This dissertation has been microfilmed exactly as received 69-4875

DONAHUE, Thomas Scott, 1940- THE PRESENT STATE OF MIDDLE ENGLISH DIALECT STUDIES.

The Ohio State University, Ph.D., 1968 Language and Literature, linguistics

University Microfilms, Inc., Ann Arbor, Michigan

^ Copyright by Thomas Scott Donahue 1969 THE PRESENT STATE OF MIDDLE ENGLISH DIALECT STUDIES

DISSERTATION Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

by Thomas Scott Donahue, B.A., M.A.

******

The Ohio State University 1968

Approved by

/-At- 'Adviser Department of English ACKNOWLEDGMENTS

I wish to express my gratitude to Professor F . Utley for guidance, and to K. J. Davis for encouragement. VITA

August 29, 1940 Born — Youngstown, Ohio 1962 B.A. with Honors, Denison University, Granville, Ohio 1962-1965 NDEA Fellow, Miami University- Ohio State University 1964 M.A., Miami University, Oxford, Ohio 1965-1968 Teaching Assistant, Department of English, The Ohio State University, Columbus, Ohio

Field of Specialization Middle English Language and Literature. Professor Francis Lee Utley TABLE OF CONTENTS

Page

Acknowledgments ii Vita iii

X Introduction: The Background of the Moore, Meech, and Whitehall Monograph 1 II Dialect Theory since the Neogrammarians 39 III Phonology, Onomastics, and Graphology: Practical Results on Middle English Studies of Modern Dialect Theory 80 IV Perspectives and Conclusions 144

Bibliography 198 CHAPTER I

INTRODUCTION THE BACKGROUND OF THE MOORE, MEECH, AND WHITEHALL MONOGRAPH

My aim in this dissertation is to give an expository and evaluative account of the present state of Middle English dialect studies, starting with the University of Michigan monograph in 1935 of Samuel Moore, Sanford B. Meech, and Harold Whitehall,^ and continuing through the succeeding variety of theoretical and practical approaches to Middle English dialect research to the present day. I intend in this first chapter to show what the scope and significance of Moore, Meech, and Whitehall's essay was, together with its preceding and ancestral essays on the subject of linguistic geography in England, and its influence on the subsequent and more modern methods of mapping phonemes and morphemes from past stages of the language. In later chapters I will treat the dialect theory of the last thirty years more extensively, stressing the fact that the notions behind dialect geography form but one of many concepts of the nature of dialect; further, I will review the practical applications which have evolved in recent scholarship

1 2

from the various dialect theories. Finally, I will offer in the last chapter a criticism of these dialect theories and methods, with the hope of establishing a scholarly perspective on them, along with some recommendations of the most productive combination of dialect concept and technique for future researchers in this field. It was not long after Matteo Bartoli, Gaston Paris, Antoine Thomas and others2 developed the principles of dialect geography that an interest in that discipline arose in England. According to Sever Pop's account in La Dialectologie,3 Reverend W. W. Skeat founded the English Dialect Society in 1873, in an attempt to organize and coordinate the collection of lexical materials in England, Scotland, Wales, and Ireland for the English Dialect Dictionary. Also in 1873, a certain emphasis was given to the interest in the geographical distribution of dialects when Prince Louis-Lucien Bonaparte gave his collection of dialect words and maps to the Society. After gatlaBsing vast amounts of data over several years, Prince Louis had plotted on a map of England the geographical boundaries identifying and separating the varieties of usage among the English peasantry into thirteen dialects: Eastern, South Eastern, South Western, Devonshire, Cornish, Western, Shropshire, Worth Western, Midland, East Midland, North Eastern, North Midland, and Northern.4 Prince Louis' stress on plotting dialect boundaries on maps was to have \

3 a long range influence on the field of English dialect study; for the moment, however, while other European scholars were producing maps of the modern dialect dis­ tribution in their respective nations, the researchers in the English Dialect Society were directing their efforts toward the publication, not of linguistic atlases, but of the Society's journal, the English Dialect Dictionary, and the English Dialect Grammar.5 Meanwhile,. Alexander J. Ellis was also collecting material on contemporary English dialects as the fifth part of his work in the Extra Series of the Early English Text Society, On Early English Pronunciation. Ellis wrote as follows of his task: The object of this treatise is to determine with considerable accuracy the different forms now, or within thet last hundred years, assumed Ey~the descendants of the same original word in passing through the mouths of uneducated people, speaking an inherited language, in all parts of Great Britain where English' is the ordinary medium of commnnicatibn between peasant and peasant.5 To achieve this, Ellis chose 971 words for comparative pronunciation purposes, an$l arranged them in a questionnaire containing fifteen sentences, with an additional seven sentences to be used in an investigation of lexical differ­ ences. Then, during a period of fourteen years, he relied upon over 800 informants from 1145 locations throughout England to provide him with an enormous amount of data from that questionnaire. From all this information, be was able 4

to draw "Ten Transverse Lines"7 from east to west across England, dividing the country into the six dialect regions of Southern, Western, Eastern, Midland, Northern, and Lowland.& Then, as Ellis plotted his material on a map, he found England to be divided into 42 dialect districts, with a total of 86 additional sub-varities or "parts of districts separately consideredamong the 42 dialects. Although Ellis' informants were laymen, and thus were not professional phoneticians, there is nevertheless a high degree of accuracy in this study: in later years, after he had time to compare his dialect divisions with the results of historical phonology, he found that "the districts thus obtained correspond very fairly with those which history, grammar, and vocabulary prescribe."10 Not long after the work of Ellis and the English Dialect Society, other scholars began to shift their attention from collecting material on contemporary English dialects to locating boundaries for the dialect phonology of the past. The interest in historical phonology as a philological discipline, of course, had been developing for d e c a d e s ; but soon after the study of pl&ce names was incorporated in dialect research, scholars began to investi­ gate past dialect boundaries with the same analytical rigor that Ellis had applied to dialect boundaries in the present. The field of onomastics had coexisted from the beginning 5 with a scholarly interest in describing exact phonolggical regions: in 1883, Henry Bradley, seeking a "source of evidence respecting the territorial limits of English dialects prior to the Norman Conquest," found that "dialect distinction" occurred in "our local nomenclature," and he concluded specifically that "the present southern boundary of Yorkshire constituted an actual dialectal frontier" in Anglo-Saxon times.12 In 1901, Alois Pogatscher followed Bradley's example and used place names to determine "Die Englische ae/5 Grenze,"13 and although Otto Ritter later quarreled with the validity of Pogatscher's source materials, 14 he nevertheless strongly approved of using onomastic methods to establish accurate boundaries. By this time, place name techniques had received the unofficial sanction of philologists in Germany and England, so that in the same year appeared Eilert Ekwall's essay on the strengths, weaknesses, and practical problems of place name methodology, in which he treated, among others, the changes in Old English a, jr, and the rounding of a in medieval Lancashire, 15 and Henry Cecil Wyld's classic studies of "The Treatment of OE y in the Dialects of the Midland and South-East counties in Middle English," and of "Old English in the Dialects of the South and Southwestern Counties in Middle English."16 Pogatscher and Wyld in particular were concerned with locating dialect boundaries as precisely as possible. 17 Then, in 1915, Alois Brandi reexamined the 6

results of Pogatscher, Ekwall, and Wyld with additional and more reliable place name sources in his Zur Geographie der altengllscheri Dialekte. 18 As place name study found more practitioners, there was a constant accumulation and transfer of phonological information between onomatologists and other specialized students of history and of historical phonology. A most important article, combining the philological knowledge of Lorenz Morsbach with an interest in the methods of Pogatcher, was written by Richard Jordan in 1910; 19 in his succinct, region by region discussion of which literary texts best demonstrate what dialect features, Jordan provided an extremely helpful guide for later dialect scholars in the selection both of important sources and salient phonological features. He doubtless influenced the later investigations of Henry C. Wyld, Mary S. Serjeantson, and J. P. Oakden, all three of whom maintained an extremely high level of systematic research on Middle English dialects. In an exmination of the "South-Eastern and South-East Midland Dialects in Middle English” Wyld, after carefully selecting /^Y^his texts, studied the occurrence, separately and in combination, of 21 phonolpgical features in the areas in question, 20 in the effort to determine the precise geo­ graphical distribution of those features, and to locate a number of literary texts whose places of origin were in dispute. Mary S. Serjeantson, writing on "The Dialectal 7

Distribution of Certain Phonological Features in Middle English," 21 relied more heavily than Wyld in the last- named article on place name research to discover the dis­ tribution of four phonological features in three general areas in England: 22 first, the counties south of Shrop­ shire, Staffordshire, Derbyshire, and Nottinghamsire; second, the West-Midlands, South, and South-Midlands; and last, the particular combinations of phonological items in the west Midlands and the Southwest. In a later article on "The Dialects of the West Midlands in Middle English," 23 Miss Serjeantson dealt in part with the vexed question of the precise nature and location of the dialect of the Pearl poet: she discussed the distribution of twenty phonological points in the West Midlands region, and she found the area to be divided into a North Western group (containing Cheshire, South Lancashire, West Derbyshire, Staffordshire, and Shropshire), a Central Western group (Herfordshire, Worcestershire, and West Warwickshire), and a South-Western group (Gloucestershire, and "perhaps a small part of West Oxfordshire”); 24 in addition, Miss Serjeantson attempted more accurately to locate twenty West Midland texts within these three groups. J. P. Oakden undertook many of the same tasks on a larger scale a few years later in the first volume of his Alliterative Poetry in Middle English. ^ In the "Dialectal Survey" portion of this volume, Oakden referred chiefly to literary texts 8

in an analysis of the distribution of 45 phonological features in Middle English;.26 in addition, he mapped iso- glosses for thirteen of these, and he then located all the alliterative Middle English Poetry within the Northern or Midland areas. In assigning texts to the Midlands, he referred to areals that apparently are accurate to the near­ est compass point: north, southeast, or southwest, and so forth, but nowhere does he give an idea of the geo­ graphical limits of these regions. Again, these monographs show the influence of the eclectic approach in Jordan's earlier article: except for Oakden's survey, all the essays blend onomastics with other methods, including a review of the pertinent scholarship; all the essays, and Oakden in particular, demonstrate that the more phonological features considered, the more accurate the picture of a given dialect will be; and further, as was Jordan, these scholars were concerned with assigning literary texts of uncertain origin to a definite geggraphical area. J. P. Oakden, in fact, was so confident of his analysis of literary texts that he revived the mapping techniques of Alexander Ellis. It is a paradox that students of the past stages of English have developed an insatiable drive for accuracy in locating phonological features and dialect boundaries despite the limitations of the source materials, while Ellis, whose sources were at least more accessible, was content to draw his isoglosses "roughly," 27 so that "the bounding lines 28 may represent a width of five or ten miles." Yet strive as they might, the earlier scholars did not achieve their desired degree of accuracy. In Hans Kurath's Plan and Bib­ liography to the Middle English Dictionary, superimposed maps of the isoglosses of Brandi and Wyld, and then of Brandi and Oakden, show that isoglosses of the same feature are never mapped the same way twice. 29 There are several reasons for this: sometimes these scholars did not attempt to draw their sources from a sufficiently limited range of time, and too large a chronological spread in their manuscript materials therefore distorted their results. More often, however, the source materials were either non-representative of an area, or too few in number, or both, so that two people working on the same problem could use different sources and present widely different solutions. 30 This is precisely why Otto Ritter and Alois Brandi redid Pogatscher's 5g/g boundary, and this also was why Mary S. Serjeantson addressed herself anew to the £ distribution in each of her studies even though Wyld had already given an extensive treatment of that vowel. By the time of Oakden's book, there was, for the sake of accurate dialect scholarship in the future, an obvious need for an inventory, analysis, and evaluation of the Middle English source materials which cure valuable for the study of dialects and their distribution; there was also an urgent need for a meticulously correct map of the dialect isoglosses at any well-defined stage in the Middle English language. Moore, Meech, and Whitehall set out to fulfill these needs in their monograph. In 1931, these scholars took "a corpus of definitely localized and dated literary texts and documents" 31 which had been prepared for the Middle English Dictionary, and used it as "the basis for a study of the distribution of Middle English dialect characteristics" (1). Among the literary texts examined for the study were three classes of material: in the first, it was required that the source manuscript be accurately dated, that it be "com­ posed in a certain place" (1), and that the original manuscript or "a contemporary (or nearly contemporary) copy written in the same locality" be available (1)• The strictness of these requirements eliminated all but six literary texts among those examined in England; in order to provide themselves with more linguistic data, then, Moore and Meech classed a second group of manuscripts according to less rigorous criteria. In this group were all those texts which were known "to have been copied (or at least owned) in a certain place" (2), and which demonstrated "highly consistent and uniform" dialect features (2). It was felt that the language in such a manuscript would show the true dialect forms of the place in question, because if the manuscript did in fact originate in another place or 11 if later scribes bad copied the manuscript from an original the language would not be consistent and uniform; indeed, Moore and Meech contended that it was virtually impossible for a scribe to copy a text of any length written in a dialect materially different from his own without substituting many forms of his own dialect for the forms that occurred in his original (2). In some cases, nevertheless, a text which was localized and yet inconsistent in dialectal forms was used if it could be shown to be a copy of an original manuscript which could also be localized, or which had a known place of composition. When these manuscripts were studied, careful note was taken of the contrasting dialectal features, and each feature was assigned to its own geographical locale. The third class of literary texts used in the study included those which were copies of an original manuscript whose place of composi­ tion and dialect were known, but which, as copies, were present in a later manuscript that could not be localized. In such cases, Moore and Meech would use the manuscript only if its language was "highly consistent and homogeneous” (2) throughout. In all, within the three categories of "localized literary text," "manuscript localized; composition not necessarily localized," and "composition localized; manu­ script not localized," the scholars used 44 literary texts. The search for documents for use in the Study began in 1930-31 as Mr. Meech compiled for the Middle English 12

Dictionary a "bibliographical survey of the writings in Middle English" (3). To this was added the substance of four earlier collections of printed documents: the English Guild Returns of 1389 and Furnivall's Fifty Earliest English Wills, published by the Early English Text Society, and in addition, Lorenz Morsbach's monographs on London documents, Pber den Ursprung der neuenglischen Shriftsprache and Originalurkunden,' all provided "a valuable nucleus of 35 non-London dialectal documents" (3)• Moore and Meech then expanded their search for printed sources to include 159 additional texts: they looked through the^catalogues of the manuscripts in the libraries of Cambridge, Oxford, and the British Museum; they made an exhaustive search of the volumes of the Early English Text Society and the Rolls Series, and of the publications of the Surtees Society and the Camden Society; they examined the publications of "the various archaeological and historical societies such as the Bristol and Gloucestershire Archaeo­ logical Society, the Shropshire Archaeological and Natural History Society, etc., etc." (3); they obtained other texts from "miscellaneous works on local history and topography" (3) and they worked through the available "documents, records, and letters" listed in the second edition of Charles Gross' Sources and Literature of English History.33 There were fewer sources of useable unprinted docu-

■ ■ ments. In part, Moore and Meech used Ancient Fetitions and 13

Early Chancery Proceedings: from 200 such documents they found only 25 texts which showed forms other than those of the London dialect, or were traceable to "the locality in which the documents originated " (4 ). In order to accumulate more data, Mr. Meech began correspondence with the town clerks of "more than one hundred towns and boroughs incorporated before 1450" (4), hoping to find early documents written in English; in reply, he obtained "forty-five unprinted documents of much more than average value" (4). In all, Moore and Meech used 266 documents, most of which could be classified as follows: petitions, chancery proceedings, wills, testaments, borough, gild, and parish records, letters, deeds, charters, indentures, leases, declarations, diocesan records, and inscriptions on stone, wood, or glass. It is immediately surprising that from the vast variety of available document material, the scholars found so little that could be used. A central difficulty with petitions, chancery proceedings, official letters, major legal instruments, and much other recorded data is that these documents were written in London, and hence showed only the London English dialect features. To cope with this problem, Moore and Meech saw early that they would have to establish the linguistic features and the geographical boundaries of the London dialect before they did anything 14 else; so, using Chambers1 and Daunt's Book of London English, the unprinted Pleas and Memoranda from the London Guildhall, Furnivall's London Wills, City petitions printed in the Rolls of Parliament, English documents from the London Letter-Books, and the records of the Grocers' and Carpenters' Companies, they developed some workable, heuristic assumptions about the London dialect. A. Mor- phemically, their assumptions proved to be quite accurate: "a study of this material showed that Except for the increasing use of their for her and them for hem? the previous theories" as to the features of London English "between 1384 and 1450 were essentially correct" (7). Geographically, they determined that the London dialect may have been found in the city itself and in a region lying north of the Thames as far as Oxford, east of the line running from Oxford to Rugby, and south of the line running from Rugby to Stamford and from Stamford to the Wash, including Middlesex, Buckinghamshire, eastern Oxfordshire, Northamptonshire, Cambridgeshire, Norfolk, and the counties east and south of those (6). To make sure that the dialect was indeed uniform throughout this region, Moore and Meech took exceptional care to accept only genuinely authentic documents, such as originals, copies made or at least preserved in the same locality, and documents of merely local interest that were almost certainly written in and for the place to which they refer (6). 15

With the characteristics and boundaries of the London region established, then, it became easier to narrow a large body of documents down to the much smaller group that could be localized within areas outside of the East-Central region. A second major problem with the documents was, as might be expected, that "English was almost nonexistent in the official papers until c. 1375, and even then it was rare until 1430" (5). As a result, 77 per cent of the 266 documents date from between 1410 and the cut-off point, 1450. Despite the late date of these texts, the scholars were .quite careful to assure that the language .in each accurately represented the forms of the area in. question. Any documents about which there was uncertainty were rejected if they fell within any of the following categories: 1. Documents not definitely localized (except superior or distinctly good docu­ ments that are localized within limits which are not close to any dialect boundary). 2. Documents whose evidence is quite contra­ dictory to that of superior or distinctly good documents of earlier date from the same place. 3. Documents not near dialect boundaries which show none of the specific characteristics of the locality as established by superior or distinctly good documents. 4. Documents in which there occur a combination . of characteristics (usually at least three) which do not belong to the locality of the document, but which occur together in some other dialect region. Such documents, however, were actually rejected on 'facts known or reasonably inferred from the document itself.' 16

5. Documents later than 1450, except Northern and occasionally Western documents that are cleaxly. dialectal in spite of their late date; local copies of documents originally written not later than 1450 were accepted if they quite evidently reflect the language of their originals rather than the language of the date' at which they were copied (6-7). After they studied all the acceptable documents, Moore and Meech "recorded statistically1’ the data which they collected (8), and then Mr. Whitehall plotted the material on outline maps of England with isophonic lines. After the plotting was complete, Whitehall undertook an investigation of the geographical or physiographical features on or near which the isophones rested. Then, with the caution that "the physiographical boundaries are an interpretation of the isophones and are therefore subject to the modifications that additional documentary evidence may bring" (32), Whitehall projected an analysis from the two-dimensional map surface to the three-dimensional properties of physical features on the land. In general, he found that the important physiographical features influencing dialect study are those which cause a region to be sparsely populated, and those which inhibit population movements and "impede external communications" (27). Specifically, he found the features to include: the irregular "relief of the land surface," (27) when, for ' o instance, a range of high hills "penetrated by few river gaps or practicable passes" serves as a "linguistic frontier" (27); the presence of unnavigable rivers, or of 17 long stretches of a river whose depth or width makes the river unfordable; extensive marshes or tracts of undrained fen or morass; certain impenetrable forests, and, any combination of one or more of these features. Once the geographical barriers are recognized, Whitehall showed that political, diocesan, and racial boundaries serve to intensify that isolation of a community which allows dialectal differences in the language to develop (27). From this scrupulously, and even fastidiously examined material, Moore and Meech extracted the following eleven phonemic and morphemic features, which they, felt to represent the major dialect regions of England beginning around the year 1350; in a northern dialect area, Old English s was retained as an unround vowel and spelled a, ai, or ay as in mar, baith, or hayl; the west Midlands were characterized by either an a or an o spelling of the Old English a if followed by m or n; in the west Midlands and the south, Old English y and eo were retained as front vowels and spelled u or ui and: eo, o, oe, u, or ue; in the south, in Kent, and in the lower southwest-Midlands, initial y occurred for f; and in Kent and the south, there was "evidence of West Saxon umlauts of the Old English ea and €ST, and evidence of West Saxon diphthongization of e after initial palatals" (6). Morphemically, Moore and Meech found the north and £he north Midlands distinguished 18 by an -es third person singular present tense ending; in the north and the northeast Midlands, the present indicative plural had an res ending; the present indicative plural hd an -eth ending, however, in Kent, in the south, and in the southwest Midlands; sal and suide for shal and shulde was characteristic of the north, the northeast Mid­ land, and the central east Midland regions; hem, ham, and horn were the pronouns for them in the east and west Mid­ lands, the south, and Kent; and finally, Moore and Meech found that the present participle with ending -nd was being pushed north, east, and west of London to the far borders of the country. When Mr. Whitehall plotted the isophones which separated these various dialect features, he found that the lines intersected in such a way as to divide England into ten geographical regions; further, each region was different from all the others because each had a unique combination of dialect features. Moore, Meech, and Whitehall then gave each region a name, with the result that their map contained the following ten Middle English dialect areas: Kentish, Southern, Southeast-Midland; Central East-Midland, Northeast-Midland, Southwest-Midland, South-central West- Midland, Northwest-Midland, and Northern. For the most part, modern students still consider these distinctive features and their corresponding dialect areas to be the products of valid and accurate research; 19

however, the "additional documentary, evidence” which White­ hall mentioned has since been collected by the scholars who are publishing the Middle English Dictionary at the University of Michigan. This information, together with the more modern linguistic attitude that there can be no rigid isophonic demarcation separating one dialect feature from another in a neighboring dialect, 35 has rendered White- hall's physiographic projection "untenable." 36 In fact, on turning to the direct continuation and culmination of Moore,. Meech, and Whitehall's researches in the "Dialect Areas” section of. Hans Kurath's Plan and Bibliography to the Middle English Dictionary, 37 one sees a simplification of both the treatment of distinguishing Middle English dialect features and the description of isoglosses. Kurath has found that Whitehall's isoglosses for the following dialect features either were inaccurate or were based on insufficient evidence: Old English S' spelled a, ai, ay; the umlauts of ea, and the palatal diphthongization of e; initial v for f; the sal and sulde, and hem, ham, and hom morphemic boundary; and the present participle ending. Kurath, of course, maps these lines whenever possible, but he offers no physiographic descrip­ tion of them. Those isoglosses which he does, describe, together with the pertinent dialect areas and the distin­ guishing phonemic or morphemic features, are as follows: the Northern dialect extends north of & a/d boundary which 20

runs from the mouth of the Humber west­ ward along the ^southern boundary of the East Riding of Yorkshirethen, through the West Riding north of the River Aire, crosses the Pennine Chain into Lancashire, and then swerves northward to the River Lune (8). Next, the general Midland dialect lies north of the -en/-eth present plural ending boundary, which moves from the estuary of the Thames westward along the southern boundary of Essex, Middlesex, Buckingham, and Oxford to the mouth of the River Cherwell, . . . north­ ward through Oxfordshire, • • • then in a northwesterly direction through southern Warwickshire, southern Staffordshire, and central Shropshire to Wales (9). The Southern dialect Kurath lists as south and southwest of this same line. The line separating the East and West Mid­ land dialects is the a/o boundary, which follows rather closely the eastern boundary of Lancashire, Derby, Stafford, and Warwick; in Gloucestershire It runs along the crest of the Cotswolds and then swerves westward to the estuary of the River Severn (9).

- \ Adjacent to the East Midland are two smaller dialect regions: the Northeast Midland lies north of the present third person singular -es/-eth boundary, which travels westward from the Wash along the southern boundary of Lincolnshire, cuts through southern Nottingham and Derby, and then runs northwestward through the middle of Stafford­ shire and southern Cheshire to the estuary of the River Dee (9). To the south is the Southeast Midland dialect, which lies 21 east of the :-en/-etb boundary, and also east of the hul/ hil boundary, which goes northward through Oxfordshire and Warwick­ shire. South of Oxfordshire it runs south­ ward through Berkshire and then, apparently, close to the eastern boundary of Hampshire; north of Warwickshire it seems to swerve northwestward through Staffordshire, eastern Cheshire and southwestern Lancashire (9). East and west of this same line lie the Southeast and the Southwest dialects, respectively. Kurath's stress throughout this section of the Plan and Bibliography is on updating the results of Moore, Meech, and Whitehall as simply and succinctly as possible: thus he describes a newer version of Whitehall's present plural -es/-en isogloss, but because in his modernized representa­ tion this isogloss does not determine a major Middle English dialect region, he maps it separately. The condensed treat­ ment in this later, volume seems to be successful, rfor Rurath devotes only two pages to the material which occupied Whitehall on pages 25-47 of the original article. Yet even after one accepts Kurath's adjustments and acknowledges the facts that dialect theory, and the cumulative analysis of Middle English sources, have changed since 1935, 1 think that a copy of the Moore, Meech and Whitehall monograph should be kept as a locus of reference,, because after all, by. definition a simplified, treatment cannot be a duplication of the effort,, the subtlety., or the results of the original product. Indeed, Kurath himself admits that 22

the survey nevertheless provides the most reliable evidence now available for setting up a tentative scheme of the major dialect areas and their subdivisions, to which texts written in regional dialects or exhibiting regional features may be a s s i g n e d . 38 This is true chiefly because of the tremendous labor Moore and Meech gave to the collection and analysis of sources. In addition, however, the effort was guided by Mr. Moore's immense knowledge of historical phonology; 39 further, these scholars supplemented their researches by consulting works on historical phonolpgy such a Luick's Historische Grammatik der englischen Sprache, 40 and by comparing their isophones to Ellis' more modern correlates in Part V of his Early English Pronunciation. Thus, Moore and Meech were able to establish which of the eleven dialect characteristics were intrusive and which were recessive, and there can be no argument about their statement that "the displacement of sal, hem, initial v, and -eth third singular is quite simply explained as being due to the intrusion of the opposite characteristics”(41). Further, in discussing the retention of Old English £, y, and eo, eo as a front round vowel, Moore and Meech's knowledge of the history of the "spontaneous unrounding of the front round vowel” along "the eastern borders of the u-region" enabled them to make a prediction about the stability of their isophone after 1450; 23

the spontaneous unrounding in the west, coupled with the transmitted influence toward unrounding from the east, must have caused perceptible changes in the boundary, and it seems probable that the line indicated by our isophone has been affected by such changes (42) . All told, Moore and Meech's researches are more valuable because they corroborated their results with the evidence of historical phonology. Nevertheless, Moore and Meech admit that their work shows certain shortcomings. Their manuscripts, despite the rigorous principles from which they were selected, are usually inconsistent to some degree: there may be a number of irregular spellings or minority forms in any given manu­ script, and there may be a mixture of London dialect forms with those of the region under examination. Even with the extreme care that was taken with the evidence, these scholars feel that many of their isophones are inadequate, simply because "they are merely the lines indicated by the material we have” (21). More literary texts and documents are needed, so that more can be learned about the various dialects. In fact, the largest drawback to dialect research, * * ' d j . T they feel, is the scarcity of texts, and with more texts examined and with their salient features plotted, "the resulting map would probably show fifty or more dialects rather than ten" (24). 24

Despite such shortcomings, Moore and Meech made a signal contribution to dialect research. Their efforts at finding useable texts for dialect work formed a model for later attempts in the. field. Further, their stress on the fact that theirs is a study of written dialect forms and the changes in the written language, and that such a study does not necessarily reflect the conditions of the spoken language, gave an impetus to the development of the important modern field of graphology. 41 This very stress led them to state a fact which may have a bearing on the origin of Standard English: there was a progressive assimilation of written English to' the London or East-Central standard which resulted in the almost complete disappearance of dialectal forms in documents written after 1450 (23). Most importantly,, however, Moore and Meech1s scholarship provided "a system of lines that divide England into ten regions, each having its distinctive combination of dialect characteristics," and these lines also "reflect at least approximate boundaries for eleven of the most important characteristics of the regional dialects of Middle English as they, existed about the middle of the fourteenth century" (23). In a sense, one can compare the work of Moore, Meech, and Whitehall with that of Alexander Ellis: in pro­ viding a network of geographically named dialect regions, 25

they achieved for Middle English the same task, albeit on a different scale, that Ellis had attempted for nineteenth century English* In most ways, however, Moore and Meech*s achievement has greater scholarly significance: the practical value which a definite dialect division has for both the Middle English phonologist and the historian of medieval England is enormous, for it provides a new kind of regularity in the study of manuscripts and their locations. Further, a vast number of more specialized studies has become possible: a scholar may now investigate the history of the sound changes over a given period of time among the texts of, say, the Northeast Midland dialect alone. A study can be made of the influence of London English on the provinces, or vice-versa. Scholars now have a fixed frame of reference within which to trace the evolution of Standard English. In addition, when enough primary sources are examined and found acceptable, it may be possible to dis­ cover how the ten dialect areas themselves have contracted, expanded, or changed in any way; from this, a scholar may be able to formulate principles on the general behavior of neighboring dialects in Middle English. Moreover, with standard dialect regions the information in works as diverse as Luick's histories of phonology and Wells* Manual of the A Writings in Middle English can be updated and made more valuable. In general,: the importance of a definitive 26

■treatment— or nearly so— of dialect areas can hardly be overestimated, for in organizing a large number of diffuse materials, such a work gives momentum to Middle English studies overall. Since 1935, there have been several criticisms of the sources and methods in Moore, Meech, and Whitehall's monograph. C. L. Wrenn 43 reviewed the study shortly after it appeared, and reported some errors in judgment by Moore and Meech: since they used Dan Michel's Ayenbite of Inwit, they should also have drawn upon the Ormulum and the Corpus Christi Cambridge manuscript of the Anerene Wiese, for as literary texts these sources show a highly consistent dialect. Wrenn felt also that their term "London English" was used too freely, and that the -nd participle ending was too close to the widely distributed Scandinavian -and ending to be considered a genuinely distinctive dialect feature; further, he showed that .their manuscript of the Poema Morale is inferior to others which are easily available. Sven Rubin felt that literary texts, in contrast to documents, do not offer enough variant forms, and hence show "serious limitations when it comes to using them for the purpose of ascertaining Middle English dialect boundaries and dialect peculiarities." 44 More recently, Bertil Sundby raised a serious objection to Moore and Meech's method: ."these scholars apparently take it for granted that the place 27 where a manuscript was kept was identical with the place where it had been written out." 45 The most extensive criticisms of the monograph, however, were made by Angus McIntosh, who is preparing a Survey of Middle English Dialects for publication in the near future; he shows that the handling of many dialect features was both hasty and "needlessly crude." 46 The behavior of y-words, despite the attention given them by Henry Wyld and Miss Ser.jeantson, still is not understood perfectly, for this vowel demonstrates as well as any that "chaque mot a son histoire;" 47 thus Moore and Meech's generalizations about this dialect criterion are inaccurate. Further, McIntosh shows that the :"hem, ham, and hom for them" feature is drastically oversimplified, for throughout England there are as many as twelve variations (hem, horn, heom, ham, j>em, fcam, j^aim, foeim, them, tham, thaim, theim) which may be of "dialectal relevance.""48 In addition, Mr. McIntosh is dissatisfied with the "choice of textual sources" in the essay. In stead of the 266 documents 49 which they analyzed, McIntosh found "at least 400" which are acceptable; further, he had developed a method, which I describe below, allowing him to use far more literary texts than the 44 originally allowed by Moore and Meech. Also, McIntosh believes that they "worked with too wide a chrono- logical spread;" 50 thus, 28

linguistic differences due to chronological factors were confused with genuine dialectal (what I call diatopic as distinct from dia­ chronic) differences, to the considerable . confusion of the whole study.51 Thus it is best to restrict the chronological range to a century, preferably 1350 .to 1450;. but if one does this, he must use as thoroughly as possible every scrap of material available to him. Mr. McIntosh's most pointed criticism of Moore and Meech, however,* pertains to the total number, of dialect features they worked with: he considers the eleven characteristics too small a number, and he believes that even Oakden's 45 items are not enough. Instead, for his Survey, he writes that "I myself started with some seventy- five items and . . . for certain purposes we have stepped up the total to more than two. hundred and sixty.n 52 . His technique for gathering well over 200 distinctive features within a given dialect is simple and straightforward: one begins by accepting the ten dialect regions described and named by Moore,. Meech and Whitehall. Next, one con­ centrates on a given region (the Northwest-Midlands, for instance), and then studies a literary text with a homo- geneous dialect from that region— let us say the manuscripts of the Pearl poet, in this case. From these texts the scholar makes a list of all those features which he may. have reason to believe distinctive.. Then, he searches for other 29 homogeneous texts within the same region and attempts to fit, or to match,, the features of the second text with the list of those from the first. As this operation (which McIntosh calls the "fit-techhique," because one attempts to fit a manuscript into a geographical area) 53 is repeated, the scholar gradually develops a long list of features, gleaned chiefly from literary texts, which may fill in an isogloss map far more reliably than any other method. The fit technique, Mr. McIntosh holds, will be a most productive way of expanding our knowledge of Middle English dialects.. From the early influence of onomastics on the tradi­ tional methods of historical phonology, through the scholarly achievements of Moore, Meech, Hans Kurath, and Angus McIntosh, the emphasis in Middle English dialect studies has been on determining with ever greater accuracy the various dialect regions and their component phonological features. Yet during the stress on the accurate mapping of features, several theoretical departures occurred in modern thinking about the nature of linguistic dialect: " the linguistic geographers, and the structural dialectologists in particular, have shown cause for many of the traditional concepts about dialect to be reconsidered. In the next chapter, I give an account of the changes which the hew dialect theories, have brought, and in Chapter III, I show what the most recent impact of those theories has. been, on Middle English dialect scholars since Moore, Meech, and Whitehall. 31

NOTES

1. Samuel Moore,. Sanford B. Meech, and Harold Whitehall, "Middle English Dialect Characteristics and Dialect Boun­ daries, " Essays and Studies in English and Comparative Literature, University of Michigan Publications in Lan­ guage and Literature^ XIII (1935), 1-61. 2. See the discussion of linguistic geography in Chapter II. 3. Sever Pop, La Dialectologie (Louvain, 1950), 909-913. 4. Pop reproduces Prince Louis' map on p. 911. 5. Joseph Wright, ed ., The English Dialect Dictionary, London, 1898-1905. Also referred to is: Joseph Wrignt, ed., The English Dialect Grammar, Oxford, 1905. This English Dialect Society published 80 monographs between 1873 and 1896. 6. Alexander J. Ellis, On Early English Pronunciation, Part V; Existing Dialectal as compared with West Saxon Pronunciation, EETS, Extra Series, LVI (1889), p. 92. 7. Ellis, p. 15. 8. The various lines, which record the pronunciations of the words "some," "the," and "house," and which also note the pronunciation of /r/, are as follows: 1— the North sum line, 2— the Southern suom line, 3— the reverted ur line, 4— the Southern teeth line, 5— the Northern dheetH~~Iine, 6— the Southern hoos line, 7— the Northern tee line, 8— the Southern sum line, 9— the Northern suom line, and 10— the lowland border. 9. Ellis, p. 3*. These districts which Ellis carefully locates and discusses at length in his volume, are as follows: western Celtic Southern, mid Celtic Southern, eastern Celtic Southern, Mid Southern, eastern Mid Southern, East Southern, northern West Southern, southern West Southern, western West Southern, South Western, North Western, West Eastern, Mid Eastern, South Eastern, North Eaatern, East Eastern, Border Midland, southern North Mid­ land, western North Midland, northern North Midland, eastern North Midland, western Mid Midland, eastern Mid Midland, East Midland, western South Midland, eastern South Midland, East Northern, West Northern, North Northern, South Lowland, eastern Mid Lowland, western Mid Lowland, southern 32

Mid Lowland, northern Mid Lowland, southern North Lowland, mid North Lowland, north North Lowland, south insular Lowland, and north Insular Lowland. 10. Alexander J. Ellis, English Dialects--Their Sounds and Homes; Being an Abridgement of the Author's Existing Phonology of English Dialects, which Forms Part V of his Early English Pronunciation (London, 1890), p. 1?1. 11. I refer here to the early and classical repositories of information in this field,, such as the following: Morsbac.h, Lorenz. Mittelenglische Grammatik. Pt. 1, Halle, 1896. Kaluza, Max. Historische Grammatik der englischen Sprache» 2 vols. Ber1in-Schoneberg, 1906-0?. The Middle English material is in volume two, pages 1-203. Horn, Wilhelm. Historische neuenglische Grammatik. Strassburg, i960. Berndt, Rolf. Einfuhrung in das Studium des Mittel- englischen. Halle, 1930. Wyld, Henry C. A Short History of English. London, 1927. Emerson, Oliver F. A Middle English Reader. New York. Wyld, Henry C. A History of Modern Colloquial English. New Ygrk, 195FT ! Luick, Karl. Historische Grammatik der Englischen Sprache. With additions by Friedrich wild and Herbert Koziol. Reprinted and published by Basil Blackwell in Great Britain by permission of Bernhard Tauchnitz Verlag, Stuttgart.. 2 vols. Cambridge: Harvard Press, and Stuttgart: Bernhard Tauchnitz,. .1964.. Wright, Joseph and Elizabeth M. An Elementary Middle English Grammar. London, 1928. Wells, John E. A~Manual of the Writings in Middle English 1050-1400. New Haven, 1926. Plus supple- ments 1-VIII, extending the coverage through July, 1941. The ninth supplement, by Beatric Daw Brown, Eleanor K. Heningham, and Francis Lee Utley, gave "Additions and Modifications to December, 1945," and was published at New Haven in 1951. I include Wells not for any discussions of phonology,, but because he treats questions of dialect in his exposition. 33

More recently, the following works have dealt with Middle English phonology: Jordan, Richard. Handbueh der mittelenglischen Grammatik. Heidelberg, 1934. Wardale, Edith E. An Introduction to Middle English. London, 1937. Brunner, Karl. Abriss der mittelenglischen Grammatik. Halle, 1938-1S5T: Brunner, Karl. Die Englische Sprache. 2 vols. Halle, 1950. See. especially Volume t , pages 85-93 and 222-230. Mosse, Fernand. A Handbook of Middle English. Translated by James A. Walker. Baltimore, 1952. Horn, Wilhelm. Laut und Leben; englische Lautgeschichte der neueren Zeit (1400-19SoT~^ Bearb. und hrsg. von Martin Lehnerl. t 2 vols. Berlin, 1954. Berndt, Rolf. Einfuhrunq in das Studium des Mittel­ englischen. Halle, 1980. Broolc, George L. English Dialects. London, 1963. See pages 55-78. Also, one should consult any "Notes on Dialect" sections present in the editions of the Early English Text Society. 12. Henry Bradley, "Dialect in English Place-Names," Academy, xxiii 563 (February 17, 1883), p; 116. 13. Alois Pogatscher, "Die Englische ae/e Grenze," Anglia, XXIII (1901), 302-09. 14. Otto Ritter, "Zur Englischen ae/e Grenze," Anglia, XXXVIII (1913), 269-75. 15. Eilert Ekwall, "Ortsnamenforschung ein Hilfsmittel fur das englische Sprachgeschichte-Studium," GRM, V (1913), 592-608. 16. These articles are found in Englische Studien, XLVII (1913), on pages 1-58 and 145-166 respectively. 17. Hans Kurath mapped Wyld's results in his Plan and Bibliography to the Middle English Dictionary. See Note 29. 18. Alois Brandi, Zur Geographie der altenglischen Dialekte, Berlin, 1915. This work is an attempt, much like Alexander Ellis', to apply the results of the study of a certain lin­ guistic period as a historical perspective on the dialect conditions of an earlier time. Just as Luick felt that earlier stages of the language could be investigated through 34

Ellis' modern dialect divisions, Brandi hoped that a study of the distribution of certain phonological features of Middle English might be helpful in a treatment of Old English dialect boundaries. This view,, of course, has been changed in recent years, for it is now recognized that dialect boundaries are highly unstable both in place and time. See the discussions in Chapter II. In other ways Brandi's work shows a most valid approach. In three sections, he discusses the distribution of Old English dialects first, according to the settlement patterns of the various Anglian and Saxon tribes (here he draws chiefly on Bede's Historia Ecclesiastica Geritis Anglorum); next, he examines the development and influence of the various dioceses of the Church in Anglo-Saxon times, and then compares the effects of "Diozesangeographie und Stammesgeographie;11 finally he draws upon such sources as the Feudal Aids and the Hundred Rolls to test Pogatscher's 35/5" line, Wyld's y line, and Ekwall's a/o criterion in Lancashire, all in the belief, that the Middle English boundaries will bear a strong rela­ tion to the location of the Old English boundaries. For a more direct treatment of the relation between Old and Middle English dialects, see Chapter IV. Hans Kurath also mapped Brandi's results. See note 29. 19. Richard Jordan, "Die mittelenglischen Mundarten," GRM, II (1910), 124-34. 20. Henry Cecil Wyld, "South-Eastern and South-East Midland Dialects in Middle English," Essays and Studies. VI (1920), 112-145. In this article, Wyld treats the changes in thir­ teen Old English vowels and eight "points of_ace±dence;" the features are as follows: ae, ae, eald (aid), ear + cons, i-mutation of a, eo, eo, io-x, y, y, ea, ea-i, ea-i; third person singular, present plural, past participle of strong verbs, infinitive, present participle, the form and occurrence of the accusative plural pronoun, third person pronoun plural, and the present plural of to be. 21. Mary S. Serjeantson, "The Dialectal Distribution of Certain Phonological Features in Middle English," ES, IV (1922), 93-109, 191-198, 223-233. 22. The features she treats are:, the fracture of ae before 1 + a consonant; the i-mutation of ea (from W. Gmc. au); Middle English /y/ from Old English eo; Old English y and y + a front consonant. 35

23. Mary S. Serjeantson, "The Dialects of the West Midlands in Middle English," RES, III (1927), 54-67, 186-203, 319-331. Following are the features she deals with:. . Old English ae-*e, a; ae—Ve/, /£/', ae (i-mutation of a-)— ye/?; y— >?; eo— »?; eo before a back consonant; ael + consonant; i-muta­ tion of ael + consonant?. Old English ear— *?; a, (o) + nasal — >?; ea + -i,-j? eag, eah; infinitive ending; present participle ending past participle ending in strong verbs; third person singular present indicative; plural of present indicative; plural of to. be; the forms of the feminine pro­ noun of the third person; and the forms of the plural pro­ noun of the third person. 24. Miss Serjeantson, RES, p. 55. 25. J. P. Oakden, Alliterative Poetry in Middle English: The Dialectal and Metrical Survey, Volume I, Manchester, l£36. The "Dialectal Survey1' is on pages 5-130. 26. The features, which are listed on pages 10-11, are as follows (I use Oakden's abbreviations): 1-0.E. a + nasal as a or o; 2-0.E. a + 1 + Cons, mutated as a or e; 3-0.E. I-and' as '-and1 or V-ond'? 4-Is M.E. 'ong' retracted to 'ung'?; 5-0.E. y as u, i, e? 6-O.E. a as a or o? 7-0.E. aw as aw or ow; 8-0 .E. ag,~0h, as aw or ow (a? or 07 )j_ 9-0.E. o as o or 0; 10-Do ai, ei, oi appear for a, e, o?; II-Does M.E. ou appear as au?;~T2-0.E. ae as tense or Tax e; 13-Is O.E. ae before d,“F, s, n, 1, r tense e£? 14-0.E. eo as e or a rounded, vowel (oe, u, ue); 15-0. E.. ea + cons, mutated as e or u; 16-Is O.E.. ea when shortened, a or e?; 17-Is M.E. e3 from O.E. eag, eah, raised to Ty ?; 18-0.E. ae as a or e? 19:=S7e . hw as wh, w or gu? 20-0.E. cw, 0. Fr. gu as qu or wh; 21-Any iniFiaT voicing of f, s?; 22-Does 'sc-' in unstressed positions appear as 's-*?; 23-Does the spirant V-ht' appear as '-ght, ' '-ht,1 ' 3t,-' etc.?; 24-Do '-us,' '-ud' occur for '-es,''-ed'?; 25-Are there any noun '-en' plurals?' 26-Is the form of the plural of "this" found as thire or these, etc.?; 27-The form of the 3rd pers. sg. fern. pron. nom.; 28-The form of the 3rd pers. pi. pron. nom.; 29-The form of the 3rd pers. pi. pron. acc.; 30-The form of the 3rd pers. pi. pron. gen/; 31-Is the adj. and adv. ending '-ly' or 'lich(e)'?; 32- The infinitive termination; 33-Does the *i-' verbal prefix survive?; 34-Does the 'i-' of the 2nd class of O.E. wk. verbs survive?; 35-The pres. part., termination; 36-The pres, indie, sg. terminations; 37-Any syncopated forms in the pres, indie, sg.?;. 38-The pres, indie, pi. terminations; 39-The form of the pres, indie, pi. of the verb 'to be1; 40-The imperative terminations; 41-The 2nd pers. sg. wek. pret. termination; 42-Is the final '-ed* of the wk. pret. 36 and past part.: unvoiced? (excluding the cases where it occurs in connection with voiceless sounds) ?; 43-Is this final d/ \ '-t' ever lost?;; 44-Are the contracted forms of take (ta), make :(ma) ever; found?; 45-Are the forms 'quilke,' 1swilke' found rather than 1whiche,' 'swiche'? Oakden on page 38. supplied, isoglosses which show the distribution of points, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 12, 13, 14, 15, 17, 19, 21, 22, 27, 28, 29, 30, 31, 32, 34, 35, 36, 38, 40, 41, 44, and 45 as listed above. 27. Ellis, Abridgement, p. 5. 28. Ellis, Part V, p. 4. 29. Hans Kurath, Plan and Bibliography to the Middle English Dictionary, Ann Arbor, 1954. The maps are on page Id. 30. The problem of sufficient and representative sources has continued to plague onomatologists: see the discussion of the English Place Name Society's later volumes in Chapter XII. 31. Moore, Meech, and Whitehall, p. 1. Subsequent refer­ ences to this source will be quoted in the text of this chapter. For brevity's sake, and in part because Mr. Whitehall's contributions to .the monograph;/ are now con­ sidered invalid, I refer to the authors hereafter as "Moore and Meech.” 32. Joshua T. Smith, and Lucy T. Smith, eds., English Gilds; the original ordinances of more than one hundred English gilds, EETS, XL (1870), Original Series. The other references cure as follows: Furnivall, Frederick J. The Fifty Earliest English Wills, EETS, LXXVIII (1882), Original Series. Morsbach, Lorenz, fiber- den Prsprung der neuenglischen Shriftsprache. . Bonn, 1866. ______. Mittelenglische originalurkunden von der Chaucer-zeit bis zur mitte des XV jahrhunderts. HildelEergTTmi ------~ *------33. Charles Gross, Sources and Literature of English History from the Earliest Times to about 1485, 2nd edition, London, 1915• 37

3.4. Raymond W. Chambers and Beatrice M. Daunt,: A Book of London English, 1384-^1425, Oxford, .1931. . Also mentioned is the Rotuli par 1 iamentorum; ut et petitiones et placita in parlieuaento (127^~l5o3i, n.p., n.a.,— Index, 1932, edited by John Strachey, John Pridden, and Edward Upham. 35. See the discussion in Chapter II. 36. Kurath, Plan and Bibliography, p. ix. 37. Ibid., pp. 8-9. This section contains maps and Kurath1s corrections and readjustments of the results of Moore, Meech, and Whitehall. References to the Plan will be quoted in the text of the chapter. 38. Kurath, p. ix. 39. I am not tempted to compile a complete bibliography of Mr. Moore's writings on phonology; yet the following well known works are indispensable for students in the field: Moore, Samuel> and Thomas Knott. The Elements of Old English. Ann Arbor, 1919. Moore, Samuel. Historical Outlines of English Phonology and Morphology. Ann Arbor, 1929. 40. See note 11. 41. See the discussion of graphology in Chapter III. 42. John E. Wells, A Manual of the Writings in Middle English 1050-1400, New Haven, 1926. Plus supplements I-VIII, extending coverage through July, 1941. Supplement IX, by Beatrice Daw Brown, Eleanor K. Heningham and Francis Lee Utley (New Haven, 1951), covers to December, 1945. The corrected information on the locations of the various texts could appear in a tenth supplement to Wells, devoted exclusively to dialect questions. Of course, to bring Wells up to date since 1945 would be a super­ human task. 43. C. L. Wrenn, "Philology: General Works," YWES, XVI (1935), 30-65. 44. Sven Rubin, The Phonology of the Middle English Dialect of Sussex (Lund, 1951), p. 8. 45. Bertil Sundby, Studies in the Middle English Dialect Material of Worcestershire Records (Bergen, 1963), p. 3. 38

46. Angus McIntosh/. "A New Approach to Middle English Dialectology," ES, XLilV (1963), p. 3. Mr. McIntosh spoke of financial problems in publishing the Survey, and he did not seem confident that the problems would be solved soon. 47. McIntosh quotes Gilli^ron (who made this statement in his 1 'Atlas linguistigue de la France) on p. 2. 00 • McIntosh, P- 49. Ibid., p. 4. in o

• Ibid., p. 5. 51. Ibid., p. 5. 52. Ibid., p . 2. 53. Ibid., p. 6. CHAPTER II

DIALECT THEORY SINCE THE NEOGRAMMARIANS

Ever since the 1860's in Germany, England, and America, there have been two divergent views about the nature o£ dialect in languages. Theoretically speaking, William Dwight Whitney, who synthesized the ideas o£ Heinrich Steinthal and August Schleicher,1 held that whether one speaks of "idiom, or patois, or dialect,, or 2 language," any tongue is an "aggregation of particulars, and each [ learner J appropriates more or less of them, 3 according to his means and ability." In his emphasis on "particulars," Whitney did not have recourse to such terns as phoneme or morpheme; yet his mention of pro­ nunciation and forms shows tha.t he anticipated the structural notions of a later time: The difference between any given dialect and another of kindred stock is made up of a multitude of sepiarate items of difference, and consists in their sum and combined effect: thus, for instance, words are possessed by the one which are wanting in the other; words found in both are differently pronounced by each, or are used in senses either not quite identical or very unlike; combinations and forms belong only . to one, or are corrupted and worn down in diverse degrees by the two; phrases occur in the one which would be meaningless in the other. 39 40

In thus suggesting some of the linguistic properties of these "particulars," Whitney gave an important early stress to a long-enduring concept about dialect: any dialect is made up of "a multitude of separate items," and is distin­ guished from another through the "sum and combined effect" of those items. Whitney's position on the real nature of dialects foreshadowed that of one segment of the Nepgrammarians of the 188C's. Among the members of this school of thought, which held in general that all phonetic change could be explained through a number of immutable, formulaic sound laws, and that any exceptions to those laws could be 5 explained as owing to the influence of analogy, there was, contrary to what one might expect, a good deal of contro­ versy. Some, like 6. I. Ascoli and P. Meyer, believed with Whitney that a "particular combination" of sounds may manifest itself as distinctive in a given geographical area; others, like H. Schuchardt and G. Paris, felt that the rigidity of the sound laws actually made all speculation about dialect rather irrevelant: since, as Paris held, in any Romance language, "Nous parlons latin," the question of what linguistic forms belonged where was not as impor- • <> tant as the study of comparative linguistics among major language groups. 7 41

Inevitably, a reaction set in against this inter­ pretation of phonetic change. Paris himself and Antoine Thomas in part led. the reaction through their emphasis on patois; these two scholars wrote a number of monographs about the vernacular in the villages and hamlets of the D French countryside in the final decades of the century. Elsewhere, other scholars became reluctant to accept what they considered to be the simplistic interpretation of phonetic reality offered, by the Neogrammarians. In Italy, Matteo Bartoli began working under an entirely different assumption and argued that the Neogrammarians menagaient d'etioler la linguistigue en 1'emprisonnant dans les formules rigides et risquaient de faire perdre contact avec les r£alit£s infiniement delicates et changeantes de la vie. Bartoli and his group "ont considere les langues comme des organismes vivants,. et ont donne le pas aux faits d'ordre historique, geographique, et social."^0 They switched the emphasis from the diachronic study of what were later to be called phonemes and morphemes, to the synchronic study of any and all linguistic phenomena available to the scholar through every means possible— period manuscripts, local manuscripts, records, documents, the transcribed speech of living informants, or any method fry which one could record linguistic data. Following the precept of Charles Nodier that "tout homme qui n'a pas soigneusement explore les 42 patois de sa langue,. ne la salt qu'a demi, scholars all over the continent developed a new discipline which even­ tually came to be known as linguistic geography . G. Wenker, \ with his work in north and central Germany; V. Bennike and M. Kristensen, with their Linguistic Atlas of Denmark; Dr. Je Te Winkel's Atlas of Northern Holland; M. Blanc qua ert' s Atlas flamande; Eessler's Atlas du nord-ouest de l'Allemagne; GilHeron's 1 'Atlas linguistique de la France, and K. Jaberg's and J. Jud's Atlas suisse-italien, 12 all made use of the cartographic method, in which numerous examples of vernacular linguistic forms were collected and then plotted on maps, in an effort to obtain as detailed as possible a picture of the spoken language at the time of the study. Despite the revolutionary zeal with which the early scholars in linguistic geography approached their labors, and despite the attractiveness of the proposal that lan­ guages be studied anew with regard to their qualities as "organismes vivants," the new school, of linguists did not turn from the major Nepgrammarian principles at all. . On the contrary, as the new group of scholars gathered data about patois and about local or non-standard forms, they were still guided by the notion that a small number of carefully evolved phonetic laws would, as before, explain the behavior of linguistic change.. At the start, these men were con­ cerned with testing their new. data against the rules evolved by the older generation of philologists. Gillieron, 43 who knew that the Neogrammarian sound laws, had been formu­ lated for the standard, literary languages of the various European capitals, could easily demonstrate that the laws did not apply to sound changes among the patois; with his new data, he could show that the old laws (such as ”ch fran^ais provenant de c devant a latin, a l'eu issu de o long tonique libre" 13 ) did not apply to the patois. But with a rigorous and systematic approach similar to that of the Neogrammarians, Gillieron sought to investigate the reasons and then to formulate rules explaining why Gaston Paris was now wrong, why a patois in reality "ne remontent pas necessaireraent au latin par filiation directs. Ultimately, the linguistic geographers, who all had been trained in Neogrammarian theories, discoveries, and conclusions, came up with a new dimension to the old idea that "chaque mot a son histoire speciale."/ 1 5 They found that in the patois they were dealing with separable languages which had fixed phonemic and morphemic systems, and which had evolved from a cumulative linguistic substratum which itself varied according to the histories of pestilence, famine, war, and the movements of peoples in a given locale fir since Roman times. They found further that the phonetic systems of the French patois, for example, had interacted with each other to varying extents over the past 1900 years, and that the influence 'of the standard language of 44

the capital was felt as if the latter were a foreign lan­ guage. Such facts, together, with the presence or absence of geographical barriers, divided a country into numerous dialect areas, in which certain sound changes may or may not have taken place at the moment of the survey. In all, they found that they had to pay a more strict attention to Reactions reciproques entre la forme et le sens et influences analogiques d'une part, phenomenes d'ordre social,, echanges de formes, voyages des mots d'autre part,, etude interne et externe du ,g langage, et point de vue statique et evolutif and that, indeed, they could not be as sure of the "limites phonetiques" as the Neogrammarians were. Once the new perspectives were gained about patois ■ and what happened to them in history, however, the lin­ guistic geographers reverted to the patterned methodology of their mentors to explain "la distribution geographique des vocables— leurs migrations, leur extension, leur recul— les rencontres, les chocs, les alterations survenues au cours de ces voyages." 18 Their emphasis involved a recon­ sideration of the techniques and the scholarship of the past, and their highest praise was reserved for the man who could incorporate in his studies the learning of the past: J. Jud, esprit methodique et clair, epris de larges^syntheses#qui reposent sur des analyses fouillees et penetrantes, a applique la methode au passe et s'est efforce, en remontant .peu a peu, de reconstituer les sires anciennes, les . voyages des mots au moyeri age, les. stapes de la rbmanisation et les couches successives du latin vulgaire, les rapports d 1 interpenetration 45

. entre le latin et lea langues voisines, enfin les aireg des mots' ant^rieurs a la romanisation. In all, the main principles of the linguistic gepgraphers followed in a direct line from the thinking of the more rigid Neogrammarians. The newer group sought to bring to bear the knowledge of the past century of the proliferation of non-standard dialects; instead of working with wholly different standard languages, as the previous generations had done, they turned their attention to non-standard lan­ guages which were grouped within smaller geographical areas. As Albert Dauzat puts it in his Geographie linguistique: II s'agit— en s'aidant des criteres geographiques et sociaux, de la psychologie, populaire, des documents linguistiques ahci&ns et recents— de retrouver les lois qui ont preside aux transforma­ tions, aux creations, aux groupements, auxQ voyages, cH la vie et a la lutte des mots. Since the Neogrammarian thought, scholarship, and methods were preserved by the early researchers in linguis­ tic geography, it is not surprising to find that in 1925 in his The English Language in America George Philip Krapp, like Whitney and Ascoli, stated that "Dialects of all kinds are^merely the convenient summaries of observers who bring together certain homogeneities in the speech habits of a group and thus secure for themselves an impression of unity." 21 Eight years later, Leonard Bloomfield formulated a conclusion about dialect that was quite similar to Ascoli's notion: to Bloomfield, a dialect "presents a unique 46

combination of forms, each of which also appears, in other combinations, in some of the neighboring localities.1,22 In the various discussions of dialect which followed Bloom­ field, the emphasis centered on the interpretation that a dialect is not necessarily composed of phonemes or morphemes thait are unique to the area in question. Otto Springer demonstrated that any of the phonemic or morphemic features of a given dialect might occur elsewhere in the language, and he too favored Bloomfield's theory that a dialect is a combination of forms. 23 Then, in an essay on literary dialect, Sumner Ives developed this principle even further: At any one moment a dialect appears as eclectic rather than homogeneous; its individuality exists in the peculiarity of its combination of features, not in the peculiarity of the discrete features themselves. In fact, there may be no single feature which is found in^that dialect and that dialect alone. Ives then attempted to define a dialect and, after consid­ ering Frapp's definition, he came to this conclusion: A local dialect is not a homogeneous set ox speech conventions that differs from other homogeneous sets of conventions in each feature, but is rather a combination of features which are individually diverse in their distribution. Altogether, then, it is apparent that with their emphasis on eleven unique phonemic or morphemic features, Moore, Meech and Whitehall were working under assumptions 47

which were contrary to the evolving theory about dialect in the first four decades of this century. Evidently, from the very beginning of their researches, they gathered and analyzed data with the eventual aim of producing a number of discrete dialect characteristics each of which was unique to an isolable geographical area; with this end in view, they of course succeeded in their task. . Xet a number of questions arise about Moore, Meech and Whitehall's approach: were they aware of the implications of the Neogrammarian concept of dialect, a concept which had been freshly rephrased by Leonard Bloomfield only two years before they published their results? Were there practical reasons for choosing those particular assumptions about the nature of dialect which are implied in their decision to deal with single and discrete features instead of a com­ bination of features— that is, might they have felt that either their materials or their time were too limited to work in any other way than the one they chose? These, of course, are unanswerable questions; yet there is a more serious one which must be answered. Fifty years before, that sect of Neogrammarians headed, by Gaston Paris and H. Schuchardt had deplored the fact that "no objective, general criteria in the selection of .features" had been developed as yet by dialect theoreticians. 26 With Moore, Meech, and White­ hall , the move toward objectivity, has been overstressed: 48 in an effort to pare away all: nonessential data, to simplify all dialect criteria down to a scant eleven features, the spirit of scientific and analytic rigor may have been carried to an extreme. Eventually, the question arises: would it not be more desirable -to have a fuller , and ulti- ‘ mately more nearly adequate, picture of the character and arrangement of the linguistic features of the various Middle English dialects? The weaknesses in Moore, Meech and Whitehall's selection of dialect criteria have been recognized by Angus McIntosh: in addition to criticizing the "depress- ingly small amount of information" 27 offered by the eleven dialect characteristics and the entirely too small "number of items which they decided to record, and then plot,"28 McIntosh pointed out that the major failing of their approach had been "the absence of any ordered theoretical basis" 29 in their methods. This criticism is quite accurate; in 1935, no one had established any practical way of dealing with the theoretical basis which Bloomfield and his prede­ cessors had established. Yet the incipient notions for working with dialects as combinations of characteristics were established in 1931, for in that year Prince N. S. Troubetzkoy developed what proved to be the foundation of structural dialectology. 30 49

Troubetzkoy was concerned with the fact that much of the mapped data published, by Gillieron and the other early scholars in linguistic geography actually failed to give a precise picture of the location and extent in square measure of many dialect features, and that the maps quite generally provided no idea of the total number of phonetic features which, when taken in combination, would distinguish one dialect,;£rom another. Current cartographic presentations then, did not give a clear understanding of the true nature of dialect: La dialectologie moderne ou la geographie linguistique soutient que chaque mot isole qui pr&sente une modification phonique possederait ses gropres limites d*extension et que par consequent les limites d'extension geographique d'une modification phonique ne pourraiLent jamais etre tracees exactement et nettement. Not content with the notion that dialect maps could be excused for not giving an accurate indication of boundaries, Troubetzkoy decided to rearrange some of the emphases in the contemporary thinking about dialectology. He felt that the dialect material which had been collected and then plotted on maps was not being typed with an appropriate amount of linguistic refinement or sophistication, and that a dialect worker should decide for himself whether or not he wanted to gather phonemic., phonetic,, or etymological material, and then act accordingly. It was only when this threefold distinction was made, Troubetzkoy argued, that 50 an eventually accurate picture about the extent and nature of a given dialect could be drawn. Once Troubetzkoy made the distinction, he strove to show as clearly as possible all the complicated ways in which phonemic, phonetic, and etymological data might be interrelated. He first pointed out that II existe une difference phonologique d 'inventaire quand un dialecte possedeun phoneme qui est inconnue d'un autre dialecte. II existe une difference phonologique de fonction quand un phoneme appar^it dans un dialecte en une position phonologique ou il n'appar&it pas dans un autre dialecte. He further developed the notion that Les differences phonetiques peuverit etre absolues, si elles concernent la prononciation d'un phoneme en toutes positions, ou limitees (autre- ment dit combine to ires) si3elles n'apparaissent qu'en certaines positions. Finally, he set forth the principles that there are phonetic and etymological sub-types which can be distinguished even further: II ya <\es differences phoniques etymologiques qui se trouvent en liaison avec des differences phonologiques de fonction. C'est-a-dire qui si dans un dialecte la^fonction d'un phoneme determine est limitee en comparaison d'un autre dialecte, cela se produit habituellement au benefice d'un autre phoneme plus empire (dans les position ou le premier phon&me ne peut pas se trouver), ce par quoi la limitation de fonction du premier phoneme se trouve pour ainsi dire compensee. Dans de tels cas on peut parler de differences etymologiques compensatoires. Mais dans d'autres cas ou les differences etymologiques ne sont liees a aucune, difference de fonction, on peut appeler. des differences. Etymologiques libres. 51

After presenting this singular taxonomy, Troubetzkoy asserts that a particular stress on identifying and gather­ ing the phonemes, which excludes outright phonetic and etymological material, will enable the scholar to plot a more distinctive set of data, and to establish the size and features of the given dialect more fully. More precisely, with this terminology and with this set of differentiations, Troubetzkoy felt that the study of the behavior of dialect features in the zones of transition between dialects, or at short distances away from the isoglosses plotted on dialect maps, could be done more easily. He actually wanted to find out what the nature of linguistic activity was in areas where two dialects occupied the same, or nearly the same ground: he wanted to know what phonemic, phonetic, or etymological features, alone or in combination, were shared— if any— and why; he wanted to see how many new, general sound laws should be formulated, if any; and finally, he wanted to test as many of the aging Neogrammarian sound laws as he could. In all Troubetzkoy felt that this refined approach would enable a scholar to pursue a number of projects at the same time on a given dialect. Prince Troubetzkoy next developed a careful, graphic method for generating data about the zones of transition and the areas immediately adjacent to plotted isogloss lines. Starting first with the assumption that phonemes differ 52 either because of demonstrable historical change or because of the influence of neighboring dialects, and second, that "un phoheme ou une combinaison de phonemes peut dans un dialecte exister ou non-mais il n'y a pas de troisieme possibilite,"/ 35 he proceeded. by superimposing maps on which were plotted the variety, of forms in which a given phonemic feature occurred: Les isolignes de chaque mot isole presentant la t mutation phonique en question doivent §tre portees k part sur des cartes specia^es et ensuite ces cartes doivent §tre superposees; sur les cartes synthetiques ainsi obtenues les isolignes communes (c'est-a-dire qui coincident)^apparaissent sous forme de traits epais et fonces, et celles qui ne coincident pas sous forme de traits minces,et pales; les zones de transition son caracterisees par une accumulation de ces traits p&les, tandis que les,,-autres regions .... sont nettes de ces lignes.36 Troubetzkoy was not precisely clear about what he hoped to find by this method of superimposing maps with plotted phonemes, although he did offer some conjectures as to the results: if one accepts the notion that a given sound simply is or is not phonemic, then it is possible that some of the previously unclear ispglosses in zones of tran­ sition actually represent phones which are in the process of becoming phonemic, or vice-versa. That is, in a given dialect, a certain allophone may have become systematically separated in a series of words from the same allophone in another series of words, through population movements, or through the sudden rise to economic dominance of a given 53 class of people; as time passes, the allophone may become distinctive of meaning in its new phonetic environment, and it will then be,, by definition, phonemic.. Or, on the other hand, two previously separated allophones may merge for any of a variety of reasons, and the resulting single sound may become phonemic. In either case, Troubetzkoy felt that an unreadable isogloss line denoted an allophone or a phoneme in a state of change, and he concluded that "1'expression 'zone de transition* a done ici un tout autre sens que quand il s'agit d ’une difference etymologique. In summing up his ideas about this new approach to dialectology, Troubetzkoy placed special stress on the value for each dialect of assembling and identifying "11 inventaire phonologique et les fonctions des divers phonemes."\ 38 He was so sure of this method that he predicted quite remark­ able results; it would be possible not merely to show the relationship between dialects that shared allophones or phonemes which were being altered in the same phonetic environments, but one might also speculate about changes which were going on beyond the. borders of the dialect under examination: "1'etude des differences phonologiques peut etre poursuivre en dehors d'une langue et meme en dehors d'une famille de langues." 39 With the method of super­ imposition, then, a researcher need not balk at studying the behavior of phonemes in the same or similar phonetic 54

environments even though the overall phonetic, morphological, syntactical, or lexical systems of the languages concerned were quite different.: If it is not an outright break, then Troubetzkoy's theories about phonemes transcending language boundaries at least show an important departure from the structuralist theories that had been developed forty years before by Ferdinand de Saussure. In a quite general way, Troubetzkoy acknowledges his immense debt to the original structuralist theories of de Saussure when, in an autobiographical statement shortly before his death,, he speaks of the force of "1'influ- ence de l'ecolet de Ferdinand de Saussure" 40 that was felt by the new generation of linguists whom he led in Moscow in 1915. Troubetzkoy, in fact, was a great disseminator of Saussurean structuralist ideas among the Prague School of linguists in the late 1920's and early 1930's; But with this notion of phonemes shared across linguistic boundaries, Troubetzkoy is moving in a new direction on his own. Ferdinand de Saussure, as Troubetzkoy did later, made a tripartite division of language study and directed his analyses among "phonetics, phonemics, and historical phonetics.It was de Saussure's contention that these three divisions,, separately and together, constituted a "synchronic self-contained system" 42 within a language. In his work with phonemics, Troubetzkoy borrowed these 55 and a number of additional ideas from de Saussure, among them the notions that "two phonemes cannot occur at once,1143 and in de Saussure's words, "Un phoneme joue, par lui-merae, 44 un role dans le systeme\ d'un etat 4 de langue." But, as Rulon S. Hells has shown, apart from showing that a phoneme is not a "signe," but more accurately a "tranche de sonorite,"/ 45 de Saussure never did make perfectly clear just how phonemes are indeed systematic. Evidently, Troubetzkoy was in part addressing himself to this question, and his solution shows that he did not think that phonemes form a closed system within a given language or a given dialect; instead, his point about shared phonemes shows that if they are in any way describably systematic, phonemes in general must belong to a system that is non-unique, and that can be shared even among "une famille de langues." It is easy to see at least three of the dominant linguistic trends of the late nineteenth and early twentieth centuries in the work of Prince Troubetzkoy. From the linguistic geographers he preserved an interest in the methods of collecting and mapping dialect data; from the Neo­ grammarians he worked on under the impulse to account for sound changes and to use new material in order to evolve new sound laws; and from Ferdinand de Saussure and his followers he advanced the research into the structural interrelationships between the phonemes, morphemes, syntax, 56

and lexicon of a given language system. All these trends he passed on to the structural dialectologists of more recent years; yet the crux concerning the imputed systemic nature of phonemes— which, in essence, is the original problem about zones of transition and unclear isogloss lines- endures to this day. Now, however, the terminology has changed a bit, and modern scholars in the field are con­ cerned with questions of the diffusion, the continuity and discreteness, and the convergence and divergence of dialect features. For twenty years after Prince Troubtezkoy1s essay, there were only two or three studies, and these in Slavic languages only, 46 which made use of his ideas on structural dialectology; it was not until the early 1950's that his notions were generally received by linguistic geographers with an increasingly rigorous attention. Even so, as the dormant subject was opened up in the pages of Romance Philology, Orbis, and Word, some time passed before the full implications of Troubetzkoy's ideas were understood. Writing in 1952, Andre Martinet thought that the following -problems should be studied anew: all questions concerning the "bilingual situations arising from language diffusion,"A *7 and "how far two structures in contact can be maintained intact, and to what extent they will influence and modify each other." 48 Martinet felt that mutual influence between 57

languages or dialec-ts was the primary cause of linguistic change, and he suggested that all inquiry about phonetics or phonemics should be based on a knowledge of the demo­ graphic facts of the dialect in question: If we try to gather information as to the causes of the diffusion of a certain language, we shall probably arrive at the conclusion that it is nothing but a by-product of the military, political, religious, cultural, economic, or simply the demographic, expansion C of the people7 . . . a language does not get the better of its competitors because of its intrinsic qualities, but because it is that of a more warlike, more fanatical, more cultured, more enterprising, or more prolific community. Two years later, Martinet elaborated upon these views by introducing the terms "convergence" and "divergence," 50 which he placed into the perspective of nineteenth century Stammbaumtheorie and Wellentheorie notions. "Divergence" is another way of describing the development of dialects as, historically, they seem to branch out from an ancient, original source— the "family-tree" theory which August Schliecher proposed in 1961-2. "Convergence" occurs when a given dialect emerges in cultural, political, or econom« ical hegemony over its neighbors— a possibility discussed by Johannes Schmidt in his "wave theory" in 1872. In the same year as Martinet's article, Uriel Wein- reich at once confirmed his position as a structuralist and disposed of any awkward doubts about the possibility of shared phonemes among languages: "a language is an organ­ ized system, unique and closed, whose members are defined 58

by opposition to each other, and not by anything outside of the system." 51 With this emphasis, Weinreich neglects or ignores one of Nikolai S. Troubetskoy1s major ideas; never­ theless, he proceeds to build on Troubetzkoy's point about compared dialects by developing new terminology and a new hypothetical perspective on the subject. Like Troubetzkoy, Weinreich felt that the zones of transition between dialects were actually areas in which a complicated linguistic flux was in progress; with his new methods, however, Weinreich believed that it is possible to undertake "the study of gradually changing systems." 52 To do this, he proposes that the researcher first, take as complete an inventory of the given dialect's phonemes as possible; and second, compile as much information as he can about the differences in distribution of these phonemes. 53 All of this implies, of course, that the researcher has taken careful account of the character of all allophones, of the facts that vowel length, or the positional or combinatory variants of the vowel may be phonemic, and of "the train of associations, oppositions, and functions that define linguistic forms." 54 Once such an inventory and compilation is made between the dialects in question, Weinreich goes on to explain that then Dialectology would be the investigation of problems arising when different systems are treated together because of their partial similarity . . . its special concern /would bg7 the study of partial similarities and differences between systems and of the structural consequences thereof. 59

To facilitate this investigation, Weinreich suggested that a heuristic comparison be made with "discrete and homo­ geneous systems that are derived from description and that represent each a unique formal organization of the substance of expression and content."56 This comparison, or the act of comparing, itself, is done with a "diasystem," which is constructed, again, by "placing discrete varieties in a kind of continuum determined by their partial similiarities;" 57 yet the peculiar value of a diasystem is that it "can be constructed ad hoc out of any number of. varieties /or, dialects/ for a given analytic purpose."58 It is its par­ ticular heuristic quality which makes the diasystem a valuable methodology for the dialectologist, and in all Weinreich was confident that his method would lead to "the formulation of a technical concept of 'dialect' as a variety or diasystem with certain explicity defining features." 59 Whether this will indeed occur or not remains to be seen; at any rate, the diasystem concept would seem to be the next logical step in manipulating the data produced through Troubetzkoy's technique of superimposing phoneme maps. Over the next few years, the theoretical foundations of structural dialectolggy began to solidify. Scholars ignored Weinreich's stricture about the inflexibility of phonemic systems, and returned to Prince Troubetzkoy's observations about shared phonemes. Edward Stankiewicz 60 noticed that the "'mixed' dialects in the border areas of Poland . . . share their phonemic systems with one area and their grammatical systems with another,"®° and upon close analysis of this problem, he first found that phonemes can form and disappear in ways anticipated by Roman Jakobson1s terms "dephonemic ization," "phonemic ization," and "rephonemicization," 61 and second, he developed a series of general rules for investigating phonemic behavior. He noticed at once that "speakers are known to take over distinctive as well as non-distinctive features from their neighbors," 62 and that there are "zones of transition in which some phonemic distinctions are not obligatory but optional, and as such endowed with social connotation;" 63 primarily, however, he found that the less stable phonemes within a zone of transition occur in what he chose to call a "mixed" dialect, and the more stable phonemes occur in "core" dialects. With this distinction between mixed and core dialects, Stankiewicz hoped to be able to classify any phonemic system, no matter how partially developed or fluctuating it might be, and he evolved another general rule which stated that the 'mixed* dialects may overlap with the 'core' dialects at some points or on some levels of their phonemic systems, but they lack the overall complex of features which enables us to view the 'core' dialects.as varieties containing certain phonemic types. 61

Still more of the earlier concepts of structural dialectology were maturing rapidly. Troubetzkoy's emphasis on the need for phoneme inventories was reaffirmed, and Stankiewicz felt that it would be necessary to go even further than that, for "It is . . . insufficient to compare systems simply in terms of their inventories; such compari- sons do not amount to more than listing." 65 Stankiewicz believed that all studies in structural dialectology should also classify the inventoried phonemes according to the linguistic; features first described and detailed by Roman Jakobson: 6 6 thus, there should be taken into account such features as the differences in consonality, compactness, gravity, voicing, nasality, and stridency between dialects. Like Weinreich, Stankiewicz. emphasized the problems of dis­ covering why and when certain phonemes change in zones of transition, although in his terminology the difficulty is one of "defining the relation between continuity and dis- creteness," 6*7 or of "defining continuity and similarity between discrete local systems, and of grouping them into higher types." 68 He believed that the structural dialec- tologist's task was to develop a method which provides for the simultaneous study of discreteness and continuity; and indeed, like Weinreich, he felt that any ad hoc method would do. If the problem involved somehow gaining "objective criteria for determining differences between linguistic systems" 69 — that is, of "stating the validity" of boundaries between dialects and of "calibrating the degree of similarity" between them— then any "fictitious construct"7 0 was permissible as long as it served the purpose. To be brief and oversimplified, then, structural dialectology became an attempt, using any means, to "determine criteria for the grouping of dialects."71 With Stankiewicz's article on the subject, many of the basic tenets of structural dialectology were established in the years since, theoreticians writing on this topic have sought to clarify a miscellany of minor points. Ob­ serving that neighboring dialects during the passing of time may evolve in structurally different ways, Jean Fourquet has reexamined certain Neogrammarian ideas about dialect, 72 and he and Pavle Ivic^ have independently arrived at the notion that the phenomena of shared phonemes, and of any features at all which might be ambivalent in two or more systems, might best be understood if we realize that "in most cases we have elements of the 'wave' and the Stamm- baum picture co-existing in the same landscape." 74 In other words, if the wave theory is taken into account, the result will show that— to push the metaphor beyond its limit— the dialectal acorns dropping off the Stammbaum- theorie oak will not produce identically structured trees. In his most recent work, Pavle Ivic repeats Troubetzkoy's 63 stress on the notion that the superimposition of phoneme maps may allow new sound laws to he formulated all the more easily; 75 further, he suggests that a statistical approach be adopted in order to give this new discipline greater objectivity: On ne pourra obtenir un groupement definitif des dialects de chaque langue avant d'avoir calculi les indices numeriques des similitudes et differences mutuelles-gbas^s sur les donnees des atlas linguistiq.ues. Ivic, finally, goes on to propose that the scope of structural dialectology is needlessly restricted by. the concentration on phonemic inventories alone; like the transformationalists, he feels that morphemic inventories could also be established as dialect criteria.^ Of course, this new approach to dialectology is not without its critics. Raven X. McDavid has argued that, in effect, despite the continuing researches on the subject we still do not have an adequate definition of a phoneme or its precise function, and that as a result, structural dia­ lectology has yet to deal accurately with the following questions: 1. Do compromise-phonemes §ls yet undescribed or unacknowledged? exist? 2. Can we by definition phonemicj.ze for more than one dialect,, or even one speaker, at a time? 3. Can we expect phonological data to fall into a single-over-all pattern £hat is, a single systemJ? 64

Obviously, speaking from within an entrenched structuralist conservatism, McDavid has remained unconvinced of many of the assumptions that most structural dialectologists have accepted since the work of Troubetzkoy. Modern researchers in this field, however, have not been troubled by the penetrating questions of the more literal-minded among current structural linguists. Instead, they plunge in and do their research anyway, confident that their results will be valid if they follow principles as thorough as those of Alf Sommerfelt: The complete structural investigation of a language consists in drawing up a phonemic inven­ tory , in explaining how the phonemes are grouped, how they may be characterized by traits common to many of them, and what the prosodic features may be; the morphophonemic system, the grammatical . role of the phonemes, and, finally, the grammatical forms themselves and the character of the vocabulary. The amount of research involved in this approach is extremely laborious and time-consuming, and the number of full-scale structural treatments of dialects is small. 80 However, shortly after his theoretical work on the subject was published, Edward Stankiewicz completed a model study using the principles of structural dialectology^ and I will turn to this now to show just how the technique works in action. The first remarkable point about Stankiewicz's essay, which he titles "The Phonemic Patterns of the Polish Dialects, A Study in Structural Dialectology,11 81 is that 65

he does treat all the dialects of Poland in an article only thirteen pages long; the information in such a study can be organized so as indeed to provide multum in parvo. Secondly, he does not give an indication of how he collected the data for his dialectal comparisons, whether by culling dialect maps, or by superimposing them, or whatever; the important methods in his inquiry instead involve a statement about phoneme inventories, a list of ancillary festures used in the comparison, and a series of further statements about the location of any dialect characteristic that needs to be discussed. In all, the article is a piece of succinct scientific reporting in which results are stressed more than methods. Stankiewicz begins by stating his purpose: "to describe the phonemic patterns of the Polish dialects in terms of their distinctive features and to compare the inven­ tories of vocalic and consonantal phonemes in various Polish dialects." 82 Since structural dialectology works through comparison, the researcher has a choice of comparing any two dialects at a time, or of selecting one dialect as a constant or a base, and then firing others with it, one at a time; Stankiewicz chose the latter method and he selected the standard Polish literary dialect as his base. He then stated the phonemic inventory of that dialect: there are 35 consonantal phonemes and 6 vowel phonemes, the former 66 divided among dental, alveolar, palatal, velar, liquid, and semi-vowel types, and the latter just listed as vocalic O O phonemes. Stankiewicz next set up a chart, or "matrix" of the linguistic features discussed by Roman Jakobson, and with­ in this matrix Stankiewicz recorded the articulatory pro­ perties of the phonemes of standard Polish. Using the matrix, then, one could qualitatively describe a given phoneme, like /pA by one or more tof the following of Jakobson's linguistic features: vocalic or nonvocalic; consonantal or noncon- sonantal; compact or diffuse; grave or acute; nasal or oral; continuous or interrupted; strident or mellow; voiced or voiceless; or, finally, sharp or plain. Thus, with these lists of phonemes in the standard dialect, of consonant types, and of linguistic features, Stankiewicz has supplied himself with a complicated set of interrelated variables which he can then use to describe all of the many non-standard dialects. With this descriptive apparatus outlined and organ­ ized, he proceeds. He shows, for instance, that in a certain locale the dialect may have 23 or 24 consonant phonemes and 9 or 10 vocalic phonemes; further, he shows that in contrast to the standard pronunciation, the alveolars may be mellow and the palatals may be strident in this dialect. In the next dialect, which may have 30 consonant phonemes and 8 vocalic phonemes, he may show that there is no phonemic distinction between "palatalized and non-palatalized labials." 84 67

Or, in another dialect, which may have 37 consonant phonemes and 5 vocalic phonemes, he may find that the velars are quite nazalized and that the liquids are vocalic. And so he continues, region by region, until he has treated the similarities and the distinctive differences among all the dialects in the country. At the end of his article, the amount of data he has worked with allows him to formulate a general rule: Maximal vocalic sub-systems are incompatible with maximal consonantal subsystems, and, conversely, minimal vocalic sub-systems are g5 incompatible with minimal consonant sub-systems. In all, the complex and highly developed descriptive system which Stankiewicz built for this inquiry has enabled him not only to type the various dialects of Poland, but also to advance an apparently solid hypothesis about the present state of all the dialects at once. Structural dialectology is at the relatively young age of 36; another approach, which is comparatively embryonic in its development, is potentially far more complicated. Drawing upon the theoretical, foundations laid by Noam Chomsky's Syntactic Structures,86 Morris Halle, writing .in 1962, suggested that the concepts of generative grammar be applied to dialect study. Halle felt that the emphasis of dialect geographers and structural dialectolpgists on the primary data of phonemes and morphemes had allowed a stagnant condition to evolve in modern dialect study, and 68 he proposed that other linguistic norms be investigated instead. Given the general definition of generative grammar as "a collection of statements, rules, or axioms which describe, define, or generate all well-formed utterances in a language and only those," 87 Halle submits that a new dialectology might concentrate "on the grammars of the dialects, i.e., on the ordered set of statements that describe the data, rather than on the data directly." 88 Within the year, Samuel Jay Keyser adopted a few generative principles in a review of Hans Kurath and Raven I. McDavid's The Pronunciation of English in The Atlantic States. 8 9 Keyser investigated the mix of data in some selected transition zones (which Kurath and McDavid call "phonemic heteroglosses") and then developed a set of two rules, which may apply in 1-2 or 2-1 order, governing the behavior of the phonemes /ai/, /au/* and /a/ in Rochester, Hew York, Winchester, Virginia, Hew Bern, Horth Carolina, and Charleston, South Carolina. 90 Then, in 1965, Sol Saporta reexamined Halle's hypotheses and tested them on the variety of noun plurals in the Castilian, Latin American, and South Chile dialects of Spanish. Saporta agreed with Halle that the "comparison of the underlying rules may be more revealing than comparison of the primary data alone.” 91 More precisely, Saporta found that dialect differences in many cases can be codified into rules which can be as simply 69 stated and yet as inflexible as the old sound laws of the Neogrammarians: The grammatical description of a given dialect may be converted into an adequate description of a related dialect by the addition, deletion, or reordering of a relatively small number of rules. Indeed, it is tempting to propose that the degree of difference between dialects is nothing more than a function of the number and type of such changes. Then, like Keyser, Saporta discovered that the sequence of generated rules may have to be altered to explain certain phenomena, so that The grammar of two speakers with different dialects will differ then, in one of two ways: either the grammars will have differ­ ent rules, or the grammars will have the same rules in a different order. In all, Saporta shows that it is quite possible to set up a few grammatical rules which a given dialect will systematically follow, and which will distinguish that dialect from all others. Generative grammar is a technique which is still developing, and any quite detailed grammatical investiga­ tion along transformational lines must develop equally from the spirit and the letter of the law as set forth by Noam Chomsky. Now and then, well before all the transformational problems have been posed, let alone solved, attempts have been made to generate rules and formulas upon historic linguistic data. But even though these researches may be premature, they should be praised and encouraged; the 70 validity of such works depends on the thoroughness with which they treat the primary data. Rodolfo Jacobson has written a dissertation in which he tries to Construct a grammar of the Middle English London dialect of the second half of the Fourteenth century in the expectation that such a grammar will eventually lead to linguistically adequate conclusions regarding the whole historical development of English. Jacobson admits that he has had to produce an "expansion" of some of the transformational concepts of Chomsky and his followers; but this of course in no way invalidates his study, for if the work is thorough, at best it will be entirely correct and even classical, and at worst it will be autotelic. The point is that if one is thoroughly grounded in the work of Chomsky, Morris Halle, and Charles J. Fillmore, 95 it is entirely possible to transpose transformational methods and techniques into any era for language study and, if one has enough data, for dialect study. Altpgether, the dialect theory of the last eighty years exists in a continuum which itself is characterized by a number of set linguistic principles. The prpgress from the combination-of-features theories to dialect geography, structural dialectology, and generative rules depends upon the assumptions that the inductive method is closer to producing the truth than any other, and that the external forms of a language imply a set of disciplined internal storms. Structural dialectology and generative grammar 71 presume first;, that a dialect is a system, and second, that the Neogrammarian principle that exceptions can be explained by research and by the future formulation of rules still holds true. As is true of any inductive processes, all of these theories depend upon a great deal of primary data in practice. Just what sort of primary data is now available for the study of Middle English dialects will be the subject of the next chapter. 72

NOTES

1. W. D. Whitney, Language and the Study of Language (New York, 1867), p. vi. 2. Ibid., P* 175 00 H 3. Ibid., P« • 4. Ibid., P* 178 5. Holger Pederson, Linguistic Science in the Nineteenth Century; Methods and Results. Trans, by John Webster Spargo (Harvard, 1931), p. 295. Pederson also speaks of the Neo- grammarian stress on the influence of loan words and spelling pronunciation on phonetic change (p. 295). 6. Edward Stankiewicz, "On Discreteness and Continuity in Structural Dialectology," Word, XIII (1957), p. 50. 7. On p. 50 Stankiewicz records Paris as having summed up the opinion of the times as follows: "In reality there are no dialects.” Paris himself doubtless did not believe this (see below). 8. Albert Dauzat, La geographie linguistique (Paris, 1922), p. 63. 9. Ibid., p . 6. 10. Ibid.-, p. 6. 11. Quoted on p. 28 of Sever Pop's La Pjalectologie, Vol. I (Louvain, 1950). Zola, Maupaussant, George Sand, and Victor Hugo, among novelists and authors of short fiction, were also concerned with recording dialect differences at this time. 12. These works are mentioned in Dauzat, pp. 15ff. 13. Dauzat quotes GiH4eron on p. 52. 14. Dauzat, p. 53. 15. Quoted in Dauzat, p. 55. 16. This quotation is found on Dauzat, p. 60, and the preced­ ing discussion is developed from Dauzat, pp. 54-60. 73 17. Dauzat, p. 57. 18. Ibid., p. 51. 19. Ibid., p. 23. 20. On p. 31, In making these brief remarks, I have relied upon the condensation' and succinctness in Dauzat's treatise. How­ ever, for the most detailed account of dialect geography yet available, one must turn to the encyclopedic compilation by Sever Pop, Ik* Dialectologie, Apercu historiaue et methodes d 1incm^tes lincnaistiaues. 2 vols., Louvain, 1950. This work discusses names, dates, and scholarly achievements and their significance throughout the history of linguistic geography. In the Introduction to Volume I is a history of linguistic geography from 1200 to the present; from 1800 on, developments in this discipline are reviewed decade by decade. In the rest of this massive work, Pop gives specific histories, to­ gether with accounts of current projects, in the field of dialectology in the following countries and languages: France, Portugal, Italy, Dalmatia, Sardinia, Romania, Germany, Switzerland, Luxemburg, Belgium, Denmark, Sweden, Norway, Great Britain, the United States, Canada, Finland, Estonia, Hungary, Greece, Albania, China, India, and Korea, with special surveys of the Franco-Provencal, Catalan, Ladin, Celtic, Slavic, Berber, Bantu, and Arb tongues. Obviously, the scope of this scholarship is enormous; the volumes could be best described as a giant's handbook of dialect geography. In addition, of major interest is the conclusion, where Pop discusses the feasibility and desirability of an International Center as a sort of clearing house for dialect information and research. 21. George P. Krapp, The English Language in America (New York, 1925), p. 226. 22. Leonard Bloomfield, Language (New York, 1933), p. 325. 23. Otto Springer, "Dialektgeographie und Textkritik," PMLA, LVI (1941), p. 1168. 24. Sumner Ives, "A Theory of Literary Dialect," Tulane Studies in English, II (1951), p. 142. 25. Ives, p. 143. 26. Stankiewicz, op. cit., p. 50. 74

27. Angus McIntosh, "A New Approach to Middle English Dialectology," ES, XLIV (1963), p. 1. 28. Ibid., p. 2. 29. Ibid., p. 1. 30. Originally, Troubetzkpy's article was "Phonologie et gdographie linguistique," in Travaux du Cercle Linguistique de Prague, IV (1931), 228-34. I.refer to the reprint of the article untranslated in J. Cantineau's translation o£ Troubetzkpy's Principles de Phonologie (Paris, 1964), found on pp. 343-50. The edition will be referred to hereafter as "Cantineau." 31. Cantineau , p. 345 32. Ibid. P- 343. 33. Ibid. P* 344. 34. Ibid. P* 344. 35. Ibid. P- 346. 36. Ibid. P* 347. 37. Ibid. P- 346. 38. Ibid. P- 348. 39. Ibid. P- 349. 40. Ibid. P- xix. 41. Rulon S. Wells, "De Saussure's System of Linguistics," in Martin Joos, ed., Readings in Linguistics (Washington: ACLS, 1957), p. 2. 42. Wells, p. 22. Co 43. Wells phrases this idea on p. 4. 44. Wells, p. 7. 45. On pp. 6-8 of his article. Wells demonstrates that de Saussure's ideas about how phonemes were intrinsically sys- timatic were quite uncertain and sketchy. Wells shows that the relevant statements which de Saussure makes are quite few, and that his most straightforward declaration^on the subject is "Les phon&mes sont avant tout des entites opposi- tives, relatives, et negatives•" (p. 7)• Wells then quotes 75

the following passage from de Saussure, ana remarks that this is the most extended treatement of the systemic nature of phonemes which de Saussure offers anywhere in the compiled Cours de linguistique: Un phoneme joue par lui-meme un role dans le systeme d'un etat de langue. Si par. exemple en grec m, £, t, etc., ne peuvent jamais figurer a la' fin d^un root, cela revient k dire que' leur presence ou leur absence k telle place compte dans la structure du mot et dans celle de la phrase. Or dans tous les cas de ce genre, le son isol^, comne toutes les autres unites, sera choisi k la suite d'une opposition mentale double, ainsi dans la groupe imaginaire a n m a , le son m est en opposition syntagmatique avec, ceux qui 1'entourent et en opposition associative avec tous ceux que l 1esprit peut suggerer, soit a n m a . (Wells, p. 8; he quotes from section . 180 b of the Cours de linguistique.) As I show in the text, Troubtedskoy's researches led him to modify de Saussure*s ideas toward a concept which holds that phonemes are shared between neighboring dialectal systems. 46. The studies are as follows: Lazrczius, G. Bevezetes a fonologiaba. Budapest, 1932". Havranek, B. "Nareci ceska," Ceskoslovenska Vlastiveda, III (1934), 84-218. Weinreich, Uriel. "Sabesdiker laen in Yiddish: a Problem of Linguistic Affinity," Word, VIII (1952), 360-77. 47. Andre Martinet, "Diffusion of Language and Structural Linguistics," RPh, VI (1952), p. 6. 48. Ibid., p. 7. 49. Ibid., p. 5. 50. Andre Martinent, "Dialect," RPh, VIII (1954), p. 2. The following discussion is developed from Martinet*s remarks on p. 10. 76

In Holger Pederson’s Linguistic Science in the Nine­ teenth Century, trans. J. W. Spargo, Harvard, 1931, there is an account ofthe impact of Schliecher's Starombaumtheorie (from his Compendium der yergleichenden Grammatik der indogermaniscnen Sprachen) on pages 272 and 311; there are remarks evaluating Schmidt's Wellentheorie (from his Pie Verwandtschaftsverhaltnisse der indogermanischen SpracEen) on page 315. — 51. Uriel Weinreich, "Is a Structural Dialectology Possible?" Word, X (1954), p. 388. 52. Ibid., p. 391. 53. Ibid., p. 393. 54. Ibid., p. 394. 55. Ibid., p. 390. 56. Ibid., p. 390. 57. Ibid., p. 395. 58. Ibid., p. 395. 59. Ibid., p. 399. 60. Stankiewicz, op. cit., p. 45. 61. Ibid., p. 58. 62. Ibid., p. 58. 63. Ibid., p. 55. 64. Ibid., p. 58. 65. Ibid., p. 53. 66. Stankiewicz's discussion is on p. 59. Jakobson develops his theories about linguistic features with M. Halle.and C. G. Fant in Preliminaries to Speech Analysis, MIT, 1952. I detail the precise linguistic features wnich Stankiewicz adopts for his purposes on p. 23 of this chapter. 67. Stankiewicz, p. 48. 68. Ibid., p. 52. 77

69. Ibid-r P- 51. 70. Ibid., p. 48. 71. Ibid., p.46. 72. Jean Fourquet, "Classification Dialectale et Phonologie Evolutive,” In Diego Catalan, ed., Miscelanea Homenaje A'Andr^ Martinet, Vol. II (Madrid, 1956), p. 6l. 73. Pavle Ivic, "On the Structure of Dialectal Differentia­ tion,” Word, XVIII (1962), p. 45. 74. Ibid., p. 45. 75. Pavle Ivic, "Importance de caracteristiques structurales pour la description et la classification des dialectes," Orbis, XII (1963), p. 120. 76. Ibid., p. 128. 77. Ibid., p. 121. 78. Raven I. McDavid, "Structural Linguistics and Linguistic Geography," Orbis, X (1961), p. 45. 79. Alf Sommerfelt, "The Structural Point of View. Applied to the Analysis of the Consonant System of a Donegal Dialect,” Celtica, V (1960), p. 112. 80. In addition to Sommerfelt's article (mentioned in note 79), the following studies may be cited: Moulton, William G. "The Short Vowel System of Northern Switzerland, A Study in Structural Dialectology," Word, XVI (1960), 155-182. • ______. "Phonetische und phonologische Dialektkarten: Beispiele aus dem Schweizer- deutschen," a report to the First International Congress of General Dialectology. _____ . "A Phonetic Change in its Phonemic Setting: Old long & in Modern Swiss German," report read at the 75th Annual Meeting of the MLA, Dec. 28, I960. 78

Moulton's work in 1960 was anticipated by. two earlier articles: Stankiewicz, Edward. "The Phonemic Patterns of the Polish Dialects, A Study in Structural Dialec- . fcology," in Morris Halle,, ed ., For Roman Jakobson (The: Hague, 1956) , 518-30. Klagstad,: Harold L., Jr., "A Phonemic Analysis of Some Bulgarian Dialects," in American Contribu­ tions to the Fourth International Congress of Slavicists (The Hague, 1958), 157-68♦ 81. Found on pp. 518-30 of Morris Halle, ed., For Roman Jakobson, The Hague, 1956. 82. Ibid., p. 518. 83. Ibid., p. 520. 84. Ibid., p. 522. 85. Ibid., p. 528. 86. Published by Mouton in The Hague, 1957. 87. Morris Halle, "Phonology in Generative Grammar," in The Structure of Language, Readings in The Philosophy of Language, ed., by Jerry A. Fodor and Jerrold J. Katz (Englewood Cliffs; Prentice Hall, 1964), p. 334. Originally, this article was in Word, XVIII (1962), 54-72. 88. Ibid., p. 342. 89. The review occurs in Lang, XXXIX (1963), 303-16. 90. Op. cit., p. 310. 91. Sal Saporta, "Ordered Rules, Dialect Differences, and Historical Processes," Lang, XLI (1965), p. 219. 92. Ibid., p.219v 93. Ibid., p. 218. 94. Rodolfo Jacobson, "The London Dialect of the Late Four­ teenth Century. A Transformational Analysis in Historical Linguistics," DA, XXVII (1967), 2141-A. 95. The book of Chomsky is mentioned in note 86 above, and the essay by Halle in note 87. The pertinent works by Fill- more are: Fillmore, Charles J. "The Position of Embedding Transformations in a Grammar," Word, XIX C1963), 208-231. ’______. Indirect Object Constructions in English and the Ordering of Transformations.- Monographs on Linguistic Analysis, T~. The Hague: Mouton, 1965. CHAPTER III PHONOLOGY, ONOMASTICS AND GRAPHOLOGY: PRACTICAL RESULTS ON MIDDLE ENGLISH STUDIES OF MODERN DIALECT THEORY

As I suggested in Chapter Two, the increasingly sophisticated development of dialect theory since the Neo- grammarians of the 1880's has had a number of effects upon the nature of traditional and modern dialect research. The new dialect theories, in a variety of forms, have influenced the current scholarship in Middle English in one obvious way: the methods of linguistic geography used by Moore, Meech, and Whitehall have produced for some scholars a newly organized, geographical frame of reference for the students of historical phonology. Further, more recent hypotheses from both the linguistic geographers and the structural dialectologists may eventually give the historical phonologist an even more accurate idea of. how phonemes change, and of what phonemes change in which environments. For those lan­ guage researchers who are trying to draw a more nearly complete picture of Middle English dialect characteristics than the one provided by Moore, Meech, and Whitehall, modern

80 dialect theory has demanded more and more extraordinary pains in the collection, and then in the analysis, .of pri­ mary linguistic data— the phonemes and morphemes of the dialects in question. In their search for more primary data, scholars have employed the analytical techniques of onomastics and graphology in order to provide more refined criteria for distinguishing one dialect system from another, and they have also adopted the attitude that no manuscript with a Middle English spelling in it can be neglected in their language researches. In this chapter, after reviewing the relation of the new theories to developments in histori­ cal phonology, I will discuss those recent methods and techniques of dialect research which the new theories have brought into use since 1935. Because the interests and the methods of the nine­ teenth century linguists have endured to this day, it is in the field of historical phonology that the greatest advances in the analysis and classification of dialect material have been made. The immense achievements of Karl Luick,^ for example, in the investigation of historical ^change in consonants and in vowel length and quality in the early stages of the English language, have had lasting influence on the nature and scope of modern language study; each year, more of Luick's hypotheses and conclusions are tested, and supplemented or readjusted by current historical 82 phonologists. In simple fact,, moreover, from most per­ spectives the very study of past dialects is largely a study of historical phonology, for it is chiefly through the various problems involved in vowel change that the features differentiating one dialect from another can be recognized and understood. Unfortunately, historical phonology and the more recent dialect theories have developed separately, and they continue to do so; yet some few advances have been made by scholars who are knowledgeable in both the tradi­ tional and modern fields of dialect study. It is on the relationship of historical phonology to the modern dialect trends of linguistic geography and structural dialectology that I will base my remarks in the first part of this chapter. After the work of A. J. Ellis and H. C. Wyld 2 and before the monography by Moore, Meech, and Whitehall, English scholars desiring a geographical organization and background in their phonological studies relied on the mention of vague regions, or directions toward those regions, whenever they wished to discuss the position or location of a sound change. Despite the research of Ellis, there was still no standard and consistent practice for naming a geographical region, nor was there a common understanding about which counties or parts of counties in England comprised what region. In an article on "The Dialects of the West Midlands 3 in Middle English," Mary S. Serjeantson deals with an 83 area the geography of which she traces in a fashion very much like Moore, Meech, and Whitehall: her North-Western, Central-Western and South-Western groups have a north-south boundary quite like that of the South-central West-Midland, North-central West Midland, and Northwest-Midland districts of the later University of Michigan monograph. However, 4 Miss Serjeantson also mentions a "north and central" region whose boundaries she does not specify, and further, she is inclined to identify dialect boundaries with the political boundaries of the various counties— a practice which more recent evidence has tended to discredit. 5 In his History g of Modern Colloquial English, Wyld himself, on the other hand, deals only with the dialect designates of Northern— 7 "including the speech of the Scottish Lowlands” — East and West Midland, Southwestern (Herefordshire, Worcestershire, Shropshire, and Oxfordshire), Central Southern (Berkshire and Hampshire), and South-Eastern (Kent and Essex). Aside from adding that each of the Midlands dialects "varies in the Q northern and more southerly areas,” Wyld gives no specific details about the boundaries of these dialect districts. Obviously, he did.not feel .it necessary to be precise about boundaries, and perhaps he believed that a rather inexact geographical framework best suited his purposes. It is surprising, however, to find that essays in historical phonology since 1935 which might benefit from a geographical perspective 9 have been indifferent and neglect­ ful, and perhaps in some cases willfully ignorant, of the work of Moore, Meech, and Whitehall. Some scholars, of course, have indeed taken advantage of that monograph. It was Harold Whitehall in an article in 1941 who (and one would not expect otherwise) first made good use of the new regional breakdown of Middle English dialects. In an investi­ gation of "The Development of Middle English u in Early Modern British and American English," 10 Whitehall traced the behavior of /tr/ in New England, and in its British antece­ dents. Rather than try to draw any direct relationship between the New England dialect and a precise geographical forebear in England,^ Whitehall dealt in general with the three "East- Midland varieties" 12 of Middle and Early Modern English, and then drew his conclusions with a degree of accuracy in the positioning of dialects that was impossible before 1935. Then, although he mentions only the "North," "South," and the "Midlands" as dialect areas, David W. Reed in 1949 referred to "forty definitely localized and dated texts between c950 and 1500 and . . . twenty non-localized texts from the 11th and 12th centuries, when few localized texts were available," in his "History of Inflectional tt in English Verbs. His method is quite similar to Moore and Meech's, and since this dissertation was written at the University of Michigan, we.-may suppose that he was referred to the dialect treatise 85

in question. Next, in 1956, Hans Kurath scrutinized materials from the "Northeast Midland and Northern areas" 14 in a dis­ cussion of "The Loss of Long Consonants and the Rise of Voiced Fricatives in Middle English." All of these scholars benefited from the greater refinement of material and method made possible by Moore, Meech and Whitehall. Other scholars, for perhaps a variety of unstated reasons, do not believe that a geographical breakdown of dialects has any value for phonological studies. 15 Thus, in his "Early Loss of /r/ before Dentals," 16 Archibald A. Hill contends, in a country by county survey and work list, that "a general sound change" 17 resulted in the pre-dental loss of /r/ between 1300 and 1500; Bertil Sundby bases his conclusions concerning "Middle English Overlapping V and W and its Phonemic Significance" 18 mostly on his researches among documents from the single county of Worcestershire; Robert P. Stockwell brings no consideration of dialect to his "The Middle English "Long Close' and 'Long Open' Mid Vowels," 19 and thus neglects the possibility that there may have been dialect variations in the behavior of Middle English mid vowels; and A., S. Liberman attempts a dis- cussion of only the East-Midland dialect (he deals with one, not the three of Moore, Meech, and Whitehall) and those regions "south of the Humber” in his essay "On the History of Middle English A and A." 20 In other words, given that 86

Moore, Meech, and .Whitehall's monograph can serve as a simple taxonomy of dialect regions, historical phonologists have still to use it to its fullest potential as a guide for detailed studies of phonemes and for corollaries to existing sound laws. Another way to state the problem is this: there is at present little coordination between the field of histori­ cal phonology and the contributions of linguistic geography to Middle English dialects. Yet this is not true everywhere; at the University of Michigan, Hans Kurath, Sherman Kuhn, and others producing the Middle English Dictionary are applying their phonological findings in a continuation of the work of Moore, Meech, and Whitehall. Kurath*s notions of the forces producing dialect change have appeared in print many times in recent years. While linguistic geography has matured as a discipline, it has expanded the scope of its inquiries beyond the range of what Kurath calls "pure linguistics" 21 — the study of phonemes and morphemes exclusively — to include eventually all those demographic facts which may bear upon the linguistic content of the dialect in ques­ tion. In the past,. &urath notes, scholars referred to sound laws in order to explain such complicated developments as how phonemes split or merge, how allophones of one phoneme come to be subsumed under another 87

phoneme, how native .phonemes are replaced by foreign phonemes, how foreign phonemes are adapted to the native system, and under what conditions foreigh phonemes are adopted. But then, in his work on the Linguistic Atlas of New England,23 Kurath found that in addition to a knowledge of sound laws, other kinds of information were needed in order better to illustrate the way in which new phonemes arise or are lost,, how the incidence of phonemes in the vocabulary changes, and how the phonic character of phonemes is modified. In some instances, the immediate drive for a change can be safely inferred from the regional and/or social dissemination of the variants; in others, a probable source of the change can be pointed out by considering demographic factors such as settlement, migration, population and cultural centers, the British background, and so on. Thus, in contrast to those who would account for all phonemic and morphemic behavior through established laws, Kurath holds that language is a historical product shaped by com­ plicated cultural and social processes, and there­ fore it cannot be wholly systematic at any time. The diachronist will want to relate the changes in language to the history of the speakers when­ ever possible; and the linguistic geographer should not stop short of interpreting heteroglosses, dialect boundaries, and speech areas in terms of demographic areas and forces of one kind or another. Kurath is not alone in his emphasis on the influence of extra-linguistic conditions upon phonetic and phonemic behavior. W. A. Grootaers and Y. Mase, for example, have studied the geographical and historical facts which have in part caused the partitioning of dialects in certain areas in Japan; 26 yet in asserting that socially and culturally all dialects are village or hamlet-centered, Grootaers and Mase have placed primary stress on their idea that the linguistic consciousness of the average villager is suffi­ cient to establish dialect boundaries. This position, of course, is quite controversial in contemporary research, and it can be applied to historical study only with extreme caution. Yet when the demographic approach is adopted as an interpretive perspective on the behavior of sounds in the past, it is most informative. When Kurath views the Moore, Meech, and Whitehall maps of England in this per­ spective, he concludes simply that The coalescence of dialect boundaries with county lines points to the importance of political boundaries in shaping the speech areas; lack of such agreement points to supra- regional factors, such as the influence of major trade centers, major trade routes, and dominant cultural centers or institutions upon adjoining speech areas. Among all the evolving relationships between his­ torical phonology and the dialect theory; since the Neo­ grammarians, none is more important than that concerning phonology and- structural dialectology. At present, the connection between these two disciplines is strong, yet complicated; it is possible to trace the development of an affinity between the two fields from Ludwig Wittgenstein's philosophy of language to Andr^ Martinet on the theoretical side, and from Karl Luick to William G. Moulton on the practical side. 28 In Wittgenstein's Philosophische Untersuchungen, many of his conclusions about the nature of language served to re-emphasize the early structuarlist ideas of N. S. Troubetzkpy. 29 For Wittgenstein, the form of a linguistic utterance is less important than the demonstrated function which the utterance performs; the precise character of what he chose to call "language games" 30 is best elucidated by studying the use that is made of the given statement. This idea, of course, reinforces the interpretation that a system must be viewed parochially or discretely, in all its communi­ cative functions. In addition, his stress on the notion that "the meaning of a word is governed by the language-game in which it figures," 31 and on the functionalist properties of speech forms which result in the view that language is "a tool with a rich variety of uses," 32 gives added force to Troubetzkoy's notion that the exact phonemic or morphemic features of a given utterance and its linguistic features exist within a definite system. With the philosophical weight of Wittgenstein's ideas thrown into the balance, — structural dialectologists have all the more assurance in their approach to language as a system of communication. The precise relationship between phonology and structural dialectology is subtle, and may be phrased as follows: if it is granted that the phonemes and morphemes of a dialect exist within a balanced system, and if it is 90

further- granted that the system itself undergoes change, the question arises~can one predict the precise phonological changes that a given dialect system will produce? 33 . Very few attempts have been made to answer this question. In a recent article on "Why phonetic Change is Regular," Isidore Dyen begins by reaffirming the systematic nature of language: he asserts that there is a dependence between the function of language as a means of communication and the regularity of phonetic change . . . thus the use of language as a means of communication forces regularity on phonetic change. Yet when Dyen attempts to generalize on the connection be­ tween dialect systems and phonetic change, he falls back on the concept that physicists and chemists use to explain the breakdown of a system: "phonetic change is one of the forms that entropy takes as it increases in the intercomraun- ication of the speakers of a lanaguage." 35 Dyen's suggestion is valuable, but it is obvious that a far more specific and less analqgical approach to this question is needed. Such an approach was attempted in 1952 by Henry M. Hoenigswald in an article entitled "The Phonology of Dialect Harrowing." 36 Hoenigswald outlined five possible ways in which a stronger dialect A may influence a weaker dialect B. In his analysis, he posits an earlier dialect x from which both A and B derive, and he deals, for the sake of illustration, with the imaginary morphemes "tern" and 91

"del." Xn the first of the five cases, Hoenigswald shows that /tem/ and /del/ may "remain phonemic ally distinct in both A and B," 37 so that the morphemes may not be phonet­ ically identical across the two dialect systems, but they are fitted in meaningfully within each separate system. Secondly, he shows that there may be a case of an incomplete sound change such that /tern/ and /del/ remain phonemically distinct in both A and B, but the pronunciation changes in such a way that /t/ and./d/ of A are both picked up as, and, in being taken over, replaced by, /t/. of B. While this process continues, dialect B will have an "elegant variant /tel/ alongside the local /del/" 39 ’ and then /del/ gradually may be driven out. Thirdly, Hoenigswald suggests a case in which "/t_/and IkJ have fallen together in A into one phone, /tJ , which is phonetically closer to the ItJ than to the 7k J of B (where the contrast remains in force)."40 In this instance, when the sound change is in process in A instead of B, there again arises in B a "standard variant /tel/" which will “compete with the local /del/?"4*" eventually, the merging of /t J and /d_/ in A will be complete and will spread to B, and /del/ in B will be replaced by /tel/. A fourth possibility involves sound changes which have already been completed before A and B come into contact: if, for instance, in B phones /t/ and /kj have merged, the words /tem/ and /del/ of A will be brought in as /tem/ and 92

/tel/ in. B. The fifth possibility is more complicated: /t7 and may have "fallen together in B into one phone more similar to A JxJ than to A /cL/ ; also, from a third source (say, from earlier 7x7) a phone similar to A /dJ yl A has come into existence in B." In this instance, a basic /t7-7dI distinction is preserved in both dialects, but in B the phone CdJ will have a duplicating effect when it is influenced by the /dJ in A: in dialect B, there is created "a standard variant" with the ThJ from A "matching the local ftj, and further, there is set up an analogical proportion," with local /tel/: elegant /del/::local /tem/: elegant /dem/. 43 In such a case, since there is no pres­ sure for sound change from A, the variant forms in B will apparently remain, as variant forms, until the social or economic hegemony of one form in B drives the others out. In general and simplified terms, Hoenigswald's five rules may be summarized thus: if two dialects A and B derive from dialect x, and if A is dominant over B, there are a limited number of ways in which phonetic change may take place if the behavior of a given set of morphemes within the two dialects is investigated. First, dialect ^ may borrow the morphemes directly from A, with the phonemic distinctions preserved, and with only allophonic variations of those morphemes occurring in B. Next, the morphemes in question may be phonemically distinct in A and B, but a 93 sound change in B brings about a phoneme merger in that dialect; subsequently, all morphemes with the phonemes in question will be changed in B as they are borrowed from A. Or, thirdly, a sound change originating in A passes to B and for a time introduces variants and competitive forms in B. Fourthly, perhaps a sound change in B precedes one in A, and as words are borrowed from A, a phonemic distinc­ tion is dropped in B. Finally, two phones may merge in B, but from another source one of the merged forms reappears. Thus, in words coming from A, there will be a renewal of phonemic distinctions based on the dialect systems before the merger, and again variants and competitive forms will appear in B. With this set of rules to guide dialect research, an historical phonologist1s task of deciding what happened when is made easier, particularly if the researcher is willing to take into account the possible influence of political, sociolpgical, and demographic factors. But even with such rules, neither the phonologist nor the structural dialectologist has a ready understanding of what specifically happens when phonemes shift and merge, and doff their old meanings and don new ones. Uriel Weinreich, Andre Martinet, and William G. Moulton have thought about this very problem, and their inquiries have produced a new concept which Moulton calls "phonological space?1): 44- as this concept develops in theory and practice, it holds promise that the researcher 94

eventually may be able to predict phoneme behavior with, paradoxically, a minimum of primary data to guide him. One of the first modern steps toward a study of the precise behavior of phonemes in an unstable language or dialect situation was taken in 1953, when Uriel Weinreich published his study of bi-lingualism, Languages in Contact. Weinreich felt that the best way to investigate what happens when a population speaks two dialects is to begin with a phonemic inventory of each system: "if the phonic or grammatical systems of two languages are compared, and their differences delineated, one ordinarily has a list of the potential forms of interference in the given contact situa- tion." 45 More specifically, in a comparison of the basically different Romansh and Schwyzertutsch languages of Switzerland, Weinreich found that phonologically speaking the nature of the "interference" between two languages or, by careful extension, between two dialects, could be des­ cribed by the following rules the terms "primary system" and "secondary system" refer respectively to a speaker's native dialect and to his acquired dialect : (1) Under-differentiation of phonemes occurs when two sounds of the secondary system whose counterparts are not distinguished in the primary system sure confused. (2) Over-differentiation of phonemes involves the imposition of phonemic distinctions from the primary system on the sounds of the secon­ dary system, where they are not required. 95

C3) Reinterpretation of distinctions occurs when the bi-lingual distinguishes phonemes of the secondary system by features which in that system are merely concomitant or redun­ dant, but which are relevant in his primary system. (4) Actual phone substitution, in the narrow sense of the term which Weinreich later defines as 'the non-customary pronunciation of an identifiable phoneme1 (p. 21) , applies to phonemes that are identically defined in two languages but whose normal pronunciation differs. Weinreich's observations on this subject carry additional authority because he is himself multi-lingual; thus his statement on morphology that "the transfer of morphemes which are as strongly bound as inflectional endings in many European languages seems to be extremely rare" 47 can be accepted with little hesitation even though he offers slight supporting evidence for his contentions. Still, the scarcity of such evidence and of specific demonstrations in Languages in Contact makes a reader wish for an even more fundamental treatment of the nature of phoneme behavior. In 1955 Weinreich's teacher, Andre Martinet, published such a book, basing it on his many years of study, teaching, and scholarship in Romance phonology, and blending in it many of the ideas from the various schools of phonological and dialect research which had developed since the Neogrammarians. / In fact, in many ways Martinet's Economie des change- / AO ments phonetiques is grounded in an eclectic formulation of the ideas of Ferdinand de Saussure, Nikolai S. Troubetzkoy, 49 and the early functionalists such as Paul Passy. 96

Primarily, the book attempts a more sustained and methodo- logical approach to those questions concerning sound change to which Saussure had addressed himself. Like Troubetzkoy, Martinet holds that languages and dialects are systems and that sound changes occur within and throughout the system in question; indeed, the stress in the Economie is that, quite simply, the basic cause of phonetic change is^ dialect borrowing. 50 With Paul Passy, Martinet believes that, basically, "une opposition phonologique utile a la compre­ hension mutuelle se maintain mieux qu'une autre moins utile" (42), and, that there is a "conflit permanent entre la tendance a 1'assimilation des phonemes voisins dans la chaine et la pression conservatrice du systeme," which means, in specific terms, that "deux sons consecutifs tendent toujours a etre assimiles . . . cette tendance est combattue par la necessite de conserver les distinctions significatives" (43). In short, Martinet adopts Passy's two fundamental rules of functionalism: phonetically speaking, 1. Le langage tend constamment a se debarasser de ce qui est superflu. 2. Le langage tend/constamment a raettre en relief ce' qui est necessaire. To this combination of ideas as a theoretical basis, Martinet added and altered some of the ideas of Otto Jesperson, Henri Frei, Charles Grammont, and such "binarists" as Roman Jakobson (whose Preliminaries to Speech Analysis is given particular scrutiny), 52 so that ultimately he presents a compendium of functionalist and structuralist thought up to 1950. Remarkably enough, Martinet even anticipated by several years Isidore Dyen's notion of "entropy"; he alludes at one point to an immeasurable force of phonetic decay which he calls a "desequilibre qui se propage a travers un systeme"' and which, in its sum of effects, has souvent un action non plus retardatrice, raais inhibitrice, et il peut etre de ceux qui sont responsables des divergences qu'on constate d'un dialecte a un autre (42). In addition to uniting a number of modern linguistic notions in the Economie,/ Martinet bases his own theoretical , ,/ contributions on several important assumptions which stem from his many years of scholarly experience. Basically, because of the influence of a "desequilibre" or entropy on phonetic behavior, he holds that language is in a slow and steady process of continuous change: "il existe, dans tout systeme phonologique et a tous les instants de son histoire, des zones ou des changements sont en preparation ou en proces" (34). 1EL times, for reasons which will be explained later, Q& *£? . the inherent instability of a system suddenly increases in its general effects, and then the language becomes all the more ready to undergo a sound change; in addition, both 98 biological and basic social or cultural forces may work at once, to reinforce one another, for since language is a human activity it is inevitably affected by human vitality and animation, and thus the forces of "moindre effort" and the "besoin de s'exprimer et de communiquer" (36) inexorably produce language change. These, of course, are quite general attitudes about what is at the base of phonetic flux; it is instead his precise theorizing first about the nature of the phoneme, and second, about the production of sounds that forms some of Martinet's more valuable contributions to the field of phonemics. When he begins his discussion of the phoneme, Martinet demonstrates briefly that that concept has had a complicated scholarly history: he traces the varying ideas on the subject from "l'hypothese realiste," to the "outil de connaissance" theory, to "1'intention phonique" emphasis, and on to the "formalistes" and those who believe in the "concept utilitaire" (32); for his personal definition, he writes II est done preferable de degager des criteres formels purement linguistiques comme delui de la pertinence distinctive. Ce critere est facile a manier et choisi de £agon a fournir des unites dont la liste ne differerait pas sensiblement de celle des 'intentions phoniques' de Baudouinr mais qui sont independantes, en th^orie de toute realite psychologique ou neuro-musculaire (32). 99

Phonemes, then, must be generally considered as "les unites distinetives" in speech (62). But on close examination, the distinctiveness of any phoneme is quite complex; for Martinet holds that, even in the idiolect of a given speaker, the phonetic character of a phoneme will "dans le meme mot, varie d'une emission a une autre" (47). This means that a phoneme has a "champ de dispersion" (47) which in all three dimensions separates the phoneme in question from neighboring phonemes produced in adjacent areas of the mouth. Across and throughout this "range of dispersion" 53 there are a great number of allophonic variants which, of course, pass for the given phoneme in speech; yet there is a "center of gravity" 54 within every range of dispersion which contains a more nearly accurate phonetic representation of the phoneme than those representations found in the allophonic variants. However, one's pronunciation of the phoneme can move away from the center of gravity, and thence among the allophonic variants, and still be understood as the given phoneme be­ cause there is a "margin of security" which retards confusion with neighboring phonemes: In the frame of a homogeneous speech community it is probable that the normal range of dis- . persion of every phoneme in a given context will not be contiguous to those of its neighbors, but that there will be a margin of security in the form of a sort of no man's land. 100

In all'/ to draw an analogy within solid geometry, a phoneme may be pictured as a sphere or a cylinder with a wobbling, vacillating center; the center, of the figure can move up and down, or back and forth, or very near to the edge, and still the physical identity of the figure remains, and is recog­ nizable. Once he describes the complexity of a phoneme in such a fashion, Martinet offers conjectures as to how phonemes interact with each other. He at first insists on the funda­ mental distinction between phonetic change and phonemic change, for it is possible for a great amount of phonetic activity to occur with no effect on the phonemic system. Further, a range of dispersion may swell and extend itself with no resulting . phonemic change, for the enlarged range of dispersion may not come into a homonymic conflict with other phonemes: a phoneme may have room for territorial expansion with no chance of conflict with its neighbors, and there is thus "une extension legitime du champ de dispersion"(48). If, how­ ever, the extension occurs among a set of neighboring phonemes whose ranges of dispersion are relatively rigid and whose margins of security are small, there will be phonemic inter­ ference and phonemic change. According to Martinet, there cure a limited number of ways in which phonemes may change. Considering the three phonemes B, A, and C, it is first of all possible that with a resistant margin of security in B and a weakening one in 101

C, A and C may merge. Secondly, it is possible that with a weakening margin of security: in A, the three phonemes may shift together in a linear fashion with B moving toward A, A toward C, and C toward a new position. Or, thirdly, it is possible that with a weak margin of security in B, A may move to a new position (undetermined as far as C is con­ cerned) and pull B toward its former and recently vacated position. Martinet presents these last two actions in graphic form and suggests terminology for the phoneme behavior. If it is established that A moves first in the sequence B— *A— * C— », then A is said to exert a "drag chain" action on B; similarly, if A moves first in A — *C, then A produces a "push chain" effect if C preserves its eg margin of security and recedes from A. In any of these cases, there is an open possibility that the rest of the phonemic system may be affected by the change in one phoneme; for, in accordance with Troubetzkoy's principles of structural dialectology, one change is felt throughout the system unless, as Martinet shows, a phoneme shift some­ where along the line does not enter into homonymic conflict with a neighbor. In the last case, no effect is passed on throughout the system. The likelihood of change in a system, however, must be considered apart from the general description of possible phoneme movements. In the first place, phoneme change 102

depends upon the amount of "functional yield" 57 in a set of contrasting phonemes: if, .for instance, a vocabulary contains a large number of words in which .the /cf^7 - /©] contrast is phonemic., then the functional yield is high, and it is quite unlikely that /df^/and will ever merge. It would follow that mergers are likely between two phonemes whose functional yield is low, and Martinet makes the cautious statement that In the case of certain mergers taking place in modern cultural language^; jfor which full data are available, the functional yield has been found to be extremely low. Yet a firm rule cannot be formulated from this observation; Martinet notes that Two neighboring phonemes will not necessarily tend to merge simply because the functional yield of their opposition is practically nil: fsj and in English are not found to approach each other in spite of the5exceptionally low yield of their opposition. Further, he remarks on a number of other factors which invali­ date any easy generalizations about functional yield: Semantic extension, word composition, and morphological reshuffling frequently afford easy solutions to the problems which may arise when a functionally important opposition is being threatened by the drifing together of two phonemes: as soon as the margin of security is invaded and danger of misunder­ standing arises, speakers will be induced to give preference to such alternative words, phrases, or.forms as will remove all ambiguity• In other words, the concept of functional yield must be con­ sidered within Paul Passy's functionalist viewpoint: if there 103

is a chance for failure of communication, a speaker's inher­ ent sense of commutative linguistic correction will make sure that an utterance is understood. In any event, it is difficult to predict phoneme behavior on the basis of the functional yield of the oppositions of a system. The picture which Martinet gives of the mechanics of phoneme interaction shows that the moving or merging and shifting of phonemes is carried on at such a high level of activity that the phonemes almost seem to have.a kinetic energy, a Brownian movement, of their own: they tendront naturellement a utiliser au mieux les-latitudes que leur offrent les organes dits de la parole; ils tendront a etre aussi distants de leurs voisins qu'il est loisible pour eux de l'etre tout en restant faciles a / percevoir (62). In an attempt to explain this amount of activity, which embodies itself in any system as "le principle de differencia tion maxima des phonemes" (62), Martinet offers his theories of phonemic series and states and of the inertia and asymmetry of speech organs. He contends primarily that, given that a phoneme has great potential for movement throughout its range of dispersion and beyond its margin of security, the best way to examine the possibilities of phoneme behavior is to study the production of phonemes within the mouth: if one compares and contrasts a phoneme with its neighbors, and then accepts the principle of least effort as the 104

underlying force behind all sound change, he may predict phoneme behavior, in the future and reveal the "evolutionary probabilitiesof phoneme relationships in the past. In order to better visualize such comparisons. Martinet sets up a pair of grids, one for consonants and one for vowels, in which he can select, a particular phoneme and judge its potential interactions with any other phoneme within the grid. His terminology and exposition are as follows: A number of consonantal phonemes characterized by one and the same articulation will be said to form a 'series1 if their other characteristic articulations can be located at different points along the air channel . . . a number of phonemes characterized by one and the same articulation at a given point of the air channel, but dis­ tinguished from one another by some other dis­ tinctive articulation will be said to form an 'order* . . . In regard to vowels, it seems more advantageous to label as 'series' a number of phonemes characterized by the same type of resonance eavities, but distinguished by differ­ ent degrees of oral aperture, and as 'order' a number of phonemes characterized by the same degree of aperture but distinguished by differ­ ent types of resonance cavities. As two examples of a consonantal series in modern English, Martinet gives the sequences^/p/, /t/, /c/,/k/ and /b/, /d/, /g/r'/g/r as two examples of consonantal orders, he gives /p/, /b/, /m/ as a labial order and /t/, /d/, /n/ as an apical order. For vowels, also in modern English, he lists /i/, /e/, /ae/ as a front series and /i/ and /u/ as a high order. 105

It is apparent from his discussion that Martinet has in mind the sort of positional breakdown of speech sounds and their place of origin that one sees in the kind of charts that accompany the sagittal section diagrams in textbooks. In such charts, the consonants are discussed from front to back either as voiced or voiceless occlusive stops and nasals (bilabials, alveolars, palatals, and velars) or as voiced or voiceless spirants (bilabials, labiodentals, interdentals, alveolars, alveopalatals, palatals, and velars); and the vowels are discussed accord­ ing to their positions in the mouth (front, central or back, and high, mid, or low), their relative tenseness or laxity, and their relative roundness or unroundness. In his divi­ sion of sounds into series and orders, Martinet, like those structural dialectologists who borrow Roman Jakobson's list of distinctive features, 64 is attempting to develop a set of terms whose various combinations may be used to describe more accurately than ever before phonemes which are changing or which have changed. The important difference between the structural dialectologists and Martinet is that the former have a tendency to adopt "le binairisme" (73), in which one of a pair of distinctive features either applies or does not apply, whereas Martinet wishes to keep all descriptive possibilities open for the sake of refinement and accuracy in giving an account of a certain phoneme. The act of 106

isolating and then applying a number, of distinctive articu­ latory properties in the description of a language is what Charles F. Hockett calls "componentia1 analysis." 65 According to Hockett,. there are two ways of developing a componential analysis: Xn one, positions and manners of articulation are alike regarded as features— all possible features are interpreted as constituting a homogeneous class except insofar as privileges of occurrence with respect to each other serves to divide them up. In the other, positions of articulation (or, rather, actively functioning articulators) are regarded as items individually capable of two or more differentiated functions, some of those functions being the same for two or more organs, some of them not. There is no question as to 'right' or 'wrong' here: the two sorts of analysis produce mutually convertible results, and the fact that neither forces itself on us to the exclusion of the other is but another aspect of the relativism or indeterminacy of phonologic pattern. Thus, as far as Hockett is concerned, Martinet1s way of describing phonemic reality is as good as but no better than any other. Yet it is not only in the method, but also in the results, of his phonemic research that Martinet's important and original contributions lie. In the fourth chapter of his text he combines his scholarly experience among Romance languages with componential analysis, and, working through the hypotheses that sound change is caused by the asymmetry of speech organs, inertia, and the economy of effort on the part of the speaker,^produces a vast number of generaliztions 107 about specific phoneme behavior in specific situations* I include a sample of these principles below. .The first statements are concerned with fronting: The palatalization of dorsals before front vowels is not entirely conditioned by the quality of the following vowel as usually assumed, but also by a general fronting of all dorsal consonantal articulations whereby post-velars become velars, velars become post-palatals, and so forth. Vocalic correlations are quite generally much better represented in the higher than in the lower orders, /oe/as a phoneme is probably rarer than /o’/ or /u/. Patterns with three series (e.g. of the /i/, /u/, and /u/ types) practically never keep these three series distinct for the lower order, which is easily accounted for by pointing out that, with maximally open jaws, the lips will be automatically retracted, and that 1 it will become difficult to distinguish between a front and a back oral cavity . . . {fthus/ /o/ and /u/ will more readily merge than /e/ and /i/, because the difference in aperture is smaller, and in general for the same number of phonemes in the front and in the back series the margins of securitygwill be narrower at the back than at the front. In the discussion of many of these principles, Martinet brings in some of Jakobson's distinctive features, such as nasality, orality, and intensity. Thus he finds a potential locus of instability in the nasality of certain vowels, for nasal vowels are never as clear as oral ones: "concomitant nasalization affects the^clarity of^vocalic gQ articulations." Further, he finds that nasal fricatives and liquid nasals are rare, and that in general, "nasality combines better with open than with closed vocalic articu- lations.” 70 As far as intensity is concerned, he remarks that 108

II y aura necessairement des conflits. ehtref 1 *utilisation oppositionnelle de l'intensite et son utilisation contrastive, conflits qui . se resoudront par des limitations variees, f pouvant aller jusqu1 a 1 *exclusion de l'intensite de l'un ou de 1 'autre domaine (129). At other times in his descriptions of phoneme be- - • -1-—n . havior, he mentions important principles in almost an obiter dictum fashion: he remarks on one occasion that a phonemic change may be felt throughout the series and orders of his componential matrix: Si /d/ est devoise, /b/ et /g/ seront sans doute devoises aussi . . . si /k/ est palatise dans certaines conditions, il est v^aisemblable que /g/ sera palatalise dans les meroes conditions (78). Later on, he interrupts his treatment of fronting to point out that a new phoneme with no opposition in the syfrtem at present may be acquired during a sound change, but that phoneme will soon acquire opposition: it will “creer un partenaire correlatif ou a evoluer pour servir de partenaire a un autre isole" (103). A little later, but within the same general topic, he remarks that "le systeme le plus stable, c'est-a-dire, phonologiquement le meilleur, sera celui dont tous les phonemes seront integres" (103). In fact, a reader encounters Martinet's generalizations wherever he looks in this chapter. Within three pages, for example, one finds comments on the following facts: an amount of phonological instability is produced because speakers tend to give the same amount of energy to all the analogous 109 differentiative units within a phonemic system (130); a complex phoneme in a unique situation (for example, /l/ as the only lateral in a language) will intrude phonetically in the pronunciation of other words— or wherever it can (131); and, the complexity of a phoneme is inversely pro­ portional to its frequency in a system, for the less it appears, the greater its distinctive value (132). For the most part,, however, Martinet develops his commentary around such set topics as Grimm's Law, the arti­ culatory or occlusive mode, and gemination. He shows that an early stage of Grimm's Law may be accounted for by progressive glottalization: Nous pouvons imaginer un passage de /t7/ a /?d J par anticipation progressive de la voix d'une voyelle suivante, et meme, en fin de compte, un aboutissement / d/ si la sonorite s'etend finalement a tout le phoneme. Ceci permettrait d'envisager le passage d'une correlation comme: p t k a p t k e t finalement k p t k p?t7 k7 7 b 7d 7g t b d g 4 II faut envisager egalement la possibilite de 1 ' extension aiix organes voisins de la tension ndcessaire k la realisation de 1 'occlusion glottale. Originellement, un resserrement pharyngal ne serait qu'un accompagnement automatique de la contraction glottale, mais il pourrait devenir 1 'element principal, puis .. unique-,_de—la marque .V-Si- nous-indiquons-l^S consonnes pharyngalisees au moyen d'un point place au-dessous ou au-dessus d'elles, nous pouvons representer comme suit une evolution possible: ptk\ptk \ ptk p’t7 k7' Ab d g / ptk avec relachement^progressif de 1 'occlusion glottale jusqu' a la voix, et enfin jusqu'a la ‘ glotte ouverte (114 ). 110

On the occlusives, Martinet remarks that the articulatory zones where the occlusives, form are not exactly those where the frictions give: the most distinct and resistant acoustic results (121), and thus the articulatory mode requires particular study: An articulatory weakening will as a rule decisively affect bilabials before the . other orders: /p / is frequently weakened to /h/or zero where ft/ and /k / are preserved . . . ThoweverJ7 a shift affecting strongly articulated consonants is likely to act more rapidly on the phonemes of the apical order . . . £further,7 a general articulatory strengthening acting upon a series of spirants may change a /F/ into ft I t but will never change an /f/ into a stop, and will simply make the articulation of fsf more energetic; />7 may remain a spirant when 18/ is made an occlu­ sive; whereas /IS/becomes a stop in strong (e.g. initial) positions but remains a spirant else­ where . But even when Martinet discusses a major topic, his dis­ cursive way of approaching a subject in this chapter leads him to give a large number of only indirectly related generalizations. He begins by referring to gemination, for instance, as a process by which frequency imposes on a system a qualitative reorganization and reduces objective duality to a unity (144); but before long, he naturally shifts his attention from the gemination of consonants to the length of the preceding vowel, and then moves on to speak of vowel length in general: XI est assez probable qu'il existe quelque rapport inverse entre la frequence des voyelles longues et leur duree (r^ele ou subjective), Ill

et que, lorsque la frequence des voyelles longues est^de mime ordre que celle des voyelles breves on ait atteint un etat instable susceptible d'aboutir a une ^limina- tion de la quantite vocalique distinctive (145) . . . la breve /voyelleJ relache son arti- . culation locale . . . CetJ les longues qui s'ouvrent . . . etant par nature plus susceptibles de duree que les voyelles fermees, les longues phonologiques tendent a s'souvrir pour maintenir leur longueur relative. (146) As I have tried to show, it is all but impossible to / summarize the Economie. Of course, for certain purposes a given section may be condensed, as William G. Moulton does in a general allusion to Martinet's methods and results: Thus it is the asymmetry of the vocal organs which is responsible for such asymmetries often found within phonological systems as: fewer low vowels than high vowels; fewer back vowels than front unrounded vowels; fewer front rounded vowels than either front unrounded or back vowels; and fewer 72 velar consonants than labial or dental consonants. Yet because of Martinet's rather desultory way of stating important principles and generalizations in the middle of paragraphs, it is hopelessly difficult to give any accurate summary of his work at all. Instead, the whole Economie itself must be regarded as a large, indivisible, and indis­ pensable repository of information about phoneme behavior. This compendium of phonological knowledge is Martinet's contribution to a special field which he calls "diachronic phonemics," a discipline which is based on observations made on "patterns” of sound change within a structuralist system, and not on the4 phonemic or phonetic "context" or morphemic , environment. 73 Martinet's approach thus is quite different from that of the traditional historical phbnologists: his is based on assumptions about economy of effort and, in general, phonetic physics; theirs is based on a vast body of knowledge of the behavior of a given sound while it is in the presence of another given sound. As I suggested earlier (see note 28), Martinet is not the first to have developed these notions of phonemic behavior within phonological space, for Karl Luick had almost precisely the same idea. More recently, William G. Moulton has applied Martinet's theories in a series of articles; 75 but the credit for first working the theories out in detail must go to Martinet. For the historical researcher in language, and for the dialectologist in particular, the Economie supplies an invaluable set of generalizations about how a given sound will act, and these principles should be used to supplement the knowledge gained by historical phonologists. For the study of Middle English dialects, Martinet's principles have an importance which is • i roughly proportional to the amount of phonemic and morphemic data available, and as more of this data is developed, the more pertinent the field of diachronic phonemics will be. 76 The evolving relationships among historical phonology, dialect geography, and structural dialectology and its off­ shoot,. diachronic phonemics, have produced several new and important linguistic perspectives and methods; yet the 113 central requisite in dialect study is that historical phonology develop in some way as large a number of distinc­ tive dialect phonemes and morphemes as possible. The most productive in this respect of all modern methods has been neither dialect geography nor structural dialectology; the discipline which has generated the most dialect features has been that particularly rigorous kind of onomastics which has been nurtured by the English Place Name Society since the 1920's, and the greatest achievements in Middle English dialect study have been made by the Society itself. From the beginning of the series of English Place Name Society publications, the main vector of linguistic inquiry was directed at producing an increasing quantity of Middle English dialect characteristics. In the Introduc- tion to the Survey of English Place Names, 77 Henry C. Wyld and Mary S. Serjeantson remark that the "phonological variations" exhibited by the various Middle English forms of the place names which they have examined, "together with others in accidence and vocabulary, are recognised as being among the chief characteristics of the regional dialects of Middle English." 78 The authors suggest that their place name materials might provide a detailed survey of the dialect characters of Middle English which should show the typical grouping of these, county by county, and an even minuter analysis of the linguistic features of smaller districts within the larger areas. 114

Wyld and Miss Serjeantson then detail certain precautions and tests which scholars might apply in working with place name data. Wyld notes first of all that many of the features appearing in a given area may seem to belong to. other areas entirely, if one accepts the present understanding of the spread of dialect characteristics: features previously known to belong to the south may appear in the west, or vice- versa, and other areas may seem to have interchanged their features. Further, some features may disappear and then reappear and then disappear again in the same area over a period of time, so that the researcher may begin to. doubt the validity of his materials. Wyld reminds us, however, that all dialects are in constant change, and further that any apparent eclecticism in a given area roust be accepted, for a combination of features in a text, which to our imperfect knowledge appears to be due to a jumble of forms from different dialects, may in reality represent a perfectly genuinggtype of English spoken in a particular area. The validity of the materials may be questioned only if one suspects that a given place name may not be spelled phonet­ ically by a local'scribe, but instead according to the conventions of the London dialect by a traveling represen­ tative of the king. Such cases do occur, of course, and as a result that place name is useless as a sample of local dialect. Nevertheless, Wyld and Miss Serjeantson declare 115

that there is enough valid and useful material to give "a consistent picture of great dialect areas,, differentiated into sub-areas which shade off, and melt, one into the other without gaps or. violent breaks in continuity."81 In their researches, Wyld and Miss Serjeantson care­ fully developed eleven phonological tests which are designed to simplify the analysis and classification of the dialect features in any geographical area of England, The tests, which guided many of the subsequent scholars in this field in their county-by-county investigations of English place names, are as follows: (i) OE ae: does it appear as e(ae), or a? Cii) OE oe (non-WS e ): has this the sound of /£/ sometimes written ea, and shortened to a, or of /e/ sometimes written ie, and shortened to e? Ciii) OE ae (a-i); questions the same as the preceding. (iv) OE y (u-i): does this appear predominantly as u (ui, uy), i, or as e? (v) OE eo: are these written e, or eo, u, Of ue, oe? (vi) OE ea-i _(WS Ie, y, I, non- WS e): does this appear as u, T, or e? (vii) OE £a-i (WS ie, y, x, non-WS e,. oe) : does this appear as u, i, e, or a? (viii) OE -ear + cons.: does this appear as -er or -ar? tix) OE -eald-, -aid-: which type survives? The former, as -eld, or the latter, as -old? (x) OE -an, -am, or -on, -om (otherwise than before nd): which type survives? (Xi) OE -eah, -eag- (Sh, Sg-): do these appear as -eh, - § 3 r (Sih, -eig-), or as -ah, ig? The authors then suggest that a scholar begin his research among such sources as "Chartularies of Abbeys, Rolls, Inquisitions and other official documents," 83 and specifically in the Ramsey Chartuiary, the Feudal Aids, and the Hundred Rolls.85 116

Later scholars, after applying the’se tests and heeding the advice that all place name forms must be checked p c against other spelling evidence so as to remove all doubt, have vastly enlarged our knowledge of the phonology of Middle English dialects. In 1925 the English Place Name Society began to.publish a county by county treatment of English place names; the editors began with a volume on Buckinghamshire, and since then they have averaged precisely one volume a year, and with the publication of the second part of the Westmorland study in 1967, they have so far ac­ counted for twenty-one of England's thirty-nine counties in forty-three volumes. 86 The place name studies become pro­ gressively more thorough as the years pass, partly because now the researchers are investigating more kinds of names (the recent volumes deal with the names of counties, towns, rivers, lakes, fields, streams, saints, people, manors, streets, roads, and houses), and partly because more primary and secondary sources are becoming available: the early volumes on Buckinqhamshire and Bedfordshire and Huntingdon ■ shire list around 120 manuscript and secondary sources; but the Society's more recent volumes apparently draw on every­ thing that is relevant in Charles Gross' The Sources and Literature of English History from the Earliest Times to about 1485, 87 plus an increasing array of general place name studies, to provide close to 500 sources for The Place 117

Names of the West Riding of Yorkshire, and then. 320 and 325 sources respectively for the investigations of Westmorland and Gloucestershire. As a result, of course, .the later volumes are much more valuable for the dialect researcher than the early ones. It is true that in the crass and inaccurate quanti­ tative point of view the sheer number of phonological dia­ lect characteristics varies from the fifteen discussed in the Buckinghamshire volume to the great complexity and detail of the fifty-seven treated in the West Riding of Yorkshire volumes. However, these figures are quite mis­ leading. The philosophy behind the "Notes on Dialect" sec­ tions evidently changes over the years from treating only those phonological features which represent systematic change, to treating systematic change and mentioning all minor variations both within the whole system and within the given changes themselves. To the more recent editors, thoroughness is the summum bonum. To try to condense here the dialect information which the editors of these volumes have gathered would be sheer folly: the phonological characteristics discussed so far would make up a volume of 105 pages, and to abstract those dialect features which pertain to Middle English alone would result in a book of perhaps two-thirds that size. Instead, I will try to give an indication of the organization of the discussions in the "Notes on Dialect" sections, and I will 118 give samples of most of the kinds of features discussed in those sections. Typically, the treatments of dialect phonology begin with accounts of the changes in the follow­ ing Old English (sometimes East Saxon or Anglian is speci­ fied ) ,. Old Norse, or Old Scandinavian vowels and diphthongs as they pass through Middle English into Early Modern M W «f Y M O W W English: ae, a, e, eo, re, o , u, ea, e(ae), I, i(o)w; also investigated cure changes in any of these vowels or diph­ thongs before nasals or before or after / r/or /w / , and particular attention is paid to the behavior of /y / as / i /, of / a/ + /id/, of /a / as an i-mutation before a nasal, and of any i-mutation before / 1 / plus a consonant. Next, any pecularities in consonants and glides are noted, such as: /f / changing to /v / or vice-versa, /w / intruding before /of, the stopping of medial /%fto /k / , the voicing of initial 7s/ , the prefixing of /r/ from the dative form of the definltevarticle, the insertion of r in the spelling as a sign of a long vowel, the change of final / s / to /s/, or the change of OE ct to ME ht. After the consonants are dis­ cussed, any change in the suprasegmentals is mentioned: most often, the authors record any shift in stress which may have occurred during the Middle English period. Then are listed any grammatical irregularities such as: a genitival us replacing es, the preservation of the final OE e or n of the dative singular, a final or medial inflectional a, the retention of weak noun inflectional forms, weak adjectival 119

inflection, us occurring for es or is. in unstressed syllables, or the presence of any of the participial forms mentioned by Moore, Meech, and Whitehall. Finally, the editors note any examples of a variety of such systematic sound changes as: prothesis of / j/, .metathesis of /r / plus vowel, or of /sk/ to /ks /, metanalysis of /a / or /an/, assimilation (OE aet coalescing with the following definite article), the substitution of phones (/v/ for /© /), the confusion of phones (ft I and /k/, / g / and /d /, /f / and /© /, /v / and tv I) , the interchange of liquids and of fh / and /f/ initially, and the loss of phones (initial /w/ before a back vowel, loss of final If /, Idj, /IT / , or /©/, or the loss of liquids or of medial In /) . This, again, is a sample of the many phonological features studied by the editors of this series. The peculiar calculus of the sample, however, is this: the presence or absence of some, or any one, of these dialect features could serve to distinguish one dialect from another. Obviously, the increase in the sheer number of phonological features studied, in comparison to the eleven tests recommended by Henry Cecil Wyld, has provided an extremely refined apparatus for the scholar interested in differentiating two Middle English dialects. Even though the work done by the English Place Name Society is of uneven value, the later volumes enable a scholar to do quite detailed studies of the 120

counties in question. I must repeat that it is impossible in this space to give a county, by county breakdown of those exact, features which the editors have found to; be character­ istic of the Middle English dialects in all areas throughout the country, for to do so would be to duplicate at least 60 pages of scholarship that already exists; instead, one must refer to the English Place Name series itself as the unique repository of this remendous amount of information. Some years after the Place Name Society began to issue its publications, it became apparent to certain scholars who were associated mainly with Uppsala and Lund Universities in Sweden, that the material on dialects in each volume was actually a by-product of the main research emphasis in onomastics, which, of course, was directed toward finding the various changes of meaning in the history of a given name. Later researchers in this field have felt that place name sources and materials would yield a far more accurate and thorough account of the phonology of a Middle English dialect if the materials were examined with the aim of sorting out the Middle English characteristics exclusively. With this intention, the scholars studied and evaluated those sources used by. the Place Name Society in order to find those collections of manuscripts and documents which would be most important to a phonologist interested in developing this more specialized field. The sources, which were gleaned in large part from Gross' guide, 88 and from the lists of the English Place Name Society itself, chiefly include official papers which contain long lists of sur­ names, such as: the Lay Subsidy Rolls, Charter Rolls, Patent Rolls, Close Rolls, Chancery Rolls various. Curia Regis Rolls, Feet of Fines, Placita de Banco, Pipe Rolls, the Hundred Rolls, Assize Rolls, Inquisitions post mortem, Inqui­ sitions Miscellaneous, The Book of Fees, Inquisitions and Assessments relating to Feudal Aids, Episcopal Registers, Monastic Cartularies, and, any local documents which might attest to the scribal practices of the county or area in question. To these sources the scholars added anything they knew to be pertinent from such lists as A Descriptive Cata- logue of Ancient Deeds in the Public Record Office 89 and the Index to the Charters and Rolls in the British Museum.90 In other words, the researchers tried to use every available source. Although they knew that they would eventually reach a point of diminishing returns as far as the relationship between the number of sources and the amount of relevant data were concerned, they acted on the general principle that the more sources, the more refined and the accurate the information about phonology would be. So far, this approach has produced four books, all by Swedish scholars: : Hjordia Bfthman's Studies in the Middle English Dialects of Devon and London (Goteborg, 1944), 122

Bertil. Sundby's The Dialect and Provenance of the Middle English Poem 'The Owl and the Nightingale; (Lund, 1950), Sveri Rubin's The Phonology of the Middle English Dialect of Sussex (Lund, 1951), and Sundby's model treatment of Studies in the Middle English Dialect Material of Worces­ tershire Records (Bergen, 1963). In each case the authors trace some or all of the Old English vowels and diphthongs through their Middle English equivalents in the counties in question: Miss Bohman deals with changes in the local V U W 4# Old English ae, ea, eo, and with the i-mutations of ea and eo; Sundby in his first study is concerned with y and the i-mutation of West GErmanic u, together with eo, ie, ea, a/o before nasal consonants, ae, and two forms of ae; Rubin states outright that his object is "to trace the development of all Old English vowels and diphthongs in the Middle English dialect of Sussex"; and Sundby in his second study does the same for the vowels and in addition discusses such consonant features as the voicing and unvoic­ ing of initial and final consonants, metathesis, and r assimilatioh. Miss Bohman asserts a subordinate thesis in ^ her work; she shows that there is a mixture of the London, East Middlesex and the official Westminster, dialect in London in the 13th and 14th centuries, and she establishes the

notion, since taken up by later scholars,92 that there is a closeconnection between the language forms of the 123

Westminster records and the evolution of modern standard English. 93 Sundby and Rubin are more closely concerned with isolating the distinctive phonological features of the dialects under analysis: in his first study, Sundby gives a rather brief treatment of Dorset and Surrey dialect characteristics, but his discussion in the later Worcester­ shire volume is much expanded and includes remarks on all the vowels and diphthongs; similarly, because of the initial thoroughness of his approach, Rubin also is able to charac­ terize all the Middle English vowels and diphthongs for Sussex. 94 The possible contributions of onomastic materials to Middle English dialect studies have been discussed in shorter monographs over the past 30 years. In 1938, Eilert Ekwall drew upon the Lay Subsidy Rolls, the Charter Rolls, and the Episcopal Registers, among others, to demonstrate that "the pronunciation o was fully established in Lincolnshire about 1300," 95 and that the a/o boundary "indicated by place- names runs somewhat to the south" of the line drawn by Moore, Meech, and Whitehall. 96 A year later, William Matthews studied Churchwarden's records in an effort to characterize the vowel sounds and consonants of the Early Modern English of Cornwall, Devon, Somerset, and Dorset. 97 Helge K^keritz used onomastic techniques to describe the nature of an.eclectic Juto-Kentish dialect in Old English. 124 in Kent, the Isle of Wight, and Surrey. 98 More recently, Gillis Kristensson examined the Lay Subsidy Rolls for Lincolnshire in 1327 and 1332, in order to investigate "the easternmost part of the a/o boundary in the early 14th century." 99 As Kristensson shows, a scholar can, through the intensive selection and scrutiny of his source material, produce a detailed solution to a quite specific problem in location and time through place name methods. In fact, as specialists in this field become more familiar with the available sources, and as more materials are discovered and found to be relevant, phonologists will be able to do research on quite small and confined geographical areas with great confidence. In a larger perspective, of course, it is not the specific problems discussed in articles but instead the volumes of the English Place Name Society, of Miss Bohman, and of Sundby and Rubin, those products of enormous effort, time, and energy, which are the towering achievements of the blend of onomastics and historical phonology. The merging of these two disciplines has multiplied our informa­ tion on Middle English dialect phonology, so that the knowledge presently at hand far exceeds that at the time of Moore, Meech, and Whitehall. Moreover, the new, rigorous attitudes concerning the analysis and use of 125 sources promises that the future knowledge of dialect fea­ tures will grow not arithmetically, but geometrically. It is obvious that the major advances in Middle English dialectology made by the English Place Name Society are phonolggical; as in the fields of linguistic geggraphy and structural dialectology, researchers in onomastics collect phonemic and morphemic data, and then locate these phonological characteristics within dialect areas. Nearly all dialect scholarship is based on phonology and the gathering of phonolggical information; yet in recent years a new discipline has been developed which offers an entirely different basis for the collection and analysis of dialect features. The new field is called graphology, and it is concerned with the study of writing as a form of language in Middle English manuscripts. Graphology, which in essence applies the "functional ist and structuralist point-of-view" 100 to the writing of language, had its beginnings in certain remarks on the differences between spoken and written language systems made in 1881 by Jan Baudoin de Courtenay.101 The sugges­ tion that writing, too may be considered a physically autonomous system lay dormant until Josef Vachek elaborated upon it in the Travaux du Cercle Linguistique de Prague in 1939. 102 Since then, the general theories have been stated by H. J. Uldall, Vachek, D. L. Bolinger, and Angus 126

McIntosh, 103 and with McIntosh came the notion that graph­ ology might be particularly useful for the study of Middle English dialects. Early in the formulation of this discipline, scholars were concerned with pointing out the parallels between writing and speech as systems: in each case, the shape of the smallest whole unit varies accord­ ing to its environment, according to its beginning, medial, or end position, and according to the "individual taste and habit" of the writer or speaker. 104 Also, each system is affected by. extrinsic factors such as the condi­ tion of one's pen or throat, and each has a finite number of characteristics which can be listed in an inventory. 105 As systems, speech and writing have a quite basic feature in common— they are rooted in sense experience: it is theoretically possible to transform any series of morphemes from any sensory field into any other sensory field, and keep them comprehensible; the only condition is that contact with the6nervous system be maintained at some point. Once this set of similarities was understood, scholars began to draw even more precise analogies. With a concept parallel to the notion of phonemic opposition, Josef Vachek felt that "written language must be based on a system of graphic oppositions capable of differentiating meanings in the given community"; 107 hence, it became possible to speak of a "grapheme" as "a member of a complex 'graphemic* 127 opposition, a member which is indivisible into smaller successive graphemic units.” 108 Eventually, the term "graph" came to correspond with "phone,” "allograph" with allophone, and "graphemes" and "polygraphs" with phoneme. 109 In a more recent paper, Angus McIntosh introduced a further refinement of concept and terminology: in distinguishing between "potential" (or lexical) phonic meaning and "actual" (or contextual) phonic meaning, he hoped to discuss such an ambiguity as the fact that graphemic

/s,.110 The relation between potential and actual phonic meaning could be studied under the head of "graphotactics, which would deal with the various classes of utterance in graphology. After taking his cue from Noam Chomsky's similar discussion, 112 McIntosh spoke of three utterance possibilities: 1. Well formed and meaningful: tinkle, chough, at, when . . . 2. Well formed and meaningless: histle, geed, incontravite, bumpless, ig, splander . . . 3. Nonsentences: ismi, vaogsl, abvgnestlho . . .113 Thus, in speaking of these "collocational" matters, 114 McIntosh develops a parallel between the new theories on graphology and the most modern developments in transforma­ tional grammar. Most of the above definitions and identifications are repeated throughout the many articles that have been written on graphology since World War II. There are. 128 however, two mongraph repositories of terms and concepts related to this subject: the first is H. A. Gleason's chapter on "Writing Systems" in his An Introduction to Descriptive Linguistics, 115 and the second is Angus McIntosh's essay on "The Analysis of Written Middle English. ^ Both Gleason and McIntosh, incidentally, devote much discussion to the question of whether or not graphs could be consid­ ered to represent the actual spoken language accurately; in the early stages of this field, in other words, there was much concern about the interrelationships of graphemes, phonemes, and morphemes. More recently, however, the graphemic system has been interpreted as an independent and autonomous visual structure, with no necessary rele­ vance to either the meaning or pronunciation of the spoken language. Graphology was put into practice long before it reached its current theoretical refinements. In 1951, Robert P. Stockwell and C. Westbrook Barritt amassed a large amount of evidence in a highly sophisticated attempt to prove that the graphs ae and ea never stand in contrastive distribution with each other within any one Old English dialect at a given time. They are allographs of a single grapheme^nd represent allophones of a single phoneme. It was not until five years later than Angus McIntosh suggested the importance of graphology to Middle English dialectology 129

there is a probability that they /the details of graphologyj can be shown to correlate much moire closely than has been realized with geo­ graphical factors, and thus throw more light than has been thought possible hitherto on problems of provenance. Thus it was that the results for Middle English of much of this theory began to appear in 1961. In an Indiana disser­ tation, John C. McLaughlin attempted in a graphemic analysis of MS Cotton Nero A. x. to establish the written idiolect of a text which, when correlated with similar studies of other written, expressions of the same language at approximately the same time, will enable the analyst to define the written dialect of a set of texts in terms of their common core and overall pattern. McLaughlin then proceeds on the principle that "a writing system provides the starting point for a set of assumptions about the spoken system which, in one way or another, it represents." 120 He also contributes to the growing list of terms in the field of graphology by combing the words "allography" and "grapheme" to form "graphoneme," which he defines as "a class of signs each member of which represents the same phoneme. In later studies in graphology, an emphasis has . arisen on establishing set and patterned modi operandi. W. Nelson Francis examined MS Phillipps 8250, f.262r. of the Chaucer Society volume of collotype facsimiles of pages from the manuscripts of Troilus and Criseyde, and decided 130 that a graphemic analysis should consist of the following steps, taken in order: first, one takes a graphemic inven­ tory of the manuscript by copying out each graph and all its varieties in shape and size, and then coding the graph to simplify future reference. Next, after the precise number of graphs has been determined, the scholar must decide which of the graphs are graphemes, and which are allographs; this involves a "close study of the distribution of the graph-types in relation to one another and to the various junctures," 122 and it must be done with extreme care. Then, one scrutinizes and tabulates the environment of each grapheme, in an effort to determine which graphemes exist in independent distribution, and which in complementary distribution, and which in free variation with which others, in Francis' words, the steps are as follows: 1. A finite repertory of graph-types is distinguished, to one or another of which every graph can be assigned on the basis of its shape. 2. A finite and smaller repertory of graphemes is deduced, and each graph-type is assigned as an allograph of one of these on the basis of its distribution and linguistic reference. 3. Facts about the distribution of the various graphemes to suit the differing needs of the palaeographer, the linguist, or the textual critic are or can be deduced' and tabulated.

Francis stresses that throughout these processes, the scholar must be as thorough as possible: in the first step, which 131

he calls the recording of "graphetic information," he insists that the researcher take account of "every detail whether or not it has linguistic significance." T O A. It is what Francis calls the "frequency and distri- bution of graphemes" 125 and their "privileges of occur- rence" 126 that, apparently, has interested the most recent researcher in graphology, Mrs. Dora Riffer-Macek. In a study of the hand of the set of Wyclif's Ferial Gospels in IIS New College, Oxford, 95, folios 1 to 31a line 5, Mrs. Riffer-Macek concentrates on the rules behind the comple­ mentary distribution of graphemes throughout the manu­ script. First, she culled a list of 58 graphs from her source, and then, after examining the "abbreviations, 'punctuation' graphs, and double alphabetic graphs," 127 she decided which letters were graphemes and which forms could be considered allggraphs. Next, after an exceedingly careful scrutiny of the manuscript, she began to isolate such pieces of information as: 128 Only the graphemes a, i, ou, and P and h in combination with an abbreviation occur as a sole grapheme in a word (133) . . . Space is thre main feature that distinguishes combinations of more than four strokes (134) . . *. Abbrevia­ tions are used as free variants of one grapheme or a sequence of graphemes. (134) Thus, her remarks on the distribution of graphemes produce quantities of data which are characteristic of the manu­ script in question; and some of this information— there 132 are close to sixty pieces of it in all— may be unique to that hand or to that manuscript. . Obviously, if there are unique graphological properties in a given source, those properties form a basis for distinguishing one source from another. In all, graphology can supply criteria of a sort which may benefit the taxonomist of Middle English dia­ lects. If the task is undertaken on a large scale, the

\ results might do as much for the grographical positioning of orthography as Moore, Meech, and Whitehall did for phonology. This is W. Nelson Francis' contention, too: the systematic analysis and tabulation of detailed structural information about a large number of hands of-known provenance can make it possible to recognize local schools of vriting-and locate them precisely in place and time. Eventually, the historical dialectologist may find graphology to be of equal value with onomastics. In this chapter, I have tried to show what practical results have been produced from the dialect theories of Bartoli, Gillieron, and Jud, and Saussure, Troubefzkoy, Stankiewicz, and IvicI contend that the ideas of the linguistic geographers and the structural dialectolqgists have supplemented those methods of historical phonology which have been practiced over the last 150 years. New perspec­ tives in linguistic geography have caused scholars to con­ sider the demogaaphic conditions in an area as they draw 133

isogloss lines. Structuralist notions have yielded the entirely new fields of "diachronic phonemics" and graph­ ology. Further, historical phonologists have adopted the highly productive techniques in onomastics to develop more phonological data than were available before. In some of these fields, researchers have yet to acknowledge the fact that important work in one discipline may be of value to the work carried on in another: historical phonologists have not always taken full advantage of the information provided by such linguistic geographers as Moore, Meech, and Whitehall. In the other fields, very little use has been made of the data generated by the new methods of scholarship: no research investigating the relationships between diachronic phonemics and the results of onomastic study has yet been done. In the next chapter, I will offer critical perspectives on the various new practical techni­ ques in dialect study, and I hope to show that more knowledge can be gained from the present state of informa­ tion and methods if one takes advantage of the fact that knowledge developed in one discipline can be applied in another. 134

NOTES

1. To cite only his major work: Karl Luick, Historische Grammatik der englischen Sprache, 2 vols., Harvard, 19 64, and Stuttgart, 1964. 2. Alexander J. Ellis, English Dialects— Their Sounds and Homes; Being an Abridgement of the Author's 'Existing Phonology of English Dialects,' which forms Part V of his Early English Pronunciation, London, 1890. Henry C. Wyld, "The Treatment of OE y in the Dialects of the Midland and SE Counties in ME," Engl. Stud., XLVII (1913), 1-58, 145-66. For further discussion of these works, see Chapter I. 3. Mary S. Serjeantson, "The Dialects of the West Midlands in Middle English," RES, III (1927), 54-67, 186-203, 319-331. 4. Ibid., p. 55. 5. See the discussion of Kurath's work later in this chapter. He feels that a scholar must consider the demography of a locale before he draws any isoglosses.

6 . Henry C . Wyld, A History of Modern Colloquial English, London, 1920. 7. Ibid., p. 27.

8 . Ibid. 9. Of course, some scholars (Dobson, for example— see below, n. 15) leave all considerations of dialect aside, for they are interested chiefly in the one or two imputed dialectal forebears of standard English: as Dobson writes, "I leave out of account those dialects which had short rounded front vowels, as unimportant for the later language." (128) Often, there are enough difficulties in the discussion of any dialect problem to occupy phonologists at great length, and thus any commentary on other dialect relationships would be mere digressions in their essays. In other cases, however, a geographical framework or background in a dialect problem would be most welcome and informative. 135

10. Harold Whitehall and Theresa Fein, "The Development of Middle English IT in Early Modern British and American English," JEGP, XL (1941), 191-213. 11. Raven McDavid shows why this is impossible in "Deriva­ tives of Middle English /oJ/in the South Atlantic Area," QJS, XXXV (1949), p. 504. 12. Whitehall and Fein, p. 209. 13. David W. Reed, "The History of Inflectional n in English Verbs," MA, IX (1949), p. 131. 14. Hans Kurath, "The Loss of Long Consonants and the Rise of Voiced Fricatives in Middle English," Lang, XXXVI (1956), p. 439. More recently, Andrew MacLeish has written a disserta­ tion in which he relied upon the Moore, Meech, arid Whitehall delineation of dialect regions to provide texts which are consistent and quite nearly pure as far as the forms of phonemes and morphemes are concerned: Andrew MacLeish, "Patterns in the Late East Midland Subject-Verb Cluster: A Quantitative Synchronic Description," DA, XXII (1961), 865. 15. It is not my purpose here to give a complete and anno­ tated bibliography of works on Middle English phonology since 1935. Instead, I wish to demonstrate that in their work very few historical phonologists, for some reason, have acknowledged the value of Moore, Meech, and Whitehall's essay as a guide to dialect regions, and to those MSS which can be localized within those regions; further, few phono­ logists have dealt with the burgeoning complexities, for sound laws and general phonetic behavior, within the newly outlined and charted dialect regions. Some other important studies in Middle English phonology which I do not'discuss in the text are: Flasdieck, H. M. "Zum Lautwert von me. e im 18 Jahrhundert," Anglia, LX (1936), 377-83. Koziol, H. "Der Abfall des nachtonigen -e ira Mittelenglischen," Anglia Beiblatt, XLVIII (1937), 307-9. Flasdieck, H. M. "Zur Lautgeschichte von me. i im Fruhneuenglischen," Anglia Beiblatt, XLIX (1938), 182-7. 136

Prins, A. A. "A Few Early Examples of the Great Vowel Shift," Neophilologus, XXVII (1942), 134-37. Bliss, A. J7 "Three Middle English Studies," English and Germanic Studies, II (1948), 41-2. Wittig, K. "Uber die mittelenglische Dehnung in offener Si^he und die Entwicklung der er Laute im Fruhneuenglischen," Anglia, LXX (1951), 47-70. Orton, Harold. "The Isolative Treatment in Living North-Midland Dialects of OE e Lengthened in Open Syllables in Middle English,""Leeds Studies in English and Kindred Languages, 7-8 (1952), 97-128. Dobson, E. J. "Middle English Lengthening in Open Syllables," TPS, 1962, 124-148. Liberman, A. S. "Some Notes on the History of Middle English /e/ and /o/ in Open Syllables," ZAA, XIII (1965), 21-24. In addition, several dissertations have appeared on subjects which range from phonology to morphology and style: Kobayashi, Eichi. "A Study of Verb Forms of the South English Legendary in British Museum MS Harley”2277," DA, XXIII (1962), 230. Mohr, Eugene Vincent. "Morphology and Syntax of AB, A Dialect of Early Middle English," DA,XXV (155*5) , 4136. Logan, Harry Millard. "The Dialect of The Life of Saint Katherine: A Linguistic Study," DA, XXVII (1966), 1340A-1341A. Shores, David Lee. "A Descriptive Syntax of the Peter­ borough Chronicle," XXVII (1967), 3859A-3860A. The titles of these dissertations demonstrate in part the character of the most recent studies in Middle English. 16. Archibald A. Hill, "Early Loss of /r/ before Dentals," PMLA, LV (1940), 308-359. 17. Ibid., p. 317. 18. ^Bertil Sui$by, "Middle English Overlapping V and W and its Phonemic Significance," Anglia, LXXIV (1956), 438-4*44. 19. Robert P. Stockwell, "The Middle English 'long Close' and 'Long Open' Mid Vowels," TSLL, II (1961), 529-38. 20. A. S. Liberman, "On the History of Middle English K and A," 101, 67 (1966), p. 67. 137

21. . Hans Kurath, "Phonemics and Phonics in Historical Phonology," AS, XXVI (1961), p. 94. 22„ Ibid., p. 95. 23. Either of the following of Kurath's works would demon- ' strate this notion: Linguistic. Atlas of New England, 3 vols., Providence, 1939-43, or A Word Geography of the Eastern United States, Ann Arbor, 1949. 24. Kurath in the AS article, p. 95. 25. Ibid. 26. I refer to the following articles: Grootaers, Willem A. "Origin and Nature of the Subjec­ tive Boundaries of Dialects," Orbis, VIII (1959), 355-84. ______. "Les premiers pas a la recherche des unites dialectales," Orbis, XII (1963), 361-80. • "La discussion autour des fron- tieres dialectals subjectives," Orbis, XIII (1964), 380-98. Mase, Y. "Une nouvelle tentative pour tracer les frontieres subjectives des dialectes," Orbis, XIII (1964), 357-79. 27. Hans Kurath, Plan and Bibliography fa the Middle English Dictionary, Ann Arbor, 1954, p. 10. 28. William G. Moulton, in his "Dialect Geography and the Concept of Phonological Space," Word, XVIII (1962), quotes on pages 24 and 25 those passages from Luick's Untersuchungen zur englischen Lautgeschichte (Strassburg, 1896)which, l^oulton feels, anticipate Ahdrd Martinet's argument in Economie des changements phonetigues. See below, note 48. 29. See Chapter II, pp 30. Justus Hartnack, Wittgenstein and Modern Philosophy, trans, by Maurlde Cranston (New York, 1965), p. 64. 31. Ibid., p. 101. 32. Ibid., p. 75. 138

33. This position, of course, derives from the reaction to Leonard Bloomfield's statement on p. 385 of Language (New York, 1933), that :"the causes of phonetic change are unknown." His statements on the subject of the causes of sound change have since been considered among the major weaknesses of Language. 34. Isidore Dyen, "Why Phonetic Change is Regular," Lang, XXIX (1963), p. 631. 35. Dyen, p. 636. The italics are mine. 36. Henry M. Hoenigswald, "The Phonology of Dialect Borrow- ing, " Studies in Linguistics, X (1952), 1-5. 37. Ibid., p . 2. 38. Ibid. 39. Ibid. 40. Ibid., p . 3. 41. Ibid. 42. Ibid. 43. Ibid., p . 4. 44. The reference is to Moulton's article, mentioned in note 28. 45. Uriel Weinreich, Languages in Contact (New York, 1953), p. 3 • 46. Weinreich, p. 19. 47. Weinreich, p. 31, 48. Andre Martinet, Economie des changements phonetiques, Berne, 1955. In the^ discussion which follows, I will give page references to this book in the text of the chapter. Quotations in English are from Martinet's "Function, Struc­ ture, and Sound Change," Word, VIII (1952), 1-32, which will hereafter be abbreviated as Word. 49. Martinet acknowledges the particular influence of Paul Passy's Etudes sur les changements phonetiques et leurs caracteres g&ndraux, Paris, 1898. For discussion of the others, see Chapter 2. 139

50. I construe this from Martinet's discussion on p. 27 of his book. 51. Martinet quotes Passy on p. 43.

52. See Chapter 2, note 6 6 . 53 . Word, p . 3. 54. Ibid. 55. Ibid., p. 4. Much of the conceptual treatment here is similar to Janet Rankin Aiken's notions of "Accord" (p. 22) and "Grading" (p. 44) in her Why English Sounds Change (New York, 1929) • For a criticism of these principles as they apply to languaaes in isolation, see Karl Jost's review in Anglia Beiblatt^ XLIV (1933), 67-71. 56. Ibid., P* 11 I-* in • Ibid., P* 8 . in 00 . Ibid., P* 10 59. Ibid. 60. Ibid. For a semantic investigation of this problem, see Edna Rees Williams, The Conflict of Homonyms in English, Yale Studies in English, Vol. 100, New Haven, 1944. 61• Ibid., p. 17. 62. Ibid., p. 13. 63. Ibid. 64. See the discussion in Chapter 2. 65. Charles F. Hockett,; A Manual of Phonology (Baltimore, 1955), p. 129.

6 6 . Hockett, p. 137. 67. Word, p. 15.

6 8 . Ibid., p. 24. 69. Ibid. 140

70• Ibid.., p. 25. 71. Ibid.f p. 27. 72. Moulton, op. cit., p. 24 73. Word, p . 1. 74. See note 28. 75. In addition to the work cited, Moulton has tested Martinet's theories in: Moulton, William G. "The Short Vowel Systems of Northern Switzerland," Word, XVI (1960), 155-82. . "Lautwandel durch innere Kausalitat:- die ostschweizerische Vokalspaltung," Zeitschrift fur Mundartforschung, XXVIII (1962), 22T-51"------76. See the discussion in Chapter 4. 77. A. Mawer and F. M. Stenton, eds., Introduction to the Survey of English Place-Names, Cambridge^ 1925. The chapter by Wyld and Miss Serjeantson, entitled "Place-Names and English Linguistic Studies," is on pages 133-42. The quo­ tations which immediately follow refer to this chapter. 00

• Ibid., p. 133. 79. Ibid. 80. Ibid., p. 134. See Chapter 2 for the features" theory. 81. Ibid., p. 135. CJ 00

• Ibid., p. 136. 83. Ibid., p . 13 5. ** 00

• Ibid., p. 139. ID 00

• Ibid., p. 141.

8 6 . The counties dealt with through 1967 . e: Buckinghamshire Bedfordshire, Huntingdonshire, Worcestershire, Yorkshire, Sussex, Devon, Northamptonshire, Surrey, Essex, Warwickshire, Hertfordshire, Wiltshire, Nottinghamshire, Middlesex, Cambridge shire and Ely, Cumberland, Oxfordshire, Derbyshire, Glou­ cestershire, and Westmorland. 141

87. Published in a second edition in London, 1915, this work is, as the title implies, an historiographer's delight.

8 8 . Gross locates and lists the contents of all the sources in his guide. 89. Published in London, 1890-1915. 90. H. J. Ellis and F. B. Bickley, eds., Index to the Charters and Rolls of the British Museum, London, 1900-12.

91. Rubin, p. 8 . 92. Notably by M. L. Samuels, "Some Applications of Middle English Dialectology," ES, XLIV (1963), 81-94. 93. Miss Bohman, p. 174. 94. As before, I choose not to reproduce the results in each of these studies, for they are contained in "Summary and Conclusion" sections in each volume: for Miss Bohman, pp. 27, 31, 52, 85, 92, 137, 155, 158, 172, and 174-177; for Sundby I, pp. 198-206; for Rubin, pp. 224-227; and for Sundby II, pp. 258-264. 95. Eilert Ekwall, "The Middle English &/& Boundary," ES, XX (1938), p. 166. 96. Ibid., p. 147. 97. William Matthews, "South Western Dialect in the Early Modern Period," Neophilologus, XXIV (1939), 193-209. 98. Helge Kokeritz, "The Juto-Kentish Dialect Boundary," AS, XVI (1941), 270-77. 99. Gillis Kristensson, "Another Approach to Middle English Dialectology," ES, XLVI (1965), p. 142. 100. Josef Vachek, "Some Remarks on Writing and Phonetic Transcription," Acta Linguistica, V (1945-49), p. 8 6 . Hereafter abbreviated as AL. 101. Josef Vachek, "Two Chapters on Written English," Brno Studies in English, I (1959), p. 7. Hereafter abbreviated Brno. 102. Josef Vachek, "Zum Problem der geschriebenen Sprache," TCLP, VIII (1939), pp. 94 f. 142

103. In addition to: Vachek's three articles mentioned above, there are these: Uldall, H. J. "Speech and Writing," Acta Linguistica, IV (1944), 11-16. Bolinger, D. L. "Visual Morphemes," Language, XXII (1946), 333-340. McIntosh, Angus. "The. Analysis of Written Middle English,* TPS, 1956, 26-55. . "'Graphology* and 'Meaning,'" Archivum Linguisticum, 13, II (1961), 107-120. References in this chapter are to Paper Six, on pages 98-110, of Angus McIntosh and M.. A. K. Halliday, Patterns of Language, Papers in General, Descriptive and Applied Linguistics, London, l96o. / . hA New Approach to Middle English Dialectology," ES, XLIV (1963), 1-11. Each of these articles contains extensive bibliographies of early writings on graphology. 104. Uldall, p. 13. 105. Ibid. 106. Bolinger, p. 333.

107. Vachek, AL, p. 8 8 . 108. Ibid. 109. Cf. Vachek, Brno, p. 27, and McIntosh, TPS, p. 43. 110. McIntosh, Patterns, pp. 104-6. 111. Ibid., p. 109. 112. Noam Chomsky, Syntactic Structures (The Hague, 1957), p. 15. 113. McIntosh, Patterns, p. 110. 114. Ibid. 115. H. A. Gleason, An Introduction to Descriptive Linguis­ tics, New York, 1961. "Writina Systems" is on pages 408- 4Z37 116. See note 103. 143

117. Robert P. Stockwell and C. Westbrook Barritt, "Some GE Graphemic-Phonemic Correspondences— ae, ea, and a," Studies in LinguisticsOccasional Papers, IV (19517, P« 7. 118. McIntosh, TPS, p. 49. 119. John Cameron McLaughlin, "A Graphemic-Phonemic Study of a Middle Enolish Manuscript: MS Cotton Nero A. x .,'1 D A , XXII (1961), pi 1618. 120. Ibid. 121. Ibid. 122. W. Nelson Francis, "Graphemic Analysis of Late Middle English MSS," Speculum, XXXVII (1962), p. 37. The fac­ simile is foundin r 7 K. Root, The Manuscripts of Chaucer's Troilus with Collotype Facsimiles of the Various Hand- writings, London, 1911. 123. Francis, p. 43. 124. Ibid., p. 36. Of the Troilus facsimile Francis remarks: "Our sample uses 30 graphemes and three junctures." (40) He then compares these results with analysis of MS Harley 3943, which he finds to have 30 graphemes and four junctures. 125. Ibid., p. 43. 126. Ibid., p. 42. 127. Dora Riffer-Macek, "The Graphemic Inventory of a Middle Enalish Manuscript," Studia Romanica et Anolica Zacrrabiensia, XXI (1966), p. 130. 128. References to this article are quoted in the text. 129. Francis, p. 46. 130. See the discussion in Chapter 2.

i CHAPTER IV

PERSPECTIVES AND CONCLUSIONS

In this final chapter, I hope to place the informa­ tion presented so far into a useful perspective: some of my statements will trace the subsequent development of the methods and theories treated earlier, and others will give practical criticisms of those varieties of dialect research. After beginning with a discussion of the geographical rela­ tionship between Old and Middle English dialects, the ( chapter continues with a section on the contributions of structural dialectology to Middle English studies^ Next, because an interest in dialects implies an interest in dia­ chronic and synchronic sound replacement, I bring up to date the theories of Uriel Weinreich, Henry M. Hoenigswald, and Andre Martinet, together with the generative concepts first applied by Sold Saporta. Lastly, I evaluate the most significant of the research techniques and ideas discussed in this dissertation, and the chapter closes with some comments on what we might expect in the future of Middle English dialect scholarship.

144 145

There is no counterpart in Old English dialect scholarship of Moore and Meech's geographical breakdown of Middle English dialects. In general, the consensus among scholars of Anglo Saxon has long been that the Old English dialect areas corresponded roughly to the settlement patterns of the Germanic tribesmen as they invaded England: the Jutes spoke what came to be known as the Kentish dialect, "embracing Kent and Surrey"? the Saxon tongue eventually was termed West Saxon, and was spoken "south of the Thornes, except Kent and Surrey"; and the Angles' speech evolved into Mercian, which was located "between the Humber and the Thames,"1 and into the two Northumbrian dialects, which may be seen in contrast in the work of the scribes Farmon and Owun, in their interlinear glosses to The Rushworth o Gospels. By name, the dialects were the Kentish, the West Saxon, and the Anglian, which included Mercian, North Mercian, and Northumbrian. Obviously, the boundaries here are about as general and ill-defined as any can be? the scarcity of texts in Old English has caused many scholars to be quite reluctant to establish definite dialect divisions, and Alastair Campbell states flatly that "it is hot possible 3 to draw a dialect map of England in the Old English period." Further, problems in dating and in localizing texts are compounded by the fact that many texts may exhibit large 146

numbers of freely borrowed forms; in addition, scholars suspect that some dialects may have vanished entirely, leav­ ing no documents as testimonies or memorials. In all, as Mr. Campbell remarks, it is safest to say that the documents we do have give only "a fair knowledge of the language in 4 use at a number of centres of culture." However, even though the older boundaries are inex­ actly known, there is some advantage in heuristically matching the Old English dialect areas with the more speci­ fic divisions outlined by MOore and Meech or by Kurath: after all, there is no reason to neglect any bearing which the University of Michigan research may have upon Old English studies. Thus, one might say that the Old English Kentish dialect became the Kentish dialect of Moore and Meech; that the West Saxon dialect became the Middle English Southern dialect; that the Mercian area split into the Southeast- Midland, the Central East-Midland, the Southwest-Midland, the South-central West-Midland, the North-central West- Midland, and the Northwest-Midland dialects; that North Mercian evolved into Northeast-Midland; and that the North­ umbrian area became the Northern dialect. Or, if one uses Kurath*s later corrections and readjustments, he might assume that the Old English Kentish dialect became the Middle English Southeast dialect;, that the West Saxon region produced the Southwest dialect; that the Mercian area 147 evolved into the Southeast Midland and West Midland dialects; that the North Mercian developed into the Northeast Midland dialect; and, once more, that the Northumbrian turned into the Northern dialect in Middle English. Simply speaking, these are the general ways in which the Old English dialect areas relate to those in Middle English. In either case, if dialect scholars bring Moore and Meech or Kurath into the picture, they would work from a basis, as far as geo­ graphical description is concerned, that is at least as firm as any presently used in studies of historical phonology.^ Of course, my suggestions that Middle English dia­ lect areas have a general correspondence to those in Old English, and that relating one known area to its presumed earlier counterpart has great value for the accuracy of phonological research, are hardly new ideas at all. During the last half-century of dialect scholarship, a few re­ searchers have used more recent dialect boundaries to hypothesize the general geggraphical bounds and the phono- . logical components of earlier English dialects. This tech­ nique, which^some linguists have termed retrodiction (as

i r - ' ' opposed to prediction), has been practiced by so reputable a scholar as Alois Brandi, who based his notions of the location of Old English dialects in his Zur Geographie der altenglischen Dialekte6 upon an onomastic investigation of Middle English boundaries. Further, Moore, Meech, and 148

Whitehall were encouraged to find that some of their iso- glosses were adequately substantiated by the more modern 7 boundaries drawn by Alexander. J. Ellis, For those scholars who trust in the. validity of the retrodiction approach, the dialect boundaries of mid-twentieth century England are being plotted in Harold Orton1s Survey of Q English Dialects? this work is far more accurate than Ellis', and thus for those who believe that modern boundaries have a precise, pertinent linguistic and historical relation to past boundaries, Mr. Orton's Survey will be the more val­ uable scholarly tool. However, one must exercise extreme caution in inter­ preting the isoglosses plotted through the methods of linguistic geography, even when the mapping technique is as thorough as that of Mr. Orton. Those early principles of structural dialectology formulated by Nikolai S. Troubetzkoy held that neighboring dialects are separated not by a rigid line passing along a physical barrier, but instead by "zones of transition" which show a blend of phonological and morphological features, and which are characterized further by the diffusion or. isolation, and the convergence or divergence, of any of a number of phones, allphones, phonemes, or morphemes. Indeed, it was because the areas occupied in common by two or more dialects could not be described precisely that Troubetzkoy developed his 149

technique of superimposing maps in order to discover more details about a given phone which may become phonemic in adjacent languages is based on the contention that boun­ daries must be considered sieve-liker and if a scholar takes the axioms of structural dialectology into account, he must acknowledge that modern isoglosses are rather tenuous con­ structs which are chiefly useful for purposes of represen­ tational or cartographical convenience. Further, if this is true of an era which has an almost limitless number of informants to draw upon, it is even more true of a time when the only linguistic witness is provided by scattered texts in a written, not a spoken, language. Researchers who depend on present isophones or isoglosses, to reach back into an earlier stage of the language are thus compounding the possibility of serious error, for this technique intro­ duces a whole new set of linguistic and geographical ambi­ guities into their investigations. Scholars using Mr. Orton*s Survey, for instance, as a basis for casting back into Middle English dialects would do well to review the ideas of the structural dialectoloaists before they, begin, ' . for the notions of Troubetzkoy and the others offer quali­ fications which are too important to ignore. The precepts regarding zones of transition are not the only legacy which Troubetzkoy bequeathed to modern dialectology: his stress on identifying dialects through 150

a combination of phonetic, phonemic, and etymological materials also has been taken up, perhaps quite indirectly, by Middle English philologists. Shortly after Moore, Meech, and Whitehall's monograph, an extremely capable German scholar provided an etymological and lexical guide bo Middle English dialects. In his Zur Geographie des raittelenalischen Wortschatzes,^ w Rolf Kaiser studied the northern and southern versions of the Cursor Mundi, the Legendaries, the Passions, Grosseteste's Chasteau d 'amour, Kindheit Jesu, and of the legend of Barlaam and Josaphat, in order to compile lists of distinctively northern and southern words in the Middle English vocabulary. Kaiser then supplemented his findings from the above sources with extensive reading among such other texts as the Pricke of Conscience, Barber's Bruce, the York and Towneley plays, John Audelay's Poems, the Owl and the Nightingale, and the Ayenbite of Inwit, so as to append additional words to his lists. As the principle results of his research, he presented his selection of "Nordworter," with etymologies and textual referencesr on pages 178-278 of his work, and then included a list of "Sudw&rter" on pages 279-291. In addition, he treated several other problems as by-products of his efforts: in Chapter III, he attempted to localize such works as Genesis and Exodus, Sir Tristrem, Horn ChiIde and Mainden Rimnild, King Horn, and the poems in the Pearl 151 group (including "St.: Erkenwald") on the basis of his vocabulary information. At the end of the Geographiehe printed a map which, in addition to showing the general Old English Northumbrian, Mercian, and West Saxon and Kentish boundaries, also attempted to locate the important authors, the places of composition, and the centers of learning in medieval England. In his stress on a geographical lexicon Kaiser developed a new philological discipline and also, secondarily, confirmed the validity of Moore, Meech, and Whitehall's general approach. On the other hand, the method to be used to locate dialects in the Middle English Dictionary, 12 the greatest compendium of lexical information on the language between 1100 and 1475, is quite unlike Kaiser's because it is indirect. Hans Kurath and Sherman Kuhn have not attempted to assign each word in the Dictionary to a given locale, simply because these scholars were treating the entire language, and most words have the same meaning in several locations. Instead, the editors, who have patterned their work after the Oxford English Dictionary, provide enough citation in all entries to enable a researcher eventually to select the predominant dialect in which a given word occurred; each entry is documented with references to author, text, line number, and editor or editions, together with the date of composition. Thus, one can look up a word 152 like eie,. then check the list of cited documents and texts in the definition, next refer to the Plan and Bibliography in order to make sure of the abbreviations in the defini­ tion, and then look up the given text in Wells' Manual for a general idea of the dialect of the word and text in ques­ tion. As far as documents are concerned, the dialect is usually apparent in the title (for example, the editors have used such documents as the Shrewsbury Assembly Book, the Book of the Abbot of Comermere, the Visitations of Religious Houses in the Diocese of Lincoln, The Statute Merchant Roll of Coventry 1392-1416, and The Chartulary of Monkbretton Priory), or, if not, then the dialect is sure to be mentioned by the modern editor of the document. Using the Middle English Dictionary to find the dialect of a common word is an involved and cumbersome process; yet the information is there in this mammoth work if one wishes to spend some time and effort to get it. Fortunately, if a person wants to trace the origin of a word that is uncommon or obviously dialectal, chances are that the very scarcity of the word will result in a rather speedy chase through very few citations, abbreviations, and documents, and one can bring his quarry down quickly. The better grammars of Middle English fulfill Troubetzkoy's suggestion about the possibility of treating dialect differences through, both phonology and morphology. 13 Yet there is one repository of information which is more convenient than these, and which in addition promises to deal with syntax in a future volume: X refer to Tauno F. Mustanoja's analysis and compilation of language research, A Middle English Syntax, Part I: Parts of Speech. 14 This work, a product of 25 years' labor, is itself a preparation for a later treatment of Middle English word order; as a result, the discussion of parts of speech in this first volume takes on its most important value as a treatise on accidence or morphology. Mr. Mustanoja deals in turn with the problems of "Gender," "Number," "Cases," "Pronouns," "Articles," "Adjectives," "Numerals," "Adverbs," "preposi­ tions," "Verbs," "Interjections," and "Conversion" in Middle English; further, throughout his discussion of these topics, he uses Moore, Meech, and Whitehall's dialect divi­ sions in giving an historical account of what change occurred where. Thus, for example, a reader finds a detailed dis­ cussion in "Cases" of which inflectional endings were levelled first, and where the levelling took place, and in "Pronouns" there is a similar treatment of the geographical location of changes in vowels or consonants. Most important­ ly, Mustanoja is thorough in his exposition: if he mentions that a linguistic change occurred in, say, Kent, he then goes on at once to discuss any analogous or homologous change 154 in all the other dialects for which evidence exists., so that one gains information about J&ent,. the Southwest,, the Midlands, and the North all in the space of a few paragraphs. In addition, he cites and locates sample texts in which the given change or characteristic is represented, and further, he discusses the Old English phonological forebear, the modern English descendant, and the possibility of foreign influence on the change in question. 15 This source, together with Kaiser's Geographie,. Kurath's Plan and Bibliography, and the work of the English Place Name Society (hereafter abbreviated as EPNS), is invaluable for the student of Middle English dialects? one looks forward to the time when Mr. Mustanoja will add to the list of Troubetzkoy's suggested categories with the treatment of Middle English word order. However there is, as one might expect, a set of practically insoluble problems involved in applying syntactic analysis to dialect study. If a scholar wishes to distin­ guish one dialect from another on the basis of the allegation that a certain syntactical pattern appears in one dialect arid not in the other, he may find that he has set himself an impossible task. Simply speaking, the problem is that of,, proving a negative proposition: the scholar would have to assert unequivocally that the given pattern did not appear in either the written or the spoken forms of the dialect in question. Of course, one could never be sure of any 155 statement to this effect,, even if he were familiar with every extant scrap of writing in both dialects--for the pattern may turn up in a text or document not yet discov­ ered. Furthermore, in these cases the written language may permit no extrapolation to the spoken language. This is not to say, however, that a dialect investigation from the standpoint of syntax is a waste of time: there may be some value in ad hoc or for-the-nonce classifications based on the material we do have, just as long as we avoid dogmatic statements. Ideally, such a syntactic inquiry would be conducted with a list of pre-established structures or patterns of word order, so that the researcher could proceed by studying the material in a certain dialect to determine the comparative frequency of the given structure, or perhaps to find whether or not the structure exists. Such a list of patterns is being established by Professor Frederick Theodore Visser in his An Historical Syntax of The English Language. 15a In this enormous feat of scholarship, Professor Visser has so far devoted two volumes and 1305 pages to an analysis of the "Syntactical Units with One Verb" patterns in the history of English; projected for future publication are separate treatments of units with two verbs, and then three or more verbs. In the first two volumes there is a list of close to 900 "types" or syntax structures, which are then fully documented with selected illustrative quotations 156

from Old, Middle, and Modern English. If the Middle English dialect scholar of the future has the time and the temerity to do so, he may choose a few of these patterns and then examine some primary material to. see whether or not there are in fact grounds for making syntactic distinctions between one dialect and another. Still another of the early emphases in structural dialectology has reappeared in the work of J. E. Conner, particularly in his article on "Phonemic Discrimination of Middle English Dialects," 16 where he discusses the possi­ bility of a phonemic inventory of a past stage of the lan­ guage. It is this scholar's contention that "medieval scribes, writing 'phonetically,' would record at least those phonetic distinctions by which native speakers of the lan- guages customarily recognized differences of meaning." 17 If one proceeds with extreme care in his analysis, the phonetic representations can then serve as the basis for a tentative phonemic inventory of a given manuscript; then, over a period of time, as he continues to compile the lists containing the phonemic inventories of all Middle English manuscripts, he may find that a particular combination of vocalic and consonantal phonemes distinguishes one manu­ script from another, and perhaps one dialect at a given . place and time from another dialect. Conner goes on to suggest that a statistical index, registering the distinctive 157 phonemic components of a large number of Middle English texts and documents, would enable a scholar to localize, to date, and to identify the dialect of a manuscript which until then had been unprocessed through the index. This method seems to me to involve more trouble and effort than It is worth: one would have to spend a lifetime in compiling the phonemic inventories of all available manuscripts in order to reduce the margins of error and ambiguity within the statistical index, and even then one perhaps would have a list not of dialects, but instead of scribal idiolects. Conner would place the scholar on a project in which all his research efforts would produce diminishing returns. Conner's particular approach to Middle English phonemics attests to the fact that many of the structuralist principles of dialect research which were fostered by Troubetzkoy are still with us in modern philological scholar­ ship. Perhaps the major contribution of structural dia­ lectology, however, has been Troubetzkoy's curiosity and concern with the kinds of phonological activity which occurs in the zones of transition between dialects. Starting with Troubetzkoy's own Grundzuge der Phonologie (which formed some of the theoretical positions in Martinet's Economie des chanaements phonetiques) 18 and continuing through the work / 1Q of Uriel Weinreich, Henry M. Hoenigswald, and Andre Martinet, scholars began to react against Leonard Bloomfield's state­ ment that 158

since a sound-change is a historical happening, with a beginning and an end, limited to a definite time and to a definite body of speakers, its cause cannot be found in universal considera­ tions or by2gbserving speakers at other times and places. Since this opinion, many of the factors mentioned by Andre Martinet— the economy of effort, the asymmetry of speech organs, the dynamic need for a speaker to express himself— have often been numbered among the actual causes of sound change. 21 In addition, there have appeared a few scholars who are interested in the anatomy of phonological change, and who are producing some remarkable formulations of phonetic and phonemic theory. In what follows, I will show how these early theories have been refined and expanded upon in recent years by Weinreich, Hoenigswald, Martinet, and William 6. Moulton. After studying the reviews and the other critical reactions to Languages in Contact, Uriel Weinreich decided in an article entitled "On the Description of Phonic Interference" to change certain emphases and to redefine and regroup some of his terms. First, he decided that a study of language contact should distinguish between the auditory and mental analysis which a speaker makes of a foreign language or dialect, and the physiological and oral rendition which he gives in speaking that language. 22 Next, in any sound change the phonological researcher must 159 distinguish among the following causes: first, the "phonic factors," involving "differences in the stocks of phonemes, in the componential analysis, and in the distributional patterns of the phonemes"; second, the "extra-phonic factors," referring to the speaker's intentional avoidance of homo- nymic conflicts, or of words which may be taboo in his native dialect; next, the "extra-linguistic factors," mean­ ing the speaker's motivations to speak clearly in order to make himself understood in the foreign tongue; and finally, the "erratic cases," comprising various unclassifiable enunciation errors and lapsi linguae, (p. 2) Further, Weinreich proposed that the phonologist examine change ac­ cording to the syntagmatic ordering— "the relations between sounds in a sequence, i.e., sounds which might occur at a given point in the spoken chain," and the paradigmatic ordering— "the sounds in the pattern, i.e., sounds which might occur in the spoken chain" (2). In this last dis­ tinction, Weinreich actually suggested a revival of the analytical approach of Ferdinand de Saussure. 23 In his particular use of the distinction, however, Weinreich felt that a consideration of the syntagmatic factors should reveal the differences in segmentation between a primary language P and a secondary language S, and that the "unit of interlingual identification should not invariably be the phoneme, but occasionally a sequence 160 of phonemes" (3), together with any "unfamiliar sequences of subphonemic segments" (4). On the other hand, para­ digmatic factors should take into account both phonemic and allophonic behavior, and Weinreich felt that the "con­ textual variants" (4) or allphonic components should actually be analyzed according to Jakobson's list of dis­ tinctive features24— that is, according to whether or not a given phoneme had "voiced, oral, fricative, apical, dental," or other phonetic or subphonemic properties (5). Thus, with the notion that "over- and under-differentia­ tion are matters of distinctive features rather than of phonemes" (6), he altered his previous definitions as follows: whenever a bilingual classifies a feature of sound as distinctive which in the original S-language is redundant, we speak of over-differentiation. In the opposite case— as when a speaker of P- Spanish analyzes the fricativity of an intevocalic English /#/ as redundant— we have under-differen­ tiation (5) . He then added the opinion that "over- and under-differentia­ tion are matters of analysis rather than rendition; even under-differentiation may lead to impeccable renditions" (6). Finally, Weinreich speculated about the results for ana­ lytical research of these modifications of his earlier theories. If sound change were studied through syntagmatic and paradigmatic methods, a scholar would be able to dis­ cuss even subtler possibilities of change, such as these; 161

1. A phoneme of S has an allophone which is subject to further segmentation in terms of language P. • ;2f A special segment is inserted into S forms by the bilingual to manifest a feature which in P does not occur distinctively in a particular sound. 3. A distinctive feature is transferred to another, existing segment of S (8-9). It is obvious, in all, that Weinreich's later article furnishes a considerable refinement of the terms and methods of analysis in Languages in Contact. In contrast to Weinreich's concern with typology, Hoenigswald has been occupied in recent years with a more precise examination of the actual phonetic and morphemic fragmentation involved in sound change. His article on "The Phonology of Dialect Borrowing" was one of a series which was to culminate in a volume entitled Language Change and Linguistic Reconstruction. 25 Hoenigswald had once remarked that "it is through positional allpphones that sound change takes place"; 26 in this most recent volume, he views the phoneme as the sum of its "positional allo- phones," and he speaks of sound change as the transfer or replacement of allophones among the phonemes in a system. Changes themselves can be classed among a number of patterns. In the first of these, there is an "absence of phonemic change" (87). The only clue that such a change may have taken place would be found in orthography, in "script and 162

also in the customary phonemicizations" (89) . In another pattern, two phonemes A and B, which are changing into the new phonemes M and N respectively, each may shed some of their allophones so that the unassigned allophones even­ tually comprise the new phoneme 0. Fourthly, the phonemes A and B may merge entirely, so that the resulting phoneme M contains the sum of all the allophones of A and B.

i Another possible pattern follows from this last: a merger may occur between, say, /h/ and /o/ in such a way that there is eventually an "unconditional loss" of /h/ from the system (91). In addition, Hoenigswald's concern with allophonic behavior leads him to treat the patterns of dependent or conditioned sound change, in which a given phone is in­ fluenced by features in its environment. In the "condi­ tioned merger with primary split" pattern (91), in which the phonemes A and B are becoming M and N as before, the allophones of A may split in such a way that some go to M and some to N, and those going to N may duplicate some of those already going to N from B, so that the two sets may combine redundantly in N. With the "conditioned loss with primary split" possibility, the phoneme A loses one of its allophones permanently (92). Next, in the "condi- ^ tioned increment with primary split" pattern the phoneme M may pick up an allophone which previously had contrasted with nothing, or with the null phoneme, 27 which exists as a locus of contrast in all systems. In his discussion of the final possibility, Hoenigswald is still concerned with the action of allophones: in treating "secondary split," he shows that the phoneme A may split so that certain of its allophones gather, merge, and shift again in becoming phoneme M, while the other allophones of A may undergo the same processes before they become phoneme N (95). All told, Hoenigswald1s breakdown of the ways in which allophones, phones and phonemes shift and recombine provides some additional possibilities for patterned sound change, and his analysis in the 1952 article should be supplemented with the exposi­ tion in this 1960 volume. . One is on surer ground, of course, when he has recourse to living speech as he deals with allophones; there is almost complete uncertainty and con­ jecture when one discusses the allophonic properties based upon a single manuscript from the past. Still, Language Change and Linguistic Reconstruction provides a refined analytical guide for dealing with phonetic change, and it may prove valuable for future discussions of dialect phonolggy. c This book is important in its general discussions of phonetic behavior, too. Hoenigswald agrees with Andre Martinet that sound change is produced through a special vari­ ety of borrowing: "sound change is generally the result of 164

internal stresses and strains within one speech community . . . its mechanics is fundamentally that of borrowing with sound substitution" (55). Further, he outlines the four "chief criteria" for studying sound change: the "replace ment pattern," or the patterns of allophone behavior de­ tailed above; the phonetic or phonemic "affinity" which "exists between the phones and morphs figuring in the replace­ ment process"? the "source for the replacement"; and the "predictability of change" (14-15). These criteria in turn form the basis for his discussion of morphological change. In addition, the various mechanics of change are illustrated with charts, figures, and diagrams, and in all cases he discusses the possible relevance of the forces of analogy and semantic shift, together with the influence of dialect borrowing. In short,.Language Change and Linguistic Recon­ struction presents a thorough discussion of change on allo- phonic, phonetic, phonemic, morphological, and semantic levels, and thus serves as the most significant source in this field since Martinet's Economie des changements phonetiques. The evolution of structuralist principles, and their application to sound change, reaches a high stage of develop­ ment in the theories of Weinreich and Hoenigswald; scholars wasted no. time, moreover, in converting theory to practice. There would be little value in drawing up in this space an 165

annotated bibliography of the structuralist monographs of the early 1960's on language, but I would refer to the current Contrastive Structure series as a practical culmina­ tion of the ideas of Weinreich and Hoenigswald, among many others. William G. Moulton's The Sounds of English and Ger­ man, Frederick B. Agard and Robert J. DePietro's The Sounds of English and Italian, and in particular, Robert P. Stockwell and J. Donald Bowen's The Sounds of English and Spanish, 28 all provide comparative structural analyses of the languages in question on a variety of levels. Such studies could serve as models for the comparisons of any two language systems; and by extension, as Einar Haugen argues in "The Analysis of Linguistic Borrowing," 29 these monographs could guide efforts in dialect study as well. In fact, if the primary data were available, the ideal phonological investigation of two dialect systems would also indicate as precisely as possible the contrastive phonetic, allophonic, phonemic, morphophonemic, and suprasegmental structures of those dialects. In the past few years, there have also been a number of practical applications of the phonological theories of Andr^ Martinet. William G. Moulton has drawn upon some materials of dialect geography in Switzerland to study the relevance of Martinet's ideas to "dialect phonemics," 30 and 166 to test what Charles F. Hockett termed the "drive toward symmetry" 31 in Martinet's notions of vowel and consonant systems* In "The Short Vowel Systems of Northern Switzer- land," 32 Moulton divided the northern region into smaller areas and then studied in paired sets the front unrounded vowels, the back vowels, the front rounded vowels, the mid and low vowels, and the high vowels of adjacent areas; he took particular notice of, for instance, the ways in which the front unrounded vowels and their allophones corresponded from one area to the next. Further, he investi­ gated the pertinence of such of Martinet's concepts as "allophonic range" between one phoneme and the next, and the "phonetic interval between the ranges of two phonemes" (178). Moulton found that neighboring vowel systems ordinarily are indeed symmetrical, and that htere is a correspondence between the location of vowels in one system and that of the vowels in the bordering dialect system. If there is asymmetry, it can be traced to the fact that "the speech under analysis is at a point of (temporal and/or % spatial and/or social) transition from one structure to another" (182). In addition, Moulton found evidence to support Edward Stankiewicz's contention about Polish dialects: 33 he observed that certain phones shift their "phonemic allegiance" (163) across boundaries. Then in 167 conclusion, he offered certain additions and qualifications to Martinet's notion of "diachronic phonemics": Synchronic phonemics assumes that phonemes are discrete, non-overlapping elements, and that the transitions from one phonemic system to another must therefore be completely sharp. Diachronic phonemics, on the other hand, assumes that during the course of time one phoneme can split into two, or two coalesce into one, and every such change implies a period of phonemic indeterminacy in part of the system. Dialect phonemics, it seems, must make both assumptions. It must seek a synchronic analysis which will reveal . . . sharp transi­ tions from one system to another . . . /and7 it must allow for gradual transitions with resulting phonemic indeterminacies (167). In a later article on "Dialect Geography and the Concept of Phonological Space," Moulton took up the study of the maximum differentiation principle 34 and the question of pho-~ netic intervals— those "areas which are topologically pre­ sent in the phonological space of a given system though not occupied by any phoneme" (23). During this research, he found the principle of maximum differentiation to be "very largely confirmed" (28) as he discovered in certain dialects such facts as the following: Both the range and quality of the allophones of /£/ are determined by its position within phonological space. Where /a/ is flanked by /£€/, it will rarely show front allophones; where it is flanked by /3/t it will rarely show back allophones; and where it is flanked by both /ae/ and /O/, it will rarely show any but central allophones. Conver­ sely, where /£/ is not flanked by /ae/, it will often show front allophones; and where it is not flanked by /£/, it will often show back allo­ phones (28). 168

In all, Moulton found most of Martinet's theories to test true, most particularly the general notion that the "dimen­ sions of phonolggical space . . . are furnished by the shape of the organs which produce speech" (24). Of course, Martinet himself has not been idle during the time these hypotheses have been examined and tried; yet in contrast to Weinreich and Hoenigswald, he has altered none of his earlier ideas at all. In recent years, he has been concerned with finding a proper method for the analysis of distinctive allophonic components or features;— "le degagement des unites distinctives et leur identification, notamment en termes de traits distinctifs." 35 Such an analysis he feels must come from within the language; any ready-made componential matrix may not apply with perfect accuracy to the given language, and Martinet rejects the distinctive features analysis of Jakobson, Halle and Fant as "un schema preetabli" (128) which was prepared by scholars who do not number among "les esprits plus soucieuxd' exactitude" (129). Whenever Martinet restates his earlier principles, however, h^ does so wj.th a directness and a clarity that have sharpened over the years: in 1960 he wrote that the precept of "maximal differentiation" holds that phonemes in a system "will tend to be as different from one another as is permitted by the organs of speech which are concerned in their production," 36 for in general "all 169

languages favor optimum contrasts" (191). A phoneme in the process of change, for whatever reason, will be "shifted in the direction where /allophonic/ variations are tolerated and away from the zone where they are inhibited" (191), because any phoneme is "modified until . . . maximal differ­ entiation is obtained" (191). A positional symmetry is thus formed and maintained, and whenever this relative stasis is disturbed, the system readjusts itself: any allophonic or phonemic "gap . . . will tend to be filled" (196), and ultimately sound change itself is the filling of old "gaps," the resultant creation of new ones, and the filling of these in turn. Significantly, this formulation of the nature of phonemic behavior stems directly from an idea which Martinet has held consistently for more than twenty years: a good many and probably most sound-changes seem to be due to an insufficient effort on the part of the speakers to distinguish between neighboring sounds. In all, he be came convinced early of the truth of the prin­ ciple of least effort, and he has stayed with his decision for a long time. As I see it, these taxonomic treatments of the n&ture of split, merger, and replacement are in the main­ stream of the kind of phonological inquiry that has been developing since the Meggrammarians. In actuality, 170

Hoenigswald and Martinet are concerned with the two physical levels on which sound change can take place: Hoenigswald described the various kinds of allophonic change, which historical phonologists have often treated as opening, clos­ ing, fronting, rounding, and unrounding, while Martinet detailed the possibilities of phonemic change, which have been traditionally discussed as assimilation, apocope, syncope, spelling pronunciation, palatalization, mutation, labialization, pharyngealization, compensatory lengthening, epenthesis, dissimilation, haplology, metathesis, and changes in vowel harmony. 38 The historical phonologists have discovered much about these varieties of sound behavior: as examples of allophonic change in Middle English, Winfred Lehmann cites the facts that Middle English /n/ had an /fj/ allophone before velars, and that an allophonic split caused Middle English /ti/ to become Modern English /a/ every­ where except after labials; moreover, the many Middle English grammars are full of examples of such phonemic changes as the syncopation of the /3/ sound in the noun plurals of Chaucer's time. 39 Such examples could be multi­ plied, but my point is that Weinreich, Hoenigswald, and I Martinet have not withdrawn into an area of tangential interest; instead, their work in allophone and phoneme inter­ action is directly related to the traditional concerns in historical phonology. 171

The significance for Middle English dialect study of Weinreich's terminology describing systems in contact, and of Hoenigswald's and Martinet's analyses of allophonic and phonemic behavior is this: such descriptive theories permit a more nearly accurate formulation of the laws of conditioned sound change, on both the phonemic and sub- phonemic levels. Mow that an historical phonologist can refer to such concepts as over-differentiation and under­ differentiation, or the varieties of primary and secondary allophonic split, or push and drag chain effects in phonemic change, he may list more precisely what Robert P. Stockwell calls the "detailed sub-rules that describe the phonetic exponents of the feature bundles" 40 of a given phoneme. In other words the phonologist may now attempt to discover the "bundling potential" (664) of a certain phone or allophone, and he may begin to describe the laws which govern the opposition of short and long vowels in a given system, or the rules which cause the doubling or diphthongization of certain vowels, or the reasons why a vowel may develop a glide (and perhaps why a /j/ instead of a /w/ glide, or vice-versa), or the causes producing variations in vowel quality— differences in height and tenseness, for example— within a system. 41 In all, when the new typology is used in combination with the knowledge of the older historical phonolggists, one can expect an 172 improvement in the codification, so to speak, of the laws of allophonic and phonemic change; then, as Stockwell promises, one can expect more "explicitness" in the descrip­ tions of "the differentiation of the bundling habits of different dialects" (664). Considered from a scientific viewpoint, the gain in explicitness is of tremendous value for language study in general and for dialect study in par­ ticular. I have chosen to present the work of Weinreich, Hoenigswald, and Martinet in such detail throughout this dissertation for two reasons: first, I wanted to show that more work has been done on the kinetics of allophonic and phonemic change than might be suggested by Charles F. Hockett's remarks on coalescence, allophonic split, and the "slow drifting about of expectation distributions" 42 The main reason, however, lies in my conviction that a modern treatment of dialect research must give some account of the major, innovative theories of sound change which are currently held by important scholars. The most recent theories of sound change, of course, may be extrapolated from the work of the genrative grammarians, and to those principles I turn now. The development of modern generative phonology begins with the ideas in Noam Chomsky's Syntactic Structures, proceeds from there through Morris Halle's The Sound 173

Pattern of Russian and through his articles "On the Bases of Phonology" and "Phonology in Generative Grammar," then receives further elaboration in Chomsky's Current Issues in Linguistic Theory, and continues in Halle and Chomsky's The Sound Pattern of English. 43 Two of the main points which are stressed in these books and articles, and which have had determining influence on subsequent researches in this field are, first, that the ultimate phonetic representation of a sentence must be obtained from the syntactic framework through the application of a series of rules, and second, that there is no intervening phonemic level at any stage during the conversion of a sentence from constituent struc­ ture to phrase marker to surface constituent structure to phonetic representation. In other words phonology is syn­ tactically based, and the phoneme is obsolete or at least moribund. More specifically, the various levels in the phonological transformational process are as follows: the constituent structure, or phrase marker representation of a sentence is converted to the surface constituent structure by means of the syntactic component— the set of rewrite rules and transformational rules whose character and function Chomsky first describes in Syntactic Structures. The surface constituent structure itself is a "string of morphemes with a labeled constituent structure, where a 174 morpheme is a terminal symbol, and a label is a nonterminal symbol appearing in the syntactic rules." 44 The morphemes in this string, the "dictionary forms" (6) of the sentence, are composed of a "sequence of phonological segments" (5), and each segment is described in terms of Halle and Jakobson's distinctive features. To these dictionary forms is applied the phonological component— a set of rules which will even­ tually produce the surface phonetic representation of the sentence. Among the rules in the phonological component are morpheme structure or redundancy rules, which determine the phonological admissibility of forms (in English, for example, they would provide that /st^mp/ is admissible and possible and /sdmamp/ is inadmissible and impossible) and phonological rules proper (which provide for such alternations as /wayf/ with /wayvz/ in the plural). In all, then, both the syntactic component and the phonological component are necessary to produce the final phonetic form. 45 Ideally, the rules in the phonological component would attain Chomsky's "level of explanatory adequacy" 46 in which the ultimate phonetic representation and all possible i/— * "contextual variations" (32) of a word could be predicted even if the word had never been heard of if it were "not part of the observed data from which this grammar is con­ structed." (33) In effect, then, producing the final phonetic form would simply be a matter of ordering the rules, 175 or of applying the rules in the correct cyclical sequence. But once the question of rule ordering is raised the possi­ bility of sound change appears, for rules in different orders will at the last produce differing surface phonetic forms. If the presence of subconscious rules in speech is acknowledged, one can see that there are many likely per­ mutations of such rules: a rule may be lost, or simplified? it may spread, or be borrowed, or be spontaneously and intrusively added in a given context from some other locus in the phonological component; or, a rule may become gen­ eralized, so as to apply suddenly in several contexts; further, a rule may be analogized, and the new rule thus formed could apply in unexpected places; in addition, rules may become reordered, or an optional rule may become obliga­ tory or vice-versa, and this new order could cause change immediately or it could present itself somewhere else in the system through analogy or generalization, or through

Kiparsky's phonology over the older methods, or vice-versa. In any event the generative technique may have great potential in future historical studies. In the study of Middle English dialects specifically, the validity of the transformational approach would increase with the greater availability of primary phonological data. It remains to be seen whether or not the classificatory method recommended by Sol Saporta, in which one dialect can be distinguished from ' 4 9 another by the variations in the ordered sequence of rules, can actually be worked out from the data supplied, for ex­ ample, in the volumes of the EPNS. As more phonological re­ search is done on Middle English dialects, and as more phonetic dialect characteristics are found and confirmed, 177 it may be possible at some future time to trace in detail the histories of the different dialects through the rules of generative phonology. To do this, of course, one would have to work in a hash of traditional and generative tech­ niques; but purist opinion to the contrary, this may prove to be a most productive way of doing dialect research, provided that one stays at one or two removes from the quick- or-dead phoneme controversy. As I mentioned in the last part of Chapter III, the graphological approach to dialect study has risen alongside the traditional and modern methods of phonological investi­ gation. Graphology, again, is the study of the written lan­ guage, a manuscript or text at a time, with the aim of extracting orthographic and distributional facts which may be used to identify the dialect of the source in question. This method may work well when one deals with texts from far corners of the country, or with those having distributional peculiarities that may reflect individual characteristics of the spoken language. However, one might have this minor reservation: with sources written in London, under auspices which may have produced a gradual standardization of both the spoken.and written languages, a scholar cannot assert with as much conviction that the scribal hand in a given manuscript represents a certain dialect. Laura Hibbard 50 Loomis, H. M. Smyser, R. W. Walpole and others present a picture of book production in the mid-fourteenth century which causes one to speculate that spoken dialects, and writing styles and orthographic features, may have been confused in ways similar to the engulfing reduplicative procedures of modern mass-production. In their reconstruction of the methods of a fourteenth-century bookshop, these scholars show that the Auchinleck manuscript must have been produced by a number of scribes working at once under the same roof, with an editorial supervisor managing the entire business. Evidently a customer would enter the shop and place an order for a book containing, say, a number of romances, legends, homilies, and humorous poems; the supervisor, or Master, would then select certain pieces of writing for the customer from the several books which were stored on shelves in the shop. Next he would ask several scribes to fill the customer's order, and the scribes, working alone or in groups, would copy out the pieces selected for the customer, perhaps from 51 a loose leaf collection of poems. Throughout this process, 52 the scribes worked in free collaboration with each other, so there was a probability that they would adopt a little of each other's poetic and orthographic styles. Thus there may have been a merging or blending, or at least a general con­ fusion, of those precise graphemic and allographic features which interest the modern dialect scholar. 179

Admittedly, the bookshops are thought to have pro­ duced vernacular works only; yet this is not to say that they were unlikely to be called in to help with official documents on occasion. The problem is, in any event, that the bookshops existed for a long time: the Auchinleck manuscript may have been produced earlier than 1350, and 53 there is a theory that John Shirley (13667-1456) may have been the Master of such a shop. Further, the London shops doubtless attracted lay scriveners from all parts of England, so that the fourteenth century publishing industry was comprised of poets, would-be poets, copyists, and hacks who spoke most of the regional dialects. In addition, the shops continuously recopied, changed, and otherwise redacted the books on their shelves in order to keep up with the latest literary fashions, and there was thus a continual in-breed- ing in the publishing business. In this "work-a-day 54 milieu" the chances were, great, I would suppose, that a homogenization of linguistic and scribal practice would take place, and the enusing styles in poetry and penmanship may have influenced the purity and the dialectal distinc­ tiveness of the orthography throughout England. In any case, before they pronounce on the dialect characteristics of a given text, graphologists should take into account the possible effects of the standardizing procedures which may have been developing in London bookshops in the late Middle Ages. 180

In all, on examining the theories of the structural- ists, and the actual research of the older phonologists and onomatologists, one is struck by the fact that scholarly practice often predates by several years the formulation of theory by one school of linguists or another. For instance, in several ways the guiding principles of the early resear­ chers in Middle English dialect phonology anticipated both the theories of the structural dialectologists and the definitions of dialect proposed by Leonard Bloomfield and 55 Sumner Ives. In 1927, Mary S. Serjeantson arrived at a conclusion similar to the view held by our contemporary structuralists: from her work among place name materials, she noted that there is no such thing as a rigid line of division between two dialect areas; every dialect is, in fact, a border-dialect, sharing5some of the fea­ tures of neighboring districts. Of course, this is a concept which Moore, Meech and Whitehall should have acknowledged before Mr. Whitehall drew the physiographic isophones for their monograph, for, as Hans Kurath has demonstrated in the Plan and Bibliography to the Middle English Dictionary, the Middle English dialect boun­ daries did not in fact occur along features of the land 57 surface. When Bloomfield remarked that a dialect shows a "combination of features," he merely repeated a discovery which Henry C. Wyld had made much earlier: in his 181 study of the "South-Eastern and South-East Midland Dialects in Middle English,Wyld pondered the nature of the "mixed" (142) dialects in thoseregions and observed that the phonemic features showed a "gradual transition" (143) from one dialect area to an adjacent one; he felt generally that a given dialect could be accurately characterized as a "combination of features," (143) and he concluded specifically that the individuality of the London dialect consists in the possession of a group of characteristics which appears to be shared in its entirety by no other dialect type (144). Examples of this apparently reversed sequence of hypothesis and practice are evident even now, for extremely capable scholars in linguistics and phonology still develop prin­ ciples of practical research before a theoretical framework is built: for instance, Robert P. Stockwell used some informal, working notions of graphology several years before Angus McIntosh developed the theories at length, 59 and further, in a study of "The Middle English 'Long Close' and 'Long Open' Mid Vowels,"60 Stockwell to some extent antici­ pated the historical precepts of Paul Kiparsky with the notion that the relation between a linguistic structure at a given date and a derived linguistic structure at a later date can be stated as a series of rules by which the later structure can be gen­ erated from the earlier, plus exceptions (529}.. 182

In all, there are enough examples of this anticipatory ten­ dency to tempt one to the conclusion that certain portions of the modern linguistic theories are not bold or inspir- ingly revolutionary; instead, they are related in an organic and evolutionary way to the massive work done in the past among texts, documents, and manuscripts of all kinds. If this is true, then the theories of the future will evolve from the practice of the present, and the best way for a researcher to make a significant contribution to the work of the future is to draw as widely as possible from 61 the ideas and the methods of the present. Of course, this point is obvious; one is always on solid ground when he knows all about what he is doing, and the surest way to proceed in Middle English dialect scholarship is to start with an overview of the present state of dialect studies, such as I have presented here, and to try the variety of approaches which may suggest themselves from this disserta­ tion. Personally, however, I as a researcher would not be abjectly grateful if I were told to experiment with all the theories and methods discussed here, with the further in-

V ^ ^ , ' junction that I take care to omit none. For this reason, I shall attempt to put these ideas and techniques in a general but succinct perspective. 183

From the beginning of any sort of dialect research project, there are two significant facts which one must keep in mind. First, the language of Middle English source materials is a written language, and after any recourse to modern dialect theory one must exercise every conceivable caution in extrapolating results to the spoken language. Next,, of basic importance to dialect investigation is, simply, the primary data: the written phones, phonemes, and morphemes of the language. Much information about the pri­ mary data of dialects is contained in the work of the historical phonologists throughout the early part of this century; nevertheless, newer techniques have provided addi­ tional phonological information, and the more productive the technique, the more valuable it is. Specifically speak­ ing, this means that for the moment onomastics is far more valuable than graphemics, and that the work of men like Luick must be supplemented from the dialect phonology sections of the volumes of the English Place Marne Society. Onomatology is so particularly productive a technique that a scholar might wish to accumulate additional phonological information on his own, perhaps to add to the dialect discussions of the early volumes of the EPNS series; if so, he might start 62 with J. P. Oakden's 45 points, and work through the items in the extensive bibliographies in the later place name 18 4 studies, or in the monographs by Hjordis Bohman, Sven 63 Rubin, and Bertil Sundby. Most scholars interested in dialect phonology are also concerned with dialect geography, or at least with the relative locations of separate combinations of phonological features. Until the Survey of Middle English Dialects of M. L. Samuels and Angus McIntosh is published, we must be content with the treatment of' Middle English dialect geo­ graphy presented in Moore, Meech, and Whitehall's study, or with the modifications of that monograph by Hans Kurath in his Plan and Bibliography to the Middle English Dictionary For additional geographical research on morphology, one should combine the information of the historical phonologists with that in the first volume of Tauno F. Mustanoja's Middle English Syntax, and for studies of syntax, the scholar should refer to the word-order patterns listed in Visser's series entitled An Historical Syntax of the English Lan­ guage. Lastly, for a geographical approach to the Middle English lexicon, one should begin by using Rolf Kaiser's Zur Geoaraphie des mittelencrlischen Wortschatzes in tandem .1 i i ,i, „ n i,*-..... c - i,i i - - - - with the Middle English Dictionary. Finally, I might stress the point made earlier in this chapter that any attempt to retrodict a relation between modern English speech and its

Middle English forms on the basis of dialect geography should be undertaken only after the utmost circumspection. 185

Once the researcher has supplied himself with an enormous amount of primary data, he should draw in addition upon the scholarship of the phonologists of the past to perform some experiments with the principles of structural dialectology, under the guidance of the model essays in the field by William G. Moulton. I will not repeat my earlier remarks on this subject, but I should stress that the structural dialectologists provide an attitude which enables the scholar to focus on the most minute of allophonic, phonemic, or morphemic changes: their admission of any ad hoc criterion for grouping or comparing and contrasting dialects permits the examination of such subtle phonic differences as, in Stankiewicz's phrase, "the relation be- tween continuity and discreteness." 64 Further, the struc­ tural dialectologists perform a most important service in stressing that languages are in continual change, and in pointing out that stability and instability in a dialect are each temporary conditions. These are only two of the more important concepts in this new field, yet I would emphasize their relevance in the study of all stages of the language, past and present. Weinreich, Hoenigswald, and Martinet proposed theories of sound change which had developed from the early structural dialectologist1s concern with sound behavior in 186 the zones of transition between dialects. These scholars' intense analyses of allophones and phonemes constitute an important advance in descriptive linguistics, for, as Robert P. Stockwell indicates, phonologists can now formulate with increased precision laws of conditioned sound change which will show why a given vowel or consonant changes instead of another. A similarly improved descriptive accuracy may result when the various criticisms and modifi­ cations of Halle and Jakobson's distinctive features theory reach their final form. 65 The practical value for Middle English dialects of the generative phonologist's concern with the addition, loss, and ordering of rules has yet to be demonstrated; 66 yet the technique shows some promise, and I mention it here together with the rest of the scholarship on sound change on the premise that any and all work in the future is likely to grow from a synoptic understanding of the work of the present. In point of fact, this premise is the raison d'etre of any present state investigation of a corpus of scholarship. Middle English dialect studies of the future will be based on an eclectic approach which will combine the most rigorous kinds of theoretical inquiry with the most productive techniques in manuscript analysis and research. Again, scholarly components of this approach exist in some stage 187 of development in the present: some, particularly the efforts of the English Place Name Society, have reached a stage of healthy maturity; others, particularly the writing of generative rules which distinguish one Middle English dialect from another, must mature with the passing of time. The researcher's task is to experiment with the various theories and techniques, in order to find and then develop as thoroughly as possible the most significant among them. 188

NOTES

1. The quoted phrases are from Joseph and Elizabeth M. Wright's Old English Grammar, 3rd ed., London, 1925, p. 4. Most historians^of English agree that the tribal division of Angles, Saxons, and Jutes forms the origin of the Old English dialect areas. There are many important discussions of the phonological differences between the Old English dialects. I have chosen not to reproduce these occasionally quite lengthy treatments; instead, I will mention the chief sources here, together with the suggestion that in the future scholars may wish to treat the phonological histories of given dia­ lects one^at a time— for instance, the changes from North­ umbrian to Northern, or from North Mercian to Northeast- Midland. In addition to Wright and, of course, Luick, the following contain important discussions of Old English dialect phonology: Sweet, Henry. "Dialects and Prehistoric Forms of Old Enalish," TPS, 1875-76, 543-569. See pp. 554-612. Siever, Eduard. An Old English Grammar. Trans, and ed. by Albert S. CooJcT’*" Boston, 1903. pp. 97-114. Huchon, Rene. Histoire de la lancrue anglaise. Tome I. Paris, 1923. See pp. r?l-l53T Wardale, Edith E. An Old English Grammar. London, 1942. See pp. 3-?. Mosse, Fernand. Manuel de 1*anglais du moyen age, I:_ vieil-anglais. 2nd ed. Paris, 1950. See pT 20. Sisam, Kenneth. Studies in the History of Old English Literatur e . Oxford, 1953. On pp. 119- T39 i s his dialect Origins of the Earlier Old English Verse." Campbell, Alistair. Old English Dialects. London, 1963. See pp. 40-5?. Brook, George L. English Dialects. London, 1963. See pp. 40-54. *" Brunner, Karl. Altenglische Grammatik nach der angel- sachsischen Grammatik von Eduard Siever. 3rd ed. Halle, 1965 . See-p p . 37-37CT. Perhaps I need not add that this list is incomplete. 2. See Brook on p. 46. 3. Campbell, p. 10. 189

4 * • Ibid. 5. Again, for the precise phonological content of the several Old English dialects, see the sources in note 1 above.

6 . Alois Brandi, Zur Geographie der altenglischen Dialekte, Berlin, 1915. - 7. See the reference in note 1, Chapter I? p. 33.

8 . Harold Orton and Eugen Dieth, ed., Survey of English Dia­ lects: The Basic Material. Vol. I, Barts 1, 2, 3: The “Six Northern Counties and the Isle of Man, ed. Wilfrid J. Ha Tri- day, Deed's, r9

11. Rolf Kaiser, Zur Geographie des mittelenglischen Wortschatzes, Palaestra , 205 ([Leipzig, ""193 TFr *"1-5X5". ~ 12. Hans Kurath and Sherman M. Kuhn, eds., The Middle English Dictionary, Ann Arbor, 1952— . “ ~ 13. For a bibliography, see Chapter I, note 11. 14. Tauno F. Mustanoja, A Middle English Syntax, Part I: Par t s_of _S^eec_h, Helsinki: Societe Neophflologfque, 1960. 15. For example, I quote from his discussion of the Genitive plural, on page 73: In the plural ME usage is more variable than in the singular. The ending -s, obviously owing to the analogy of the gen. singular ending in -s, is first generalised in the North and North Midlands, but even in these areas it does not entirely supersede the old s-less ending? in Ormulum, for example, -s_ is still a minority ending. In the South the genitive plural in -s is infrequent in early ME, but becomes gradually commoner, particularly in personal nouns. In Kent the -s ending is practically non-existent all through the period, although it predominates in the adjoining SE Midland area. The ending -en[e), found in the W Mid­ lands and particularly in the South (including Kent), is a survival of the OE weak ending -ena, used even for strong masculines and neuters in Northumbrian and late West Saxon. This is relatively common, e.g., in the Digby and Brussels MSS of Aldhelm glosses (late 11th century); cf. A. Napier, OE_GLosses, Oxford 1900, pp. 41-2, note on line 1557. 16. J. E. Conner, "Phonemic Discrimination of Middle English Dialects," Rice Institute Pamphlets, XLIV (1957), i, 17-32. 17. Conner, p . 2. 18. See the discussion in Chapter II. References to Troubetzkoy's Grundzuge can be found in Martinet's Economie on pages 18, 39, 46, and passim. 19. See Chapter III for the discussions of these scholars. 20. Leonard Bloomfield, Language, New York, 1933, p. 388. 191

21. See the discussion on page 224 of R. H. Robins, A Short History of Linguistics, London, 1967. 22. Uriel Weinreich, "On the Description of Phonic Inter­ ference," Word, XIII (1957), p. 2. Subsequent page refer­ ences to thisarticle will be included in the text of this chapter. I intend to follow the same policy in the dis­ cussions of Hoenigswald, Moulton, Martinet, Kuroda, and Chomsky, which follow in paragraphs below. 23. See Robins on page 214.

24. See Chapter II, note 6 6 . 25. Henry M. Hoenigswald, Language Change and Linguistic Reconstruction, Chicago, 1960. In this volume Hoenigswald discusses in full his ideas on Internal Reconstruction, a technique in which the phonemes, morphemes, and lexemes of the past stages of a language are inferred from the exclusive study of a recent synchronic stage of that lan­ guage. The underlying principle is that the phonological residue of past linguistic changes can be isolated in the current language, and it is thus possible to reconstruct earlier forms from the present. Following is a bibliography of writings on Internal Reconstruction: Hoenigswald, Henry M. "Internal Reconstruction," SiL, II (1944), 78-87. Bonfante, Giuliano. "On Reconstruction and Linguistic Method," Word, I (1945), 83-94, 132-161. Hoenigswald, Henry M. "Sound Change and Linguistic Struc­ ture," Lang, XXII (1946), 138-43. .. . "Diachronic Sound-Charts, A Tech- nique to Represent Sound-Change," SiL, VI (1948), 81-94. ______. "The Principle Step in Comparative Grammar,Lang, XXVI (1950), 357-64. "The Phonology of Dialect Borrow- ing," SiL, X (1952), 1-5. Marchand, J. W. "Internal Reconstruction in Seneca," Lang, XXXV (1959), 477-95. Chafe, W. L. "Internal. Reconstruction in Seneca," Lang, XXXV (1959), 477-95. 26. Found on page 364 of the fifth item in the above biblio­ graphy . 27. Hoenigswald gives examples of this contrast on pages 92-93. 19 2

28. These texts were published by the University of Chicago Press, in Chicago and London, in 1963, 1965, and 1965, respectively. 29. See the first few pages of Einar Haugen, "The Analysis of Linguistic Borrowing," Lang, XXVI (1950), 210-231. 30. See William G. Moulton, "The Short Vowel Systems of Northern Switzerland," Word, XVI (1960), 155-182. Moulton uses the term "dialectphonemics" on page 167. 31. Charles F. Hockett, A Course in Modern Linguistics, New York, 1958, p. 451. 32. This is one of a series of studies by Moulton. See the additional references in Chapter II, note 80.

33. See Chapter II, notes 6 and 80. 34. William G. Moulton, "Dialect Geography and the Concept of Phonological Space," Word, XVII (1962), 23-32. Martinet's maximum differentiation principle is explained in Chapter III of this dissertation on page 103. 35. Andre Martinet, La Linguistique Synchronique, Paris, 1965, p. 127. 36. Andre Martinet, Elements of General Linguistics, trans. by Elisabeth Palmer, London, 1964, p. 191. 37. Andr4 Martinet, Phonology as Functional Phonetics, London, 1949, p. 24.

3 8. I owe this distinction and the subsequent discussion to Winfred Lehmann, Historical Linguistics; An Introduction, New York, 1962, pp. 159ff. 39. See Lehmann, pages 149 and 153, and also see note 11 in Chapter I of this dissertation. 40. From page 664 in Robert P. Stockwell, "An Overall Pattern in English Phonology," in Horace G. Lunt, ed., Pro­ ceedings of the Ninth International Congress of Linguists, The Hague: Moulton, 1964. Stockwell's essay is on pages 663-669. 41. I owe this discussion to Stockwell, pp. 665-666. 193

42. Hockett, op. cit., p. 443. Hockett elaborates on this on pages 201-204 of his article "Sound Change," Lang, XLI (1965), 185-204. 43. The bibliography is as follows: Chomsky, Noam. Syntactic Structures■ The Hague, 1957. Halle, Morris. The Sound Pattern of Russian. The Hague, 1959. ______. "On the Bases of Phonology," and "Phono- logy in Generative Grammar," are on pages 324-333 and 334-352 respectively in Jerry A. Fodor and Jerrold J. Katz, The Structure of Language, Engle­ wood Cliffs: Prentice-Hall, 1964. Chomsky, Noam. Current Issues in Linguistic Theory. The Hague, 1-64. ______, and Morris Halle. The Sound Pattern of EnglishT" New York, 1968. 4 4. This is quoted from Sige-Yuki Kuroda, Yawelmani Phonology, Cambridge: MIT, 1967, p. 3. 45. In his article on "Some General Properties of Phono­ logical Rules," Lang, XLIII (1967), 102-118, Chomsky speaks in greater detail: he assumes that the phonological component of a grammar is a system of rules that assigns to each syntactic structure a phonetic representation— or, in the case of free variation, several such representations; that the phonetic representation is a matrix in which rows are associated with features of a universal phonetic system, and entries indicating the position of the segment in question along the phonetic scale (possibly a two-valued one) defined by the feature in question; that the 'syntactic structures' to which phonological rules apply are surface structures, i.e. labeled bracketings of strings of formatives; and that a string of formatives within labeled brackets is itself a matrix in which rows are associated with the features of the universal phonetic system extended (a) by the feature /j-fc- segment that distinguishes segments from junctures, (b) by a feature system for junctures, and (c) by certain 'diacritic features' that express either idiosyncrasies of particular formatives and classes of formatives, or phonologically relevant syntactic properties of these formatives, assigned by 194

convention as features of the particular segments. I will refer to such representation of strings of formatives as 'phonological representation* (103). In this same article, he concludes that It seems that the rules of phonology are linearly ordered and apply in a cyclic manner, as determined by surface structure, forming the phonetic represen­ tations of larger units from the ideal phonetic repre­ sentation of their underlying constituents, by a fixed set of principles. Certain rules apply only once in the cycle, by virtue of the fact that they are restricted to the level of words (118). 46. Chomsky, Current Issues, p. 31. 47. I owe this discussion of the possible permutations of phonological rules in sound change to Mr. David Stampe of the Linguistics Department of Ohio State University. 48. Paul Kiparsky, "Sonorant Clusters in Greek," Lang, XLIII (1967), p. 619. Kiparsky's conclusion in fact is that "certain rules, accounting for a relatively well- understood part of Greek phonology, generalize to other more obscure parts of Greek phonology." (634) I do not intend these remarks on generative phonology to sound glib, for as is generally known, generative studies ordinarily contain a tremendous amount of intelligent appli­ cation of principle by some quite capable scholars. Some­ times such application results in a reevaluation of the evi­ dence of the structuarlists. For instance, in a privately circulated paper on "Linguistic Universals and Language Change," PEGS, 1967, Kiparsky reinterprets the data in William G. Moulton's 1960 article to show that the symmetry in a given system is "phonetic rather than phonemic symmetry" (20) . More importantly, Kiparsky in this paper shows that if one is content to adopt heuristic devices, it is possible to look at dialects without knowledge of their historical origin . . . t o explain any ordering difference between them as due to the adoption of a spreading rule at one position in the sequence of rules in dialect 1 and at some other position in dialect 2 (26) . 195

He also notes that a pair of rules may come to be differently ordered in different dialects by another wave-theory effect. For example, if rule A spreads from West to East, and rule B spreads from East to West across some dialect area, then, if the two rules are critically ordered with respect to each other, the Western area will end up with the order (A, B), and the Eastern area with the order (B,A) (26). Still later, Kiparsky adds this commentary on dialects: The typical form of rule addition is the borrowing of rules among adults; simplification typically occurs in the learning of language by children. An interesting consequence of this is that isoglosses formed by the spread of rules over a speech territory should form large, coherent dialect areas, whereas those formed by simplification should be characteris­ tically discontinuous because of independent develop­ ment of the same change in several speech communities. The historically interesting isoglosses, therefore, should be based on the presence vs« the absence of rules and not on differences in the form and order of shared rules (34). 49. See the discussion and note 91 in Chapter IX. 50. I refer to the following articles: Root, R. K. "Publication before Printing," PMLA, XXVIII (1913), 417-431. Brusendorff, Aage. The Chaucer Tradition, Copenhagen, 1925. See pp. 54ff. Tatlock, J. S. P. "The Text of the Canterbury Tales in 1400," PMLA, L (1935), 100-139. Bennett, H. S. "The Author and His Public in the Fourteenth and Fifteenth Centuries," E&S, XXIII (1937), 7-24. Manly, John, and Edith Rickert. The Text of the Canter­ bury Tales. Chicago, 1940. See Volume I, pp. 24, 60, 72, 119, 203, 225, and 423. Loomis, Laura H. "The Auchinleck Manuscript and a Possible London Bookshop of 1330-1340," PMLA, LVII (1942), 595-627. Walpole, R. N. "The Source MS of Charlemagne and Roland and the Auchinleck Bookshop,1* MLN LX (1945), 22-26. 196

Smyser, H. M. "Charlemagne and Roland and the Auchinleck MS," Speculum, XXI (1946), 275-288. 51. Smyser, p. 282. 52. Walpole, p. 22. 53. Bennett, p . 19. 54. Loomis, p. 622. 55. See the discussion in Chapter II. 56. Mary S. Serjeantson, "The Dialects of the West Midlands in Middle English," RES, III (1927), p. 57. 57. See note 10 in this Chapter. 58. Found in E&S, VI (1920), 112-145. 59. For McIntosh on Graphology, see the discussion in Chapter Ill I refer to the article by Stockwell and C. Westbrook Barritt entitled "Some OE Graphemic-Phonemic Correspondences— ae, ea, and a," SiL;- Occasional Papers, IV (1951), 1-39. 60. Found in Texas Studies in Literature and Language, II (1961), 529-538. 61. There is ample precedent for work on just this level of intense and energetic application: Robert P. Stockwell shows it in the works cited in notes 59 and 60, and in his "The Phonology of Old English: A Structural Sketch," SiL, XIII (1958), 13-24, and "Structural Dialectology: A Proposal," A S , XXXIV (1959), 258-68. So does Bertil Sundby in his Studies in Middle English Dialect Material of the Worcester­ shire Records, Oslo, 1963. 62. See Chapter I,note 26. 63. See the discussions in Chapter III. 64. See Chapter II, note 67. 65. I will not go into these here, but a sample criticism is Peter Ladefoged's suggestion that phenologists do away with the binary treatment of features by introducing the full "number of different integers" which would specify the "number 197 of contrasts which a given language uses" within a given parameter.. Thus a phoneme /A/ would have such a designa­ tion as /2 -6/,meaning that it has, say, 2 out of the 6 contrasts in graveness available within the language. See p. 73 of Ladefoged's Linguistic Phonetics, Los Angeles: UCLA, 1967. Ladefoged details his preference for a new distinctive feature system on pages 74-90.

6 6 . Halle and Chomsky's Sound Patterns of English is to cover "the evoluation of the English vowel system beginning with Hart and going through John Wallis, Christopher Cooper, and John Batchelor to the present" (p. 7 86 in Halle and S. Jay Keyser's review of Bror Daniellsson's John Hart's Works on English Orthography and Pronunciation, in Lang7 XLX1I (1967), 773-7 86); yet there is still nogenerative study as yet based, for instance, on the information in the volumes of the English Place Name Society. BIBLIOGRAPHY

Agard, Frederic*, r'., and Robert J. DiPietro. The Sounds of English and Italian. Chicago, 1965. Aiken, Janet Rankin. Why English Sounds Change. New York, 1929. Bennett, H. S. "The Author and His Public in the Fourteenth and Fifteenth Centuries," E&S, XXIII (1937), 7-24. Bemdt, Rolf. Einfuhrung in das Studium des Mittelenglischen. Halle, 1960. Bischoff, Bernhard. "The Study of Foreign Languages in the Middle Ages," Speculum, XXXVI (1961), 209-24. Bliss, A. J. "Three Middle English Studies," English and Germanic Studies, II (1948), 41-2. Bloomfield, Leonard-. Language. New York, 1933. Bohman, Hjordis. Studies in the Middle English Dialects of Devon and London. Goteborg, 1944. Bolinger, D. L. "Visual Morphemes," Lang, XXII (1946), 333- 340. Bonfante, Giuliano. "On Reconstruction and Linguistic Method," Word, I (1945), 83-94, 132-161. Bradley, H. "Dialect in English Place-Names," Acad., XXIII (563), Feb. 17, 1883, 116. Brandi, Alois. Zur Geographie der altenglischen Dialekte. Berlin, 1915. Bronstein, Arthur J., and Esther R. Sheldon. "Derivatives of Middle English o in Eighteenth- and Nineteenth- Century Dictionaries," AS, XXVI (1951), 81-89. Brook, Geroge Leslie. English Dialects. London, 1963.

198 199

Brunner, Karl. Abriss der mitte lenqli schen Grammatik. Halle, 1938-1953. . Die Englische Sprache. 2 vols. Tubingen, 1960-62. • Altenglische Grammatik nach der angels&ch- sischen Grannnatlk von Eduard Sievers. 3rd ed. Halle, 1965. Brusendorff, Aage. The Chaucer Tradition. Copenhagen, 1925. Bulbring, Karl. Altenglisches Elementarbuch. Heidelberg, 1902. Campbell, Alistair. Old English Grammar. Oxford, 1959. Chafe, W. L. "Internal Reconstruction in Seneca," Lang, XXXV (1959), 477-95. Chomsky, Noam. Syntactic Structures. The Hague, 1957. . Current Issues in Linguistic Theory. The Hague, T964T------"The Current Scene in Linguistics: Present Directions," CE, XXVII (1966), 587-95. "Some General Properties of Phonological Rules," Lang, XLIII (1967), 102-128. Connor, J. E. "Phonemic Discrimination of Middle English Dialects," Rice Institute Pamphlets, XLIV (1957), i, 17-32. Dauzat, Albert. La gfeographie linguistique. Paris, 1922. Dieth, Eugen. "A New Survey of English Dialects," E&S, XXXII (1946), 74-104. "Middle English Lengthening in Open Syllables," TPS"(19F2), 124-148. Dyen, Isidore. "Why Phonetic Change is Regular," Lang, XXXIX (1963), 631-7. c- 200

Ekwall, Eilert. "Ortshamenforschting ein Hilfsmittel fur das engl. Sprachgeschichte-Studium," GRM, V (1913), 592-608. "• "The Middle English a/ff Boundary," ES, XX (1938), 147-68. Ellis, Alexander J. On Early English Pronunciation, Part V , Existing Dialectal as Compared with west Saxon Pronun­ ciation. EETS, Extra Serj.es, LV1. London, 1889. . English Dialects— Their Sounds and Homes. London, 1890. Emerson, Oliver, F. A Middle English Reader. New York, 1915. Fillmore, Charles J. "The Position of Embedding Transforma­ tions in a Grammar," Word, XIX (1963), 208-231. ______Indirect Object Constructions in English and the Ordering of Transformations. The Hague, 1965. Flasdieck, H. M. "Zum Lautwert von me. | im 18. Jahrhundert," Anglia, LX (1936), 377-83. "Zur Lautgeschichte von me. i im Frlihneuen- glischen,1U Anglia Beiblatt, XLIX (1938), 182-7. Fodor, Jerry A., and Jerrold J. Katz, eds. The Structure of Language. Englewood Cliffs: Prentice-Hall, 1964. Fourquet, Jean. "Classification dialectale et phonologie evolutive." Pages 55-62 in Diego Catalan, ed., Miscelanea Homenaje A Andrd Martinet, Vol. II, Madrid, 1958. Francescato, Giuseppe. "Structural Comparison, Diasysterns, and Dialectology," Zeitschrift fttr Romanische Philologie, LXXXI (1965), 484-491. Francis, W. Nelson. "Some Dialect Isoglosses in England," AS, XXXIV (1959), 243-250. "Graphemic Analysis of Late Middle English MSS,“ Speculum, XXXVII (1962), 32-47. 201

French, W. H. "Dialects and Forms in Three Romances," JEGP, XLV (1946), 125-32. Gleason, Henry Allan. An Introdaction to Descriptive : Linguistics. New. .York, .1961. Greenburg, Joseph H. "Synchronic and Diachronic Universals in Phonology," Lang, XLII (1966), 508-17. Grootaers, Willem A. "Origin and Nature of the Subjective Boundaries of Dialects," Orbis, VIII (1959), 355-84. "Les premiers pas i. la recherche des unites dialectales," Orbis, XII (1963), 361-80. ______. "La discussion autour des frontieres dialectales subjectives," Orbis, XIII (1964), 380-98. Gross, Charles. Sources and Literature of English History from the EarTiest Times to about 1485. 2nd edition. London, 1915. Halle, Morris. "Phonology in Generative Grammar," Word, XVIII (1962), 54-72. ^ Hartnack, Justus. Wittgenstein and Modern Philosophy<, Trans, by Maurice Cranston. New York, 1965. Haugen, Einar. "The Analysis of Linguistic Borrowing,” Lang, XXVI (1950), 210-31. "Dialect, Language, Nation," American Anthro- pologist, LXVIII (1966), 922-935. Hill, Archibald A. "Early Loss of fr] before Dentals," PMLA, LV (1940), 308-359. Hockett, Charles F. A Manual of Phonology. Baltimore, 1955. • . A Course in Modern Linguistics. New York, 1958. Hoenigswald, Henry M. "Internal Reconstruction," SiL, II (1944), 78-87. "Sound Change and Linguistic Structure," Lang, XXII (1946), 13 8-43. 202

__ . "Diachronic Sound-Charts, A Technique “ to Represent Sound-Change," SiL, VI (1948), 81-94. "The Phonology of Dialect Borrowino," SlL, X (1952>, 1-5. ______. Language Change and Linguistic Recon- "“struction. Chicago, i960. ” “ ~ Horn, Wilhelm. Historische neuenglische Grammatik. Strass- burg, 1908. ” ~~ * ~ ~ ___ ■______. Laut und Leben; englische Lautgeschichte der neueren Zeit (IflFO-rSSO T. BearF. und" hrsg. von Mar tin- Lehnert. ~2* vols. Berlin, 1954. Huchon, Rene. Histoire de la lanaue anglaise. Tome I. Paris, 19 2 J. “ Ives, Sumner. "A Theory of Literary Dialect," Tulane Studies in English, II (1951), 137-82. Ivid, Pavle. "On the Structure of Dialectal Differentiation," Word, XVIII (1962), 33-53. ______. "Importance des caracteristiques structurales pour la description et la classification des dialects," Orbis, XII (1963), 117-31. Jacobson, Rodolfo. "The London Dialect of the Late Four­ teenth Century; a Transformational Analysis in Historical Linguistics," DA, XXVII (1967), 2141-A. Jacobson, UIf. Phonological Dialect Constituents in the Vocabulary oF Standard- EnalTsHT Lund^Studres in EngTIshr XXXI 11963 f, 1-33S'. Joos, Martin, ed. Readings in Linguistics. Washington: ACLS, 1957. Jordan, Richard. "Die mittelenglischen Mundarten," GRM, II (1910), 124-34. ______. Handbuch der mittelenglischen Grammatik. ~Heidelberg, 1934. - - - Kaartinen, Anja, and Tauno F. Mustanoja. "The Use of the Infinitive in A Book of London English 1384-1425," NM, LIX (1958), 179-92. 203

Kaiser, Rolf. Zur Geoaraphie des mittelenalischen Wortschatzes. Palaestra, 2o5 (l9~J7) ,“T-318.~ Kaluza, Max. Historische Grammatik der Englischen Sprache. 2 vols. Berrin-Schoneberg,i506-07. ~ Kennedy, Arthur G. : A Bibliography of Writings on the English Language f r om the Beg inn mg of Printing “"to ~the Enid oF 1 ^ 2 . "Cambridge and”~New Haven, T5TT. Kiparsky, Paul. "Linauistic Universals in Lanauaae Chanae," PEGS, 4/21/67, 1-43. "Sonorant Clusters in Greek," Lana, XLIII TTS’STJ"^ SI9-635. -- K^keritz, Helae. "The Juto-Kentish Dialect Boundary," AS, XVI (1941), 270-77. Koziol, H. "Der Abfall des nachtoniaen -e im Mittelenalischen," Anglia Beiblatt, XLVIII (1937),"307-09.------Krapp, George P. The Enalish Lanauaae in America. New York, 1925. " Kristensson, Gillis. "Another Approach to Middle Enalish Dialectology," ES, XLVI (1965», 138-56. Kuhn, Sherman M. "The Dialect of the Corpus Glossary," PMLA, LIV (1939), 1-20. Kurath,. Hans, and Sherman M. Kuhn, eds. The Middle English Dictionary. Ann Arbor, 1952— . ~ _ - _ Kurath,. Hans, Margaret S. Palmer, and Richard L. MeKelyey. Middle Enalish Dictionary; Plan and Bibliography. Ann Arbor,I ■ I I 1954.I I I II I ■ I II I I I...... ' Ill nil II I* I III!...... I I — Kurath, Hans. "The Loss of Long Consonants and the Rise of Voiced Fricatives in Middle Enalish," Lana, XXXII (1956), 435-45. "Phonemics and Phonics in Historical Phonology," AS, XXXVI (1961), 93-100. Kuroda, Siae-Yuki. Yawelmani Phonoloay. Cambridge: MIT, 1967." 204

Lehmann, Winfred. Historical Linguistics: An Introduction. New York, 1962. ~ Levin, Samuel R. "Neaative Contraction: An Old and Middle English Dialect Criterion," JEGP, LVII (1958), 492-501. Liberman, A. S. "Some Notes on the. History of Middle English CeJ and .Co 2 in Open Syllables," ZAA, XIII (1965), 21-24. "On the History of Middle English sc and a ," m “ 67 “11966), 66-71. Loomis, Laura H. "The Auchinleck Manuscript and a Possible London Bookshop of 1330-1340," PMLA, LVII (1942), 595-627. Luick, Karl. Historische Grammatik der Englischen Sprache. With additions by Friederich wild and"Herbert KozfolT Reprinted by Basil Blackwell in Great Britain by per­ mission of Bernhard Tauchnitz Verlag, Stuttgart. 2 vols. Cambridge: Harvard Press, and Stuttgart: Bernhard Tauchnitz, 1964. Lunt, Horace G. Proceedings of the Ninth International Congress of Linguists7 Cambridge, Mass., August 27-31, 1962. Janua Linguarum, Ser. Maior, XII. The Hague, 1964. McDavid, Raven I. "Derivatives of Middle Enalish o: in the South Atlantic Area," QJS, XXXV (1949), 496-504. _ __ . "Structural Linguistics and Linguistic GeograpHy,^ Orbis, X (1961), 35-46. McIntosh, Anaus. "The Analysis of Written Middle English," TPS, 1956, 26-55. _ . "A New Approach to Middle English “bfarectoTogy," ES, XLIV (1963), 1-11. •______, and M. A. K. Halliday. Patterns of language, Papers in General, Descriptive and Apprie

MacLeish, Andrew. "Patterns in the Late East Midland Subject-Verb Cluster: A Quantitative Synchronic Description," DA, XXXI (1961), 865. Manly, John, and Edith Rickert. The Text of the Canterbury Tales. Chicago, 1940. “ ” . Marcband, J. W. "Internal Reconstruction of Phonemic Split," Lang, XXXII (1956), 245-53. Marckwardt, Albert ;H. "An Unnoted Source of English Dia­ lect Vocabulary," JEGP, XLVI (1947), 177-82. Martinet, Andrd. Phonology as Functional Phonetics. London, 1949. ” "Diffusion of Language and Structural Linguistics," RPh, VI (1952), 2-13. ______. "Dialect," RPh, VIII (1954), 1-11. ______. feconomie des changements phonetioues. Berne, ±955. Elements of General Linguistics. Chicago,

rg-6T7 — *----- ______. La linguistigue synchronigue. Paris, 1965. Mase, Y. "Une nouvelle tentative pour tracer les fronti^res subjectives des dialectes," Orbis, XIII (1964), 357-79. Matthews, William. "South Western Dialect in the Early Modern Period," Neophilologus, XXIV (1939), 193-209. Mawer, A., and F. M. Stenton, ed. Introduction to the Survey of English Place-Names. Cambridge, r9T2’57 Moore, Samuel, Sanford Brown Meech, and Harold Whitehall. "Middle English Dialect Characteristics and Dialect Boundaries," Essays and Studies in English and Compara- tive Literature, University of Michigan~Pubrrcatron T n Language and-Literature7'~XIII(1935) , ~T-617 ~ Morsbach, Lorenz. Mitteleiiglische Grammatik. Pt. I. Halle, 1896. c Mosse, Fernand. Manuel de 1'anglais du moyen ^ge, I: vieil-anglais ■ 2nd”ed. ParTs, ISTS’OT 206

______. A Handbook of Middle English. Translated --Ey~Jarnes A. Walker . ~BaTtiinore, “ Moulton, William G. "The Short Vowel Systems of Northern Switzerland," Word, XVI (1960), 155-82. __ _ . The Sounds of English and German. Chicago, ”T5T6’5‘. “ ~ " ~ ~ Mustanoja, Tauno P. A Middle English Syntax? Part X; Parts of Speech. SocidttT~N£ophl'IoToaTque, XXIII_(T96'5y,

Oakden, J. P. Alliterative Poetry in Middle English; The Dialectal and"Metricar Survey7 Vol. I. Manchester, I5T07------Orton, Harold. "The Isolative Treatment in Living Midland Dialects of OE £ Lengthened in Open Syllables in Middle English," Leeds Studies in Enalish and Kindred Languages, 7-8 (lS’SSTj 97-128. * ____ , and Eugen Dieth, eds. Survey of English Dialects: The Basic Material. Vol. 17 Parts"T7~27* 3: The Six Northern Counties and the Isle of Man, ed. wrr£red-j7 Halliday, Leeds, voIT IV, Parts 1, 2: The Southern Counties, ed. Martyn P. Wakelin, Leeds, rsrsy. - - - Patch, H. R., and R. J. Menner. "A Biblioaraphy of Middle English Dialects," SP, XX (1923), 479-95. Pederson, Holger. Linguistic Science in the Nineteenth Century: Methods andT Results. Translated by John Webster Spargo. HarvarcT7_EF3l. Peterson, Paul W. "Dialect Groupina in the Unpublished Vercelli Homilies," SP, L (19153), 559-65. Poaatscher, Alois. "Die Enalisqhe 5£/£ Grenze," Anglia , XXIII (1901), 302-309." Pop, Sever. La dialectologie. 2 vols. Louvain, 1950. Pulgram, E. "Structural Comparison, Diasystems, and Dia­ lectology," Linguistics, IV (1964), 66-82. Heed, David W. "The History of Inflectional n in English Verbs," MA, IX (1949), 131-32. - Riffer-MaiSek, Dora. "The Graphemic Inventory of a Middle English Manuscript," Studia Romanica et Analiea Zagrabiensia, XXI (1966) , T27-35’. ~ 13 Ritter, Otto. "Zur Enalischen ail/e Grenze," Analia, XXXVIII (1913), 269-75. Robins, R. H. A Short History of Linguistics. London, 1967. Root, R. K. "Publication before Printina," PMLA, XXVIII (1913), 417-431.------Rubin, Sven. The Phonology of the Middle English Dialect of Sussex. Lund 7 1§5T7 ” Samuels, M. L. "Some Applications of Middle English Dia­ lectology," ES, XLIV (1963), 81-94. Saporta, Sol. "Ordered Rules, Dialect Differences, and Historical Processes," Lang, XLI (1965), 218-224. Serjeantson, Mary S. "The Dialectal Distribution of Certain Phonoloaical Features in Middle Enalish," ES, IV (1922),'93-109, 191-198, 223-233. ' "The Dialects of the West Midlands in MiddTe“EngTTsh," RES, III (1927), 54-67, 186-203, 319-331. Sievers, Eduard. An Old English Grammar. Translated and edited by Alberfs. Cook"7“ Bos€bn7"”1903. ______. Altenglisch Grammatik nach der. engelsachsischc GrainmaEDc von E. Sievers. Neubearbeitet von K. Brunner. Tubingen, 1965. Simko, Jan. "A Criterion of Relationship between Lanouaaes," ZAA, V (1957), 191-99. ^ Sisam, Kenneth. Studies in the History of Old English Litera­ ture. Oxf ord, 19‘53” “ ~ 208

Sommerfelt, Alf. "The Structural Point of View Applied to the Analysis of the Consonant System of a Doneaal Dialect," Celtica, V (1960), 107-115. Sprinaer, Otto. "Dialektaeoaraphie und Textkritik, " PMLA, LVX (1941), 1163-1178' Stankiewicz, Edward. "The Phonemic Patternings of the Polish Dialects, A Study in Structural Dialectology," pp. 518-530 in Morris Halle, ed.. For Roman Jakobson, The Hague, 1956. “ ______' . "On Discreteness and Continuity in Structural Dialectology," word, XIII (1957), 44-57. Stockwell, R. P. and C. Westbrook Barritt. "Some OE Graphemic-Phonemic Correspondences— ae, ea, and a," Studies in Linguistics; Occasional Papers, IV (1951), 1-39 o Stockwell, Robert P. "The Phonology of Old English: A Structural Sketch," SiL, Xlli"(1958), 13-24. ______"The Middle English 'Long Close' and PLong Open1 Mid Vowels," Texas' Studies in Literature and Language, II (1961), !T5sF^38. ~~ ______. "An Overall Pattern in English Phonology," pages ?6’5’-6‘69 ”in Horace G. Lunt, ed., Proceedings of the Ninth International Congress of Linguists, TEe” Hague, 196?. ~ " ~ ______, and J. Donald Bowen. The Sounds of English and"~SpanisTET. Chicago, 1965. “* Sundby, Bertil. "Middle English Overlapping y and ft and its Phonemic Significance," Anglia, LXXIV (1956), 438-444. ___ __ . Studies in Middle English Dialect Material o3T~tfie Worcestershire"'Records. Oslo, 1963. Sweet, Henry. "Dialects and Prehistoric Forms of Old Enalish," TPS, (1875-76), 543-569. Tatlock, J. S. P. , The .Text of the Canterbury Tales in 1400," PMLA, L (1935), 100-139. 209

Troubetzkoy. N. S. Principles de phonoloaie. Translated by. J. Cantineau. Paris, .1964. Uldall, ,H. J. "Speech and Writina," Acta Linguistica, IV (1944), 11-16. "■ Vachek, Josef. "Some Remarks on Writing and Phonetic Transcription," Acta Linguistica, V (1945), 86-93. ______. "Two Chapters on Written English," Brno Studies in English, I (1959), 7-34. Visser, Frederick Theodore. An Historical Syntax of the English Language, Part One:~ Syntactical Units with One Verb. £ vols. Leiden, 19'63”Tvdrr ITT and 19‘S’d (Vol. II f. Walpole, R. N. "The Source MS of Charlemagne and Roland and the Auchinleck Bookshop," MLN, LX^llSTdSTl ZZ-Z6. Wardale, Edith E. An Introduction to Middle English. London, 1937. ~ ~ ~ ______. An Old English Grammar. London, 1942. Watson, Georae. "Dialect Survivals of Analo-Saxon Inflection," JEGP, XXXV (1936), 44-60. Weinreich, Uriel. Languages in Contact. New York, 1953. "Is a Structural Dialectoloay Possible?" “Word7~X TldS4), 388-400. "On the Description of Phonic Interference," Word! XIII (1957), 1-11. Wells, John E. A Manual of the Writings in Middle English 1050-1400. New Haven,”19267" Plus "supplements I-VIII, extending coverage through July, 1941. Supplement IX, by Beatrice Daw Brown, Eleanor K. Heningham, and Francis Lee Utley (New Haven, 1951), covers to December, 1945. ?' Whitehall, Harold, and Theresa Fein. "The Development of Middle English u in Early Modern British and American Enalish," JEGP, XL (1941), 191-213. 210

Whitney, William Dwight. Language and the Study of Language. New York, 1867. " “ Williams, Edna Rees. The Conflict of Homonyms in English. Yale Studies in English, V6l7 100. “New Haven, 1944 . Wrenn, C. L. "Philology: General Works," YWES, XVI (1935), 30-65. ‘ Wriaht, Joseph. The Enalish Dialect Dictionary. London, 1898-1905. “ ______;____ . The English Dialect Grammar. Oxford, x905. __ . Old English Grammar. London, 1908. . An Elementary Middle English Grammar. ” London, 15*23 ~ “ *' ” ______and Elizabeth M. Old English Grammar. 3rd “ed. London, 1925. “ “ "* " Wyld, H. C. "The Treatment of OE y in the Dialects of the Midland and SE Counties in ME," Enalische Studien, XLVII (1913), 1-58, 145-66. ““ ______. A Short History of English. London, 1927, "South-Eastern and South-East Midland Dialects Tn Middle English," E&S, VI (1920), 112-145. ____ _ . A History of Modern Colloquial English. London, r9*5'6".