The history of final vowels in english

The Early Schwa Deletion Rule 162 Chapter 7 Schwa preservation in Late Middle English as a prosodie phenomenon 171 7.1.. A n examination of the evidence for phonemic identification of th

The History of Final Vowels in English

Topics in English Linguistics

The History of Final Vowels

© Printed on acid-free paper which falls within the guidelines of the

A N S I to ensure permanence and durability

Library of Congress Cataloging in Publication Data

Minkova, D o n k a , 1944 —

The history of final vowels in English : the sound of

muting / D o n k a Minkova

p cm — (Topics in English linguistics ; 4)

Includes bibliographical references and index

ISBN 0-89925-784-4 (alk paper)

1 English language —Vowels I Title II Series

The history of final vowels in English : the sound of muting /

D o n k a M i n k o v a — Berlin ; New York : M o u t o n de Gruyter,

1991

(Topics in English linguistics ; 4)

ISBN 3-11-012763-6

N E : G T

© Copyright 1991 by Walter de Gruyter & Co., D-1000 Berlin 30

All rights reserved, including those of translation into foreign languages N o part of this

book may be reproduced or transmitted in any form or by any means, electronic or

mechanical, including photocopy, recording, or any information storage and retrieval

system, without permission in writing f r o m the publisher

Printing: Ratzlow-Druck, Berlin

Binding: Lüderitz & Bauer, Berlin

Contents

Preface ix

Chapter 1

Schwa in the history of English 1

1.1 Problems involved in the present study 1

1.2 Forces at play in schwa loss 2

2.2 Lorenz Morsbach (1896): Mittelenglische Grammatik 16

2.3 Karl Luick (1921-1940): Historische Grammatik der

engli-schen Sprache 20

2.4 Richard Jordan: (1934/1968) Handbuch der mittelenglischen

Grammatik 23

2.5 Wilhelm Horn/Martin Lehnert (1954): Laut und Leben

Eng-lische Lautgeschichte der neueren Zeit (1400 — 1950) 24

2.6 Karl Brunner (1962): Die englische Sprache, ihre

geschichtli-che Entwicklung Vols I, II 24

2.7 Joseph Wright and E M Wright (1923): An Elementary

Middle English Grammar 25

2.8 Fernand Mossé (1949): Manuel de l'anglais du moyen âge 26

2.9 Jacek Fisiak (1970): A Short Grammar of Middle English 27

2.10 Peter Erdmann (1972): Tiefenphonologische Lautgeschichte

der englischen Vokale 27

3.4 Unetymological final -e 55

3.5 Elision in hiatus 62 3.6 Rhyme evidence 69 3.7 Scribal errors, editorial corrections 71

3.8 Graphically unmarked evidence 75

Chapter 4

Phonological aspects of schwa loss 87

4.1 The phonemic status of schwa in Middle English 87

4.2 The Middle phonemic inventory 91

4.3 Distribution 104 4.4 Phonetic parameters Distinctive features 106

4.5 Phonological correlates of schwa loss 109

4.6 Some consequences 114

Chapter 5

Morphological aspects of schwa loss 125

5.1 The morpho-syntactic functions of schwa 125

5.2 The morphologization of schwa loss 133

5.3 Classification and ranking of factors for schwa loss 134

5.4 Syntactic correlates 140

5.5 Extralinguistic factors 146

5.6 Hierarchy of factors within individual word classes 149

Chapter 6

Early schwa deletion as a prosodie phenomenon 155

6.1 Early schwa loss data 155

6.2 The shared properties of early schwa loss 158

6.3 The rhythmic phrasing of early schwa deletion 159

6.4 The Early Schwa Deletion Rule 162

Chapter 7

Schwa preservation in Late Middle English as a prosodie phenomenon 171

7.1 Final -e in weak adjectival inflexions 171

7.2 The weak adjectives in Chaucer 172

7.3 The weak adjectives in non-Chaucerian Late Middle English 173

Contents vii

7.4 Accounts of adjectival schwa preservation 175

7.5 The rhythmic nature of final -e 177

7.6 The metrical structure of adjectival phrases 178

7.7 The proposal reviewed: objections 186

7.8 Conclusion 187 Bibliography 193

Preface

Schwa has always been the p o o r relation of the stressed vowels in English historical phonology The academic searchlights have been directed on the developments of the vowels in stressed position M o s t studies in the field do m a k e some reference to the changes of unstressed vowels, yet the supposedly predictable genesis and unidirectional fate of the ubiquitous schwa-type vowels have attracted very little attention Seen in a wider context, however, the reduction of unstressed vowels to schwa and its subsequent loss in final position has had a m o r e p r o f o u n d effect on the phonology, prosody, and g r a m m a r of the language than any individual qualitative or quantitative change of the stressed vowels This b o o k is an attempt to bring together traditional philological records and knowledge

of the ways and means of schwa loss and reexamine them in relation to the entire language system

The first three chapters of the book deal with the d a t a base: how we know t h a t unstressed vowels in final position were reduced and lost, what

m e t h o d s and resources can be used in this context, how we classify and interpret the various types of textual evidence Separating out poetic f r o m non-poetic schwa loss, graphic f r o m inferred evidence, loss of non-morphemic, root-final schwa f r o m loss of inflexional schwa, proves useful

in establishing causes and consequences of the process on all levels

C h a p t e r 4 addresses rarely asked questions a b o u t the synchronic status

of schwa in Middle English A n examination of the evidence for phonemic identification of the unstressed vowel in final position suggests t h a t in terms of traditional phonemic analysis schwa has always been part of the phonemic inventory of English In Early Middle English there was a match between the underlying, or lexical, representation of schwa, and its surface realization Instances of early deletion are analysed as post-lexical rules operating first within a n a r r o w set of environments, and later across the board in word final position By the end of the Middle English period, after c 1400, schwa is disallowed as an underlying segment in absolutely final position except in a peripheral subset of markedly foreign borrowings Optional insertion rules operate in very specific lexical items and in some clearly defined prosodie contexts

C h a p t e r 5 covers the morphological correlates of the change In Middle English as many as eighteen different grammatical functions can be associated with the m o r p h e m e represented by <-e> in writing This

suggests that lack of morphological distinctiveness is a m a j o r factor in

schwa loss in those word classes, such as the nouns and the verbs, in which schwa figures prominently in the inflexional paradigm It is argued that the familiar generalized statements concerning the morphological attrition of schwa should be refined substantially by referring not to the lexicon as a whole, but to separate word classes The causes and rate of loss in each class turn out to be different The chapter offers a detailed hierarchization of individual factors and their potential importance in schwa loss for each word class

Chapters 6 and 7 introduce arguments from metrical phonology into the account of schwa loss Schwa loss affected the language continuously for at least three centuries; it is impossible to get a clear and systematic notion of what items were subject to deletion at the peak of the change, around 1250 — 1350 However, if we look at the very early instances of deletion, as well as at the instances of late preservation of schwa, a clearer pattern emerges Early schwa deletion can be defined in terms of a specific metrical foot structure — a definition which extends to later cases of deletion and eventually embraces all occurrences of word final loss At the other end of the chronological span, Late Middle English, we find schwa surviving consistently only in one isolated syntactic context, the weak declension of etymologically monosyllabic adjectives The motiva-tion for the preservation is neither functional nor segmental; it can be read off the metrical structure of the constructions involved and can be seen as an instance of the principle of eurhythmy Both chapters address schwa loss from a new perspective — the perspective of prosodie organiza-tion of language Both chapters demonstrate that the operations of schwa loss should be constrained within two domains: the domain of the foot and the domain of the clitic group

The customary task of acknowledgements and thanks is not easy in the case of a study which draws on the achievements of traditional philological scholarship and on modern linguistic theory Indebtedness is not quantifiable, and one paragraph will inevitably fall short of expressing

my gratitude to all who have helped bring the project to this stage Work

on this book was done in three universities: Edinburgh, Sofia, and U C L A The mixture of background can be my only excuse for the blend of philology and at least two kinds of linguistics I have used in my analysis

I acknowledge gratefully the generous assistance of the U C L A Senate Research Committee since 1985 for the final research and writing Two great philologists and teachers, Professors Marco Minkoff and Maria Rancova, encouraged and supported me in my early steps in the profes-sion To their memory this book is gratefully dedicated Among the many

Preface xi

colleagues to whom I feel indebted I want to mention specially Roger Lass, Heinz Giegerich, John Anderson, Edmund Gussmann, Bruce Hayes; they have shown me how exciting and complicated historical phonology can

be, though no study could do justice to their collective wisdom, nor should any of them be implicated in my blunders My closest professional model and associate Robert Stockwell has tried in vain to teach me how

to keep effusiveness and verbosity in check; to his salutary influence the readers owe the elimination of one third of the original bulk of the book

My sons Dimo and Lubo, the culprits solely and jointly responsible for all infelicities in the book, have shared me with schwa far too long — their only reward must be that no other teenager in the world would know what this weird word means Finally, for years I have described

my task to friends and colleagues as "everything you always wanted to know about schwa but were afraid to ask" What started out as a blind date with the topic developed into a relationship which has enriched me;

I hope I have enriched it More questions remain than have been asked

or answered; this is only a beginning

Chapter 1

Schwa in the history of English

1.1 Problems involved in the present study

The relative significance of various factors, phonological, morphological, tic, sociolinguistic, and stylistic in relation to the causes and consequences of the loss of final schwa in English has remained an unresolved and controversial issue

syntac-in the philological literature of the last 150 years This study is an attempt to clarify some recalcitrant aspects of the change and offer a hypothesis defining schwa loss in terms of the more general linguistic properties of English, diachron-ically and synchronically

Provisionally, before going into the phonetic and terminological details in Chapter Four, I will use the term schwa to mean, vaguely, the unstressed, reduced, mid, central, murmured, indeterminate, colorless vowel, the usual sound of the first and last vowel of America, for which I use the IPA mid central vowel sym-bol [s] Briefly, word-final schwa in English is generally assumed to have origi-nated from neutralization of the oppositions between a number of distinctive and stable sounds occurring prior to and during the time span between the ninth and the eleventh centuries Its subsequent history, i.e between the eleventh and the fourteenth centuries, is characterized by non-distinctiveness, instability, and pro-

gressive disappearance; by the second half of the fifteenth century the letter <-e>

was no longer pronounced, except in a handful of recognizably foreign words The spelling evidence for Old English (OE) suggests that the distance between the vowels /e/, /a/, /u/, and /o/ in unstressed syllables was fairly well preserved throughout the Classical period (excluding the /a/ and /o/ neutralization, and pos-sibly other short vowel neutralizations before nasals) 1 After 950, in Late Old English (LOE) and in Early Middle English (EME) this difference was no longer

maintained; the grapheme <-e> began to be used in all post-tonic syllables, no matter whether they belonged to basically monomorphem«: words, e.g OE guma 'man', ME gume; OE wundor 'wonder', ME wunder, OE lufu 'love', ME luve; OE stede 'place', ME stede; or whether they were syllables functioning as inflex- ional morphemes, e.g mac-ode 'made', ME makede, OE waer-on 'were', ME weren; OE hwal-as 'whales', ME whales; OE dag-um 'to the days', ME dagen; aris-αδ 'they rise', ME ariseth, OE scip-a 'of ships' ME shipe etc.^ The high vow-

els in some derivational affixes, especially the /i/, appear to have resisted the weakening; thus /i/ shows no evidence of reduction and typically survives in

suffixes such as -ίξ, -ic, -isc, -ing; while -ung alternates with -ong Morphemic

identification is probably responsible for the consistent preservation of the

unstressed vowel in words ending in -o(c)k: parrok, hassok, buttok, bullok ^

There is no written documentation of reduction in a limited set of lexical items, which behave indiosyncratically most likely for reasons of association with a

specific register (ME bishop, abbot) - no morphological justification can be

found for the apparent persistence of the original vowel in their final syllables

1.2 Forces at play in schwa loss

Standard textbook descriptions of final vowel reduction in Late Old English and

Early Middle English attribute the process, manifested by the adoption of <-e>

spelling in post-tonic syllables to the following factors:

(1) Phonetic changes in progress: phonetic reduction and its corollary - the

neutralization of phonemic distinctions - resulting from the concentration of stress on the root syllable, leading to a series of important grammatical changes, and above all to a radical decrease in the number and loss of inflexions

(2) Morphological changes in progress; neutralization of the original differences

between the inflexional morphemes of nouns, verbs, etc., renders these morphemes functionally deficient, and promotes the appearance of new means of expressing grammatical relations This leads to redundancy and eventual loss of the final unstressed vowel in major class words

(3) Phonetic and morphological changes in progress simultaneously; in this view

phonology and morphology are in a counterfeeding relationship: the absence of stress prominence renders inflectional morphemes non-distinctive phonologically, and therefore redundant functionally; loss due to the absence of function in some instances extends to cases where schwa is not a grammatical unit, contributing to the phonological spread

of the change Some variant of this compromise is the most frequently encountered description of the loss I believe this to be the only correct account, yet it needs additional defense, expansion, and refinement

(4) In addition to the system-internal factors in (1) - (3), the developments of the

post-tonic syllables in Middle English have been causally linked to the

Forces at play in schwa loss 3

extralinguistic factor of the existence of a large bilingual community in areas of the North and North East Midlands after the tenth century This study surveys the evidence for the changes in post-tonic syllables in Middle English In the light of this evidence I examine the plausibility of the views summarized in (1) - (4) There is little discussion in the philological literature of the phonological properties and behavior of /a/ in Middle English, its phonemic status, or its place and function within the prosodie system of the language The last two chapters will address the issue of schwa loss in terms of the prosodie patterns and preferences of Early and Late Middle English

Providing explanations of language change has been the concern of most major schools, branches, currents in both philology and linguistics I state briefly my indebtedness to the various hypotheses which may have direct bearing on the history of schwa The background assumptions and the descriptive and analytical techniques I have used are self-selected and derived from the following contributions to the historical study of language:

1.3.1 Neogrammarian

The notions of sound law and analogical change are crucial for the Neogrammarian model In that framework sound change is characterized by a regularity which elevates it to the status of a "law" Phonological change is also understood to proceed mechanically, affecting all items exhibiting a particular phonetic configuration; it is presumed to be gradual and impervious to direct observation The famous Neogrammarian identification of sound laws with universal natural laws has been questioned and rejected on the basis of the now obvious fact that the regularity of phonological change is never absolute Post Neogrammarian

studies have revealed intricate ties between sound change and morphology, syntax, and word formation - this now axiomatic contingency compromised the notion of

"mechanical" sound change The gradualness hypothesis, although it overlaps with our ideas of change through lexical diffusion, does not coincide with it, and leads to unacceptable notions of pervasiveness, and has been challenged on this and other scores The same is true of the idea of "unobservability" of sound change, which was rendered obsolete by the excellent sociolinguistic studies of the last couple of decades

Schwa loss is clearly a phonetic process From that point of view the principle

of regularity supports the assumptions concerning the predictable relationship between lack of stress as the conditioning factor, and weakening and deletion of the final vowel Under a magnifying glass, however, the detail of textual evidence

reveals an uneven rate of disappearance of -e in Middle English, which suggests a

morphological interference with the process of loss Even this initial basic knowledge would preclude any attempts to formulate the change in terms of a Neogrammarian sound law

Analogy is another tenet of the Neogrammarian school It is a principle which applies to the patterning of higher-than-the-phoneme units, i.e it is most frequently appealed to in morphology, syntax, word formation Analogy has the effect of counteracting phonological change The extent to which this principle can be called upon to explain the causes and mechanisms of language change presents a complex and controversial problem and there is a vast body of literature on analogy This model of accounting for language change has been used

by scholars since the sixteenth century The treatment of analogy in the Neogrammarian school allows it to be a strategy which "tends to become a terminological receptacle, devoid of explanatory power, a catchall for irregularities in the operations of 'regular' sound laws" (King 1969: 127), and therefore in more recent diachronic studies it has been overshadowed by newer proposals and only mentioned in conjunction with such proposals as a peripheral factor Many instances of paradigmatic leveling unrelated to analogy have been quoted in support of the idea that this explanatory model should either be abandoned or resorted to only in very special cases (King 1969: 129-30) The related idea of proportionality, however, has had many proponents Studies by Kuryfowicz (1947), Manczak (1980), Anttila (1977), etc have attempted to formulate and explicate the "laws" and tendencies of analogy and have succeeded

in eliminating some of the insidious vagueness discrediting this model

Analogy enters into the discussion of schwa deletion for two reasons First, schwa loss is a thoroughgoing morphological change, and morphology is one of the language domains where analogy is agreed to be most powerful and most

Remarks on method 5

effective Second, traditionally, sound change and analogy are seen as standing in a restorative relationship, i.e "grammatical structures rendered opaque, or under the threat of being rendered opaque, by sound change are likely to be 'repaired' by analogical change" (Bynon 1977: 44) Yet, analogy is not necessarily only an antidote used to reintroduce order into a paradigm into which sound change has inducted lack of uniformity At least partially, after the levelling of inflexions, the Middle English picture is one of sound change and analogy both working in the same direction - a situation unaccounted for by the standard Neogrammarian views of the interplay between sound change and analogy ^

1.3.2 Functionalism

Even if one accepts analogy as a valid and interesting model of language change, its explanatory power is confined to the spread of already existing forms; it obviously feeds on already existing forms and cannot create new forms of its own accord Functionalist accounts, on the other hand, have been concerned with the actual origin of innovative linguistic forms Scholars in this tradition, adopting the idea of language as a more or less rigid hierarchy of mutually dependent constituents, have contributed to the elucidation and understanding of the relations obtaining among the various linguistic units within the language system The limitations of this approach are by now widely recognized (cf Lass 1980a), yet it has been fruitful in stimulating further research and in prompting alternative hypotheses In the history of schwa in English a certain degree of

"functionality" has to be incorporated into the account of its phonemic development - useful and insightful notions such as that of phonemic contrasts and neutralization are functionalist, and they are central to the explication of the

synchronic relations of -e in Middle English

Diachronic functionalism, the recourse to explanations of language change by invoking the metaphor of survival of the fittest, of linguistic units as healthy, or unhealthy, as superior or inferior in performing their duty, is more suspect Language change does not proceed irrespective of, or in spite of, the needs of communication The familiar adage "communication alone shapes language" (Martinet 1960: 191) is an alternative way of saying that language is first and foremost a social phenomenon Considered in the narrow context of schwa loss,

though, extreme functionalism would imply that if at any point in its history -e

had been perceived as absolutely necessary, if its omission would have impaired the communicative value of the utterance, its ultimate loss without prophylaxis

or further therapy would have been inconceivable ^

Another ramification of the functionalist model concerns the reliance on a statistically computed functional load of a unit, which is then assumed to be in some way explanatory in terms of the unit's proneness to various diachronic processes In spite of the initial attractiveness of the idea of an empirically based explanatory strategy, its application encounters too many serious problems for it

to be adopted without further comments or reservations

Functional load, ultimately a teleological concept, is measured by Prague School phonologists by the number and frequency of the minimal oppositions into which a given phoneme can participate, a procedure supposed to predict the diachronic stability or instability of a phoneme This type of diagnostic runs into difficulties both with reference to the actual mechanics of computing functional load, and with reference to the interpretation of the results within a more general picture of the history of a language How does one select the correct amount of linguistic material providing the necessary and sufficient context for the evaluation of the functional load of one phoneme? Does the idea apply to mor-phology and syntax as well as to phonology? Is it phonemes, allophones, or features that one measures the functional load of? These and other questions have prompted extensive criticism of the whole notion, cf King (1967), Bynon (1977:

87-89), Lass (1980a: Chapter Three) I limit my "functionalist" discussion of -e

to an evaluation of its phonemic and morphological status within the entire system of the language, and I agree from the start with the cautionary note that

"it may be altogether inappropriate to look for a motivation of phonological change which is based on the structural status of phonological units" (Bynon 1977: 88)

Another ramification of the structuralist model is the view that "the phonological evolution of a language may be seen as a continuous effort to maintain a state of balance between inertia on the one hand, and communicative needs on the other" (Bynon 1977: 89) Economy is a general linguistic

"collective-psychological" factor which has been described as "basic in the analysis of the mechanism of language change" (Polivanov 1931: 54) It is common sense to assume that, especially in more relaxed and colloquial registers, speakers can take shortcuts, and their language would be subject to variability and change attributable to "inertia" or "human laziness" These notions can all too easily be confused with the now totally rejected idea of diachronic developments

as instances of "deterioration" In any case, evoking "economy", in the sense of

"shortcuts", and not as "deterioration", as anything more than a marginal phenomenon in historical linguistics is an extreme position which may be dangerously loaded in the context of a change which amounts to deletion, as is the

resulted in what has been described as a rapprochement between the generativist

and the traditional philological views, cf the discussion in Bynon (1977: 145) Accumulation of diachronic data in phonology has confirmed the assumption that "the motivation of phonological changes depends rather on performance-related targets than on purely formal considerations" (Bynon 1977: 138) The psychological reality and the analytical usefulness of the phoneme, initially regarded as uninteresting by the early generative phonologists, have regained legitimacy as minimal contrastive underlying units, albeit in the guise of

"underlying segments" The surface attestations of underlying forms can be derived with the help of rules which operate on those forms Stipulation of the conditions for the operation of the rules translates any difference between a postulated underlying form and its realization into a predictable, rule-governed change Stability and pervasiveness of a particular type of innovative surface realization may suggest that there has been a reanalysis of the underlying linguistic "system" Little has been done in studies of schwa loss from this point

of view; as a very preliminary outline of the problem one can say that during Middle English schwa loss was initially a phonologically and morphologically motivated rule The spread of the rule caused reanalysis of the underlying structure of the language By the end of the period a schwa whose sounding is guaranteed by e.g the metrical pattern of a poetic piece, must be accounted for in terms of insertion by rule, rather than as a deletion

1.3.4 Socially motivated change

Language is a social phenomenon; it cannot be understood or described adequately, let alone "explained", without reference to the social context in which it functions Sociolinguistic studies have contributed substantial insights into our comprehension of the mechanisms of language change, and of sound change in particular Martinet's extreme position (1960: 188-189) that "only internal causality concerns the linguist" has long been shown to be too limited; the complete autonomy of the so called "internal" and "external" factors has also been rejected, see Vachek (1962), Avrorin (1975: 25-28),** Berezin and Golovin (1979: 189), Bynon (1977: Chapter Five)

The extent to which the sociolinguistic parameters of a dead language can be established depends on the availability of texts representing various stylistic registers Some marginal information can be gleaned from knowledge of the regional origin of a particular text, as well as from familiarity with the personal history of the author Such information is not readily accessible for Middle English; in the absence of first hand comments and descriptions of change in progress by contemporaries, we have to be satisfied with very broad and possibly unilluminating observations on the "social" character of schwa loss One would assume, for instance, that in its incipient stage schwa loss could have a phenomenon correlating with speech unaffected by familiarity with the spelling system, "uncontaminated" by education Conversely, since the written form of the language never abandoned the symbol for schwa, one might speculate about the possibility of a period of time when pronouncing final schwas could have been a mark of a literate, highly conservative variety of English One line of inquiry which can prompt interesting observations about the rate of the change in relation

to the sociolinguistic background would be taking into account the register of the textual source, i.e one would expect alliterative poetry to be more conservative than accentual-syllabic poetry, narrative type texts to be more advanced than non- narrative texts However, the balance and interplay between "internal" and

"external" forces in language change in the case of schwa loss can be sought more profitably in the introduction of another parameter - outside linguistic influence

Remarks on method 9

former group.9 Evidence that linguistic developments can be generated by the contact between languages is most readily available in the sphere of vocabulary, but there is also abundant literature on the manner in which continuous exposure

to and simultaneous use of two languages can influence a speaker's phonology, morphology, syntax In the conditions of prolonged bilingualism on the British Isles, both Anglo-Scandinavian, and Anglo-Norman, one can expect some evidence

of the varying rates of schwa loss depending on the linguistic situation of the area from which a given text originates A direct influence from either Scandinavian or

Norman French on the loss of -e in phonological terms cannot be postulated, since there is no parallel development in Scandinavian, and as for the French -e muet, it

began to disappear from the end of words under certain conditions much later than

it did in English The morphological history of {-e} is universally assumed to have

been influenced, at least to some extent, by the contact between the different languages spoken on the British Isles More specifically, there is a long tradition

of associating the disintegration of the inflexional system of English has to Anglo-Scandinavian bilingualism (Jespersen 1938 [1962]: §79; Ilyish 1973: §275 etc.) According to this widely accepted view absolute accuracy of the grammatical form is not essential in the process of communication because of the genetic relationship between the language of the settlers and the language used by the native population The phonetic identity or similarity of the roots of words is taken as a catalyst, and in some studies, as a sufficient condition for basic understanding to take place between the speakers of English and the Scandinavian settlers More recently, detailed research of creolization processes has provided a new framework within which the contact between Scandinavian and English can

be described Middle English has been analyzed in terms of a system which cannot

be derived directly from its Anglo-Saxon predecessor (Fisiak 1977: 249) This new system, the Middle English interlanguage or creole (the two terms are used synonymously here), emerged under the decisive influence of Scandinavian Poussa (1982: 83) summarizes the first stages in the development of (Chancery) English

as follows:

866 Old Mercian Old Danish

918 East Midland Creole

1016 Hybridization with Late West Saxon and Mercian

1016 Stabilization as spoken koiné in London and the Midlands

1066 Further Danish influence, hybridization with London dialect(s)

1066 French influence (mainly written, apart from lexis)

14th c English patriotic revival Immigration to London from the

Midlands

The earlier and familiar Jespersenian view, and the more recent, creolization account of the way in which contact with other languages affected the history of English are fully compatible They both cover the attested dialectal distribution

of morphological changes in Middle English; the creolization hypothesis,

however, emphasizes processes which take place in the spoken form (Poussa 1982:

70), undoubtedly an important and neglected area of historical analysis Stylistic considerations are also taken into account in the study of creóles Creoles are generally characterized by loss of grammatical gender and simplification and loss

of the grammatical inflexions; assuming the existence of a Middle English creole

can help us sharpen our understanding of the very different behavior of -e in

English and in the other Germanic languages given identical phonological conditions of reduced stress

1.4 Analytical procedures

The brief introductory notes in 2.1 on the major ways in which the origins and nature of language change have been approached in the literature, as well as their applicability to the Middle English material must be supplemented by an outline

of concrete procedures of analysis Once again my point of departure is fairly familiar Important methodological principles for the study of language change were laid down in the classical study by Weinreich, Labov and Herzog (1968: 183-188) I reproduce their Empirical Principles for the Theory of Language Change below:

(1) Determine the set of possible changes and possible conditions for change (The

Constraints Problem )

(2) Trace the ways in which the changing feature is embedded in the linguistics

structure (The Embedding Problem [a]).^

(3) Consider all factors, "stimuli and constraints both from society and from the

structure of language" which influence change in relation to one another

(The Actuation Problem) Heterogeneity underlies change An overall

view of the process of change is best achieved through a multiplicity of approaches (186).^

This cursory survey of some basic background assumptions suggests that one cannot expect a simple and unequivocal answer to the question of why a certain change has occurred in language The only purely internal factor about whith we

Analytical procedures 11

can be positive with respect to schwa is analogy; analogy, as noted, however, provides a satisfactory explanation only of the spread of already existing new forms, i.e the source of the proportional model for the analogy should be available before it can start affecting more and more forms in the l a n g u a g e

External factors, sociolinguistic, or contact-induced processes, must be brought into the picture; they can provide significant additional information in the account

of schwa loss, especially in reference to the surface morphological composition of previously inflected grammatical classes: nouns, adjectives, and verbs

The discussion of theories of the origin and mechanisms of language change presented here has not been concerned with the "actuation" of change, in terms of transmission patterns as described by scholars studying language acquisition Change can be attributed to the way children and adults acquire and modify language This particular aspect of the "transition" of change is unrecoverable in the written records; in the absence of any relevant historical information attempts at making reasonable use of the notions of "actuation" and

"transmission" in an acquisitionist sense is futile

Emphasis in this study will be laid on "change", which brings to the fore the dynamic aspect of language "Change" is a concept which comprises two separate and discrete stages in the development of any diachronic process: the stage of innovations, and the stage of actual change, the latter involving "the generalization, or codification of innovations" (Andersen 1974: 22)

At the same time an analysis which aims to uncover the network of functional relations within the limits of one historical period must necessarily assume the existence of an idealized, relatively homogeneous, synchronic bulk of language Each Middle English text examined will be taken as representing one such synchronic state Synchronic states can then be compared to each other to reveal the dynamism, instability, variability, potential or actualized "change" in a diachronic perspective This theoretical position, ideally achieving a perfect compromise between the synchronic and diachronic approaches, has gained universal acceptance and requires no special justification

Notes

Chapter 1 Schwa in the history of English

1 Angled brackets < > are used throughout the text for graphemes; squares [ ]

enclose allophones; slashes / / enclose phonemes Italicized -e is used as a

cover notation for any final -e, prior to defining it as a grapheme, a phoneme,

an allomorph, or a morpheme Curly brackets { } are either morphemic or enclose members of a set

2 The length of the stressed vowel in the forms quoted in the text has been left unmarked if it is irrelevant to the discussion

3 See Jordan (1974: § 133-135), also Luick (1921-1940: § 440)

4 This type of mutual enhancement of sound change and analogical change is discussed in Zirmunski (1954: 38 ff.): "It is not always [the case] that

phonetic changes disorganize the grammatical system " Here Zirmunski

refers specifically to the regular phonetic reduction of final unstressed syllables as a very significant factor in the development of the inflection in the Germanic languages In this case the phonetic process eliminates the discrepancies within the old grammatical system and thus interacts with morphological analogy to create a new, more unified, and consistent system Zirmunski believes also that analogical tendencies "contribute to the improvement and perfection of the grammatical rules of a given language in accordance with special internal laws of its development" (1954: 28) I take

"improvement" and "perfection" to imply that a morphologically uniform paradigm is somehow superior to a formally heterogeneous one, an interpretation which renders this statement unacceptable Of some interest is another point which Zirmunski makes in this connection,however, namely that internal, i.e inflectional analogy restores the unity within the paradigm

of a word, separating more rigorously the root morpheme as the bearer of the objective (material) meaning of the word (1954: 31) This is an important typological observation about word structure in the Germanic languages and

in English in particular; morphological analogy related to the weakening to

and loss of {-e} results in realignment and concentration of the semantic and

grammatical information on the root

5 The controversy over the legitimacy of functionalist arguments and explanations continues (see Lass 1987a, 1987b; Samuels 1987)

6 The type of "economy" discussed here should be differentiated from the unfortunately synonymous, but highly desirable economy of linguistic description within the generative framework The reference here is to functionalist economy, an idea which most linguists have dismissed as untenable Yet as recently as 1981 a whole book (Krustev 1981) was devoted

to the absurd task of "explaining" a series of changes, phonological, morphological, syntactic, derivational and even graphic, from the principle of economy This extreme and oversimplified view cannot be found in the works

Notes 13

of A Martinet, one of the leading proponents of the theory of economy as a trigger of phonetic change Krustev himself is apparently aware of the unacceptability of such an extreme position (1981: 7) yet he proceeds with the analysis of selected data which is supposed to convince the reader of how strikingly "economical" the Bulgarian language is Similar examples in other languages would be easy to quote; this would defeat the idea of the

"exceptionality" of the Bulgarian language with respect to economy An equally impressive body of "uneconomical" changes in any one language renders the effort futile

A more acceptable position with respect to economy has been defined in Budagov (1974: 36-37), who lists eight points which either refute the notion altogether, or caution the linguist against mixing up "economy" with

"linguistic differentiation", "economy" with "regularity", "abstract linguistic economy" with "economy in the individual speech act" etc Also, Budagov's discussion points out an apparent contradiction between the principle of economy and the tendency in language to maintain a dynamic equilibrium between form and meaning (optimally one form - one meaning), and between simplicity and complexity

7 Cf King (1969: 62-64, 84): "Linguistic change is change in competence, not change in performance·, it is change in the grammar, not originally change in

the output of that grammar" (1969: 84, italics King's)

8 Avrorin characterizes the dependence between the structure of language and its functional aspect in the following way: (1) Although this dependence is unquestionable, it is neither absolute nor automatic; (2) The dependence is not total, and (3) It is unidirectional, irreversible, i.e language function is not determined by language structure, while the opposite is true (1975: 26-27)

9 The type of contact-induced account introduced here is a subset of the more general category of "external" explanation which "seeks motivation in terms of something that exists outside of and independently from the grammar" (Kenstowicz 1981: 431)

On the definition of "functional" in relation to "external" see Avrorin (1975: 23-33) In his classification the structural development of language is called forth and regulated by two kinds of stimuli: (a) internal laws which exist by virtue of the necessity of regulation of the linguistic structure and b) various extralinguistic, more precisely extra-structural factors of social nature The first type influence the structure of language directly and are localized exclusively in it; the influence of the second type of factors is

exerted to a greater or lesser extent through the mediation of the functional aspect of language

10 "It is argued that the fundamental changes which took place between standard literary Old English and Chancery Standard English: loss of grammatical gender, extreme simplification of inflexions and borrowing of form-words and common lexical words, may be ascribed to a creolization with Old Scandinavian during the Old English period The Midland creole dialect could

have stabilized as a spoken lingua franca in the reign of Knut Its

non-appearance in literature was due initially to the prestige of the OE literary standard" (Poussa 1982: 84)

11 The nature of the linguistic material analyzed in this study precludes, or at

least limits severely observation of ongoing change, the study in vivo;

therefore I have not included several otherwise (synchronically) important lines of investigation, namely:

(1) Trace the distribution of the change through successive age levels of the

population (The Transition Problem)

(2) Determine the degree of social correlation in the process of embedding of

the change in the social structure (The Embedding Problem [b])

(3) Determine directly the level of social awareness as a property of

linguistic change (The Evaluation Problem)

12 Another point underlined in Weinreich, Labov and Herzog's general principles for the study of language change is that: "Linguistic and social factors are closely interrelated in the development of language change Explanations which are confined to one or the other aspect, no matter how well constructed, will fail to account for the rich body of regularities that can be observed in empirical studies of language behavior" (Weinreich, Labov and Herzog 1968: 188)

13 For a discussion of this issue see Hogg (1979: 58)

in comments concerning the loss of schwa in English What follows below is a survey, and commentary, on some popular and influential statements which pro-vide the input to my own findings and interpretations of this specific issue 1 The choice of representative sources is based on the following criteria:

(1) Author's interest in the chronology and causation of the developments in

unstressed syllables

(2) Originality in the presentation of the textual data and in the interpretation of

the motivation for the change

(3) Impact of the book on subsequent scholarship in the field; an inevitably

subjective perspective on the author's scholarly stature and the prestige and popularity of his work

It should be noted that a number of original and influential books on the history of English do not concern themselves specially or at all with the developments in the unstressed syllables This makes (1) an automatic and self-evident first criterion (2) is prompted by the fact that many of even the best historians of English repeat previous authors, sometimes expanding the textual references, yet without suggesting a different interpretation The danger of subjectivity in (3) is not great, because it is contingent upon (1) and (2), on the availability and originality of opinions on the history of schwa

2.2 Lorenz Morsbach (1896): Mittelenglische Grammatik

Lorenz Morsbach's Mittelenglische Grammatik is one of the first influential

Middle English grammars, concentrating solely on the phonetic changes between the twelfth and the fifteenth centuries Morsbach proposes a more precise subperiodization of Middle English into:

(1) Early Middle English - c 1100 to c 1250

(2) Central Middle English - c 1250 to c 1400

(3) Late Middle English - c 1400 to 1500

These chronological labels are only approximate, yet they can be linked to clusters of linguistic changes, the availability of texts, the development of the literary tradition, and to real extralinguistic events They provide a useful frame

of reference Most of the developments discussed in this study occurred during the first two periods, Early and Central Middle English, but the first half of the fifteenth century still shows some survivals of schwa in final position, so reference will also be made to Late Middle English

Mörsbach takes the existence of [-a] as an uncontroversial fact about the vocalic system of Middle English He describes it as a "weakly sounding,

unclearly spoken -e vowel" (1896: VIII), but in spite of its assumed similarity

with [e] he is consistent in using the special phonetic symbol [a] for it

The chronology of the changes affecting the [-a] is presented as follows: (1) c.1100 toe 1250

The changes of unstressed -e are listed by Morsbach under the "basic features" [Hauptmerkmale] of the period (1896: 4) During that period unstressed -e is

"still preserved in many cases"; this latter observation represents the characteristically noncommittal tone of the accounts familiar from the textbooks As early as 1896, and to the present, schwa loss has been an area of

research where qualifications of the type often, mostly, frequently, etc mark the

limits of precision with which the process is described Such vagueness is inherent

in the nature of the material, and I will not be able to avoid it any more than ous scholars have done, yet the level of precision can be improved by looking closer into the patterns exhibited by individual morphological classes On a more positive note, Mörsbach clearly underlines that by 1100, and certainly long before

previ-1250, the final [-a] in a number of words was no longer pronounced; and sometimes it was not even written

Lorenz Morsbach (1896): Mittelenglische Grammatik 17

(2) c 1250 to c 1400

This is the period during which the process of schwa loss reached its climax From

"occasionally disappearing" [-a] became a sound which was only "occasionally preserved" in the South; in the North it was "almost completely" silent In spite

of the ambivalence of this early account, there emerges a picture of rapid and pervasive loss; any survivals of schwa after this period of 150 years can be regarded as aberrant and have to be analyzed separately

In the section covering c 1250 to c 1400 Morsbach also remarks on the frequent occurrence of cliticization attested in the spelling of very early texts

(1896: §48), especially with pronouns, prepositions, adverbials, etc., thus: delder

= de elder, 'the elder', nart = ne art 'are not', tane = to âne, 'to one', kiddit = kidde it 'made it known', caldis = calde es 'is called' The recognition of this

phenomenon is an important initial step in establishing the significance of such forms for the reconstruction of the spoken language Morsbach recorded the evidence, but offered no comments on the implication of such spellings for the interpretation of the prosodie organization of the speech of the period

The criteria on which Morsbach relies for classifying the reduction of vowels

to [-a] and their loss in final position are syllable stress within the word and phrasal stress within the sentence: (1) Weakly stressed position within the word (1896: §67-§81), and (2) Weakly stressed position in the sentence (1896: §81-

§85)

Group (1) includes cases of reduction and shortening in derivational suffixes

(e.g -hod 'hood', -döm '-dom', -les '-less', -lich(e) '-ly' ,-ere 'er', -ig '-y' etc.), in

which some reduction is assumed on the basis of later changes, but for which there

is no indication of the reduction in the spelling One would assume that the consistent spelling of these suffixes is related to their independent status in the morphological inventory of the language; also, it may suggest the preservation of

a certain amount of secondary stress, or at least the potential for it Morsbach does not comment on the morphological autonomy and stressability of these items; he only enumerates them and remarks on the possibility of short and long vowels coexisting in such suffixes until Late Middle English The assumption of different levels of stress in derivational suffixes and the second elements of compounds on the one hand, and in inflexions, on the other, relates stress to the accompanying changes of vowel quality and quantity, and Morsbach's observation, though left undeveloped, is undoubtedly germane to the issues raised in this study Chapter Four returns to the discussion of this problem

Vowels in fully unstressed syllables make up the remaining part of Morsbach's group (1) Such vowels were weakened already in Old English and

"not infrequently disappear completely" (1896: §68) Morsbach seeks the

"natural" explanation of this process in the accentual organization of the language, resulting in concentration of the main expiratory force on the first syllable of non-compound words, which is usually the root syllable The less energetic pronunciation of unstressed syllables makes them susceptible to reduction and loss This is seen as a gradual process which can be inhibited by various tendencies, lying in human nature [Psyche], in the construction of the speech organs [im Organismus der Sprachwerkzeuge], as well as in the mechanism

of language transmission [Überlieferung], True to the Neogrammarian model, Morsbach paid homage to the importance of analogical factors in bringing about the changes in unstressed syllables, but he jeopardized the acceptability of this particular argument by admitting that "the power of analogy can either preserve

or destroy" (1896: §68).^ With respect to the issue of regularity of sound change Morsbach was more extreme than most scholars in that school, more Neogrammarian than Paul, whose view of the nature of sound laws is more moderate Though compelled by the nature of the evidence to accept variation and

"irregularities", Morsbach was clearly trying to present a uniform picture of the changes.^

Morsbach treats separately instances of schwa loss in final syllables

following syllables bearing secondary stress (1896: §72), thus ME laffdig < OE hlafdige 'lady', ME allmess < OE aelmesse 'alms' etc.), and concludes that in such trisyllabic words the apocope of final -e could depend on the quantity of the root syllable: -e disappeared after a long root syllable holi 'holy', hardi 'hardy', but manie 'many', bisie 'busy', i.e a short root syllable does not favour the loss The

reference here is specifically to Kentish, but a similar dependence is observed also

in the South West (Robert of Gloucester).^

In the section describing the ultimate loss of -e in final position (1896: §75), Morsbach notes that both Old English final -e and the "analogical" -e (whose

existence in itself is an indication of loss, but Mörsbach seems to ignore this obvious inference), began to disappear in Early Middle English Morsbach's

reference to an "analogical" -e is potentially confusing: it can be interpreted as

both the unetymological insertion of schwa, and as the replacement of other vowels by schwa The more likely inference from his discussion in general is that that what he meant was the "analogical" schwa was the one resulting from neutralization He believes that such "analogical" influences were behind the change initially, and later the development was carried through under the impetus/pressure [Wucht] of phonological factors, more specifically strong stress

on the root syllable The uneven geographical spread of the change is recognized:

Lorenz Morsbach (1896): Mittelenglische Grammatik 19

the earliest instances of loss are recorded in Scotland and the North Midlands, whereas in Kent, and in the South in general, the process took much longer

It should be noted that in a later work Morsbach's position on the impoverishment of the inflexional system in Middle English adumbrated later

functionalist and creolist oriented treatments In his 1913 Grammatisches und psychologisches Geschlecht im Englischen he wrote: "Den ersten Anstoss zu

dieser Bewegung gab die Erweiterung der ursprünglich engeren Kasus-b e d e u t u

η g e η und infolgedessen die Vermischung und der teilweise Zusammenfall der Kasus (Synkretismus)" [The first stimulus for this development came from the broadening of the originally narrow case meanings and, as a result of that, the mixing and the partial collapse of the cases (syncretism)] In other words, in his later work Morsbach believed that some functionally defined cause antedates the phonological and analogical changes of Middle English (Morsbach 1913: 7)ß Two points emerge from this earliest comprehensive survey of the issue: First, the difficulty of establishing the chronology and localizing the change, ensuing from "the lax metrical structure of a lot of poets, the influence of foreign patterns, poetic conservatism and some others" (Morsbach 1896: §75, Anm.) Second, in spite of some obvious and persistent problems with the evidence, definite statements about schwa loss have been attempted Thus, in the South:

"final -e is generally preserved intact in Kent until the middle of the 14th c." (Morsbach 1896: §80) These conclusions are mainly based on evidence from The Ayenbite of Inwyt (1340), a prose text in which the preservation of final -e may be

scribal and traditional, another issue which requires elaboration

There is no agreement among scholars on the significance of the presence or

absence of scribal <-e>: we are warned not to assume that absence of final -e in the

dative case, the weak declension of adjectives, the past tense of weak verbs etc., is

indicative of loss of -e, presumably meaning general schwa deletion in

non-morphological contexts (Luick 1921-1940: §473, Anm 2; Jordan 1968 [1934]:

§141 Anm 2, etc.) On the other hand the grammatical significance of this morpheme is admittedly allowed to be totally suppressed or greatly obscured in

poetry To maintain that presence of -e in some morphological functions signifies preservation of schwa, while absence of -e from inflexions does not signify loss,

without any further attempt to qualify or justify these statements, is confusing These questions will be addressed in the following chapters, after I have reviewed the textual evidence in terms of chronology and type

The picture of spread of schwa loss one can extract from Morsbach is the following: Between 1200-1300 loss affects the North; from 1250-1350 the change spreads into the Midlands, and between 1300-1400 it affects the language

of the South This is a rough summary: as pointed out earlier, Morsbach describes

the final -e with vague approximations such as "mostly preserved", "largely

lost", etc In that form his observations are very much in keeping with the results

of later studies of this development

Finally, and in a separate section, Morsbach treats the changes of the unstressed vowels in words which are likely to appear under reduced sentence

stress Old English long vowels tend to get shortened in this position (OE butan 'but', OE us 'us', OE tzning 'any', etc.) There is also an interesting commentary on

the cases of elision, which Morsbach attempts to interpret in phonetic terms (1896: §84) He describes elision as a transition of the type of movement the vocal cords make in producing a vowel from "glottal catch" in OE to "clear glottid"

or, perhaps, "gradual glottid" in Middle English This account is within a frame

of reference no longer used by phoneticians today, yet it is of interest as a precursor of more recent attempts to define the original conditions, the domain, and the effect of wide spread elision, as well as its articulatory nature.^

23 Karl Luick (1921-1940): Historische Grammatik der englischen Sprache

The Historische Grammatik by the Austrian scholar Karl Luick, of which only the

phonology was completed, has been enormously influential in forming the views

of subsequent generations of philologists and historical linguists In this monumental work, as well as in his many individual studies of texts and specific phonological processes in the history of English, Luick departed from the atomistic tradition of his predecessors and focused his attention on the interrelation and overall implications for the system of individual phonetic changes.and dialectal variants Luick was concerned with the directionality and causality of sound change, and with the organization of the individual phonetic units into larger units - syllables and words His 1898 treatment of the quantitative changes in the history of English is a fine example of an almost precocious view on the optimality of syllable structure as the trigger of quantity adjustments Luick was meticulously detailed in his exposition: the wealth of information based on textual evidence in his historical phonological writings has remained unsurpassed to this day

Nearly 60 pages of the Historische Grammatik are dedicated to vowel changes

in unstressed position, the most ever written on this subject in one book The developments he covers fall within the period from the eleventh to the fifteenth century Vowels in pre- and post-tonic syllables are treated separately, in great detail and are accompanied by numerous textual references Separate sections

KarlLuick (1921-1940): Historische Grammatik der englischen Sprache 21

discuss the changes of the unstressed syllables in native and Scandinavian words

on the one hand, and in loan words of French and Latin origin, on the other.^ Luick dates the initial reduction of unstressed vowels uncontroversially: it started in Old English and preceded Middle English, because the spelling with <-

e> was firmly established in all Middle English texts proper (1921-1940: §440) Unlike Morsbach, he did not attempt to distinguish the -e resulting from the

reduction of various vowels from the short vowel phoneme /e/ in stressed position (The phonemic slashes are not Luick's, but his overall treatment of phonetic entities would justify their use in reference to his work) The reduced vowel became by far the most frequent and the most typical vowel in unstressed syllables in Middle English

Luick's section on the development of vowels in syllables outside the main accent opens with a statement of two tendencies shared by stressed and unstressed vowels One is the tendency towards simplification, the other is the multiplicity and diversity of changes of both types of vowels, attributable to the influx of numerous loan words in the language (1921-1940: §439) Luick stays within the limits of phonetics/phonology proper; there is no discussion of the morphological motivation or significance of the reduction, which would imply that he regarded absence of stress as the primary impulse for the change The scenario which unfolds following the basic assumptions is familiar: absence of stress causes reduction, reduction is the first step towards complete loss Luick supplies a very thorough documentation of the environment in which loss was most frequent, he covers in detail the chronology and the geographical distribution of the change Hiatus, frequently accompanied by orthographic evidence of cliticization, is

assumed to have been the first environment in which loss of a weak final -e set in

Schwa loss in hiatus in Middle English was a continuation of an Old English

process Luick quotes MSS of the 10th c which exhibit forms such as wen ic Ί hope', scegdic Ί said' for wene ic, sœgde ic (1921-1940: §452) The observation is

not new; what is interesting here is the original comment Luick offers in the discussion of the change in this environment He writes: "Later (i.e after Old

English) elision became more frequent and more general as a result of the eration of the speech tempo; [italics DM's] even before Orm's time each post- tonic -e was lost in the spoken language before a following vowel or the h - of

accel-weakly stressed words" (1921-1940: §452) This change was not generally reflected in the spelling; as usual spelling remained conservative

This is a very early attempt to establish a relationship between the written forms and their prosodie significance in the literature on schwa in the history of English Elision, in Luick's opinion, was so regular, "that it must have had a basis

in the spoken language" (1921-1940: §452 Anm.) This quite intriguing

observation has not been followed up in the literature I believe that it deserves more attention; some of the discussion in Chapters Six and Seven will focus on the possible importance of schwa loss evidence for our understanding of the prosodie organization of speech

Another environment in which loss of final unstressed -e occurs very early is

weak position of the whole word within the limits of the sentence This development affects auxiliaries, conjunctions, various non-major class words, and

is fully attested by the twelfth century (1921-1940: §454 1.) Some examples of

the change are: when 'when' for OE hwanne/hwenne, als 'also, as' for OE ealswa, but/bot 'but' for OE butan, wer 'were' OE wœron etc

Trisyllabic words are also likely to undergo the change at a very early stage,

thus: lafdi 'lady', OE hlcrfdige, almes(s) 'alms', OE celmesse, orrest 'struggle', OE orresta, etc Referring to this particular instantiation of early schwa loss Luick

brings up the issue of causation and once again suggests "acceleration of the speech tempo", especially in colloquial speech, as the reason for the suppression of secondary stress, in the case of derivatives, which leads to a loss of the final syllable.**

Luick proposed a relatively early dating for the change in loan words from French and Latin, namely during the thirteenth and the beginning of the fourteenth century, contingent upon reduction of the secondary stress in words

such as mater e 'matter' Since -e was the only post-tonic syllabic vowel in Old French, it was replaced by the native -e and shared its fate, thus sir 'sir' for sire, dam 'dame' for dame,frer 'friar' for frere etc Unlike other authors who place the loss of French final -e earlier of later than the corresponding loss in the native

component, Luick dismisses the possibility of separate treatment of the loan vocabulary in the spoken language on the basis of evidence from versed The

appearance of longer preservation of unstressed final -e in French words he

attributes to either a peculiarity of slow and careful speech, with secondary stress

still realized, as in words of the type cumpanie 'company', or as possibly

misleading scribal testimony, due to adherence to the French spelling models (Luick 1921-1940: §461)

Schwa loss in native words in general is traced under the heading "complete

loss" [gänzlicher Schwund] of final -e in a separate detailed section (§ 473) Luick

defines "complete loss" as the development which deletes the schwa from the

unstressed syllable of disyllabic words, as in fare, helpe, love, tonge, etc., i.e the

lexical items that do not normally fall into the groups outlined above His description is very definite both with respect to the chronology and the geographical distribution of the process: "The loss is completed in the North by the time of the appearance of our first texts, therefore it was carried through

Karl Liúck (1921-1940): Historische Grammatik der englischen Sprache 23

during the thirteenth century there" (Luick 1921-1940: §473), a statement qualified by the expected comment on the slow rate at which spelling catches up

with the changes in the spoken language The graphic history of the final -e in

Luick's outline includes also the emergence of the familiar practice of

preservation or unetymological addition of the letter <-e> to the end of words containing a long vowel + a simple consonant, thus: fare 'fare', wife 'wife',

wine 'wine', föte ' foot'

In summary, Luick's account of the changes in final unstressed syllables emphasizes the specific contexts for early disappearance of schwa: hiatus, lack of prosodie prominence within the phrase or the sentence, and trisyllabic structure The explanation proposed for this process is purely phonetic The intriguing reference to "acceleration of the speech tempo", repeatedly mentioned as the factor causing reduction and loss in and of unstressed syllables, is not elaborated further or substantiated in any way Taken at its face value the assertion has a rather bizarre implication — the argument can be reduced to the statement that at some point in the history of the language its bearers must have been seized with some inexplicable urgency to speak faster Coming from a scholar of Luick's stature and erudition, though, the observation cannot be ignored; indeed as I hope

to be able to show later, some of the apparent absurdity of Luick's early intuition

is dispelled is we place the discussed phenomena in the appropriate prosodie perspective

2.4 Richard Jordan: (1934 [1968]) Handbuch der mittelenglischen Grammatik

Richard Jordan's treatment of the processes in unstressed syllables is quite detailed.'® Jordan differs from previous scholars in the way in which he traces the

chronology of the loss of final -e in relation to the conditioning environments In

his view it is the syllable quantity that determines the course of the loss in

trisyllabic words: final -e disappeared as early as the twelfth century after a long first syllable, thus laffdig 'lady', OE hlkfdige, allmess 'alms', OE œlmesse, etc

As regards the complete disappearance of final -e in all environments, Jordan

proposed a somewhat earlier date than other scholars, especially for the Northern

dialects, i.e all final -e's are lost in the North already in the thirteenth century

(1934 [1968] : §138)

Jordan underlined the morphological implications of the change At the end of the section covering loss in final position he writes: "Not to be confused with

phonetic loss of final e, but functional in nature is the early Middle English loss

of e in dat sg as a result of fusion with acc into an objective case" (Crook 1974:

§141) No further comment is offered on the possible interrelation between the two developments

2.5 Wilhelm Horn/Martin Lehnert (1954): Laut und Leben Englische

Lautgeschichte der neueren Zeit (1400-1950)

The period covered by Horn and Lehnert's famous book only marginally overlaps

with the assumed chronological span of schwa loss By 1400 final unstressed -e is

generally believed to have been lost in speech in most environments Questions concerning the chronology, dialect distribution, the relevance of syllabic structure, etymological origin etc in relation to this development are not discussed in Horn and Lehnert's work In one sense, however, their treatment is radically different from all previous accounts: they provide the clearest function-

alist statement concerning the factors leading to loss of final -e in English Horn and Lehnert formulated a hypothesis, according to which the loss of -e should be

attributed first and foremost to its functional insufficiency [Funkrionsarmut] The account refers to the fact that in Middle English the expression of syntactic relations was increasingly taken over by non-morphological means This

independently motivated change rendered the final -e redundant In support of this

hypothesis Horn and Lehnert remind us of the familiar examples of Old English

loss of -e in prepositional phrases, e.g to dœg 'today', instead of to dœge, cet ham 'at home', instead of œt hame (1954: §306) The argument is not pursued any

further; later in the book the authors remark that a particularly heavy stress on the root syllable is a condition for the loss of the vowel in the following syllable In this respect the account repeats existing accounts which rely on reduction under weak stress.^

2.6 Karl Brunner (1962): Die englische Sprache, ihre geschichtliche

Ent-wicklung Vols I, II

Karl Brunner's account of schwa loss differs from that proposed in previous studies in the following points:

(1) The beginning of the change is dated somewhat earlier than in most other

historical surveys Brunner refers to schwa loss as a change taking place

"gradually, from the twelfth century onwards" (1962, 1: 348)

(2) Brunner draws attention to the metrical evidence for loss of -e in the earliest

poetic works of the period, and notices rhymes of words in which the

Karl Brunner (1962): Die englische Sprache, Vols I, I 25 final -e appears to be retained with words in which a final -e is not justified etymologically In rhymed verse written in the North the -e is

assumed to have been deleted already around 1300 (1962, 1: 349) Brunner's attempt to build into his account evidence from rhymes is quite unusual

(3) According to Brunner's account the Late Old English syncretism between the

dative and the accusative case is a development preceding the loss of -e in

time, but not conditioned by it; the early date of the elimination of the morphological differences between the two cases is not regarded as being

in a causal relationship with the subsequent phonological loss of final -e

2.7 Joseph Wright and E M Wright (1923): An Elementary Middle English Grammar

This very influential textbook covers the development of the vowels in unaccented syllables in English in considerable detail Special attention is given to both the morphological significance, and the phonological aspect of the change

Concerning the chronology of the loss of final -e the authors state that " it is

only possible to fix approximate dates for its loss" (1923: §139) Judgments about the date of the loss are based primarily upon evidence from accentual-syllabic verse, but Wright and Wright add a caveat against exaggerating the importance attached to metrical evidence They approach such evidence with caution on the grounds that poetry in general allows conservative language patterns, and thus some features of the "poetic" language may lag behind the developments in the normal spoken language The dating of the loss as suggested

by the Wrights is as follows:

(1) In the Scottish and Northern dialects it had ceased to be pronounced in all

forms by about the middle of the thirteenth century

(2) In the dialects of the Midlands final schwa had ceased to be pronounced in all

forms by about the middle of the fourteenth century, but the loss of final

-e in pronunciation began in some parts of this large area at a much earlier

date (1923: §141)

Wright and Wright take an interesting position on the interrelation between the

quantity of the stressed syllable and the loss of -e They assume final -e to have

"disappeared or rather ceased to be pronounced earlier in disyllabic forms with a

short stem syllable than in those with a long stem syllable" (1923: § 1 4 1 ) N o

concrete textual evidence is quoted to support this statement

Wright and Wright comment also on the way in which final -e was lost across

the vocabulary "In all dialects it disappeared in pronunciation earlier in nouns and verbs than in adjectives, and earlier in the strong than in the weak declension

of adjectives" (1923: §141) This interesting observation is made almost in passing, and the possible reasons for this particular sequence of events are not discussed The context of the account remains strictly segmental phonological Like Jordan, the Wrights believe a long stem syllable to have accelerated the loss

in trisyllabic words The discrepancy between the assumed effect of the quantity

of the stem syllable on di- and trisyllabic words in their account has not been cussed, or even noticed, in the literature If the observation could be substantiated,

dis-it would present a challenging phonological problem

Finally, a feature original with their textbook, Wright and Wright propose

different rates of attrition for the final -e in French and in native words, compare

to the comments on that issue made by Luick discussed in section 2.3 They suggest

that in French words, especially after st, ce [-s] and after vowels, e.g best(e),

tem-pest(e), plac(e),foli(e) etc.), the final -e disappeared in pronunciation earlier than

in words of English origin (1923: §230)

2.8 Fernand Mossé (1949): Manuel de l'anglais du moyen âge

In the section dedicated to the history of the unstressed vowels Mossé offers the following comments: phonetically the vowel developing in final position is

"vague" (1949: §36), loss of -e began in unaccented form words: whanne, whan 'when', Sanne, dan 'then', bute, but/bot 'but', etc The change was gradual,

affecting the North during the thirteenth, and the Midlands and the South in the course of the fourteenth century

Unlike the Wrights who speak of a "short stem syllable" as favoring the change in disyllables, Mossé believes that loss occurred first in disyllabic words with a short root vowel No evidence is quoted to support this claim, however, and one can only speculate that this might have been an impression based on later

spelling, i.e the lack of <-e> in the Modern English spelling of words descended from the -VCe type (e.g sone 'son')

Fernand Mossé (1949): Manuel de l'anglais du moyen âge 27

2.9 Jacek Fisiak (1970): A Short Grammar of Middle English

Fisiak's Grammar provides an explicitly phonemic account of sound change in the

history of English vowels Fisiak assumes the existence of a separate phoneme /a/

in Late Old English which remains unchanged in Middle English, and is

represented in spelling by <-e> One special property of this phoneme is that it is

subject to a distributional constraint: it can only occur outside primary and secondary stress (1970: 36) Though the possibility of analyzing schwa as a member of a separate subsystem for the unstressed vowels does not come up, Fisiak recognizes that "in a weakly stressed position Middle English had two phonemes /i/ and /a/" (1970: 46).13

2.10 Peter Erdmann (1972): Tiefenphonologische Lautgeschichte der

englischen Vokale

The account proposed by Erdmann proceeds from the assumption that schwa loss

is not exclusively a phonological problem, but it falls also in the sphere of morphology (1972: 221) He states, obviously disregarding the account proposed

by Horn and Lehnert, that "so far the history of the unstressed syllables has only been given a one-sided treatment from a phonological point of view, without a more detailed consideration of its grammatical relations and interrelations" (1972: 234) Against this background he proposes an analysis of the unstressed vowels prior to the eleventh century as morphophonemes Within Erdmann's framework only one syllabic element remains in unstressed position after the eleventh century This single syllabic element can no longer be described as a morphophoneme because it participates in no alternations From the eleventh century on /a/ is described as a "Silbenträger unbestimmter Qualität" (1972: 227) In the fourteenth century this unstressed vowel falls together with zero There is no discussion of the possibility of /a/ maintaining a contrastive status in Middle English

At the end of his book Erdmann indicates that schwa loss can be treated in terms of grammaticization of sound processes He relates the history of the unstressed vowel to Open Syllable Lengthening in Middle English The information carried by the unstressed vowel is not lost when the vowel itself is lost; open syllable lengthening has the effect of "preventive sound change" through which syntagmatic information is paradigmatized and which makes