handbook of

Varieties of English. Vol. II: Morphology and syntax, Bernd Kortmann and Edgar W. Schneider (eds.), 303–318. Berlin/New York: Mouton de Gruyter.

Comrie, Bernard 1989² Language Typology an Linguistic Universals. Chicago: University

of Chicago Press. Corbett, Greville 1991 Gender. Cambridge: Cambridge University Press. Crystal, David 1995 The Cambridge Encyclopedia of the English Language. Cambrigde:

Cambridge University Press. Curme, George O. 1962 English Grammar. New York: Barnes & Noble. [rpt.; 1925; 1947]. Curzan, Anne 2000 Gender categories in early English grammars: Their message to the

modern grammarian. In Gender in Grammar and Cognition. I Approaches to gender. II Manifestations of gender, Barbara Unterbeck (part I), Terttu Nevalainen and Mirja Saari (eds.), 561–576. Berlin/New York: Mouton de Gruyter.

Dartnell, George Edward and Rev. Edward Hungerford Goddard 1893 A Glossary of Words Used in the County of Wiltshire. London:

Oxford University Press. [Publications of the English Dialect Society 69; Vaduz: Kraus Reprint Ltd., 1965].

Dekeyser, Xavier 1980 The diachrony of gender systems in English and Dutch. In Historical

Morphology, Jacek Fisiak (ed.), 97–111. The Hague: Mouton. Downes, John 1986 A Dictionary of Devon Dialect. Padstow: Tabb House. Du Pree, Victor 1969 MUNFLA Tapes 70-003/C626-33. Elliott, Ralph 1984 Thomas Hardy’s English. Oxford: Blackwell.

Page 372: A Comparative Grammar of Brtiish English Dialects

362 Susanne Wagner

Ellis, Alexander J. 1889 On Early English Pronunciation. Vol. V. Existing dialectal as

compared with West Saxon pronunciation. London: Trübner. Elworthy, Frederic Thomas 1877 An Outline of the Grammar of the Dialect of West Somerset.

London: Trübner & Co. [Publications of the English Dialect Society 19; Vaduz: Kraus Reprint Ltd., 1965].

1886 The West Somerset Word-book. London: Trübner & Co. [Publications of the English Dialect Society 50; Vaduz: Kraus Reprint Ltd., 1965].

Feener, David Cyril 1983 MUNFLA ms 83-289/11pp. Fischer, Andreas 1976 Dialects in the South-West of England: A Lexical Investigation.

Bern: Francke.Francis, W. Nelson 1983 Dialectology. An Introduction. London/New York: Longman. Fudge, Jesse 1969 MUNFLA Tape 70-037/C751. Halpert, Herbert and John D.A. Widdowson 1996 Folktales of Newfoundland – The Resilience of the Oral Tradition.

St. John’s: Breakwater. Hancock, Ian F. 1994 Componentiality and the creole matrix: The Southwest English

contribution. In The Crucible of Carolina. Essays in the Develop-ment of Gullah Language and Culture, Michael Montgomery (ed.), 95–114. Athens/London: University of Georgia Press.

Hirooka, Hideo 1980 Thomas Hardy’s use of dialect. no place: Shinozaki Shorin Press

(Japan).Hopper, Paul J. and Elizabeth Closs Traugott 1993 Grammaticalization. Cambridge: Cambridge University Press. Howe, Stephen 1996 The Personal Pronouns in the Germanic Languages. Berlin/New

York: Mouton de Gruyter. Huddleston, Rodney and Geoffrey K. Pullum 2002 The Cambridge Grammar of the English Language. Cambridge:

Cambridge University Press. Hynes, Derek 1977 MUNFLA transcript of Tape 77-187/C2654 and ms, 55pp.

1965a

1965b

Page 373: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 363

Ihalainen, Ossi 1985a He took the bootle and put ‘n in his pocket: The object pronoun it in

present-day Somerset. In Focus on: England and Wales, Wolfgang Viereck (ed.), 153–161. Amsterdam: Benjamins.

1985b Synchronic variation and linguistic change: evidence from British English dialects. In Papers from the 4th International Conference on English Historical Linguistics, Amsterdam, 10-13 April 1985, Roger Eaton , Olga Fischer, Willem Koopman, and Frederike van der Leek (eds.), 61–72. Amsterdam: Benjamins.

1991 On grammatical diffusion in Somerset folk speech. In Dialects of English. Studies in grammatical variation, Peter Trudgill and Jack K. Chambers (eds.), 104–119. London/New York: Longman.

Ivany, R.M. 1974 MUNFLA transcripts of Tapes 74-039/C1520-21. Jennings, James K. 1869 The Dialect of the West of England, particularly Somersetshire.

London: John Russell Smith. Jones, Charles 1988 Grammatical Gender in English: 950–1250. London: Croom Helm. Jones, Malcolm and Patrick Dillon 1987 Dialect in Wiltshire. Trowbridge: Wiltshire County Council. Kastovsky, Dieter 2000 Inflectional classes, morphological restructuring, and the dissolution

of Old English grammatical gender. In Gender in Grammar and Cognition. I Approaches to gender. II Manifestations of gender,Barbara Unterbeck, (part I), Terttu Nevalainen and Mirja Saari (eds.), 709–727. Berlin/New York: de Gruyter.

Keenan, Edward L. 1987 Variation in Universal Grammar. In Universal Grammar: 15 Essays,

Edward L. Keenan, 46–59. London: Croom Helm. Keenan, Edward L. and Bernard Comrie 1977 Noun phrase accessibility and Universal Grammar. Linguistic

Inquiry 8: 63–99. Klemola, Juhani K. 1996 Non-standard periphrastic do: a study in variation and change. PhD

thesis, Department of Language and Linguistics, University of Essex.

Kruisinga, Etsko 1905 A Grammar of the Dialect of West Somersetshire: Descriptive and

Historical. Bonn. [Bonner Beiträge zur Anglistik, Heft 18]

Page 374: A Comparative Grammar of Brtiish English Dialects

364 Susanne Wagner

Labov, William 1972 Sociolinguistic Patterns. Philadelphia: University of Pennsylvania

Press.Lake, Sharon 1977 MUNFLA transcript of Tape 78-024/C3162. Lehr, Genevieve 1977 MUNFLA transcript of Tape 78-006/C3321. Leisi, Ernst and Christian Mair 19998 Das heutige Englisch: Wesenszüge und Probleme. Heidelberg:

Winter. Linn, Michael D. 1993 Appendix: resources for research. In American Dialect Research,

Dennis R. Preston (ed.), 433–448. Amsterdam: Benjamins. MacKay, Donald G. and Toshi Konishi 1980 Personification and the pronoun problem. Women’s Studies

International Quarterly 3: 149–163. Marcoux, Dell R. 1973 Deviation in English gender. American Speech 48: 98–107. Marten, Clement 1973 The Devonshire Dialect. Exeter: Clement Marten. Mathiot, Madeleine and Marjorie Roberts 1979 Sex roles as revealed through referential gender in American

English. In Ethnolinguistics: Boas, Sapir and Whorf Revisited,Madeleine Mathiot (ed.), 1–47. The Hague: Mouton.

Mifflin, Robert 1972 MUNFLA transcript of Tape 72-184/C1228. Mitchell, Bruce 1985 Old English Syntax. Oxford: Clarendon. Moore, Samuel 1921 Grammatical and natural gender in Middle English. Publications of

the Modern Language Association of America (PMLA) 36: 79–103. Morris, Lori 1991 Gender in modern English: The system and its uses. PhD thesis,

Université Laval, Quebec. Orsman, H.W. (ed.) 1997 The Dictionary of New Zealand English. Auckland: Oxford

University Press. Orton, Harold 1962 Survey of English Dialects - Introduction. Leeds: Edward Arnold. Pawley, Andrew 2002 Using he and she for inanimate referents in English: questions of

grammar and world view. In Ethnosyntax. Explorations in Grammar

Page 375: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 365

and Culture, Nick J. Enfield (ed.), 110–137. Oxford: Oxford University Press.

2004 Australian Vernacular English: some grammatical characteristics. In A Handbook of Varieties of English. Vol. II: Morphology and syntax, Bernd Kortmann and Edgar W. Schneider (eds.), 611–642. Berlin/New York: Mouton de Gruyter.

Perry, F.C. 1921 The dialect of North Somerset. Transactions of the Yorkshire Dialect

Society 4 (22): 17–41. Poplack, Shana 1979 Function and process in a variable phonology. PhD thesis,

University of Pennsylvania, Philadelphia. 1980 The notion of the plural in Puerto Rican Spanish: Competing

constraints on (s) deletion. In Locating Language in Time and Space, William Labov (ed.), 55–67. New York: Academic Press.

Poplack, Shana and Sali Tagliamonte 1993 The zero-marked verb: Testing the creole hypothesis. Journal of

Pidgin and Creole Languages 8: 171–206. 1996 Nothing in context: Variation, grammaticization and past time

marking in Nigerian Pidgin English. In Changing meanings, changing functions. Papers relating to grammaticalization in contact languages, Philip Baker and Anand Syea (eds.), 71–94. London: University of Westminster Press.

Potter, Mary C. and Linda Lombardi 1998 Syntactic priming in immediate recall of sentences. Journal of

Memory and Language 38: 265–282. Powell, Clifford 1976 MUNFLA Tape 76-290/C2894. Quinton, Leith D. 1972 MUNFLA transcripts of Tapes 72-199/C1154-55. Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech, and Jan Svartvik 1985 A Comprehensive Grammar of the English Language. London:

Longman. Ramson, W.S. (ed.) 1988 The Australian National Dictionary. Melbourne: Oxford University

Press.Ross, Alan R. C. 1936 Sex and gender in the Lindisfarne Gospels. Journal of English and

Germanic Philology 35: 321–330. Ryan, Diane 1985 MUNFLA ms including transcript of Tape 85-154/C12247.

Page 376: A Comparative Grammar of Brtiish English Dialects

366 Susanne Wagner

Ryan, Patricia F. 1978 MUNFLA transcript of Tape 79-341/C4719 and manuscript, 71pp. Sankoff, David and Suzanne Laberge 1978 Statistical dependence among successive occurrences of a variable in

discourse. In Linguistic Variation. Models and Methods, David Sankoff (ed.), 119–126. New York: Academic Press.

Scherre, Maria Marta Pereira and Anthony J. Naro 1991 Marking in discourse: ‘Birds of a feather’. Language Variation and

Change 3: 23–32. Siemund, Peter 2001 Pronominal gender in English. A study of English varieties from a

cross-linguistic perspective. Postdoctoral thesis, Fachbereich Philosophie und Geisteswissenschaften, Institut für Englische Philologie, Freie Universität, Berlin.

2002 Animate pronouns for inanimate objects: Pronominal gender in English regional varieties. In Anglistentag 2001 Wien - Proceedings,Dieter Kastovsky, Gunther Kaltenböck, and Susanne Reichl (eds.), 19–34. Trier: Wissenschaftlicher Verlag.

forthc. Pronominal Gender in English: A Study of English Varieties from a Cross-Linguistic Perspective. London: Routledge.

Simms, Garfield 1975 MUNFLA Tapes 75-164/C2829-31. Squires, Shirley 1978 MUNFLA transcript of Tape 78-427/C4220 and manuscript, 43pp. Svartengren, T. Hilding 1927 The feminine gender for inanimate things in Anglo-American.

American Speech 3: 83–113. 1928 The use of the personal gender for inanimate things. Dialect Notes 6:

7–56. 1954 The use of feminine gender for inanimate things in American

colloquial English. Moderna Sprak 48: 261–292. Szmrecsanyi, Benedikt 2004 Persistence phenomena in the grammar of spoken English. PhD

thesis, Englisches Seminar, Albert-Ludwigs-Universität, Freiburg. Tree, Jean E. Fox and Paul J.A. Meijer 1999 Building syntactic structure in speaking. Journal of Psycholinguistic

Research 28: 71–92. Trudgill, Peter 19992 The Dialects of England. Oxford: Blackwell. Viereck, Wolfgang, Karin Viereck and Heinrich Ramisch 2002 dtv-Atlas Englische Sprache. München: dtv.

Page 377: A Comparative Grammar of Brtiish English Dialects

Gender in English pronouns 367

Visser, Frederikus Th. 1963 An Historical Syntax of the English Language. Part I: Syntactical

units with one verb. Leiden: Brill. Wagner, Susanne 2001 Pronoun exchange – a feature of English dialects? Unpublished

manuscript. 2003 Gender in English pronouns. Myth and reality. PhD thesis,

Englisches Seminar, Albert-Ludwigs-Universität, Freiburg. http://www.freidok.uni-freiburg.de/volltexte/1412

2004a English in Cornwall – a re-evaluation. Paper presented at the 15th Sociolinguistics Symposium, Newcastle upon Tyne, April 1-4 2004.

2004b ‘Gendered’ pronouns in English dialects – A typological perspective. In Dialectology Meets Typology: Dialect Grammar from a Cross-Linguistic Perspective, Bernd Kortmann (ed.), 479–496. Berlin/New York: Mouton de Gruyter.

2004c English dialects in the Southwest: morphology and syntax. In AHandbook of Varieties of English. Vol. II: Morphology and syntax,Bernd Kortmann and Edgar W. Schneider (eds.), 154–174. Berlin/New York: Mouton de Gruyter.

Wakelin, Martyn F. 1975 Language and History in Cornwall. Leicester: Leicester University

Press. 19812 English Dialects: An introduction. London: Athlone Press. 1986 The Southwest of England. Amsterdam: Benjamins. Wales, Katie 1996 Personal Pronouns in Present-Day English. Cambridge: Cambridge

University Press. Welsh, Kinsley 1971 MUNFLA transcripts of Tapes 71-131/C1033-35. Widdowson, J.D.A. 1963 MUNFLA transcripts of Tapes 63-002/C0005, C0010-11. Wiseman, Sherman 1976 MUNFLA Tapes 76-295/C2913-14. Wood, Brian 1971 MUNFLA Tapes 72-089/C1187-88. Yother, Larry 1971 MUNFLA Tapes 73-046/C1958-59, C1961.

Page 378: A Comparative Grammar of Brtiish English Dialects

Index

Accessibility Hierarchy 2, 21, 48–70

adverbial relative clause 50, 112 affix loss 173–175 agreement

verbal see verbal concord analogy 177–178 analytical genitive 59, 111 and-coordination 61, 79 animals (as referents) 253–259

in CanE 270–271 apparent-time variation 158

BE see also verbal concord finite be forms 136, 140, 142 generalized am 140 generalized is 141

Belfast 181–186 British National Corpus (BNC) 14,

23, 112, 256, 353

Cheshire 138 Chester-Wash line 138 clauses

existential 129, 138, 156–161 interrogative 129, 161, 165–168 locative inversion 129 relative 130, 161, 164 tags 161, 166–167

cleft 62–70, 112, 130, 164 pseudo-cleft 62–67, 70, 93

cliticization 177 Cognitive Grammar 192–193 concord see verbal concord ‘connector’ which 21, 88–93 Construction Grammar 193 contact

Old English and Brythonic 173 Old English and Scandinavian 176

coordinate construction constraint 79

copies 70, 116 Cornwall 218, 220, 239–241, 245–

246, 251, 291, 312–322, 327, 330–338, 354, 358

Cumberland 135, 141, 150

Derbyshire 138, 150 Devon 211, 239, 241, 244–245,

251, 315–324, 327, 330, 333– 338, 354, 358

Dorset 216–218, 241, 246, 315–318, 337, 358

Durham (county) 135, 150

Early Modern English 40, 57, 178 elided preposition 45, 51, 88, 90, 93 English Dialect Grammar 29 entrenchment 192–193 existential 40, 63–71, 75, 76, 91

focussing 62–64 Folktales of Newfoundland 242,

248–251, 259, 342–346, 350 fossilization 171 Freiburg English Dialect Corpus

(FRED) 8–22, 132–133, 140–142, 154, 212, 319–338, 353 (dis)advantages of orthographic

transcripts 10–11 corpus design 11–13 principles of compilation 4–6 role of oral history 6–7, 13–15

frequency 192–193 fronting 45–47, 93 see also pied-

piping gapping 2 see also zero subject

relative clause

Page 379: A Comparative Grammar of Brtiish English Dialects

Index 369

genderhistory of, in English 225–226 in PrDE 227–235 in the world's languages 222–225

gender assignment in American English 261–270 in Tasmanian Vernacular English

274genitive avoidance 59–62 grammaticalization

of existential there 156

habitual 131 have

forms in -n 140 Herefordshire 140 Hierarchy Constraints 48–49 historic present 146

in Irish English 146

idioms 171 inherent variation 191 Irish English 146

Lancashire 131, 138, 147–150 Late Modern English 40 left dislocation 61–62 Lexical-Functional Grammar 186–

189‘lexically empty’ antecedent 62, 65–

70, 112 Lincolnshire 144, 150, 178 Lowman Survey 29–31, 35, 54

markedness 180–181, 186–189, 256, 275, 287, 342

microparametric variation 181–186 Middle English 40, 42, 58, 65, 75–

76, 91, 112, 225–226 dialect boundaries 162 Midlands dialects 175 Midlands verbal systems 140 northern dialects 173–179

southern dialects 176 modification 62, 67, 88 MUNFLA 242, 247–251, 339–340

narrative present see historic present negation 2–3

and verbal concord 151, 156, 160 negative polarity items 184 Newfoundland 211–214, 235, 242,

247–250, 253, 264, 272–277, 280–284, 293, 338–342, 345, 350–352

Newfoundland English, varieties of see Newfoundland

nonreduction 21, 85–86 Northern Ireland see Ulster Northern Ireland Transcribed Corpus

of Speech (NITCS) 23, 111, 132, 144, 149, 158, 170

Northern Subject Rule 124–199 seealso verbal concord Celtic substrate hypothesis 173 conserved only with forms of BE

150, 169 definition 128 extension to BE 178 extension to forms of BE 149 history 173–179 markedness relations 180–181 Position-of-Subject Constraint

129, 131, 144–146, 185–190 southern boundary 178 southern limits 162–164 Type-of-Subject Constraint 129

Northumberland 135, 142, 150–151 Nottinghamshire 144 noun

count 211, 214–222, 277–280, 293–296, 303, 305, 310–313, 326–327, 340, 346–351

mass 211, 216–220, 249, 277–280, 293, 302–303, 312, 314, 347–351

Page 380: A Comparative Grammar of Brtiish English Dialects

370 Index

Old English 65, 213, 224–226, 234 Northumbrian dialect 173–177 West Saxon dialect 177

Old Norse 174, 176

paratactic (construction) 30–31, 40, 60–61, 73–75, 92–93

partitive genitive 43–44, 47–48, 75–76, 90, 112

personal(ity)/nonpersonal(ity) 41–45

personification 232–235, 252–254, 262, 264, 269–270, 306, 308, 310, 356–357

pied-piping 45–46, 76 see also fronting

Position-of-Subject Constraint seeNorthern Subject Rule

possessive have or get construction 60–61

Present-Day English 212–213, 224–227, 230, 233–235, 338

Principles-and-Parameters theory 146, 181–186

pronoun retention 21, 52, 71–79, 95–96

pronounsdemonstrative 165 indefinite 165, 168 strong subject forms 183

referent types abstract referents 269, 339–340,

343–345body parts 309–310, 330–331 feminine 311–312 man-made objects 294–306, 320–

328, 339, 342 masculine 293–311 nature 306–309, 328–330

restrictive(ness)/nonrestrictive(ness) 38–41, 77–78, 86, 96, 111

resumptive (pronoun) 48, 60, 70–86, 88, 90–96

Scotland 154 singular there’re 156–159

Scots 170 see also Ulster Scots Older 150

SED see Survey of English Dialects she

non-referential 251, 259–261, 268, 271–272, 340, 344, 349, 351

Shropshire 138, 140 Somerset 215–218, 239–245, 251,

280–283, 291, 315–322, 327–328, 330–338, 354, 357–358

Southwest England 211–214, 275, 287, 293, 319, 339, 342, 346, 350–352, 359 see also West Country

Staffordshire 138 Standard English 2–3, 22, 28, 38,

41, 46–47, 50, 55, 58, 63–68, 76, 78, 81, 84, 94–96

standard variety see Standard English

stochastic grammars 191 structural markedness 72 Survey of English Dialects 355 Survey of English Dialects (SED) 4,

31–34, 37, 54, 56, 59, 94, 111, 132–135, 138, 144, 161, 213–214, 220–221, 235–243, 247, 250–251, 319–322, 325–327, 331–337, 345, 349, 351, 355 Basic Material 285–292 Fieldworker notebooks 292–319

Tape-Recorded Survey of Hiberno-English Speech (TRS) 132–133

Teesside 141–142 thae (demonstrative) 165 there (existential marker) 129, 156–

161

Page 381: A Comparative Grammar of Brtiish English Dialects

Index 371

they as demonstrative – see thaeloan from Scandinavian 177

thou–you distinction 135–136, 142–144

topicalization structures 62–68 Tyneside 151 Type-of-Subject Constraint see

Northern Subject Rule

Ulster 144, 149, 154, 170 communal variation 159 singular there’re 156–159 Ulster Scots 157, 159

Ulster Scots 170

Varbrul 132, 160, 170 verbal concord

formalist analyses 146, 179–190 in ‘Central North’ 135, 141–142,

150, 154 in ‘Lower North’ 137, 155, 158 in ‘Upper North’ 136, 142 in East Midlands 136 in Middle English 162, 173–179 in Northwest Midlands 136–142,

150in Old English 173–177 in Scotland 138, 154, 170 in Southwest England 147, 179 in Ulster 138, 149, 154, 170, 181–

186in West Midlands 136, 142 -n forms 136, 138–140, 151 past tense, in Old English 149 present-tense be 136, 140–142 was–were 136, 142, 149–155, 158 with coordinated verbs 131, 144–

146

with demonstrative pronouns 165 with existential clauses 129, 138,

156–161with habitual clauses 131, 147–

148with historic present 146 with indefinite pronouns 165, 168 with intervening adverbs 130, 146,

160, 182 with inversion 129, 161, 165–168,

182with locative inversion 129 with negation 151, 156, 160 with relative clauses 130, 164 with thou 142–144

vernacular primitives 179

Welsh 173 West Country 211, 218, 221, 244–

245, 250, 254, 264–265, 272–273, 277, 291, 293, 305, 310–320, 323, 330, 332, 336–342, 346–351 see also Southwest England

Westmorland 135, 150 Wiltshire 217, 239, 242–244, 251,

299, 315–322, 325, 327, 330, 335–338, 358

Worcestershire 140 Word Grammar 189–190

York 151, 170 Yorkshire 131, 138, 147–150, 166

zero relatives see zero subject relative clause

zero subject relative clause 21, 29, 35–39, 48, 55, 62–70, 77, 79, 84, 89, 94–95, 111–112

Page 382: A Comparative Grammar of Brtiish English Dialects
Page 383: A Comparative Grammar of Brtiish English Dialects
Page 384: A Comparative Grammar of Brtiish English Dialects