French Sound Structre PDF

French Sound Structure
FRENCH
STRUCTURE
Douglas C. Walker
2001 Douglas C. Walker. All rights reserved.
University of Calgary Press

2500 University Drive NW
Calgary, Alberta
Canada T2N 1N4
www.uofcpress.com
Further information relevant to the use of this book and to the study of French phonology in
general may be found on the following www site: http://www.ucalgary.ca/~dcwalker/
National Library of Canada Cataloguing in Publication Data
Walker, Douglas C.
French sound structure
Includes bibliographical references and index.

ISBN 1-55238-033-5
1. French language Phonetics. 2. French language Pronunciation.

3. French language Phonology. I. Title.
PC2135.W34 2001 441.5 C2001-910584-3
We acknowledge the nancial support of the Government of Canada through the Book
Publishing Industry Development Program (BPIDP) for our publishing activities.
All rights reserved. No part of this work covered by the copyrights hereon may be repro-
duced or used in any form or by any meansgraphic, electronic or mechanicalwithout
the prior written permission of the publisher. Any request for photocopying, recording,
taping or reproducing in information storage and retrieval systems of any part of this book
shall be directed in writing to CANCOPY, Suite 1900, One Yonge Street, Toronto, Ontario,
M5E 1E5.
We greatfully acknowledge the International Phonetic Association for allowing us to repro-

duce the IPA chart found on page viii. For more information please contact John Esling at
esling@uvic.ca or visit www.arts.gla.ac.uk/IPA/ipa.html
Printed and bound in Canada by Hignell Book Printing.

This book is printed on acid-free paper.
Page and cover design and typesetting by Kristina Schuring.

Table of Contents
Sound Charts and Transcription . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii
Abbreviations and Symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .x
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi
1. The Object of Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1

1.0 Preliminaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1
1.1 The French Language . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 A Brief Historical Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6
2. Key Descriptive and Theoretical Concepts . . . . . . . . . . . . . . . . . . . . . 11

2.0 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.1 Orthography and Pronunciation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2 Morphological and Lexical Notions . . . . . . . . . . . . . . . . . . . . . . . . . . . .13
3. Basic Descriptive Units and Domains . . . . . . . . . . . . . . . . . . . . . . . . . .21

3.0 Introduction: The Segment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .21
3.1 The Syllable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .23
3.2 The SF Phonological Word . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .30
3.3 The Phonological Phrase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .31
3.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .37
4. Vowels and Semi-vowels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .41

4.0 The Vowel System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .41
4.1 Vowel Length . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .42
4.1.1 Lengthening Consonants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.1.2 Intrinsically Long Vowels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .44
4.1.3 The /' ':/ Opposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .45
4.1.4 Supplementary Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .47
4.2 The Mid Vowels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .48
4.2.1 Mid Vowels in Final Open Syllables . . . . . . . . . . . . . . . . . . . . . . . . . .49
4.2.2 Mid Vowels in Final Closed Syllables . . . . . . . . . . . . . . . . . . . . . . . . .50
4.2.3 Mid Vowels in Nonnal Closed Syllables . . . . . . . . . . . . . . . . . . . . . .51
4.2.4 Mid Vowels in Nonnal Open Syllables . . . . . . . . . . . . . . . . . . . . . . .52
4.2.4.1 Analogy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .53
4.2.4.2 Vowel Harmony . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .54
4.2.4.3 The loi de position . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .55
4.2.5 Grammatical Consequences of the Constraints on Mid Vowels . . . . .57
4.2.6 Orthography and Pronunciation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .59
4.2.7 The Potential Merger of /n/ and // . . . . . . . . . . . . . . . . . . . . . . . . . . .59
4.3 The Low Vowels /a/ and /#/ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .60
4.4 Nasal Vowels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .62
4.4.1 The Merger of /'/ and / . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .63
4.4.2 Distribution of Nasal Vowels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .64
4.4.3 Alternations between X and VN . . . . . . . . . . . . . . . . . . . . . . . . . . . . .65
vi
4.4.4 History and Orthography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .73

4.4.5 Dialects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .75
4.5 Schwa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .76
4.5.1 Orthographic Representations of Schwa . . . . . . . . . . . . . . . . . . . . . . .77
4.5.2 Distributional Constraints on Schwa . . . . . . . . . . . . . . . . . . . . . . . . . .77
4.5.3 The Phonetic Realization of Schwa . . . . . . . . . . . . . . . . . . . . . . . . . . .78
4.5.4 The Deletion of Schwa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .80
4.5.4.1 Schwa in the Phonological Phrase . . . . . . . . . . . . . . . . . . . . . . . . . .80
4.5.4.2 A Rhythmic Constraint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .84
4.5.4.3 Epenthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .85
4.5.4.4 The Tendency Towards Irregularity . . . . . . . . . . . . . . . . . . . . . . . . .86
4.5.4.5 Stylistic Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .88
4.5.4.6 The Stabilization of Schwa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .90
4.5.5 Alternations Involving Schwa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .92
4.5.6 Dialects and History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .95
4.6 Semi-vowels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .100
4.6.1 Orthographic Representations of the Semi-vowels . . . . . . . . . . . .102
4.6.2 The Phonology of the Semi-vowels . . . . . . . . . . . . . . . . . . . . . . . . . .103
4.7 Further Effects of the Phonological Phrase . . . . . . . . . . . . . . . . . . . . .106
4.8 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .107
5. Consonants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
5.0 The Consonant System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
5.1 Geminate Consonants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .130
5.2 Nasal Consonants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .132
5.2.1 The // /nj/ Interchange . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .132
5.2.2 The Importation of /0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .134
5.2.3 Nasal Assimilation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .135
5.3 Voicing Assimilation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .136
5.4 Aspirate-h . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .140
5.4.1 Other Types of Aspiration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .146
5.4.2 Historical Comments Regarding < h > . . . . . . . . . . . . . . . . . . . . . . .147
5.5 Final Consonants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .148
5.5.1 Stable Final Consonants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .148
5.5.2 Latent Final Consonants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .152
5.6 Linking Phenomena: Enchanement and Liaison . . . . . . . . . . . . . . . . .160
5.7 Liaison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .160
6. Prosody . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .177
6.0 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .177
6.1 Stress and Rhythm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .177
6.1.1 Phrasal Stress . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .178
6.1.2 Emphatic Stress . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .181
6.2 Intonation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .182
6.3 Colloquial Constructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .187
vii
7. Around the Phonological Periphery: Playing with Language . . . . .191

7.0 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .191
7.1 Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .191
7.2 Acronyms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .195
7.3 Reduplication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .199
7.4 Word Games: Verlan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .200
Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .207
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .213
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .227
viii
ix
Standard French Vowels and Semi-vowels

front back
spread round round
semi-vowels L w
high K [ W
higher-mid G 1 Q nasal

lower-mid ' n ' n
low C # #
Standard French Consonants
labial apical palatal velar uvular

obstruents
stops
voiceless p t k
voiced b d g
fricatives
voiceless f s 5
voiced v z <
sonorants
nasals m n 0
liquids l *
*In International Phonetic Association (IPA) notation, [] represents a voiced

uvular fricative rather than a liquid. (A separate diacritic symbol is available
to indicate a liquid realization.) For convenience, however, we will use the
symbol [] throughout this work, while classifying [] with the liquids on
functional (largely phonotactic) grounds, and bearing in mind that the actual
phonetic realizations of r-sounds in French show bewildering complexity
(including uvular trills, velar fricatives, apical trills and aps, and, needless to
say, the uvular fricative used here).
x
Abbreviations and Symbols

Indicates that items from this dataset are included on
the CD-ROM
SF Standard French
CF Canadian French
MF French of the Midi (the southern part of France)
OF Old French
CL Classical Latin
X any segment
C any consonant
V any vowel
X any nasal vowel
G any glide (semi-vowel)
L a liquid (/l/ or //)
N any nasal consonant
schwa; a neutral, lax, unstressed vowel subject to deletion
placed before a vowel to indicate that it is stressed
the null symbol; no sound is pronounced
m a syllable
. syllable boundary
+ morpheme boundary
# word boundary
|| phonological phrase boundary
#__ word-initial position
__# word-nal position
__. in an open syllable
__C. in a closed syllable
|| __ at the beginning of a phonological phrase
__ || at the end of a phonological phrase
_ obligatory liaison
(|) optional liaison
| prohibited liaison
() encloses optional material
<> encloses specically orthographic representations
< derives from (historically)
> becomes (historically)
* indicates an incorrect or impossible form
m., f., sg., pl. masculine, feminine, singular, plural
ind. subj. indicative, subjunctive
Note on translations: Most of the French examples cited have been translated
into one of their English equivalents, except in certain cases where the mean-
ing is transparent (e.g., incorrect incorrect) or where the meaning is irrel-
evant because the discussion bears on exclusively phonological issues.
xi
Preface
The material that follows reects my efforts over the past several years to
understand the fascinating complexities and the theoretical implications of the
sound patterns of French, and to communicate them to my classes in a com-
prehensive and comprehensible fashion. As a consequence, I owe signicant
debts of gratitude to the many students who, through their questions and com-
ments, have prodded me to clarify both my thinking and my presentation. I
hope the result is a detailed, well-illustrated, and useful description of the pro-
nunciation of Modern Standard French, incorporating occasional comments
on regional and social variation, on abbreviatory processes and word play, on
certain historical phonological changes that continue to be reected in the con-
temporary language, and on the interdependence of phonology and morphol-
ogy in an appropriate manner.
This work is intended primarily for university students studying French, not
as a practical guide to pronunciation improvement but as a discussion of the
sound system of the language. It is written in a way that presupposes little
or no formal training in linguistics proper (other than some familiarity with
basic terminology and with phonetic notation, to which students are normally
exposed independently). The work should also provide data of interest to stu-
dents of linguistics, where discussions of French phonology (schwa, liaison,
nal consonants, and aspirate-h, in particular) have played a major role in
attempts to resolve certain theoretical matters. Finally, there should be some
material of relevance to those members of the general public with an interest
in the nature of the French language, since pronunciation is rarely considered
in any detailed way in the general introductory handbooks of French.
Now that the text is complete, I must also acknowledge the stimulation pro-
vided by the community of scholars working on French phonology, a domain
that provides seemingly endless fodder for the theoretical cannons of the day,
and that, in a more neutral and (at least potentially) less contentious manner,
fascinates and challenges those seeking to understand it in all its heterogene-
ity. I trust this work, complemented by a relatively detailed list of references,
furnishes an appropriate way to begin to confront the challenges involved.
Among French phonologists, two names require special acknowledgement:
Jurgen Klausenburger and Yves-Charles Morin. Their well-informed and
insightful studies of a vast range of French phonological issues provide models
that none can ignore and that all would do well to emulate. More importantly
in this context, however, they were kind enough to provide detailed and con-
structive criticism of virtually every element of this presentation, and it is
immensely improved as a result. ces deux collgues, un grand merci.
Preparation of this book has beneted from the advice of a group of stu-
dents who suffered through its earlier versions and who have given me insight-
ful feedback. Thanks are due to Shauna Haas, Sarah Johnson and Meghan
McIntyre, and also to Anne Marie Hallworth-Duez, laboratory instructor
extraordinaire, who subjected previous versions of this manuscript to a meticu-
lous reading. The six speakers, Georges Blary, Jean-Bernard Gauthier, Etienne
xii Preface
Grang, Sverine Lamontre, Fleur Larocque, and Eileen Lohka, who lent their
voices to the CD ROM, deserve special acknowledgment for adding a bit of
reality to alleviate the potentially abstract and arid theoretical discussions in
the text.
Both directly and indirectly, this work has been supported in a variety of
ways: by grants from the Social Sciences and Humanities Research Council
of Canada that have allowed me, over the years, to pursue the study of
French linguistics; by a fellowship from the Camargo Foundation in Cassis,
France, during which the work was brought into focus; and by a sabbatical
leave from the University of Calgary. More immediately, I must acknowledge
the generous nancial support received from Alberta Learning and Canadian
Heritage, through the Canada-Alberta Agreement on the Ofcial Languages
in Education 1999-2000, and a Fellowship from the University of Calgary
Learning Commons that provided for technical support in the preparation of
the CD-ROM. The Learning Commons team, led ably by Kathy Schwarz,
Instructional Designer, included Greg Phillips, Sound Engineer, Lane Turner,
Audio Technician, Michelle McGrath, Graphic Artist, and Programmers Julian
Wood, Robert Purdy, Ashley Rollke, and Rob Loh, Testers Mike Walker and
Eric Rogers, and Production Assistant Gord Southam. At the University of
Calgary Press, Joyce Hildebrand edited the text with great acumen; Kristina
Schuring spent many hours on the details of design; and Walter Hildebrandt,
John King and Tim Au Young provided very helpful general support from the
outset of the project. My sincere thanks to each of these organizations and to
all of the individuals involved. And nally, with much gratitude to Tracy, Cris,
and Dave, sine quibus non.
1
Chapter 1
The Object of Description
1.0 1.0 Preliminaries

While the title of this book, French Sound Structure, appears relatively
straightforward, it also provides a starting point for general discussion of a
number of concepts that will allow us to place the details to follow in a larger
context. It will be useful, for example, to understand in some detail what the
term French refers to. Sound, moreover, is at once too general and too spe-
cic. It is not just any sounds that are of interest, but that subset of possible
human sounds that play a variety of roles in the linguistic system we call
the French language. Nor is it just French sounds in isolation that we need
to examine, but sounds as constitutive of words, as conveyors of different
types of meaning, as participants in different phonological and grammatical
processes. Sound systems are, in other words, highly structured: classes of
sounds, general phonological processes, grammatical consequences are all
notions that will be important in the descriptions to follow.
1.1 1.1 The French Language

Originating in the Latin spoken in Northern Gaul and establishing itself as
a recognizably independent language in the early Middle Ages,1 the French
language is now spoken worldwide. It enjoys ofcial status in several dozen
countries (including France, Belgium, Switzerland, Luxembourg, Canada,
Cameroon, Ivory Coast, French Guiana, Madagascar, and French Polynesia)
and is a widely used medium of communication or education in many others
(such as Algeria, Morocco, Tunisia, much of sub-Saharan Africa, and Vietnam).
It is one of the ofcial languages of the United Nations and of numerous other
international organizations. Internationally, French numbers approximately 75
million speakers who count it as their native language and upwards of an addi-
tional 150 million who use it readily for communicative purposes. The French
language has long been associated with the prestige attached to French scien-
tic, literary, and cultural contributions and continues, despite increasing pres-
sure from English, to play a major role on the world stage.
2 Chapter 1
The wide geographic distribution of the French language is understandably

correlated with linguistic differences. The French of Lige is not that of
Marseilles, Geneva, Montral, Port-au-Prince, Algiers, Dakar, or La Runion.
In fact, even within the borders of France itself,2 Alsace, Picardie, Normandie,
Touraine, Bourgogne, Auvergne, Provence, and other regions offer recog-
nizable local varieties. The French language, then, is geographically diver-
sied, perhaps even to the extent that widely separated versions pose dif-
culties of mutual comprehension. To this geographic diversity may be added
social diversity. The speech of la haute bourgeoisie or even a residual aristo-
cratie will differ from that of the middle class, recent immigrants, technocrats,
labourers, or the SDF (sans domicile xe, the homeless). The speech of the
young differs from that of older generations. (Reection on the use of slang,
nicknames, tutoiement, or profanity, for example, provides immediate conr-
mation of generational differences.) The speech of men differs from that of
women. Each profession enjoys its own jargon. Finally, we must recognize
differences correlated with various speech situations, where the nature of the
context requires (or favours) different degrees of formality, different styles or
registers. Usage differs in courtrooms, classrooms, family rooms, and locker
rooms, and expressions appropriate in one context may bring censure or deri-
sion in another.
Against this background of dramatic regional, social, and stylistic variation,
what is the analyst (or the student) to do? Fortunately, there exists a widely
accepted response to this challenge, adopted by linguists and speech commu-
nities alike. The solution involves identifying a geographically, socially, and
stylistically circumscribed variety and using it as a reference point. This refer-
ence point, usually called a standard language or, in the case of French, le bon
usage, is considered by its speakers to be the most appropriate variety for use
in formal and educational contexts. It is traditionally employed in the media
and codied in grammars.3 Often, it is the object of attention of a national
linguistic legislature such as the Acadmie franaise. Thus, the object of
discussion here will be Standard French (abbreviated SF), that variety of the
language identied most often with the speech of the Parisian middle class
(la bonne bourgeoisie parisienne; le Parisien cultiv) when its members are
engaged in polite conversation. Warnant (1987), in a discussion of the reasons
underlying his choice of recommended pronunciations, puts it as follows: 4
Gographiquement, nous avons choisi de consigner la prononciation

dun des franais de Paris et, dans un sens large, de la rgion parisi-
enne. Paris est sans conteste le centre de la trs grande majorit des
activits intellectuelles et culturelles de la France. Nous pensons plus
particulirement celles qui se manifestent par et dans la parole.
Socialement, nous avons choisi de recueillir la prononciation des

intellectuels et des gens cultivs, qui, dans des domaines extrme-
ment divers, sexpriment dordinaire non seulement avec clart, mais
encore avec lgance.
The Object of Description 3
Nous avons aussi choisi de ne dcrire que le parler dune gnra-

tion, celle qui compte actuellement entre 20 et 45 ou 50 ans. Nous ne
risquons pas ainsi, dune part, de proposer pour demain lusage dune
prononciation dj vieille dans lun ou lautre de ses traits. Nous vi-
tons, dautre part, de donner en exemple une prononciation non encore
dnitivement installe.
Despite Warnants efforts to circumscribe the object of his description, his

comments still do not rule out alternative pronunciations for one and the
same form (and, in fact, he often lists alternates particularly where schwa is
involved). Martinet and Walter, much more sensitive to the great heterogene-
ity of current SF pronunciation, approach the matter pragmatically rather than
normatively in this way (1973: 1617):
Est-ce dire que toutes les prononciations quon peut entendre soient
galement recommandables? Et dabord quest-ce quune prononcia-
tion recommandable? Il semble quen cette matire lidal soit de
ne rien faire qui attire lattention et la dtourne ainsi de la comprhen-
sion de ce qui est dit. Les bonnes prononciations sont celles qui pas-
sent inaperues, les mauvaises celles qui soudain vous rappellent, ne
serait-ce qu un niveau trs infrieur de la conscience, que votre
interlocuteur est de telle origine, nationale, gographique ou sociale.
Ce que recherche ltranger qui apprend une autre langue, cest
prcisment dviter que les formes quil emploie rvlent sa non-
appartenance la socit dont il cherche utiliser lidiome.
While these quotations give us a very good general idea of our descriptive
target, they still leave room for a certain exibility. First, even within an
unmarked, standard pronunciation, there will exist variants, often related
to age differences, between innovative and conservative realizations. These
distinctions have certain descriptive consequences, some of which we will
address below. Secondly, despite our concentration on Standard French, we
may occasionally wish to venture into an exploration of nonstandard territory,
particularly the domain of popular or colloquial speech. It would be appropri-
ate, then, to review briey the nature of other registers. It is important to rec-
ognize, however, that stylistic (and social) variation forms a continuum and
that levels of usage (and the distinctions made by analysts) may grade subtly
from one into another. That being said, we may think of at least the following
general distinctions:
(1) Levels of usage (registers) 5

(a) formal (niveau cultiv, soign, soutenu): implies a formal or
perhaps ofcial speech situation in which careful attention is paid
to the organization and delivery of the message; may include
archaisms of various types; normally conservative rather than
4 Chapter 1
innovative (although highly formal speech can also be innovative

under certain circumstances, as in the phenomenon of liaison sans
enchanement [Encrev 1988], which characterizes in particular the
public pronouncements of certain politicians).
(b) standard (franais courant, usuel, commun, ordinaire): the
unmarked register recognized as the norme by native speakers; the
usage described by Martinet and Walter above.
(c) familiar: used in less formal situations, primarily in oral
communication between family members, friends, or those of similar
status.
(d) popular (trs familier, vulgaire): spontaneous, unmonitored speech
replete with phonetic reductions, errors (e.g., avoir for tre as
auxiliary, lack of agreement, absence of subjunctive), slang, and often
profanity.
The following graded examples, moving from formal to familiar or popular,

may give an indication of the passage from one level to another:6
(2) (a) O demeurez-vous? Where do you reside?

O est-ce que vous habitez? Where do you live?
Vous habitez o? You live where?
(b) Voil une voiture qui fait sensation. This car is creating a
sensation.
Elle est sensationnelle cette voiture. This is a sensational car.
Elle est sensas cette bagnole. What a terric car!
(c) Que dites-vous? May I ask what youre
saying?
Quest-ce que vous dites? What are you saying?
Vous dites quoi? Youre saying what?
Quoi? What?
Hein? Huh?
(d) Elle sest prise de lui au premier regard.
She became enamoured of him at rst sight.
Elle est tombe amoureuse de lui ds leur premire rencontre.
She fell in love with him on the rst date.
Elle est alle samouracher de ce petit voyou.
Shes head over heels in love with the guy.
(e) importuner to importune

ennuyer to bother
enquiquiner, emmerder to bug
(f) Je vous prie de ne pas fermer la porte.
May I ask you not to close the door?
Ne fermez pas la porte, sil vous plat.
Please dont close the door.
Fermez pas la porte.
Dont close the door.
(g) Qui fut le premier ministre? Who was the rst prime
minister?
Qui tait le premier ministre? Who was the rst prime
minister?
Ctait qui le premier ministre? Who was he, the rst prime
minister?
(h) Je nai pas vu ce lm. I havent seen this lm.
Moi, ce lm, je lai pas vu. This lm, I havent seen it.
(i) Pourquoi dites-vous cela? Why do you say that?
Pourquoi que vous dites a? How come you say that?
(j) lhomme avec qui ils sont venus the man with whom they came
lhomme quils sont venu avec the man they came with
While our concern in this work will be primarily with the phonology of the
standard register, it is important to realize that the concept French encom-
passes both extensive phonological variation and the use of other mechanisms
(e.g., vocabulary choice, extensive syntactic modications) to permit subtle
adjustment of usage to t a rich diversity of speech situations.
Although in what follows we will concentrate on the Standard French of
Paris or the le-de-France, it is also evident, given the great geographic diver-
sity in French, that one can recognize regional standard languages (and even
include Paris as one among many regions). That is to say, the notion of stan-
dard, in terms of an unmarked register recognized as the norme by native
speakers, is not limited to any specic region. In social or stylistic terms, stan-
dard can be applied to a variety recognized as prestigious within any commu-
nity. One often sees reference, for example, to Standard Canadian French, just
as the standard in English-speaking Canada is dened with respect to a style
signicantly different from the Queens or BBC English in Great Britain.
6 Chapter 1
Minor lexical or phonetic differences aside, in fact, Standard Canadian French

and Standard Parisian French are highly similar the differences only become
dramatically signicant in the popular registers, and even here we would still
have no difculty in recognizing both vernaculars as inherently French, rather
than as separate languages.
Finally, although we have spoken of standard and norme, the speech we
refer to by these terms is in a direct way a historical accident. The prestige
attached to the standard associated with Parisian French reects the political
and economic importance of Paris, mirroring its role as capital of France,
focus of intellectual activity, seat of the government, industrial centre, and
so on. Had French history been different, the dialect of Dijon, Poitiers, or
Toulouse might just as easily have become the standard. Thus, in strictly lin-
guistic terms (in tems of the linguistic structures involved), the forms of one
dialect are no more worthy or unworthy than those of another. This is not
to say that one cannot make value judgements about certain forms or styles,
nor that these judgements are unimportant or without consequence. Opinions
about speech forms, or particularly about their appropriateness to the context
in which they are used, are crucial in all societies. But they reect the social
values attached, at a limited set of times and places, to language and its use,
and constitute judgements about the users, rather than an evaluation of the
inherent merit of one pronunciation or one structure over another.
1.2 1.2 A Brief Historical Review

As the foregoing remarks have indicated, SF is a product of a combination
of historical circumstances. The French of Paris in the Middle Ages is mark-
edly different from (and perhaps incomprehensible to) that of a speaker in the
court of Louis XIV or a Parisian lycen of our days. Here we will review, very
briey, some of the key events and stages leading to the current situation.
Restricting our discussion to historical times (roughly from the eighth cen-
tury B.C. onwards), the territory we now call Gaul was at that period inhab-
ited by a variety of Celtic tribes who occupied lands previously settled by
Ligurians, Iberians, and Aquitanians, among others. Eventually, the Celts took
(or were given) the name of Gaulois, lending as well their name to the land
they occupied: La Gaule.7 From about 125 B.C., Roman presence in southern
Gaul (la Gaule transalpine) increased, and subsequently Caesar led a conquest
of the remaining Gallic territory (5850 B.C.). Little by little, aided by the
importance of the Roman administrative and educational structures, the Latin
language replaced the earlier Celtic dialects, at least in the urban centres. Latin
became the language of Gaul, but Latin itself, through the centuries, evolved
into distinct spoken and written versions that increasingly diverged from each
other. Following the fall of the Roman Empire in A.D. 476, local varieties mul-
tiplied and a linguistic frontier solidied between the more heavily Romanized
south and the northern regions, now occupied and governed by the Francs, a
Germanic tribe from the region of the Rhine. The langue doc or occitan in the
south and the langue dol in the north became increasingly distinct.
In the north, the Francs assimilated linguistically to the local Latin-speaking
population, but not without a certain amount of inuence on the lexicon8 and
on pronunciation, the latter usually attributed to the heavy stress accent of their
Germanic speech. Gaul remained, in other words, a Romance-speaking terri-
tory, even following the Viking invasions of the early eighth century. These
normands (hommes du nord) adopted in turn the Romance vernacular, even-
tually exporting it to England following the Norman Conquest of 1066. In
England, Anglo-Norman literature ourished, and French played a dominant
role in British administrative and cultural life for several centuries.
By the mid-ninth century (Serments de Strasbourg) and certainly by the
beginning of the tenth (Cantilne de sainte Eulalie) the indigenous Romance
language of the north of France had diverged sufciently from the earlier
spoken Latin that one can speak of the birth of the French language (franais,
language of the Francs). At this time, however, social and political conditions
were inuenced not so much by a unique national capital but by strong
regional centres, so one must inevitably speak of regional dialects including,
among others, picard, champenois, anglo-normand, bourguignon, louest and,
needless to say, the francien of the le-de-France. At the Old French stage,9
many of the regional courts rivalled that of Paris, and literature in these dia-
lects was easily the match of that written in francien.
With time, however, French kings extended their military and political
domain much further throughout the territory, and the language of the Ile-de-
France beneted from increased prestige: francien was on its way to becom-
ing the national standard. In 1539, under Franois Ier, the Ordonnance de
Villers-Cotterts made French (i.e., francien) the ofcial language (replacing
Latin) for all court orders and judgements. In 1549, Du Bellay, on behalf of
La Pliade, published the manifesto Dfense et illustration de la langue fran-
aise, a work, along with those of Rabelais, leading to much linguistic innova-
tion, innovation aided by the explosion of literary works following the inven-
tion of printing. The increasing importance of Paris had further linguistic con-
sequences. As Bonnemason (1993: 28) puts it, lEtat monarchique fait de la
langue franaise son affaire. Le pouvoir politique est Paris, la langue est celle
de Paris et elle sera codie et rgente. The Acadmie franaise, created in
1635 by Richelieu, codied the orthography and published its rst dictionary
in 1694. The Grammaire de Port-Royal of 1660 established a general stan-
dard to be met by even the greatest writers. Linguistic prescriptivism took rm
root.
Despite the inuence and prestige of Paris, however, regional dialects
(patois) and regional languages persisted. Still, the industrial revolution, the
development of science and technology, the importance of the writings of
the philosophes, extensive exploration, and colonization all contributed to
the expansion of French and to the suggestion of le franais comme langue
universelle. Regional languages and the patois suffered under the uniformiz-
ing pressures of the Revolution and the imposition of French as the general
8 Chapter 1
language of schooling, pressures that exist to this day. Nor is the role of the
mass media negligible as a standardizing force. Nonetheless, alongside the
national standard, linguistic variation in the form of both dialects and distinct
regional languages remains characteristic of contemporary France, a variation
that is sometimes discouraged ofcially and sometimes (as in some recent leg-
islation inspired by the European Union) ofcially supported. And the stan-
dard language itself is a mixture containing a foundation of words from the
original Latin source, supplemented by early Celtic and Frankish contacts
(among others), a technical and learned vocabulary necessitated by corre-
sponding intellectual or industrial developments, and loans from the many lan-
guages with which its long history has brought it into contact. Against this
complex background, it is consequently not difcult to grasp why the notion
of Standard French is at the same time both an arbitrary and ever-changing
construct and a useful reference point. In any event, SF, as described above
in all its complexity, is the object of this work. Before we undertake a study
of the specic sound structures of SF, however, it is necessary to provide
some further, theoretically oriented detail concerning the concepts to be used
in a description of this material. This brief orientation is the task of the next
chapter.
Notes
1. The Serments de Strasbourg from 842, a document conrming a political alliance
between two of Charlemagnes grandsons concerning the partition of his empire,
is often given as indicating the birthdate of French since it contains the earliest
surviving text in the vernacular of Gaul.
2. Despite what may be thought, France itself is not linguistically homogeneous:

other languages spoken to a greater or lesser degree within continental French
territory include Catalan, Occitan or Provenal, Italian, Basque, Breton, Flemish,
and the Germanic dialects alsacien and lorrain. Recent immigration has also sig-
nicantly increased the number of speakers of Arabic.
3. See Morin (1999) for a detailed and up-to-date survey of this question, incorpo-
rating the signicant term le franais de rfrence, a designation perhaps pref-
erable to Standard French or le franais standard. We retain the latter on the
basis of its wide familiarity. Morin also provides much detail concerning pro-
nunciation variation within various normes, as well as the difculties inherent in
the notion of norme itself. Martinet (1990) addresses some of the same issues.
In this context, spoken media have become more tolerant of a range of
regional accents, provided that they are not too marked, as witnessed, for exam-
ple, by the popularity on national television of the literary critic Bernard Pivot
with his Burgundian accent (in Apostrophes and Bouillon de culture). In a paral-
lel fashion, national television in Great Britain is also more accepting of regional
pronunciations, as indicated by the presence of Welsh, Scottish, or Yorkshire
readers of the national news. In North America, the situation on national televi-
sion news does not yet appear to be as diverse.
4. Warnant, interestingly enough in this context, is Belgian. The quotation is taken

from the cover of his pronouncing dictionary but is compiled from material on pp.
xxxxi.
5. Muller (1985: 225262) provides an excellent survey of these issues, with copi-
ous examples. Guiraud (1969: 24) contrasts the two poles of usage in the fol-
lowing indicative terms: Bref, il est lgitime de distinguer deux formes limites
de la langue conditionnes par un ensemble de variables complexes dont les unes
tiennent lhistoire, la culture, la socit, les autres aux conditions de la com-
munication ou la nature du message. Ainsi sopposent populaire/bourgeois,
relch/soutenu, libre/prscriptif, spontan/stylis, oral/crit, hrditaire/savant,
dialectal/national, expressif/cognitif, locutif/prdicatif, naturel/cultiv.
6. Examples are drawn from Batchelor and Offord (1993a, 1993b) and Muller
(1985).
7. Celtic traces in French are few, but include some sixty surviving words: sapin
r, chne oak, lotte monksh, bouc goat, mouton sheep, chemin road,
dune dune, druide Druid, etc., as well as several place names, perhaps most
notably Paris, from the Celtic tribe Parisii.
8. Frankish lexical remnants are much more numerous than Celtic, including France
itself, le pays des Francs. Additional Frankish words include banc seat, bl
wheat, bois wood, choisir to choose, cruche pitcher, danser to dance,
framboise raspberry, garder to keep, gurir to cure, guetter to watch, hache
axe, har to hate, honte shame (and numerous other aspirate-h words), jardin
garden, lcher to lick, marchal marshall, orgueil pride, regarder to look
at, soupe soup.
9. Old French is commonly divided into two periods, Early Old French, from the
middle of the ninth to the end of the eleventh century, and Later Old French,
from the beginning of the twelfth to the middle of the fourteenth (Pope 1934: 9).
Middle French comprises the fourteenth, fteenth and sixteenth centuries. Later
Old French constitutes a period of spectacular literary and cultural development.
11
Chapter 2
Key Descriptive and Theoretical Concepts
2.0 2.0 Introduction

The preceding chapter discussed the general notion French, noting the ambi-
guities and complexities inherent in treating a domain with the long history,
extensive geographic range, and sociolinguistic complexity that characterize
la francophonie. Here, we will focus on the remaining words in the title of
this book, sound and structure, reviewing briey a number of concepts or
distinctions that will prove useful in the description to follow.1 First among
these is the need to explore the difculties arising from the nature of French
orthography.
2.1 2.1 Orthography and Pronunciation

Clearly, there exists some type of systematic relationship between French
orthography and the corresponding ways of pronouncing French words. It is
not the existence, but rather the nature of this relationship that is of concern:
how regular is it? what types of exceptions are found? where does it fail to
inform us of the appropriate pronunciation? Unlike languages such as Spanish,
where the link between writing and speaking is close and direct, French pres-
ents a number of difculties in interpreting phonologically the orthographic
representations of words. Consider the examples in (1).
(1) Orthographic puzzles

(a) one sound many spellings
/k/: coup, occuper, kilo, qui, cinq, grecque, acqurir, echo,
ecchymose, khalife
/s/: si, cible, soixante, assez, a, science, balbutie
/a/: art, l, grce, drap,2 femme, paonne, habituer, Jeanne
/o/: vlo, aube, eau, eaux, cte, hte, hauteur, heaume, faux
12 Chapter 2
(b) one spelling several sounds

< e >: rue , breton //, serpent /'/, effacer /e/, femme /a/, en /#/
(This problem can extend to sequences of letters: < ent > [couvent
as /kuv/ or /kuv#/], < ai > [faisan /fz#/; faisceau /f'so/; faillir
/faji/], for example.)
(c) silent nal consonants
plomb, caoutchouc, noeud, clef, poing, fusil, drap, donner, gros,
salut, six, nez, il est, exact, gars, corps
(d) etymological spellings
doigt /dwa/ < CL digitus, sept /s't/ < CL septem, vingt /v'/ < CL
viginti, compter /knte/ < CL computare, baptme /bat'm/ < Greek
baptisma, corps /kn/ < CL corpus, sculpter /skylte/ < CL sculpere,
sculptus
(e) contextually determined variation
deux petits chiens /dpti5j'/
cinq petits chiens /s'kpti5j'/
dix postes /dipnst/
dix emplois /diz#plwa/
plomb plombier /pln/ /plnbje/
fusil fusiller /fyzi/ /fuzije/
exact exactitude /'gza/ /'gzaktityd/
(f) stylistic variation
je ne sais pas
/<ns'pa/
/<ns'pa/
/<s'pa/
/5s'pa/
/5'pa/
These examples demonstrate the ambiguity of certain aspects of French

orthography the lack of a one-to-one relationship between spelling and pro-
nunciation. It is often claimed that ideally there should exist only a single pos-
sible pronunciation for each sequence of letters and a unique orthographic rep-
resentation for each different pronunciation. As indicated by the symbols we
have already used, there are notational systems that implement this approach,
Key Descriptive and Theoretical Concepts 13
namely the symbols of the International Phonetic Association (IPA), which we

will use (more or less) consistently in what follows.3 IPA notation, in other
words, is appropriate because it identies consistently and specically the pro-
nunciations with which we are dealing.
We should specify, however, that the symbols may be used in different
ways. If we wish to indicate precise phonetic detail, such as the assimilation
of voicing in plan, pronounced [RN#], or the light and dark < l > of Early
OF las [las] versus els ['NU]), the symbols will be enclosed in square brackets
and will indicate the phonetically ne-grained units traditionally called allo-
phones. More frequently, however, the symbols will refer to more abstract
phonological units, enclosed in slant lines and omitting much predictable (and
potentially distracting) phonetic detail: /pl#/ rather than [RN#], for example.
These more abstract representations can occasionally involve different degrees
of abstractness, depending on whether they indicate units (normally words)
in isolation as opposed to combinations of words in larger contexts to which
various processes have applied, or whether they indicate formal as opposed to
informal or rapid speech pronunciations. In the discussions to follow, we will
use slant lines for both degrees of abstractness but indicate, where necessary,
the degree of abstractness (or the type of unit) involved.4
2.2 2.2 Morphological and Lexical Notions5

As we have already remarked, sounds combine to form words, the fundamental
units in the composition of phrases and sentences and ultimately of communi-
cation in general. As a result, it will be useful, and often necessary, to under-
stand some of the basic terminology associated with word structure, beginning
with the notoriously difcult notion of word itself.
Word refers, in fact, to four signicantly different types of unit, which we
will distinguish as follows: orthographic words, phonological words, lexical
items, and word forms. Orthographic words, no doubt the most supercially
familiar, are those written between spaces in texts. In that sense, je ne le sais
pas I dont know contains ve orthographic words, while quelques vne-
ments incomprhensibles several incomprehensible events contains three.
Phonological words are words pronounced as a single unit, whatever the
number of orthographic words involved. Thus ces enfants these children,
nous arrivons we arrive, or even je ne le sais pas constitute single phono-
logical words in French they are pronounced without interruption and with
a single accent on the nal syllable. Lexical items, as the name implies, are
those words that serve as dictionary entries, having an identiable meaning
and grammatical role and a relatively constant phonological shape. In this
sense, petit and savoir are lexical items. Word forms are the realizations of
lexical items in specic morphological or syntactic contexts. Thus, the lexical
item petit small represents the four orthographic words petit, petits, petite,
petites, and has the word forms /pti/, /pti/, /ptiz/, /ptiz/, /ptit/, /ptit/, /ptitz/,
/ptitz/, depending on the inectional categories of number and gender (as well
14 Chapter 2
as syntactic and phonological context). Savoir to know is even more com-

plex: as a lexical item it encapsulates the full range of multiple inected forms
such as sais, savons, savaient, saches, sauriez, smes, and so on, as well as
the nonnite forms savoir, sachant, su. In what follows, we will often be con-
cerned with the relationships between orthographic and phonological words
and between lexical items and word forms.
As the preceding examples make clear, words are more complex than might
at rst be thought. In fact, one can pursue the structural analysis of words
much further, using concepts such as root and afx, inection, derivation, and
compounding. A root is the minimal common part of a lexical item that occurs
in all the word forms realizing that item and that bears its central meaning. To
roots may be added afxes (prexes and sufxes) of two main types: inec-
tional and derivational. Within larger structures, inectional afxes signal the
grammatical properties of the word form in question. In French, nouns and
adjectives are inected for number (singular/plural) and gender (masculine/
feminine), as in grand, grands, grande, grandes, although the phonological
realizations of these categories are far from consistent. Verbs are inected for
a variety of categories: number (singular/plural), person (rst/second/third),
tense/aspect (present/past/future/imperfect/conditional), and mood (indicative/
subjunctive), as indicated for savoir above. Derivation, on the other hand, is
concerned with the creation of new lexical items based on the root in question.
From grand large one may derive by afxation grandet, grandelet, gran-
dissime, grandiose, grandeur, grandement, grandir, grandissement, agrandir,
agrandissement, and so on. Inection in French is accomplished through suf-
xation, while derivation employs a rich system of both prexes and sufxes.
Additional word formation processes are also found in French, including
compounding, conversion, and various minor mechanisms (abbreviation,
acronyms, reduplication, to which we will devote separate treatment).
Compounding involves the creation of a new word by combining two exist-
ing roots or word forms: sang-froid, aigre-doux bitter-sweet, savoir-faire,
porte-parole spokesperson, and so on. Conversion (called drivation impro-
pre in French) entails the transfer of a lexical item from one grammatical
category to another with the consequent attribution of new functions. Thus,
nouns can function as adjectives (un culot monstre a huge amount of nerve,
une rponse mi-gue mi-raisin an answer half g, half grape); adjectives
as nouns (mon petit my child, le priphrique the ring road, les jeunes
youth); verbs as nouns (un devoir a duty, le savoir knowledge) and so
on.6 Derivation, compounding, conversion, and other processes can all partici-
pate in the formation of new words (neologisms), processes that constantly
expand and renew the lexicon of a language. The degree of activity of any
individual process is in part a reection of its productivity, that is, the degree
of freedom and frequency with which it operates, a freedom that can vary over
time.7 Thus, sufxation with -ie or -ure (courtoisie courtesy, magistrature
magistracy) is no longer productive, while -age (dopage doping, stockage
storage), -erie (billeterie cash dispenser, sweaterie sweater shop) and
-ique (informatique computer science, robotique robotics), for example,

continue to be used in the frequent creation of new lexical items.
Neologism is one way in which lexicons are expanded. Borrowing is
another. Throughout its history, French has borrowed extensively recall the
Celtic and Frankish examples of the rst chapter. To those earliest loans,
one may add testimony of the close contact with Provenal (troubadour,
amour, ballade, salade, tapenade); extensive Italian borrowings, many from
the Italian inuence of the Renaissance (gazette, alarme, brigantin, baguette,
piano); various Arabic words, often entering through Spanish (algbre, chiffre
numeral, alcool, girafe, abricot apricot); highly diverse words from the lan-
guages of the French colonies or overseas contacts (cacahute peanut, patate
potato, tabac tobacco, cannibale, tomate, cobaye guineapig, banane,
iguane); and the massive and controversial importation of English items char-
acteristic of the latter part of the twentieth century and beyond, an importation
so signicant it has led both to dictionnaires des anglicismes and to legislation
attempting to stem the ow.8
French is also characterized by a special type of borrowing where the lan-
guage draws on its own history: the massive importation of learned loan
words. Learned words (mots savants) are those that French has imported from
Latin (and Greek) subsequent to the Early Old French stage and that have, as
a consequence, failed to be subject to the many sound changes that led to the
formation of the basic French vocabulary. This type of borrowing has led to
the formation of doublets, or pairs of words from the same source language,
a learned word alongside its popular counterpart (mots savants mots popu-
laires). Doublets, illustrated in (2), raise a number of interesting problems, of
which the most important involves the issue of related word forms: are the
phonological and semantic similarities between such words sufcient to con-
clude that they represent the same lexical item?9
(2) Doublets
afiger afiction afict afiction
angle angulaire angle angular
bte bestial beast bestial
bouillir bullition to boil boiling
cendre incinrer cinder to incinerate
contredire contradiction to contradict contradiction
cte costal coast coastal
coupable culpabilit guilty guilt
croire crdibilit to believe credibility
diable diabolique devil diabolical
16 Chapter 2
double dupliquer double to duplicate

cole scolaire school scholarly
entier intgre entire complete
t estival summer summery
fte festival holiday festival
genou gnuection knee genuection
got dgustation taste sampling
mcher mastiquer to chew to masticate
peuple populaire people popular
pied pdestre foot pedestrian
poudre pulvriser powder to pulverize
poumon pulmonaire lung pulmonary
recevoir rception to receive reception
recouvrer rcuprer to recover to recuperate
restreindre restriction to restrain restriction
royal rgal royal regal
sret scurit safety security
However we decide to approach these examples in a discussion of French pho-

nology, it will often be the case that special categories of words (loan words,
learned words, proper nouns, highly frequent forms, and so on) will require
separate treatment, either because of their specic properties or their excep-
tional behaviour. These distinctions between different types of lexical items,
in other words, will be useful in the detailed descriptions of subsequent chap-
ters.
Finally, we must return to a set of issues associated with the relationship
between lexical items and word forms. A lexical item is, in an important
sense, the basic representative of what we rst called a word. Word forms,
in contrast, are the specic realizations of those items in particular contexts.
Word forms, as well as their component parts roots and afxes often
differ in their phonological shape; when the differences are systematic and
not restricted to a single unit or a single morpheme, we will call them alter-
nations.10 Alternations occur when stems and afxes (inectional or deriva-
tional) are combined, when words vary because of the context in which they
occur, or when stylistic or other variation takes effect. The examples in (3)
illustrate some of the alternation types to be discussed below.
(3) Phonological alternations

je pars nous partons /pa/ /pat+n/ I leave we leave
je dors nous dormons /dn/ /dnm+n/ I sleep we sleep
nous venons ils viennent /vn+n/ /vj'n/ we come they
come
nous crevons ils crvent /kv+n/ /k'v/ we wear out
they wear out
gros grosse /go/ /gos/ big (m./f.)
bon bonne /bn/ /bnn/ good (m./f.)
sot sotte /so/ /snt/ silly (m./f.)
drap draperie /da/ /dap+i/ sheet drapery
son sonner /sn/ /snn+e/ sound to sound
honneur honorer /nn/ /nnn+e/ honour to sound
clair clart /kl'/ /kla+te/ clear clarity
peuple populaire /ppl/ /pnpyl+'/ people popular
je pars jarrive /<pa/ /<aiv/ I leave I arrive
il le voit il la vu /illvwa/ /illavy/ he sees it - he
saw it
petit gamin petit enfant /ptigam'/ /ptit#f#/ small
une menace la menace /ynmnas/ /lamnas/ menace
These alternations may be regular, systematic, predictable, phonologically

plausible and productive (jarrive je pars), or the opposite irregular and
unproductive (lart /la/ art le hart /la/ hart). The contexts in which
they occur may be dened in phonological, morphological, or syntactic terms,
or the alternations may be totally idiosyncratic. (Needless to say, intermedi-
ate possibilities also exist.) Once again, the conditions in which we nd these
alternations will need to be identied as we undertake a description of the pro-
nunciation of French, a description for which the discussions of the rst two
chapters have prepared us and to which we now proceed.
18 Chapter 2
Notes
1. This work does not pretend to be an introduction to phonetics, to general phonol-
ogy, or to contrastive French-English analysis, for which many excellent manu-
als exist. For the rst two areas, see Davenport and Hannahs (1998), Roca and
Johnson (1999) or Goldsmith (1995). For the third, LeBel (1990, 1991), Ostiguy
et al. (1996), and Picard (1987) all provide comparisons of English and French
pronunciation.
2. This example represents many others ending in /aC(C)/ where the nal conso-
nants are not pronounced: tabac, exact, ras, tat, gars, etc. The same situation
arises with virtually all vowels in word-nal syllables.
3. There exist other notational systems but that of the IPA is the best known. For
details concerning notation, see Pullum and Ladusaw (1986) or the works cited
in footnote 1 above. The standard French dictionaries (Larousse, Le Robert, etc.)
as well as the major pronouncing dictionaries (Warnant, Martinet and Walter,
Lerond) all use IPA notation.
4. In other words, the description presented here will ignore for the most part
the distinction between a phonetic and a phonemic or phonological description.
Normally, the symbols used will represent SF phonemes, and we will refer spe-
cically to the greater detail of phonetic variation as the need arises. In more theo-
retical terms, we will not normally distinguish phonemic from morphophonemic
representations, nor lexical from post-lexical forms. Those familiar with the lit-
erature will recognize that this discussion has a strongly concrete orientation.
5. Material in this section borrows heavily from Trask (1999), an excellent hand-
book for anyone interested in a survey of basic linguistic concepts.
6. See Bchade (1992: 136139) for a convenient summary of conversion pro-

cesses.
7. Inectional processes may also be distinguished as to their productivity. For

example, the formation of plurals in -aux from -al is nonproductive, while the
formation of feminines by sufxation of -e (professeure, auteure) is widely used.
8. See Walter (1997) for a detailed and fascinating exploration of the diverse com-
ponents of the French lexicon. Picone (1996) provides an excellent discussion of
the current dynamics of French borrowing and neologism.
9. For discussion of the issues involved, see Dell and Selkirk (1978), Walker (1975)
or Zwanenburg (1983). In general, learned words are longer, contain character-
istic consonant clusters, lack schwa, are formed with a distinct set of learned
afxes, and are more semantically specialized. We ignore here a large set of
items derived from Greek rather than Latin (e.g., aptre apostolique apostle
apostolic, hypnotique hypnotic, amnsie amnesia, etc.).
10. Technical terminology in this area abounds; we refer here to the traditional con-
cepts of morpheme, allomorph and morphophonemic alternation.
21
Chapter 3
Basic Descriptive Units and Domains
3.0 3.0 Introduction: The Segment

The sound structure of a language is usually described in terms of discrete
sound units called phonemes or segments, as well as the contexts or
domains within which these units function. The next two chapters will pres-
ent in some detail the French segmental inventory, divided into the two pri-
mary groups (a) vowels and semi-vowels (or glides) and (b) consonants. In this
chapter, we will present more general information concerning the descriptive
framework to be used.
The starting point for any description involves the establishment of the pho-
nemic inventory. SF phonemes (or segments) are presented in tabular form
below, a form organized according to the articulatory properties of the seg-
ments in question. Needless to say, we will have much more to say concerning
these sounds and their functioning in the following pages.
(1) SF vowels and semi-vowels

front back
spread round round
semi-vowels L w
high K [ W

lower-mid ' n ' n
low C # #
22 Chapter 3
(2) SF consonants

obstruents
stops
voiceless p t k
voiced b d g
fricatives
voiceless f s 5
voiced v z <
sonorants
nasals m n 0
liquids l
Despite the monolithic impression left by these symbols, the division into
vowels and consonants and the description of the two sets in articulatory terms
already indicate that we may also need to deal with smaller, subsegmental
entities. These units are called phonological features, and features also play
a fundamental role in phonological descriptions.1 Vowels, for example, are
described as syllabic (in themselves, they may constitute syllables), acous-
tically sonorous, and produced with an unimpeded airstream, in functional,
acoustic, and articulatory terms, respectively. Consonants, in contradistinc-
tion, are nonsyllabic, acoustically less intense, and produced with some type
of blockage in the articulatory passage. In addition to giving content to the
segmental symbols of (1) and (2) (or to the more general classes of vowels and
consonants), features play another exceedingly important role in phonologi-
cal descriptions: they allow sounds to be grouped into classes on the basis of
shared properties. In much of the phonology, it is classes or groups of sounds,
rather than individual sounds, that behave in a unied way, undergo certain
processes, or dene the context for certain phonological operations. In SF,
for example, all high vowels, not just /i/, become glides under certain circum-
stances; mid vowels (higher- and lower-mid alike) are subject to certain distri-
butional restrictions; all nasal consonants may block the appearance of preced-
ing nasal vowels; all liquids allow the denition of certain syllable types; and
so on. As a result, we will make use of the articulatory features of (1) and (2)
(high, nasal, velar, fricative, etc.), plus others as the need arises, in the descrip-
tive chapters to follow.
Basic Descriptive Units and Domains 23
3.1 3.1 The Syllable

Segments are grouped together into the next higher unit in a phonological
hierarchy: the syllable. Syllables, too, play a fundamental role in descriptions,
since the distribution of segments is normally dened with respect to their
position in the syllable (plus, on occasion, with respect to additional units) and
since the syllable is normally the unit referred to for the assignment of stress or
tone. Vowels, since they are by denition syllabic, may constitute syllables by
themselves (e.g., o /u/, on /n/, /a/, un //, etc.). In fact, the number of syl-
lables in a (phonetic 2) sequence is equal to the number of vowels it contains.
Going beyond this minimum, things become more elaborate, since complex
syllables are formed by the conjunction of vowels and consonants. (Except
in rare and peripheral circumstances, no SF consonant may form a syllable
by itself.) Within syllables, consonants may precede the vowel (the syllable
nucleus), in which case the syllable is open (vie /vi/, vous /vu/, etc.), or follow
it, in which case the syllable is closed (le /il/, houle /ul/). Needless to say, the
nucleus may be both preceded and followed by one or more consonants (vite
/vit/, vote /vut/, tarte /tat/). The distinction between open and closed syl-
lables is signicant in SF, which is often said to be an open syllable language
(i.e., one that favours open syllables). While this is no doubt true (as it is true
of the majority of languages, open syllables being the most frequent cross-
linguistically), we will also see certain productive processes in French that
create closed syllables. Be that as it may, certain key phenomena in French,
particularly distributional constraints on vowels, depend on the open or closed
nature of the syllables involved.
Syllables are structured into a beginning, middle, and end, technically an
onset, (vocalic) nucleus, and coda. This structure may be diagrammed as in (3),
where x represents a segment occupying the syllable position in question.
(3) A simple syllable structure

m
onset nucleus coda
x x x
It is likely, however, that syllables have a more complicated, hierarchical

structure in which the nucleus and the coda group together to form a rhyme,
as in (4).
24 Chapter 3
(4) A hierarchical syllable

m
rhyme
onset nucleus coda
x x x
Rhymes are necessary because the nucleus and coda often function together as
a single unit, and many phonological processes make reference to the structure
of the rhyme as a signicant entity (e.g., in discussions of heavy versus light
rhymes or of phonotactic restrictions).
Finally, the onset, nucleus, and coda may themselves be complex, or branch-
ing, in terms of the type of diagrams in (3) and (4). The degree of complexity
of each unit is limited, as is the nature of the segments that may occur there
(consonants but not glides are excluded from the nucleus, for example). A
more elaborated syllable structure is given in (5), although the number of
branches emanating from each node remains to be determined.
(5) A more complex syllable
rhyme
onset nucleus coda
x x... xx x x...
A central task of any phonological description will be to identify possible

syllable types in the language in question, both in terms of general structure
as outlined in (5) and in terms of the specic combinations of vowels, semi-
vowels, and consonants permitted. We will pursue the rst subject below; the
second will be surveyed in the discussion of vowels and consonants in chap-
ters 4 and 5 respectively. We should rst note, however, that general structures
such as (5) offer additional possibilities for phonological representation, pos-
sibilities that lead us into the realm of the abstract.
In considering the onset or coda positions, many phonologists allow empty

nodes to occur in lexical representations, such nodes subsequently being lled
in the course of phonological derivations. One such case from French involves
latent or liaison consonants, those consonants which are normally silent but
pronounced when the following word begins with a vowel, and which may
move from the coda position of one word to an empty onset of the following
word, other conditions also being satised. 3 In such approaches, not only may
syllabic positions not be lled, it is also possible to have specic segments
(represented by x in the structures above) appear in representations without
being linked to syllabic nodes. If such segments are not subsequently linked
to some node in the course of phonological operations, a general convention
will delete them. Liaison consonants, in this view, would be unlinked when
liaison takes place, these consonants become attached to the onset of the fol-
lowing word. In the absence of liaison, they disappear. Schwa, as well, has
been treated as an unlinked vowel, deleted if no linking to a nucleus occurs,
pronounced if there is a node for it to attach to. Much discussion continues
about these enriched representational possibilities, but for our purposes here
it will sufce to proceed to a description of possible syllables in SF without
empty nodes or unlinked segments. Examples of French syllables are given in
(6), where C, V, and G represent consonants, vowels and glides, (semi-vowels)
respectively.
(6) French syllable types (nal schwa is not pronounced)

V /u/ ou or
CV /vi/ vie life
CCV /t'/ trs very
CCCV /sky/ scru(pule) scruples
VC /nt/ honte shame
VCC /'t/ tre to be
VCCC /ast/ astre star
CVC /pa/ par by
CCVC /piz/ prise taken
CCCVC /st's/ stress stress
CVCC /pnt/ porte door
CVCCC /mikst/ mixte mixed
CVCCCC /d'kst/ dextre right side
CCVCC /p'sk/ presque almost
CCCVCC /stikt/ strict strict
26 Chapter 3
GV /wi/ oui yes

GVC /wat/ ouate cotton
GVCC /w'st/ ouest west
VG /aj/ ail garlic
CGV /pi/ puis then
CCGV /twa/ trois three
CGVC /dj't/ dite diet
CCGVC /tit/ truite trout
CGVCC /sj'</ cierge candle
While not exhaustive, this list gives some indication that French allows a rich
variety of syllables. Such lists can be misleading, however, since they do not
distinguish between syllable types that are frequent, widespread cross-
linguistically, and unmarked and those that are not. In the former case, we
would retain for French at most the following major syllable types: V on /n/,
CV bon /bn/, CCV trop /to/, VC me /am/, VCC Est /'st/, CVC par /pa/,
CCVC blonde /blnd/, CCVCC triste /tist/, GV oui /wi/, and CGV bien /bj'/,
with open syllables and, in particular, CV syllables being considered the most
simple. Nonetheless, the examples in (6) allow for the establishment of a gen-
eral syllable template for SF, given in (7), where either of the terminal ele-
ments C or G in any position is optional.
(7) A maximal French syllable template
rhyme
onset nucleus coda
CCC GV CCC
Even here, however, further comments are needed. The template in (7) allows
for a complex nucleus including a glide preceding the vowel, but any post-
nuclear glide will be considered as part of the coda. In other words, the rst
C of the coda should be taken as also encompassing the possible inclusion of
a glide. Moreover, the internal constraints within the onset and coda are also
very severe. In triconsonantal onsets, for example, the rst consonant must
always be /s/; biconsonantal onsets are usually /s/ plus a voiceless obstruent
or a nasal, or an obstruent plus a liquid. Similar restrictions, exemplied in
some detail in chapter 5, apply to codas. Finally, we should note that language-
independent considerations also inuence syllable structure in general, the
most sonorous elements are found closest to the nucleus, the least sonorous at
the periphery of the onset or coda. Such evidence is important in dening so-
called sonority or strength hierarchies among segments. These hierarchies, in
turn, help in the denition of syllabication processes, to which we now turn.
The establishment of syllable templates identifying possible syllable types
is only one approach to syllabication. A related issue involves the division
or parsing of sequences of segments into their respective syllables. Should a
VCV sequence, for example, be syllabied V.CV or VC.V? This topic is cen-
tral to an analysis of French, since French is usually said to be an open syllable
language one that favours open syllables wherever possible. The proper divi-
sion of VCV, then, is V.CV, where the rst syllable is open. But what of more
complex cases, such as VCCV, VCCCV, or even VCCCCV sequences? The
process of syllabication here depends on several principles, outlined in (8).
(8) Syllabication principles

(a) favour open syllables
(b) respect possible onsets, as dened by the template
(c) maximize onsets, minimize codas
Let us illustrate the application of these principles in the syllabication of

progressively more complex sequences. In the VCV case, the result is clear:
V.CV, since that is the only result yielding an open syllable (and minimizing
the coda). In the case of VCCV, two possibilities arise (once the obviously
incorrect VCC.V is eliminated): VC.CV and V.CCV. Both types are found
in French, and the distinction rests on principle (8b). If CC is a possible syl-
lable onset (normally indicated by the appearance of these consonants in word-
initial position as well), then VCCV is syllabied V.CCV, maintaining open
syllabication. If CC is not a possible syllable onset, then the sequence must
be syllabied VC.CV. This distinction is illustrated in (9).
(9) Contrasting syllabications

V.CCV VC.CV
aprs a.p' arpent a.p# *.pV
after acre (approximately)
attribut a.ti.by artiste a.tist *.tV
attribute artist
28 Chapter 3
acropole a.kn.pnl arcade a.kad *.kV

acropolis arcade
abrutir a.by.ti arboral a.bn.e.al *.bV
to exhaust arboreal
blouir e.blu.i alphabet al.fa.b' *.lfV
to dazzle alphabet
aplatir a.pla.ti alpin al.p' *.lpV
to atten alpine
agrandir a.g#.di acteur ak.t *.ktV
to enlarge actor
adresse a.d's accelrer ak.se.le.e *.ksV
address to accelerate
aiglon '.gln lmer l.me *.lmV
eaglet to lm
afuent a.y.# aptitude ap.ti.tyd *.ptV
tributary aptitude
We must add some supplementary restrictions to this discussion of VCCV syl-

labication, linked, at least indirectly, to the question of learned loan words.
First, obvious learned loans should be excluded from providing justication
for certain clusters. The existence of words like pneumatique, psychologie,
ptrodactyle, tmse disjunction, ctnaires comb jellies, knout knout,
phtisie consumption, and so on with initial /pn, ps, pt, tm, kt, kn, ft/ does
not justify syllabifying hypnotique, absent, capter to capture, rythmique,
affecter, strychnine or caftan as /i.pnn.tik/, /a.ps#/, /ka.pte/, /i.tmik/, /a.f'.kte/,
/sti.knin/, or /ka.ft#/. A VC.CV syllabication is clearly justied in these
cases, and some common-sense means of excluding rare and peripheral forms
from a central role in the phonology is warranted.
More signicantly, the widespread occurrence of initial /sC/ clusters (sport,
stop, scolaire, smoking dress suit, snob, sphre, slip underwear) raises prob-
lems for syllabication, since, despite the frequency of many of these forms
and the general prevalence and acceptability of /sC/ sequences word-internally
as well as initially, the preferred syllabication appears to be Vs.CV: espoir
hope /'s.pwa/, mystre mystery /mis.t'/, mosque mosque/mns.ke/,
and so on. (In word-internal positions, /s.C/ is consistently used by Warnant
[1987], for example.) One piece of evidence supporting this syllabication
might be claimed to be found in the behaviour of the vowels /o/ and /n/. Since
/n/ appears to be favoured over /o/ in unstressed closed syllables in SF, the
obligatory appearance of /n/ in such forms as bosquet grove /bnsk'/, costaud

strong /knsto/, costume /knstym/, hostile /nstil/, postal /pnstal/, etc. could
argue for a /s.C/ syllabication here. But even this argument is weakened by
the appearance of both /o/ and /n/ in a few forms like caustique, cosmos, and
so on, and the obligatory /o/ in words spelt with < au > such as saut-de-lit
negligee and sauvegarde safeguard, all with /o/ in closed syllables since
/.dl/ and /.vg/ are impossible syllable onsets. More importantly, since /n/ is also
possible in open syllables, nothing prohibits the syllabication /bn.sk'/, and
so on. However, what remains to be explained in this latter case is the impos-
sibility of /o/ in many words like postal, since /o/ is normally permitted in such
structures.
Nor do the related mid vowels /e '/ help here, despite the frequent claim
that /e/ is excluded from closed syllables in French. 4 If such an exclusion
were valid, we would not expect to nd /e/ in words such as festin if these
words were syllabied /es.CV/ since they would violate the exclusion of
/e/ from closed syllables. Hence, the argument goes, the proper syllabication
is /V.sCV/. Unfortunately for this proposal, SF does contain many words
of the form /esCV/ rather than /'sCV/, whether syllabied /e.sCV/ or /es.CV/,
and the behaviour of the mid vowels is consequently irrelevant in discussions
of the syllabication of /sC/ clusters. 5 It is perhaps reassuring to know that
cross-linguistically /s/ (and on occasion other fricatives) raise similar prob-
lems with respect to the structure of consonantal clusters, phonotactic restric-
tions, and the apparent violation of expected patterns.
The preceding discussion has dealt with VCCV syllabication and illus-
trated the role of principles (8a-b). It is when we turn to VCCCV sequences
that the role of (8c) is most evident. Consider the words in (10).
(10) Word-internal triconsonantal clusters 6

portrait portrait
ltrer to lter
surplus surplus
surplis surplice
spectral spectral
perdrix partridge
cercler to ring
directrice director (f).
meurtrier deadly
arbrisseau shrub
marbrer to marble, mottle
administrer to administer
30 Chapter 3
orchestral orchestral
escrime fencing
esclave slave
esprit mind
muscl muscular
embarquement boarding
forcen deranged
harcel harassed
Here we are confronted with two possibilities: VC.CCV or VCC.CV. Often,

the CC. will form a perfectly acceptable nal cluster and .CC an acceptable
initial cluster. In orchestral, for example, nal /Vst./ occurs in ouest, test,
veste, etc., while initial /.tV/ is widespread (trs, trois, trop, and so on).
Principle (8c) dictates the syllabication /n.k's.tral/, however, since this
results in a minimal coda at the expense of a more complex onset. The other
forms behave in an analogous fashion: if both a complex coda and a complex
onset are possible, 7 minimize the coda and choose the complex onset.
We see, then, how templates and syllabication principles interact to dene
both possible syllables in French and the proper division of longer sequences
into appropriate syllabic units. Before moving to the next section, we should
consider briey one further potential unit in the phonological hierarchy, a
combination of syllables into a unit known as the foot. Feet are well known
in poetry, where they are used to represent relationships between strong and
weak (accented and unaccented) units in accounting for the rhythmic patterns
of sequences: iambic feet start with weak syllables; trochaic feet with strong,
for example. The foot has been extended into phonological analyses to indi-
cate not just strong and weak rhythmic patterns but more general relationships
between strong and weak units so-called full vowels versus schwa in French,
to name one obvious possibility. However, because SF is usually seen as a
language where syllables are isochronous and of equal strength (except for the
last syllable in a phonological phrase), the foot is of less relevance to our cur-
rent discussion, and we will not use it further here. 8 This allows us to pass to
a consideration of the word in French.
3.2 3.2 The SF Phonological Word

In his classic article Le mot est-il une entit phontique en franais, Pierre
Delattre questions the necessity of the word as a phonological unit in French.
The question arises because two of the fundamental phonological attributes of
SF, stress placement and the operation of liaison and enchanement (the link-
ing of a word-nal consonant with the initial vowel of a following word), are
apparently dened in terms not of words but of phonological phrases. Stress

in French, for example, normally falls on the last syllable of the phonological
phrase, and words in nonnal position have no word stress independently of
that affecting the last syllable of the phrase. When liaison and enchanement
are considered, we also see that the word loses its independence, in that word-
nal consonants are resyllabied to the beginning of the following unit. This,
too, is seen as demonstrating lack of independence for the word in French and
minimizing its importance in the description of SF.
Recent work, however, some of which we will discuss in following chap-
ters, indicates that in spite of Delattres concerns, the word must be retained
as a unit in the description of French phonology.9 Many phonotactic con-
straints, for example, make reference to the word, such constraints being vio-
lated within larger units. Thus, if we consider words in isolation, nasal vowels
do not normally precede nasal consonants in French (an isolated word such
as /mnn/ is ill-formed),10 but /mnn/ occurs both in liaison (mon ami my
friend /mnnami/) and in phrases (mon nez my nose /mnne/ (compare Monet
/mnne/)). Various phonological processes glide formation and the behaviour
of nasal vowels in particular also refer to the notion of phonological word, as
Hannahs (1995b) has demonstrated.11 Thus, high vowels may become glides
when followed by another vowel within words (il joue hes playing /il<u/
vous jouez youre playing /vu<we/) but not when the second vowel is in a
separate word (il joue encore hes still playing /il<u#kn/, not */<w#kn/).
In addition to the word, phonological descriptions may also use a constitu-
ent known as the clitic group. Typically, clitics exhibit phonological behav-
iour intermediate between that of afxes and independent words. They are, for
example, more syntactically independent than afxes but are attached phono-
logically to their host words. Subject and object pronouns in French (je, tu, le,
les, lui, etc.) are characteristic examples of clitics. Clitic groups are larger than
phonological words but smaller than phrases. Because the potential effects of
clitic groups are still not rmly established in analyses of SF (pace Hannahs
1995a) and because such units operate in ways highly similar to phonological
phrases, we will not use the concept here, but work instead with the next and
perhaps most fundamental unit in French phonology, the phonological phrase.
3.3 3.3 The Phonological Phrase

In a prosodic hierarchy leading from the syllable through the foot, word, and
clitic group to the phonological phrase, the phrase plays a major role in any
description of French. As we will see, a number of the prototypical aspects
of French phonology stress and rhythm, vowel length, syllabication, the
behaviour of schwa, intonation patterns are described with reference to
phrasal boundaries. In general terms, as the name itself suggests, 12 phonologi-
cal phrases are larger than words, and manifest both grammatical (syntactic
and/or semantic) and phonological relationships. Phonological phrases thus
make a link with grammatical structure, although they are not always perfectly
32 Chapter 3
aligned with it. That being said, we will use both phonological and grammati-
cal attributes in discussing the phonological phrase, identifying rst the mini-
mal phrases that is, those phrases that may not be broken down into smaller
(phrasal) units.
In phonological terms, phrases are pronounced between pauses (at least
potential pauses), are characterized by nal stress and specic intonational
patterns, and rarely exceed six syllables in length. Over longer stretches of
speech, moreover, there is a tendency to have phrases of roughly equal length
(cf. Wioland [1991: 3738]). In grammatical terms, phrases are correlated
with major lexical categories (nouns, verbs, adjectives) and their modiers or
dependents (articles, pronouns, adverbs, etc.). These are the units that consti-
tute minimal phrases those that may not be subdivided, even in slow speech.
They are illustrated in (11).
(11) Phonological phrases

(a) verb-centred: single verbs (including auxiliaries, present
participles, or innitives) plus accompanying subject or object
pronouns and negations
je pars Im leaving.
le voit-il Does he see it?
on vous en parlera Theyll speak to you about it.
je ne lavais pas I didnt have it.
je te les envoie Im sending them to you.
donne-lui-en Give him some.
ne lui en donne pas Dont give him any.
nous ne le voulons plus We dont want it any more.
ne ten fais pas Dont worry about it.
le sachant knowing it
me le rpter repeat it to me
ne plus me le rpter not repeat it to me any more
nen sachant plus rien not knowing any more about it
ne vous en donnera-t-il pas Wont he give you any?
(b) noun-centred: nouns plus preceding (not following) determiners or
modiers
un lment an element
trois chiens three dogs
ses voisins his neighbours

mes deux enfants my two children
leurs anciens camarades their former comrades
quelques instants a few moments
plusieurs beaux tableaux several pretty paintings
de bons amis good friends
toutes ces petites maisons all these little houses
(c) adjective-centred: adjectives plus preceding adverbs
trs lgant very elegant
bien aimable very nice
assez complexe rather complex
toujours prsent always there
souvent en retard often late
vraiment mesquin really stingy
rcemment mari recently married
pas trs loin not very far
si petit so small
(d) prepositional phrases
sans lui parler without speaking to him
en arrivant upon arrival
pour les vrier in order to check them
aprs tous les autres after all the others
dune autre fois of another time
pour ne pas le gner in order not to bother them
pendant une heure for an hour
en me les enlevant taking them from me
(e) subordinators and complementizers plus the following word
que vous aimez that you like
quil a pris that he took
quand il viendra when he comes
an quelle le prenne so shell take it
34 Chapter 3
si on pouvait if we could
depuis que Claude est l since Claude has been there
pendant que nous dormions while we slept
parce quon lexige because we have to
dont on parlait about which we were speaking
ce quil voulait what he wanted
These minimal phrases are so named because (in normal speech) they must
be pronounced as single units without interruption. For je ne le vois pas, for
example, one cannot say *jene le vois pas, je ne levois pas, *je ne le vois
pas, and so on (where indicates a pause), or *trsimportant, *la pro-
chainelection, *sansy aller, *quandil viendra, and so on. Phrases of
varying syllable length, minimal and non-minimal, are given in (12).13
(12) Phonological phrases of increasing syllable length

(1 syllable) o where
oui yes
tiens hold (this)
zut drat
(2 syllables) bonjour hello
jamais never
part-il Is he leaving?
au secours Help!
tout de suite right away
demain See you tomorrow.
daccord Okay.
(3 syllables) attention attention
dans une heure in an hour
cest toi Its yours.
quand vient-il When is he coming?
chez le mdecin at the doctors
rappelez-moi Call me back.
samedi until Saturday
oui daccord yes, Okay.
(4 syllables) cest impossible Its impossible.

on va se promener Were going for a walk.
jallais me baigner I was going to go swimming.
la semaine passe last week
gnralement generally
je ne le vois pas I dont see it.
pendant une heure for an hour
merci beaucoup Thanks a lot.
cest a nest-ce pas Thats it, isnt it.
(5 syllables) veuillez mexcuser Please excuse me.
jai mal la tte I have a headache.
cest au bout de la rue Its at the end of the street.
ne pars pas tout de suite Dont leave right away.
en avez-vous trop Do you have too much?
aux Etats-Unis in the United States
cet avertissement this warning
(6 syllables) rendez-vous midi meeting at noon
il faudra en parler We have to speak about it.
pourquoi pas tout Why not in a little while?
lheure
vous ny arriverez pas You wont get there.
luniversit at the university
il y fait toujours froid Its always cold there.
considrablement considerably
As mentioned above, not all of these phrases are minimal phrases. Minimal
phrases may be combined into longer sequences depending on a variety of
factors, among which are the rate of speech (faster speech favours longer
phrases), stylistic considerations (formal speech favours shorter phrases), and
emphasis (phrase breaks may be used to mark items to be emphasized).
Needless to say, a further controlling ingredient involves length: as we have
seen, phrases rarely exceed six or seven syllables under normal circumstances.
Again, under normal circumstances, the divisions between phrases will cor-
relate with major syntactic divisions, as in (13) where the symbol (||) is here
used to indicate an optional break.
36 Chapter 3
(13) Longer phrases composed of two (or more) minimal phrases 14

(a) verb nucleus plus verb
Je ne lavais pas (||) entendu. I hadnt heard it.
Il prtend (||) ne pas le vouloir. Hes pretending not to want to.
Tu devrais (||) lui en parler. You should speak to him
about it.
Veuillez (||) ne pas me dranger. Please dont bother me.
(b) noun phrase plus verb or vice-versa
Le match (||) sest bien termin. The game ended well.
Le professeur (||) nen sait rien. The professor knows
nothing about it.
Il nous raconte (||) des btises. Hes telling us nonsense.
Elle promet (||) de nous suivre. She promises to follow us.
Connais-tu (||) ce jeune homme? Do you know this young man?
(c) noun plus following modier
un journal (||) hebdomadaire a weekly newspaper
une maison (||) de campagne a country house
lautobus (||) de la ville the city bus
un appartement (||) louer an apartment to rent
une histoire (||) pas trs amusante a not very funny story
(d) verb plus following modier
Elle me le dit (||) assez souvent. She tells me quite often.
On vous rappellera (||) Well call you today.
ds aujourdhui.
Nous lui en parlons (||) We talk to him about it
constamment. constantly.
Il faut y aller (||) tout seul. You have to go there alone.
Ils se parlent (||) avec difcult. They speak to each other
with difculty.
The decision to combine shorter phrases into one longer one may have addi-
tional phonological consequences. To begin with, there will be only one
stressed syllable rather than two: un appartement || louer an apartment to
rent /napatm# || alwe/ versus un appartement louer /napatm#alwe/.
Moreover, since syllabication does not usually cross phrase boundaries,

removal of a boundary may require resyllabication (enchanement or liai-
son): Paul lavait vite || oubli Paul had quickly forgotten it. /pnllav'vit ||
ublie/ versus Paul lavait vite oubli /pnllav'vi.tublie/ or je ne lavais pas
|| entendu I hadnt heard it. /<nlav'pa || #t#dy/ versus je ne lavais pas
entendu /<nlav'pa.z#t#dy/.15 Finally, deletion of schwa is sensitive to phrase-
initial versus phrase-internal position, so joining two phrases into one may
often result in schwa deletion: elle promet || de nous suivre She promises
to follow us. /'lpnm' || dnusiv/ versus elle promet de nous suivre
/'lpnm'dnusiv/. We will discuss each of these, plus additional phrase-
related phenomena such as intonation patterns, in more detail in later
chapters.
3.4 3.4 Conclusion

This chapter has introduced the primary elements to be used in the descrip-
tion of SF that follows. These elements include the set of segments or pho-
nemes of SF, dened in terms of articulatory features. Features, in turn, serve
to dene larger or smaller classes of sounds on the basis of shared properties
all vowels, nasal vowels, voiceless fricatives, liquids, and so on. Such classi-
cations are crucial in phonological descriptions because classes of segments,
rather than individual sounds, typically participate in the general phonological
processes of a language or nd themselves subject to phonotactic constraints.
Classes also serve to dene the segmental contexts in which these phonologi-
cal processes occur.
Processes or constraints are not dened exclusively in segmental terms,
however. We will often need to refer to domains larger than those of individual
segments or classes of segments. Segments, in other words, are part of a pho-
nological or prosodic hierarchy which includes (for our purposes) syllables,
words, and phrases. Many constraints on the distribution of segments in SF
refer to syllable or word structure, for example, and the prototypical phenom-
ena of liaison and enchanement are dened crucially with respect to sylla-
bles as well. Finally, as the preceding section has indicated, the phonological
phrase is of utmost importance in any description of French. Let us begin this
description with a discussion of the vowel system.
Notes
1. In fact, from certain points of view features may properly be seen as the funda-
mental units of phonological analysis. Feature theory is a specialized domain in
its own right, one we cannot hope to examine in detail here. For some discussion,
see Clark and Yallop (1990), Clements and Hume (1995), Davenport and Hannahs
(1998) or Kenstowicz (1994).
38 Chapter 3
2. The qualication phonetic is needed because certain more abstract representa-

tions may represent semi-vowels as vowels, the latter subsequently being modi-
ed. Note also that, unlike their English counterparts (e.g., bottle, butter), French
liquids may not be syllabic.
3. For a detailed treatment of liaison, see Section 5.6.1.
4. For such a claim, see Price (1991: 56). Tranel (1987: 52) also excludes /e/ from
nal closed syllables, a claim countered by such loan words as cake /kek/ or mail
/mel/. We will return to this matter in the discussion of mid vowels in chapter 4,
section 4.2.
5. Martinet and Walter (1973) contains many examples showing both /e/ and /'/ in
such contexts, including exprs specially /eksp' 'ksp'/, exploit /eksplwa
'ksplwa/, exact /egza 'gza/, where the syllabication can hardly be /e.ksp'/,
/e.ksplwa/, or /e.gza/.
6. We will ignore the less frequent VCCCCV possibilities (e.g., extra- rst-rate
/'ksta/, abstrait abstract /apst'/, expliquer to explain /'ksplike/), which
allow for either VC.CCCV or VCC.CCV (/'k.sta/ or /'ks.ta/, /ap.st'/ or
/aps.t'/, /'k.spli.ke/ or /'ks.pli.ke/), depending on where /s/ is assigned. Principle
(8c) favours the former option. We should also note that in popular speech, words
like expliquer and exprimer to express are reduced to espliquer and esprimer,
with simplication to a triconsonantal cluster.
7. This condition leads to a discussion of embarquement (where, as in forcen and

harcel as well, the schwa has exceptionally deleted to create the triconsonantal
cluster). In embarquement /#bakm#/, a syllabication /#.ba.km#/ is impossible
because /km/ is not a possible onset. Here, then, we must complicate the coda:
/#.bak.m#/. This data is taken from Juilland (1965).
8. See, however, Selkirk (1978) or Bullock (1995) for work that makes crucial use
of the foot in analysing aspects of French phonology the behaviour of schwa in
particular.
9. Lyche and Girard (1995), in particular, revisit this question in detail.
10. Exceptions such as ennui boredom /#ni/ or vnmes we came /v'm/ are dis-
cussed below in Section 4.4.2.
11. See also Rochet (1977) or Lyche and Girard (1995). It is perhaps worth recalling
that the concept word is complex and notoriously difcult to dene in cross-
linguistic terms (cf. Trask [1999: 34244] for a concise discussion). It is necessary
to distinguish, particularly in French, between orthographic words (those written
between white spaces) and phonological words (those pronounced as single
units), with the latter of primary importance. For our purposes here, Hannahs
denition of the French phonological word as consisting of either prexes or
stems plus all associated sufxes is most useful.
12. Also known in the literature on French as groupe rythmique, groupe accentuel,
groupe de soufe, groupe respiratoire, groupe intonationnel, albeit with occa-
sional subtle distinctions among the various units. If the syntactic or semantic
underpinnings of the unit are involved, the term groupe syntaxique is also found
(e.g., in Bchade [1992: 60]).
13. Many of these examples are from Wioland (1991: 89), a work that provides
excellent material concerning phonological phrases. Note the discrepancy in these
examples between the orthography and the pronunciation arising from the dele-
tion of schwa, a topic to be discussed in some detail in chapter 5.
14. Examples are based primarily on material from Mueller et al. (1968: 4445).
15. Such liaison examples can be contradicted by liaison sans enchanement (Encrev
1988), as in /<nlav'paz#t#dy/ or by parenthetical insertions (un robuste || mais
petit || enfant a sturdy, but small, child /nbyst | m'pti | t#f#/ [Pichon 1938:
123]). In the rst instance, no resyllabication has taken place, we simply see
the exceptional appearance of a liaison consonant in an abnormal position in a
marked style. In the second, the phrase boundary (and associated pause) remain,
but a liaison consonant still appears in the appropriate syllabic (phrase-initial)
position.
41
Chapter 4
Vowels and Semi-vowels
4.0 4.0 The Vowel System

Traditionally, SF is described as having a system of sixteen vowels, twelve
oral and four nasal, as tabulated in (1) below (see also 3.0 above). To this must
be added the semi-vowels or glides /j w/, corresponding to the three high
vowels /i y u/ respectively.
(1) SF vowels and semi-vowels (glides)

semi-vowels L w
high K [ W
lower-mid ' n ' n
low C # #
(2) Examples (in word-level contexts)

#__ __# __C.
/i/ ivre drunk vie life vite quickly
/y/ une a vue sight jupe skirt
/u/ outre besides vous you vote vault
1
/e/ t summer fe fairy cake cake
// euro Euro jeu game jene fast
/o/ autre other sot silly saute sudden change
/'/ tre to be mais but dette debt
// heure hour jeune young
/n/ or gold fort strong
42 Chapter 4
/a/ art art chat cat patte paw

/#/ ge age mt mast pte pastry
// (sur) ce this
/'/ impair uneven n end sainte holy
// humble humble brun brown emprunte he borrows
/n / ongle nail bon good honte shame
/# / entre between dent tooth banque bank
#__ C__ V__V __#

/j/ iode iodine bien well balayer sweep paille straw
// huit eight puis then 2
/w/ ouest west couette bunches
In comparative or typological terms, this vowel system is relatively rich. In

particular, it requires four degrees of tongue height and contains both front
rounded vowels and nasal vowels.3 All vowels, with the exception of //
(schwa, to which we return in detail below), are tense and nondiphthongized.
There exist a number of complex phonological problems associated with this
system, problems that we will treat in turn.
4.1 4.1 Vowel Length

Historically and dialectally, long vowels have played and continue to play a
signicant role in French. In SF, however, their role is signicantly reduced.
Long vowels are largely contextually determined, and the remnants of an earlier
long-short opposition are minimal. The contextually determined long vowels
occur in stressed closed syllables. Stress and syllable structure, in other words,
provide the two fundamental conditions that must be present for vowel length
to appear. Given the close relationship between stress and phonological phrases,
moreover, long vowels are further restricted to phrase-nal position in SF.4
(3) Constraints on contextually determined long vowels

Long vowels occur in:
(i) stressed syllables
(ii) closed syllables
If the foregoing conditions are met, two types of long vowels occur in SF: (i)
those induced by the lengthening consonants /v z < /5 and the cluster /v/, and
(ii) the intrinsically long vowels / o #/ plus the nasal vowels /'n#/.
Vowels and Semi-vowels 43
4.1.1 4.1.1 Lengthening Consonants

The effects of the lengthening consonants are shown by the following con-
trasts:6
(4) Lengthening versus nonlengthening consonants

/v f/ vive [vi:v] lively (f.) vif [vif] lively (m.)
grave [ga:v] solemn graphe [gaf] written form
/z s/ vise [vi:z] she aims vice [vis] vice
rase [a:z] he shaves race [as] race
/<5/ bouge [bu:<? she moves bouche [bu5] mouth
cage [ka:<] cage cache [ka5] he hides
juge [<y:<] judge ruche [y5] hive
/ l/ vire [vi:] ledge ville [vil] city
pour [pu:] for poule [pul] hen
mre [m':] mother mle [m'l] he mixes
/v f/ vivre [vi:v] to live chiffre [5if] gure
ouvre [u:v] she opens souffre [suf] he suffers
If the syllable containing a coda that begins with one of these consonants
is not word-nal, no lengthening occurs: berceau [b'.so] cradle, partir
[pa.ti:] to leave, guirlande [gi.l#:d] garland, for example, all have short
rst vowels.
With the exception of the cluster /v/, there is one further constraint on the
lengthening process: any additional consonant following a lengthening conso-
nant blocks the appearance of a long vowel. Thus, despite the presence of //
and even a second lengthening consonant, the vowel is short in words of the
rst column in (5) (compare the lengthened vowels of the second column):
(5) Blocking of vowel lengthening

moderne [mnd'n] modern modre [mnd':] she
moderates
verte [v't] green (f.) vert [v':] green (m.)
herbe ['b] grass air [':] air
auberge [ob'<] inn au pair [op':] au pair
rserve [ez'v] reserve serre [s':] greenhouse
cierge [sj'<] candle er [fj':] proud
44 Chapter 4
parte [pat] leaves (subj.) part [pa:] leaves (ind.)

quatorze [katnz] fourteen tort [tn:] wrong;
orge [n<] barley or [n:] gold
courte [kut] short (f.) court [ku:] short (m.)
lourde [lud] heavy (f.) lourd [lu:] heavy (m.)
purge [py<] purge pur [py:] pur (m.)
4.1.2 4.1.2 Intrinsically Long Vowels

As noted above, the so-called intrinsically long vowels in SF include / o #/
and the nasal vowels /'n#/. These vowels are lengthened in stressed (i.e.,
word-nal) closed syllables, no matter what the nature of the consonant clos-
ing the syllable, as indicated in (6), with a set of contrasting open syllable
examples included as well.
(6) Intrinsically long vowels

// jene [<:n] fast jeu [<] game
/o/ saute [so:t] sudden change saut [so] jump
/#/ pte [p#:t] pastry dgt [deg#] damage
/'/ feinte [f':t] feint n [f'] end
// dfunte [def:t] deceased (f.) dfunt [def] deceased
(m.)
/n/ honte [n:t] shame on [n] one
/#/ rampe [#:p] ramp rang [#] row
Given this phonological behaviour, there have been various attempts to dene
a natural class of intrinsically long vowels in SF on the basis of a shared pho-
netic feature or property (e.g., vowel tension). It would appear, however, that
these efforts will remain unsuccessful, and that the heterogeneous nature of
this class of vowels is to be attributed to a number of independent historical
events such as nasalization and compensatory lengthening, to which we will
refer briey below. Finally we should note that although the two higher-mid
vowels // and /o/ are instrinsically long, the third member of this class, /e/,
is rare in nal syllables, occurring virtually exclusively in relatively recent
English loan words, where it may be pronounced either long or short. There
may be, in other words, a principled reason for this exclusion. In SF, unlike
other dialects such as CF, /e/ is highly marked in nal closed syllables. If there
are no native SF words in /XeC./ (note the pronunciation of the English loan-
words discussed in footnote 36 above), the conditions for /e/ to be realized as

an intrinsically long vowel would not arise word-internally.7
4.1.3 4.1.3. The /' '/ Opposition

Conservative SF preserves one length distinction in a limited number of forms
between a long /'/ (excluding those lengthened by a following lengthening
consonant) and short /'/ (or all other short vowels). The list of words with /'/
varies from author to author, but normally contains some 250 forms (Walter
1976: 126133), among which the most notable include the following:
(7) Long /'/

/'/ /'/
ble bleats belle beautiful
bte animal bette chard
blme pallid
brme bream
caisse box
chane chain chne oak
conqute conquest
crte crest
enqute inquiry
tre (n.) being tre (v.) to be
vque bishop
fate summit faites you do
fte holiday
fentre window
gne discomfort
grle hail
matre master mtre/mettre metre/to put
natre to be born
patre to graze
pche peach pche shing
prtre priest
46 Chapter 4
reine queen Rennes (city name)

scne scene
tte head tte sucks
tratre traitor
Unlike the two preceding types of long vowels, long /'/ is distinctive in SF,
and as a consequence, examples of the vowel can be found in unstressed syl-
lables, normally provided the syllable is word-nal.8 Examples of this occur-
rence are found in (8).
(8) Unstressed (phrase-internal) long /'/

cest une fte importante Its an important holiday.
un matre-chanteur a black-mailer
il a reu une grle de balles He suffered a hail of bullets.
This opposition is increasingly unstable in SF, the long vowel being excep-
tional, so many speakers now have only one vowel /'/, the short version, with
the words in (7) therefore showing assimilation to the regular pattern.
To conclude this description of vowel length in SF, we may note that length
is not cumulative. That is, intrinsically long vowels or long /'/ followed by
a lengthening consonant are not doubly long; length in rose [oz], chaise
chair [5'z] or genre kind [<#], in other words, is comparable to that in
hausse rise [os], chane [5'n] or branle swing [b#l]. The following sen-
tences, where length appears only in the nal syllable of the phrases, illustrate
the role of stress in conditioning vowel length.
(9) Stress and vowel length

il faut y vivre [viv] You have to live there.
il faut vivre l [vivla] You have to live there.
regarde comme elle saute [sot] Look how she jumps.
quest-ce quelle saute bien [sotbj'] Doesnt she jump well?
il est rouge [u<] Its red.
il est rougetre [u<#t] Its reddish.
quelle honte [nt] What a disgrace.
quelle honte profonde [ntpnfnd] What an awful disgrace!
4.1.4 4.1.4 Supplementary Comments

With the exception of the /' '/ distinction, vowel length is fully predictable
in SF, so length is not customarily indicated in phonological notation. Nor are
there consistent orthographic indications of long vowels none in the case of
those arising from lengthening consonants and only marginal traces in the case
of /' '/ and the intrinsically long vowels. It may, however, be worth com-
menting on the role of the symbol , the accent circonexe. Historically, this
diacritic was introduced to indicate loss of vocalic hiatus or of an earlier con-
sonant, usually /s/. This loss normally resulted in a lengthened vowel, with the
resulting long short distinction often playing a morphological role (indicating
singular plural or masculine feminine distinctions, for example). With the
virtual elimination of these distinctions from SF in the twentieth century,9 the
link between the circonexe accent and length has also been lost, since there
are long vowels with no accent or a different accent (e.g., many intrinsically
long /o/s as in zone; the long vowels in caisse box, gne discomfort, etc.)
and accented vowels that are short (e.g., prte ready (f.)). There remain,
nonetheless, certain words where the circonexe serves as an indication of the
intrinsic length of a vowel, especially /#/, where it is an orthographic trace of
an earlier long-short distinction that has now become one of vowel backing,
or where it reminds us of etymologically related (learned) words that retain the
earlier consonants.10 Each of these cases is illustrated in (10).
(10) The role of the accent circonexe

(a) < > as an indication of /#/11
ge /#</ age
pre /#p/ bitter
bt /b#/ he beats
grce /g#s/ grace
mt /m#/ mast
ple /p#l/ pale
pte /p#t/ pastry
(b) orthographic distinctions
cne /kon/ cone conne /knn/ stupid (f.)
cte /kot/ coast cote /knt/ mark
hte /ot/ host hotte /nt/ basket
jene /<n/ fasts jeune /<n/ young
le ntre /lnot/ ours notre /nnt/ our
tche /t#5/ task tache /ta5/ mark
48 Chapter 4
(c) learned nonlearned pairs

pre /#p/ bitter asprit /aspeite/ bitterness
bte /b':t/ animal bestial /b'stjal/ bestial
crote /kut/ crust croustillant /kustij#/ crusty
fte /f':t/ holiday festoyer /f'stwaje/ to feast
fort /fn'/ forest forestier /fn'stje/ forest (adj.)
got /gu/ taste dgustation /degystasj/
sampling
hpital /npital/ hospital hospitalier /nspitalje/ hospital
(adj.)
le /il/ island insulaire /'syl'/ insular
matre /m':t/ master magistral /ma<istal/ masterly
mle /m#l/ male masculin /maskyl'/ masculine
Finally, we should note that vowel length has different manifestations in dialects
other than SF. On the one hand, the longshort distinction is preserved in certain
regions (especially where there is a following schwa); on the other, the phonetic
realization of vowel length may be dramatically different. CF, for example, pre-
serves the /' '/ distinction much more rmly than does SF, and long vowels
(which may also appear in nonnal syllables) are characteristically realized as
diphthongs.12 Historically and dialectally, in other words, vowel length in French
is signicantly more complex than this SF survey would lead us to believe.
4.2 4.2 The Mid Vowels

The set of mid vowels in French is composed of three related pairs: /e '/,
/ /, and /o n/. In SF, these vowels are subject to a variety of distribu-
tional constraints (or neutralizations) and exhibit, in certain contexts, a per-
plexing variability. (The situation in some other dialects, Midi French (MF) for
example, is much more straightforward.) We will rst consider a conservative
standard pronunciation before reviewing additional questions.
The two contextual factors that play a role in constraining the mid vowels
involve position in the word and syllable structure: nal versus nonnal (or
stressable versus unstressable) syllables and open versus closed syllables. In
combination, these factors give four syllable types to be considered: nal
open, nal closed, nonnal open and nonnal closed. These four possibilities
and the vowels that appear in each are summarized in tabular form in (11).
(11) Distribution of the mid vowels

Nonnal Final
Open e o e o
' n '
Closed (e) () (e) o

' n ' n
4.2.1 4.2.1 Mid Vowels in Final Open Syllables

In nal open syllables, there is an opposition between the front unrounded
vowels /e '/, but among the rounded vowels, only higher-mid // and /o/ are
permitted; lower mid // and /n/ do not occur in word-nal open syllables in
SF.13 The exclusion of // from this context accounts for the alternation we
see in pairs of words like oeuf egg /f/ oeufs eggs //, boeuf ox /bf/
boeufs oxen /b/, oeil eye /j/ yeux eyes /j/, while the exclusion of /n/
yields alternations like sotte silly /snt/ sot /so/, chaotique /kantik/ chaos
/kao/, and so on. The /e '/ distinction is of great morphological importance,
since it serves to distinguish, among others, certain future forms from the
corresponding conditionals (donnerai donnerais will give would give),
the innitive and past participle of rst conjugation verbs from the imperfect
(donner / donn donnait / donnaient to give/given was/were giving), and
so on. There are also numerous words, illustrated in (12), that illustrate the dif-
ference between the two vowels in nal open syllables.
(12) /e/ versus /'/ in nal open syllables

/e/ /'/
clef key craie chalk
donner to give cossais Scottish
t summer effet effect
poigne handful fort forest
et and haie hedge
fe fairy fait fact
pr meadow prs near
livrer to deliver livret booklet
ouvrier worker quai quay
valle valley vrai true
50 Chapter 4
Although the match between orthography and pronunciation is far from con-
sistent in French, there are several patterns that allow for prediction of /e/
versus /'/ in this context. Final orthographic < , e, ez, er > (where the < r > is
not pronounced, as in innitives) are normally realized as /e/. Final < ai, aiC,
aC, eC or C > are normally pronounced /'/, with the exception of the words
et /e/ and and les /le(z)/ the, and the verbal ending < -ai > /e/, as in jai I
have, je mangeai I ate, je mangerai I will eat.14
For the sake of completeness, (13) illustrates // and /o/ in nal open syl-
lables in a variety of orthographic guises; recall that // and /n/ are excluded
from this context.
(13) Final // and /o/

/o/ //
beau handsome jeu game
hros hero ceux those
vos your peut she can
animaux animals noeud knot
zro zero queue tail
eaux waters voeu wish
4.2.2 4.2.2 Mid Vowels in Final Closed Syllables

In word-nal closed syllables, the situation is the converse of the preceding.
The rounded pairs of vowels, front / / and back /o n/, are distinct, while
only the lower-mid unrounded /'/ occurs, /e/ being excluded from nal closed
syllables in SF (with the exception of loanwords such as ale /el/ discussed
elsewhere).
(14) // versus //, /o/ versus /n/ in nal closed syllables

// //15
jene fast jeune young
neutre neuter seul alone
meute pack peur fear
beugle to moo aveugle blind
meugle to moo veuve widow
creuse hollow (f.) deuil bereavement
/o/ /n/
saute change sotte silly (f.)
Beaune (region) bonne good
fausse false (f.) fosse ditch
paule shoulder molle soft (f.)
paume palm pomme apple
16
cte coast cote mark
In fact, for the front rounded vowels, the number of words involved in this
opposition is very low, since the appearance of // in this context is infre-
quent. Standard references normally list no more than a dozen items, while //
occurs freely, and many speakers replace // with the lower-mid partner. There
are, moreover, more specic distributional constraints on these two vowels:
only // occurs before the lengthening consonant /z/, as in creuse or the femi-
nine derivational sufx -euse (chanteuse singer [f.]), while only // appears
before // (peur fear or the sufxes -eur as in chanteur singer [m.], gran-
deur size).17 The distinction between /o/ and /n/, on the other hand, is stable,
although analogous distributional constraints are also in evidence. Specically,
only /n/ may occur before // and // (store blind, ivrogne drunk), while
only /o/ appears before /z/ (rose).
4.2.3 4.2.3 Mid Vowels in Nonnal Closed Syllables

Before treating nonnal open syllables, we may deal easily with the least com-
plex situation: nonnal closed syllables, where the lower-mid vowels /'n/
normally appear. This neutralization of the higher-mid/lower-mid distinction
leading to the absence of /e o/ from nonnal closed syllables foreshadows
the discussion of the loi de position to occur below and is illustrated in (15).
(15) /'n/ in nonnal closed syllables

/'/ // /n/
fertile fertile heurter to strike dormir to sleep
personne person meurtrier deadly colporteur pedlar
plerin pilgrim seulement only cordial cordial
18
festival festival effeuillement leaf fall nocturne nocturnal
There is, however, one complication in this area: the possibility, for some
speakers, of /e/ in nonnal closed syllables. (This minor effect is indicated by
the parentheses surrounding the /e/ in the table in [11] above.) Such examples
52 Chapter 4
arise in three primary contexts: (i) closed syllables created by the deletion of
mute-e:19 cleri celery /sel.i/, dmesure excessiveness /dem.zy/, meri
emery /em.i/, ennemi enemy /en.mi/, vnement event /even.m#/, mde-
cin doctor /met.s'/, and so on; (ii) sequences of /esC/ (assuming
the syllabication /es.C/): espoir hope /es.pwa/, destin destiny /des.t'/,
digestion /di.<es.tjn/, festin feast /fes.t'/, gestuel gestural /<es.t'l/, mod-
estie /mn.des.ti/, and so on; and (iii) occurrences of < ex >: exact /eg.za/,
examen examination /eg.za.m'/, excellent /ek.se.l#/, exotisme exoticism
/eg.zn.tism/, exploit /eks.plwa/, extra /eks.ta/, textuel /teks.t'l/, and many
others. That this pronunciation of /e/ in closed syllables is restricted to
unstressed position is indicated by the behaviour of related words: digeste
easily digested, geste gesture, modeste, texte, where /e/ is impossible.
Moreover, not all speakers of SF accept such pronunciations, allowing only
/'/ in nonnal closed syllables. As a result, we will continue to see /'/ as the
primary realization of the mid front unrounded vowels in this context.
4.2.4 4.2.4 Mid Vowels in Nonnal Open Syllables

If one were to indicate a single standard pronunciation for the mid vowels in
nonnal open syllables, it would no doubt be /e n/, as in (16):
(16) Mid vowels in nonnal open syllables

/e/ // /n/
cder to give up pleuvoir to rain poteau post
tmoin witness jeudi Thursday pome poem
fcond fertile heureux happy modle model
gnreux generous deuxime second folie madness
However, by far the greatest amount of variability and hence complexity in the
pronunciation of the mid vowels occurs in this context, and consistent gener-
alizations are difcult to establish. Paradoxically, the pronunciation of a vowel
as higher-mid or lower-mid in such syllables has very little effect on the iden-
tication of words (unlike the distinction between saute sudden change and
sotte silly (f.) or cre create and craie chalk, for example) or on com-
munication in general. In addition to simple free variation, the realization of
the vowels in nonnal open syllables is inuenced by several further factors,
sometimes conicting, sometimes mutually supportive. These factors include
analogy (the persistent inuence of the root vowel in derived forms), vowel
harmony, and (perhaps) the loi de position. We will deal with each in turn.
4.2.4.1 4.2.4.1 Analogy

French has a relatively rich system of inectional and derivational morphol-
ogy in which sufxes are added to roots to indicate a change in inectional
category or to form new words, as in il embauche nous embauchons (root:
embauch- hire) or bleu bleuir, bleutre (root: bleu blue). If a vowel-initial
sufx is added to a stem ending in a vowel or a single consonant (as in the pre-
ceding examples), the sufx will bear the stress and the root vowel will then
be found in a nonnal open syllable.20 As indicated in (17), there is a strong
tendency in SF to preserve, through analogy, the root vowel in derived forms,
even if this contradicts the indications in (16) above.
(17) Preservation of root vowels

gu ford guable fordable /e/ /e/
th tea thire teapot /e/ /e/
laide ugly laideur ugliness /'/ /'/
beau handsome beaut beauty /o/ /o/
gros large grosseur thickness /o/ /o/
bosse hump bossu hunchbacked /n/ /n/
cole school colier student /n/ /n/
roche rock rocher rock /n/ /n/
This pressure is particularly strong when it involves the vowels / /; if the

base contains either of these vowels, that vowel is normally preserved in both
verb conjugations and derived words, as in (18).21
(18) // and // in roots

feuille leaf feuillu leafy // //
elle cueille she gathers cueillir to gather // //
il pleure hes crying pleurer to cry // //
jeune young rajeunir to rejuvenate // //
beurre butter beurrer to butter // //
il creuse hes digging creuser to dig // //
deux two deuxime second // //
elle jene shes fasting jener to fast // //
heureux happy heureusement fortunately // //
bleu blue bleutre bluish // //
54 Chapter 4
In both (17) and (18), we see that this analogical pressure has produced
forms that contradict the initial generalization in (16) above: raideur stiff-
ness /'.d/, not (or at least rarely) */e.d/; beaut beauty /bo.te/, not
*/bn.te/; feuillu broad-leafed tree /f.jy/, not */f.jy/; and so on.
4.2.4.2 4.2.4.2 Vowel Harmony

Vowel harmony is a type of assimilation in French in which the vowel of the
stressed syllable inuences the pronunciation of a mid vowel in a preceding
open syllable. If the stressed vowel is lower-mid or low (/' na #/), it will
favour a lower-mid realization of the vowel in the open syllable preceding it.
If it is higher-mid or high (/i y u e o/, especially /i y e/), it will, conversely,
favour the appearance of a higher-mid vowel. Vowel harmony is never obliga-
tory; it is characteristic of less rather than more formal speech; and it affects
primarily the front unrounded pair of vowels /e '/, although its inuence on the
other mid vowels is not unknown. Vowel harmony is illustrated in (19).
(19) Vowel harmony

(a) higher-mid vowels through harmony
aigre bitter /'/ > aigri /e/ /i/
bte animal /'/ > btise /e/ /i/
22
terre earth /'/ > territoire /e/ /i/
clair clear /'/ > clairer /e/ /e/
fte holiday /'/ > fter /e/ /e/
tte head /'/ > ttu /e/ /y/
peur fear // > peureux // //
aveugle blind // > aveugler // /e/
coeur heart // > coeurer // /e/
oeuvre work // > oeuvrer // /e/
dot dowry /n/ > doter /o/ /e/
code code /n/ > dcoder /o/ /e/
froce ferocious /n/ > frocit /o/ /i e/
(b) lower-mid vowels through harmony
aimer to love /e/ > aimable /'/ /a/
tiez were (2pl.) /e/ > tait /'/ /'/
pcher to sin /e/ > pcheur /'/ //
jene fasts // > jeneur // //

il pleut its raining // > pleuvoir // /wa/
meule stack // > meulette // /'/
gros fat /o/ > grossesse /n/ /'/
roder to break in /o/ > rodait /n/ /'/
dos back /o/ > dossard /n/ /a/
rose rose /o/ > rosette /n/ /'/
4.2.4.3 4.2.4.3 The loi de position

The concept of loi de position refers to a constraint on the pronunciation of
mid vowels that correlates their degree of opening (higher-mid or lower-mid)
with syllable structure: higher-mid vowels in open syllables; lower mid vowels
in closed syllables. Applied without exception, the loi would imply the exis-
tence of only a single series of mid vowels (e.g. /E / in the approach of
Taulelle [1989]) with two series of allophones: [e ] in open syllables, ['
n] in closed. In SF, unlike the situation in dialects of the Midi, the term
loi de position is somewhat of a misnomer, however, since it does not apply
with anything approaching law-like regularity, as the details given above make
clear. Nonetheless, there are cases where patterns reminiscent of the loi de
position emerge in SF, and we will outline them briey here. Perhaps the clear-
est case involves nonnal closed syllables where the loi is normally respected:
only lower-mid vowels are found, as in fertile, heurter, portrait (with the
exception of the effects of /z/, which requires a preceding //: curieusement
curiously /...z.m#/, and the few variable examples of /e/ discussed above).
In the other three contexts with which we have been dealing (nal open syl-
lables, nal closed syllables, nonnal open syllables), certain mid vowel pairs
follow the expectations of the loi de position while others contradict them. We
see, for example, regular /e / in nal open syllables, contradicted by the
appearance of /'/ in this context as well, or regular /' n/ in nal closed syl-
lables, contradicted by the appearance of // and /o/ as well. In nonnal open
syllables, our initial or unmarked expectations outlined in (16) above showed
regular /e/ and //, contradicted by the preferred appearance of /n/, but as the
survey of the effects of analogy and vowel harmony demonstrates, the loi is
more often honoured in the breach in this context. The regularity of the behav-
iour of the mid vowels in MF, compared to the much more complicated situa-
tion in SF, is exemplied in (20) for those contexts where SF permits a distinc-
tion between the higher- and lower-mid vowels: that is, in nal open syllables
for /e '/, nal closed syllables for / /, /o n/ and in open non-nal syl-
lables for all mid vowels. MF, in other words, fully implements the loi de posi-
tion, allowing only /e o/ in open syllables (nal or nonnal) and only /' n/
in the corresponding closed syllables.
56 Chapter 4
(20) Mid vowels in SF and MF

SF MF
donner to give /dnne/ /done/
donnait was giving /dnn'/ /done/
jene fasts /<n/ /<n/
jeune young /<n/ /<n/
saute jumps /sot/ /snt/
sotte silly (f.) /snt/ /snt/
rose rose /oz/ /nz/
errer to wander /'e/ /ee/
pleurer to cry /ple/ /ple/
beaut beauty /bote/ /bote/
bott with boots /bnte/ /bote/
To conclude this section, we may note that analogy, vowel harmony, and the
loi de position may interact, sometimes reinforcing each other, sometimes con-
tradicting each other. Where two or more pressures work in concert, the pro-
nunciation of the forms in question is likely to be stable. Where they conict,
variability is the norm, as the rich data in Martinet and Walter (1973) conrm.
In (21) we see examples where at least two factors work together to favour a
certain pronunciation; in (22), one of these factors is contradicted by another
(or by more than one other), and as a result, variable pronunciation is likely
to occur. It would appear, if ranking is needed, that analogical pressures to
preserve the root vowel and the tendency to pronounce /n/ in nonnal open
syllables are the two strongest tendencies, while vowel harmony and the loi de
position are weaker and often inoperative.
(21) Mutually supporting pressures (A, VH, LP, O = analogy, vowel

harmony, loi de position, and tendency for /n/ in nonnal open syllables,
respectively)
beaut beauty /bote/: A with beau, VH with /e/, and LP all favour
/o/; contradicted by O
ct side /kote/: A with cte, VH with /e/, and LP all favour /o/;
contradicted by O
bleuir to turn blue /bli/: A with bleu, VH with /i/, and LP all favour
//
essuyer to wipe /esie/: VH with /i e/ and LP favour /e/
(22) Conicting pressures

sottise foolishness /sntiz/: A with sotte, O; contradicted by VH, LP
/sotiz/: A with sot, VH with /i/, LP; contradicted by O
mairie city hall /m'i/: A with maire; contradicted by VH, LP
/mei/: VH with /i/, LP; contradicted by A with maire
matresse mistress /m't's/: A with matre, VH with /'/; contradicted
by LP
matrise mastery /metiz/: VH with /i/, LP; contradicted by A with
matre
fter celebrate /f'te/: A with fte; contradicted by VH, LP
/fete/: VH with /e/, LP; contradicted by A with fte
jaunet yellowish /<nn'/: O, VH with /'/; contradicted by LP
/<on'/: A with jaune, LP; contradicted by O, VH
jeunesse youth /<n's/: A with jeune, VH; contradicted by LP
/<n's/: LP; contradicted by A with jeune, VH
oeillre blinkers /j'/: A with oeil, VH; contradicted by LP
As these examples illustrate, it is not surprising that the pronunciation of

mid vowels in nonnal open syllables continues to provide the motivation
for much discussion (not to say confusion) in the treatment of the SF vowel
system. Speakers of MF, on the other hand, instantiate a much less compli-
cated structure.
4.2.5 4.2.5 Grammatical Consequences of the

Constraints on Mid Vowels
The preceding sections, particularly 4.2.1 and 4.2.2, have identied strong
constraints on the appearance of certain mid vowels in SF: the general absence
of /e/ from nal closed syllables and the corresponding absence of / / from
nal open syllables. In a number of cases in morphologically complex items
(those involving sufxation), these constraints require adjustments in the pro-
nunciation of words. On some occasions, particularly involving the vowels /e
'/, these adjustments are indicated in the orthography using the letters < >
and < >, as in (23), where /e/ in an open syllable in the rst column changes
to /'/ in the closed syllable on the right.23 Needless to say, many rst conjuga-
tion verbs exhibit this phenomenon.
58 Chapter 4
(23) Alternations with /e - '/ (orthographically indicated)

/e/ /'/
cder il cde to give up he gives up
possder elle possde to possess she possesses
rvler il rvle to reveal he reveals
lpreux lpre leprous leprosy
rgler rgle to settle/adjust rule
svrit svre severity severe
systmatique systme systematic system
Other cases where /'/, /o/, and // replace /e/, /n/, and // are given in (24);
note, in particular, the consequences of an informal process of abbreviation of
words.
(24) Height adjustments in the mid vowels

lger /e/ lgre /'/ light (m.-f.)
premier /e/ premire /'/ rst (m.-f.)
prisonnier /e/ prisonnire /'/ prisoner (m.-f.)
pnible /e/ peine /'/ painful pain
agrgation /e/ agreg /'/ aggregation (degree)
aprs-midi24 /e/ aprm /'/ afternoon
bnce /e/ bnef /'/ prot
idiote /n/ idiot /o/ idiotic (f.-m.)
sotte /n/ sot /o/ silly (f.-m.)
roter /n/ rot /o/ to belch belch
hebdomadaire /n/ hebdo /o/ weekly
elles peuvent // elle peut // they can she can
ils veulent // il veut // they want he wants
dgueulasse // dgueu // lousy
boeuf // boeufs // ox oxen
oeuf // oeufs // egg eggs
oeil // yeux // eye eyes
4.2.6 4.2.6 Orthography and Pronunciation

Although it is not the purpose of this study to present in detail the ortho-
graphic-phonological correspondences characteristic of SF, we saw above that
the /e ' / opposition was usefully, though not perfectly, correlated with cer-
tain orthographic signals: nal < , e, ez, er > (where the < r > is not pro-
nounced, as in innitives) are normally realized as /e/; nal < ai, aiC, aC, eC
or C > as /'/. To this may be added the observation that orthographic < e >
before two written consonants is also generally /'/. There are, however, well-
known instances where the pronunciation contradicts the orthography (vne-
ment event /ev'nm#/ or mdecin doctor /m'ts'/, for example) or where over
time the orthography has been inuential in modifying speech patterns (so-
called spelling pronunciations): cet this as /s't/ rather that the former alter-
nation among /st/, /s't/, and /set/, ressui lair as /si/ or /'si/, gageure
difcult undertaking as /ga</ rather than /ga<y/, indemne unharmed as
/'d'mn/ rather than /'d'm/ and so on.25 To complete the picture, we may also
note that the letter sequence < eu > provides the virtually exclusive representa-
tive of the vowels // and //; that word-nal < eu > is only // (jeu game),
as is < e >(jene fast); and that the pronunciation of these mid vowels is
very largely determined by the loi de position. Exceptions to this distribution
are mainly those few cases of // in closed syllables, a set of forms best memo-
rized (other than those required by a following /z/). As for /o/ and /n/, < o >
without an accent mark represents /n/ except in absolute nal position or pre-
ceding /z/, while /o/ appears most frequently as < >, < au >, or < eau >.
4.2.7 4.2.7 The Potential Merger of /n/ and //

In a classic article entitled Cest jeuli, le Mareuc! Martinet (1958) discusses
a tendency for many French speakers to pronounce the lower-mid back vowel
/n/ in a fronted manner, to the extent that it overlaps with //26 in both stressed
(Maroc) and unstressed (joli) syllables. Further examples of the phenomenon,
taken from some 450 items in Walter (1976: Chapter VIII), are given in (25).
(25) Fronted realizations of /n/ (overlap of /n/ and //)

boulotte plump (f.)
conome thrifty
sole sole
sotte silly (f.)
anonyme anonymous
auto car
60 Chapter 4
beaujolais Beaujolais wine

blocage freeze, block
mauvais bad
motion motion
phontique phonetic
potage soup
projet project
tlphoner to telephone
volage ckle
The fronting of /n/ initially appeared to be a variable but generalized phenom-

enon, insofar as the limited amount of available data permitted such a conclu-
sion. According to Walter (1976), however, fronting is not now expanding in
her data base, but it is stable, at least for older speakers, and may easily be
heard in the spoken media. Hence, it would be currently inappropriate to speak
of any threatened merger between /n/ and //, but legitimate to point out the
potential for overlap. Factors that were originally advanced to account for the
shift include a crowded back vowel space with pressures to distinguish the
four vowels /u o n #/, as well as the small number of words distinguished by
the /n / opposition. These pressures have no doubt been reduced by the
weakening of the /a #/ opposition in favour of /a/ (see below), leaving just
three vowels to induce any pressure. Whatever the cause (or the ultimate out-
come) of the fronting of /n/, we mention it here not just because it reects a
further complication among the mid vowels, but also because it interacts with
another important vowel merger in SF. As we will see in a later section, the
merger of schwa with // (or //) has signicant consequences for any analysis
of SF. If /n/ is also to be considered in this context, the possibilities for confu-
sion increase dramatically. In fact, Walter (1976) cites several examples of just
such confusion: for example, agneler to lamb [anle], bourrelier saddler
[bunlje], reblochon kind of cheese from Savoie [nbln5n], and others, where
the original schwa is pronounced [n] or []. Nonetheless, the extension and
impact of this phenomenon are minor and perhaps moribund at this stage,27and
it is mentioned in passing only because of its presence in the general literature
on the vowel system.
4.3 4.3 The Low Vowels /a/ and /#/

Conservative SF has two low vowels, a relatively more front /a/ and a back /#/,
distinguished in forms such as those given in (26).28
(26) /a/ versus /#/

/a/ /#/
art art pre bitter
bague ring bche tarpaulin
balle bullet bas (n./adj.) bottom/low
cas case bt packsaddle
amme ame glas knell
foire fair hte haste
lac lake las weary
lave lava mle male
mare pond mt mast
masse mass ple pale
page page Pques Easter
rouage cogwheel pte pastry
soit so be it phrase sentence
stable stable ras short
toile cloth sabre sabre
In general, there is little unanimity among speakers as to which words in SF

contain /a/ and which contain /#/, and most analysts note a signicant tendency
towards the elimination of the back vowel /#/ in favour of /a/. In very con-
servative speech, however, certain words that, for the majority of speakers,
contain /a/ are pronounced with /#/: affres pangs/torrents, ge age, havre
haven, for example.29 For many other forms, the conservative speakers repre-
sented by several of the informants in Martinet and Walter (1973) give highly
heterogeneous responses, and the generalizations to be extracted are rare. One
of the most consistent is the appearance, with the exception or rare /#/, of
/a/ preceding //. A second tendency sees the vowel of root words preserved
in derived forms: pre bitter pret bitterness, cble cable cbler
to cable with /#/, amme ame enammer iname with /a/, and so on,
much as in the discussion of analogy affecting the mid vowels. Words contin-
ing < oi > (/wa/ or /w#/) are roughly evenly split between the two pronuncia-
tions, and many of them are highly variable between individuals (moi me and
mois month as both /mwa/ and /mw#/, for example). Certain sufxes show
favoured pronunciations in this general context: -tre has a preponderance of
/#/; -age, -al and -oir favour /a/; while -able, -ation, and -ois are variable.30
62 Chapter 4
Finally, there is a correlation, far from perfect, between orthographic < >
and /#/, as indicated already in the forms in (26).31 Historically, one of the rea-
sons for this correlation is clear. As we saw earlier in (10), an accent circon-
exe may often be an indication of an earlier deleted segment (consonant or
vowel), deletion normally resulting in compensatory lengthening of the pre-
ceding vowel. In this respect, the /a #/ opposition was earlier one of length,
/a a/, and was only subsequently converted into one of vowel quality (with
traces of lengthening remaining in the status of /#/ as an intrinstically long
vowel in nal closed syllables). Again, we can nd nonlearnedlearned pairs
of words providing evidence of the earlier deleted consonant.
(27) < > < as > in related words

pre bitter asprit bitterness
bton stick bastonnade beating
crotre to grow croissance growth
mcher to chew mastiquer to masticate
mle male masculin masculine
Pques Easter pascal paschal
ptre shepherd pasteur shepherd/pastor
As Walter (1976) and others have reported,32 the /a #/ distinction is variable
and unstable in SF. A merger in the direction of /a/ is well underway for
younger speakers, with the result no doubt eventually to be a vowel system
with only a single low vowel. In fact, certain recent pronouncing dictionaries
no longer note the difference, using the single symbol /A/ (representing a low
central vowel akin to [a]) for both earlier sounds.33 Given these circumstances,
it would appear appropriate for those learning the SF vowel system to conform
to the innovating pattern and to function with the single vowel /a/ in their lin-
guistic performance.
4.4 4.4 Nasal Vowels

The conservative variety of SF that is the starting point for our discussion con-
tains four nasal vowels,/n'#/, usefully captured in the mnemonic phrase
un bon vin blanc. While this phonological notation of the four vowels reects a
long tradition in French linguistics, it is not without problematic aspects, since
in detailed phonetic terms the actual pronunciation of nasal /'/ and /n/ in SF
is closer to [3] and [] respectively (not unlike the vowels in the English
words cant and dont).34 We see this problem in pairs such as bon bonne
or vain vaine, which we will transcribe phonologically as /bn bnn, v'
v'n/, even though the normal pronunciation is [b bnn, v3 v'n].35 Despite
this discrepancy between notation and pronunciation, we will retain the tradi-
tional symbols, not just because of tradition, but also because of alternations
(to be discussed below) between nasal vowels and sequences of oral vowels
followed by nasal consonants, where (some of) the oral vowels corresponding
to /'/ and /n/ are not // and /o/, but /'/ and /n/. Nonetheless, this difference in
the articulation of these nasal vowels bears remembering, if for no other reason
than the need for accurate pronunciation.
4.4.1 4.4.1 Merger of /'/ and //

Perhaps the most notable happening among the SF nasal vowels is the ongo-
ing merger of // with /'/, leading to a three-vowel system /'n#/.36 For
many speakers, particularly the younger generations, this merger is complete,
although the data in Martinet and Walter (1973) show that conservative pro-
nunciations are likely to persist for some time. Representative words illustrat-
ing the merger (from among the very small overall number of words with //
in French) include those in (28):
(28) The vowel // as /'/

jeun /C<'/ fasting
aucun /ok'/ no, not any
brun /b'/ brown
chacun /5ak'/ each
commun /knm'/ common (m.)
dfunt /def'/ defunct (m.)
emprunt /#p'/ borrowing
humble /'bl/ humble
lundi /l'di/ Monday
opportun /npnt'/ timely (m.)
parfum /paf'/ perfume
quelquun /k'lk'/ someone
un /'/ a/one
Various authors (e.g., Martinet [1955]) have seen in the paucity of forms with
// the reason for this merger: the low functional yield of // (the small
number of minimal pairs in which it participates) means that its disappearance
would have little effect on the functioning of the system. (It would lead to
little homophony.) Against this may be set the very high frequency of certain
64 Chapter 4
members of the list in (28) (e.g., un, aucun, quelquun, lundi), a frequency that
may well counterbalance the lexical rarity of //. Be that as it may, we see here
an additional example of a progressive reduction of the number of distinctions
in the vowel system of SF.
4.4.2 4.4.2 Distribution of Nasal Vowels

Nasal vowels in French are subject to a number of word-internal distributional
constraints that exclude these vowels from certain well-dened contexts and
that invite further discussion. In general tems, while nasal vowels appear
in most phonological contexts in SF, they may not occur preceding other
vowels, nor may they appear preceding nasal consonants. We may note this
as *XX and *X. Thus, while anne year /ane/ and bont goodness /bnte/
are well-formed, bonheur happiness and bonnement just, quite simply, if
pronounced */bn/ and */bnm#/, are not. There are, however, certain classes
of exceptions to these restrictions. Consider rst words such as Panhard (an
older model of car) /p#a/, enhardir to embolden /#adi/, enharnacher to
harness /#ana5e/, each with /#/ preceding a vowel. In such forms, we see the
effects of the h aspir, which, here and elsewhere in French, provides a phan-
tom consonant that conditions a variety of exceptional behaviours. The second
and more general type of exception also involves the prex en-/em- /#/, as in
enregistrer to record, entasser to pile up, embellir to make attractivewith
the roots registre, tas, beau/belle. When this productive prex is added to a
word beginning with a nasal consonant, the pronunciation of the prex as /#/ is
retained, and we nd a nasal vowel preceding a nasal consonant, as in (29).37
(29) /#/ preceding nasal consonants

emmagasiner /# + m/ to store up
emmailloter to bandage
emmancher to put a handle on
emmler to tangle
emmnager to move in
emmener to take away
emmerder to bother
emmieler to sweeten (with honey)
emmurer to wall up
enneig /# + n/ snow-covered
ennoblir to ennoble
ennuager to cloud over
Finally, in a few forms, the addition of en- to vowel-initial words also contra-
venes the above constraints, since we nd enamourer to become enamoured
of /#namue/, enherber to plant with grass /#n'be/, enivrer to intoxicate
/#nive/ with the sequence /#nV/ (compare panafricain /panafik'/ with the
nasal-nal prex pan- /p#/). This situation is similar to that arising with nasal
vowels in liaison contexts (contexts that go beyond word boundaries), a dis-
cussion that will occur below. For the time being, we note simply that the *XX
and *X constraints on nasal vowels remain robust, with the exception of two
clearly circumscribed cases, each of which is subject to independent explana-
tion.38
4.4.3 4.4.3 Alternations between X and VN

Nasal vowels or, more specically, alternations between nasal vowels and a
corresponding oral vowel followed by a nasal consonant play an important
role in French morphology. These alternations occur in inectional morphol-
ogy (e.g., in the distinction between masculine and feminine forms of nouns,
adjectives, and determiners, and between the singular and plural as well as the
indicative and subjunctive forms of verbs), in derivational morphology (e.g.,
when a root ending in a nasal vowel is followed by a vowel-initial sufx or
when a prex ending in a nasal vowel precedes a vowel-initial root), and in
certain close-knit syntactic contexts involving liaison. Each of these types is
illustrated in (30).
(30) Alternations involving nasal vowels

(a) inection
(i) masculine feminine (determiners, adjectives, nouns)
/ yn/ 39
un une a/one
aucun aucune no/none
chacun chacune each
commun commune common
brun brune brown
opportun opportune timely
/n nn/
bon bonne good
breton bretonne Breton
baron baronne baron
lion lionne lion
66 Chapter 4
/# an/
artisan artisane craftsperson
catalan catalane Catalan
paysan paysane peasant
roman romane Roman
/' 'n/; /j' j'n/
le mien la mienne mine
chien chienne dog
canadien canadienne (many sufxed forms in -ien)
europen europenne (sufxed forms in -en)
certain certaine (many sufxed forms in -ain)
plein pleine full
sain saine healthy
vain vaine vain
/' in/
copain copine pal
cousin cousine cousin
fminin fminine feminine
n ne thin
latin latine Latin
/' i/
bnin bnigne benign
malin maligne shrewd
(ii) singular plural (verbs)

/j' j'n/
tient tiennent hold
vient viennent come
/' '/
craint craignent fear
peint peignent paint
/w' wa/
joint joignent join
/# 'n/
comprend comprennent understand
prend prennent take
(iii) indicative subjunctive (verbs)

/j' j'n/
tient tienne hold
vient vienne come
/' '/
craint craigne fear
peint peigne paint
/w' wa/
joint joigne join
/# 'n/
comprend comprenne understand
prend prenne take
(b) word formation (derivation or conversion)

(i) root + sufx
/# an/
an anne year year
clan clanique clan clannish
plan aplanir at to level
volcan volcanisme volcano vulcanism
/n nn/
bon bonier good to improve
colon colonie colonist colony
information informationnel information informational
jargon jargonnesque jargon full of jargon
rayon rayonnage shelf shelving
68 Chapter 4
/ yn/
commun communal common communal
opportun opportuniste timely opportunistic
tribun tribunat tribute tribunate
un unime one rst
/ ym/
parfum parfumerie perfume perfumery
/' en/
chien chinage dog tax requiring vassals to
raise a certain number of dogs
sain assainir healthy to clean up
serein srnit serene serenity
plein plnitude full fullness
/' an/
humain humanitaire human humanity
main manier hand to handle
/w' wa/
poing poignard st dagger
bouquin bouquiniste book bookseller
clin cliner cuddly to cuddle
chauvin chauvinisme chauvinistic chauvinism
divin divinit divine divinity
n nesse thin sharpness
mesquin mesquinerie stingy stinginess
pkin pkinois Beijing Pekinese
vin vinier wine convert into wine
(ii) conversion (noun to verb)

/# an/
ruban enrubanner ribbon trim with ribbons
/n nn/
faon faonner way to shape
station stationner station to park
/' am/
faim affamer hunger to starve
/' em/
essaim essaimer swarm to swarm
/' en/
frein freiner brake to brake
/ n/
jeun jener fasting to fast
/' '/
bain baigner bath to bathe
ddain ddaigner disdain to disdain
/' - in/
chemin cheminer road to walk along
jardin jardiner garden to garden
/w' - wa/
soin soigner care to care for
tmoin tmoigner witness to witness
(iii) prex + root

/nn/ /nnn/40
non-belligrant non-belligerant non-agression nonagression
non-combattant non-combatant non-assistance failure to assist
non-conformisme non-conformism non-engag nonaligned
non-croyant non-believer non-tre non-being
non-paiement non-payment non-intervention nonintervention
non-violence nonviolence non-usage non-use
70 Chapter 4
/p#/ /pan/
panchromatique panchromatic panamricain panamerican
pangermanisme pan-germanism paneuropen pan-European
panthisme pantheism panhellnique panhellenic
panslavisme panslavism panislamisme panislamism
/bj'/ /bj'n/
bien-dire eloquence bien-aim beloved
bienfaisant benecial bien-tre comfort
bien-pensant right-thinking bienheureux blessed
/sikn/ /siknm/41
circonlocution circumlocution circumlunaire circumlunar
circonscription district circumnavigation
circumnavigation
circonvenir to circumvent circumpolaire circumpolar
42
/#/ /#n/
emballer to pack up enivrer to intoxicate
embellir to beautify enrober to coat
embourgeoiser to become entraner to carry along
middle class
emmener to take away envisager to envisage
empaqueter to wrap up
empitement encroachment
emprisonner to imprison
encadrer to frame
encercler to encircle
endolorir to make painful
/'/ /im/
imbattable unbeatable immatriel immaterial
imbrl unburnt immatriculation registration
imbuvable undrinkable immigrant immigrant
immangeable inedible immobile immobile
immanquable impossible to miss
impair uneven/odd immodeste immodest

impensable unthinkable
/'/ /in/
inchang unchanged inachev unnished
incorrect incorrect inexactement inexactly
indcis indecisive inhumain inhuman
inexible inexible ininterrompu uninterrupted
insonoriser to soundproof inopportun untimely
intolrable intolerable inoubliable unforgettable
invendable unsellable inutile unuseful
(c) liaison43
/#/ /#n/
en partant leaving en arrivant arriving
jen prends deux Ill take two jen ai deux I have two
/n/ /nn/
on part were leaving on arrive were arriving
mon cousin my cousin mon oncle my uncle
bon professeur good professor bon lve good student
/'/ /'n/
bien parti left bien arriv arrived
rien trouv nothing found rien apport nothing brought
Each of these sets of examples presents its own complexities. We will deal
with them in turn, moving from less to more complex. As we have already
seen, these data sets illustrate alternations between a nasal vowel (left-hand
column) and an oral vowel followed by a nasal consonant (right-hand column).
In some cases, the alternation appears to be correlated directly with a change in
grammatical category; in others, it is phonologically conditioned (for the most
part) depending on whether a consonant or a vowel follows the nasal vowel.
Thus, in (30a, i-iii), the nasal vowel occurs in the masculine, the singular, or
the indicative, while the oral vowel-nasal consonant sequence is in the femi-
nine, the plural, or the subjunctive. Such relationships between phonological
length or complexity and morphological markedness are widespread cross-
linguistically and have given rise to extended discussion in the theoretical lit-
erature dealing with morphological structure and typology.44
72 Chapter 4
If the general pattern X X is clear in (30a), the specic correlates of the

nasal vowel are not. Only in the case of /n/ is there consistency. The three
remaining nasal vowels do not allow their alternating counterparts to be pre-
dicted, as the set of correspondences in (31) makes clear. That is, neither the
oral vowel nor the nasal consonant partnered with the nasal vowel is unique,
although we may identify preferred patterns.
(31) X X alternations
n nn
yn
ym
n
# an
# 'n
# am
' 'n
' '
' en
' em
' in
' i
' an
' am
j' j'n
w' wa
From the correspondences in (31), /n/ consistently pairs with /nn/ (bon

bonne), // generally pairs with /yn/ (aucun aucune), /#/ with /am/ in
adverbs (constant constamment constantly) and /an/ in adjectives or nouns
(paysan paysanne peasant), and /'/ with /'n/ (sain saine healthy). This
domain remains, however, one in which the morphological consequences of
nasalization, to say nothing of the alternations involving /'/, are rich in their
complexity.
When we consider the forms in (30b-c), on the other hand, the role of
the following segment is more evident. The nasal vowel appears in word-
nal position in (30b, i-ii) or preceding a consonant in (30b, iii); the oral
vowel nasal consonant sequence normally appears when a vowel follows, as
in faon faonner to shape /fasn fasnne/, inchang unchanged inach-

ev unnished /'5#<e ina5ve/.45 This pattern is regular in all of the words
formed from derivational sufxes in (30b, i) (and with the rarer consonant-
initial derivational sufxes such as -t: bon bont goodness /bn/ /bnte/).
It is also found with the regular prexes non- and pan-, with circon- (although
the absence of this prex preceding vowels makes the question moot), and in
most instances of in-. There is a signicant class of exceptions to this pattern,
however. First, bien does not denasalize when added to a vowel-initial root
(/bj'n/, not */bj'n/ for bienheureux blessed, etc.) in violation of the
*X constraint. The same is true for the prex en- in enivrer /#nive/, although
this is the single example in Warnant (1987) of such a structure, en- appar-
ently being largely limited to consonant-initial roots.46 The liaison of words
ending in nasal vowels perhaps falls into this same category, as the cases of
en in, on one, bien well, rien nothing and mon my demonstrate in
(30c), along with aucun no/not a and commun common (aucun homme
/oknnm/), although constraints across word boundaries in French (as is the
case with liaison) are typically less strong than those within words. Of addi-
tional liaison forms, ancien ancient/former, bon good, certain certain,
plein full, and prochain next are regular, in that the nasal vowel consistently
denasalizes: bon ami good friend /bnnami/, en plein hiver middle of winter
/#pl'niv'/, and so on.47
Finally, we must comment on the behaviour of the productive prex in-.48 It
is, for the most part, regular: /'/ before consonants, /in/ before vowels. There
is, however, one set of exceptional cases. In a small number of words begin-
ning with /m/ or /n/, we nd not the expected preconsonantal form /'/, as in
immanquable impossible to miss /'m#kabl/, but rather /i(m)/ or /i(n)/, where
the bracketed (m, n) indicate the possibility of a geminate consonant: immobile
/i(m)mnbil/, innommable unspeakable /in(n)nmabl/. Thus, despite a number
of generalizations that are applicable to the behaviour of nasal vowels, we
must conclude that their behaviour is governed by a complex interaction of
phonological, morphological, and idiosyncratic factors. A brief discussion of
the history of nasalization in French may shed further light on the matter.
4.4.4 4.4.4 History and Orthography

Much of the heterogeneity of the preceding discussion may be attributed to
the complex interaction of several historical phonological processes that com-
bined to produce the X X alternations. These processes include the nasal-
ization of vowels preceding nasal consonants, the lowering of nasal vowels,
denasalization, and the loss of nasal consonants following nasal vowels. While
we cannot hope to do justice to this rich and complex matter here,49 some idea
of how nasal vowels developed is helpful in understanding the current situa-
tion in SF. Briey, the evolution followed this general path:
74 Chapter 4
(a) all vowels (and diphthongs) preceding nasal consonants were

nasalized, with (according to a now somewhat contested traditional
view) the lower vowels nasalizing rst and most strongly
(b) the resulting nasal vowels became lower or more back in
articulation, again beginning with the lower vowels
(c) nasal consonants in coda (syllable-nal) position were deleted
(d) denasalization affected those nasal vowels where the following
nasal consonant had not been deleted
If we consider typical Gallo-Roman or early OF forms such as n ne
sharp or conte story, rst pronounced /n/ /n/ and /knnt/, the processes
described above would have the following effects:
(32) Historical phonological processes

initial form /n/ /n/ /knnt/
nasalization HKP HKP knnt
denasalization n
lowering f'n
loss of nasal consonant f' knt
current form f3 n() knt()
The interaction of these processes explains, in historical terms, the current dis-
tribution of nasal vowels in particular the constraints on their distribution
we saw in (4.4.2) and the nature of the orthographic representation of these
vowels. Note rst that nasal vowels are normally found at the end of words or
before oral consonants, but not before vowels or nasal consonants: XC, X#; VV,
V are perfectly regular and produced by these (or other) processes, while *XV,
*X are not. In order for XXto arise, the rst vowel would have to nasalize and
the nasal consonant delete, but we have seen that the nasal consonant does not
delete when a vowel follows. (XX requires denasalization of the rst vowel,
not nasal consonant deletion.) In order for Xto arise, the nasal consonant
would need to remain, but we have seen that such consonants delete follow-
ing nasal vowels, provided the vowels remain nasal. Lowering and backing
also account for much of the variation in the basic vowel in many of the X
X pairs in (31). When nasal /'/, //, and /#/ correspond to oral /in/, /yn/, and
/an/ respectively (n ne, un une, an anne), the effects of these two pro-
cesses on the nasal versus the oral vowels are evident. This situation illustrates
how synchronic alternations as well as restrictions on the distribution of seg-
ments arise through the interaction of historical phonological processes.50
It is interesting that SF orthography reects these constraints only indirectly,
the orthography being conservative in nature and more akin to the situation in
the rst line of (32). In order to interpret the orthography correctly, we must
apply the processes in (32): a vowel followed by a nasal consonant at the end
of words or before another consonant is nasal, and the nasal consonant is not
pronounced; a vowel followed by a nasal consonant (or double nasal conso-
nant) followed by a vowel is oral, and the nasal consonant is pronounced, as
in bon /bn/, bonne /bnn/, bont goodness/kindness /bnte/, and bonasse easy-
going /bnnas/. There are, needless to say, exceptions to these orthographic
tendencies, themselves produced by further historical phenomena such as the
deletion of schwa or the incorporation of loan or slang words, such as cane-
ton duckling /kantn/, not */k#tn/ (compare canton district/township /k#tn/);
hameon sh-hook /amsn/, not */#sn/; stencil /st'nsil/, not */st#sil/; album
/albnm/; binse /bins/; clamser /klamse/; and so on.51 Nonetheless, there are
many regularities to be extracted from the structure of the French orthographic
system, and these regularities often signal antecedent historical events.
4.4.5 4.4.5 Dialects

To conclude the discussion of nasal vowels, we may outline briey the situa-
tion in two additional dialects, MF and CF, where the behaviour is strikingly
different from SF. MF is somewhat simpler, in that the realization of the nasal
vowels is closer to their oral counterparts and the loss of the nasal consonant
following nasal vowels has not fully taken place. The degree of nasalization
of the vowel may also be much less. The vowels /'/, /#/, and /n/ are often
pronounced closer to ['], [a], and [n], respectively, and in word-internal posi-
tion, a nasal consonant, homorganic to the following consonant, persists fol-
lowing the nasal vowel. In word-nal position, on the other hand, the inserted
nasal consonant is [0]. MF and SF pronunciations of representative words are
contrasted in (33).52
(33) Nasal vowels in MF and SF

MF SF
simple [s'mpl] [s3pl] simple
dingue [d'0g] [d3g] crazy
danser [FCPUG] [d#se] to dance
attention [CVCPULn0] [at#sjn] attention
enn [Cnf'0] [#f3] nally
bien [bj'0] [bj3] well
In CF, by way of contrast, the situation is more complex. CF makes a signi-

cant distinction between its phonetically long and short vowels. Short nasal
vowels (those occurring in open syllables) have shifted frontward or rotated
76 Chapter 4
clockwise in the phonological space, while long nasal vowels have diphthon-
gized (along with long oral vowels).53 These two sets of vowels are contrasted
with those of SF in (34).
(34) Nasal vowels in CF and SF

CF SF
on [n] [] one
honte [nwt] [t] shame
(le) vent [v3] [v#] (the) wind
il vente [ivCwt] [ilv#t] its windy
vin [v'] [v3] wine
quinze [k'jz] [k3z] fteen
dfunt [def] [def'] late/dead
il emprunte [jCp]t] [il#p't] he borrows
Thus, from a historical, dialectal, and sociolinguistic point of view, nasal

vowels in French provide a rich, varied, and challenging domain of investiga-
tion in which continuous phonological innovation means that ongoing atten-
tion is always merited.
4.5 4.5 Schwa

Perhaps no topic in the study of French phonology (with the possible excep-
tion of liaison) has generated as much discussion as schwa, from both a
descriptive and a theoretical point of view. Even the terminology is change-
able: schwa, e-muet, e-instable, e-caduc, e-fminin, and so on are used to
indicate the segment in question. These terms refer to a vowel whose precise
phonetic properties are variable, whose normal orthographic representation is
the unaccented letter < e >, and whose primary interest lies in the fact that its
pronunciation or suppression depends on a wide variety of phonological, mor-
phological, syntactic, and stylistic factors. In discussions of this vowel, dis-
tributional constraints, phonetic detail, retention or deletion, stylistic effects,
orthographic correspondences, dialectal variation, inectional and derivational
alternations, and the interaction of phonological and nonphonological factors
are all relevant. The following sections will attempt to bring some order into
this descriptive miscellany.54
4.5.1 4.5.1 Orthographic Representations of Schwa

As many of the designations indicate (e-muet, etc.), we are dealing almost
exclusively with interpretations of the letter < e > in French, as in petit
small, premier rst, samedi Saturday, tudiera will study, vie life, vite
quickly, presque almost, grande big, and so on. There are two minor
exceptions to the schwa < e > link: the rst vowel of the word monsieur
Mister, where the < on > represents schwa, and the sequence < ai > in two
additional roots, faisan pheasant and faire to do. With faisan, in addition
to the root itself, a number of derived forms also show < ai > as schwa: fai-
sander, faisand(e), faisandage, faisanderie, and faisandeau. With faire, the
situation is somewhat more complex. Derivational forms built on faire have
schwa: faisable, faisant, bienfaisant, bienfaisance, satisfaisant. In addition,
certain inectional forms have the same vowel: faisons and faisait (plus the
rest of the imperfect), in contrast to the future and the conditional, both with
schwa as < e >: ferai, ferait.55 These minor exceptions aside, the correlation
between < e > and schwa is very strong. The frequency of the letter < e > in
French, moreover, guarantees that we will be confronted with a range of issues
involving schwa at virtually every stage of our study of French phonology.
4.5.2 4.5.2 Distributional Constraints on Schwa

As was the case with previous vowels, we may identify a variety of constraints
on the appearance of schwa in SF words. (We leave aside for the moment the
behaviour of schwa in phonological phrases.) For example, no SF word begins
with schwa in absolute initial position. Nor may schwa occur immediately
next to another vowel, in either preceding or following position (with one class
of exceptions to be discussed shortly). In addition, at the level of the word in
SF, schwa does not occur in closed syllables unless as the result of a process
where one of the schwas in adjacent syllables has been deleted. Schwa may not
be accented.56 Finally, although this is not precisely a distributional restriction,
the presence of schwa in sequences such as VC# does not serve to provide
an open penultimate syllable for certain phonological processes in French that
are sensitive to syllable structure. Using the symbol // to represent schwa, we
may represent these constraints more formally as in (35).
(35) Distributional constraints involving schwa

(i) *#
(ii) *V
(iii) *V
(iv) *C.
(v) *p
(vi) VC.#
78 Chapter 4
The rst of these constraints accounts for a set of orthographic conventions in

French. In orthographic terms, no word in French begins with an unaccented
< e > unless that letter is itself followed by the letter < x > or by two or more
orthographic consonants.57 Thus, we nd bahir to dumbfound, gorger to
slit the throat of, pinard spinach, and so on with initial /e/ and examen
examination, effet effect, essentiel essential, tre to be, and so on with
/'/, but never *epinard, *xamen or *xamen, and so on. The reasons for
this are now clear. If the initial < e > followed by a single consonant were
unaccented, it would be pronounced //, in violation of the constraint against
initial schwa. Conversely, no accent is needed in words like examen, essaim
swarm, and descendre descend since the < x > or double consonants make
the pronunciation clear.58
As for the second and third constraints, vocalic hiatus is frequent
in French, except when schwa is involved: schwa-vowel or vowel-schwa
sequences are blocked, with the exception of the recurrent inuence of aspi-
rate-h: dehors outside /dn/, rehausser to heighten /ose/ (and derived
forms: e.g., rehaussement), Dehrain (proper noun ) /de'/, and so on.
Other than such cases (which are very rare), this pair of constraints is rmly
anchored in SF speech patterns.
The fourth constraint, often discussed in the context of a process called
closed syllable adjustment, requires that schwa be found initially only in
open syllables within words. Thus, while pesant heavy /p.z#/ is well-
formed, personne person */p.snn/ is impossible; only /p'.snn/ is per-
mitted. Exceptions to this constraint arise only in words with sequences of
schwas in consecutive syllables redevance tax, redemander to ask for
again, Genevois Genevan, etc. where deletion of the second schwa gives
/d.v#s/, /d.m#de/, /<n.vwa/, respectively, meaning that the rst schwa
is then in a closed syllable. This phenomenon becomes much more apparent
in a phrasal context, a topic that will occupy us below. Finally, despite the
status of schwa as a vowel, sequences like VC# often do not behave as if
they were normal bisyllables (V.C#), but rather as if the initial syllable were
closed (VC.#). This is shown by the fact that vowels that are excluded from
closed syllables (e.g., schwa itself, as well as /e/) cannot appear in the context
__C#. Thus, we nd mre /m'()/, never */me()/; mne /m'n()/, never
*/men()/ or */mn()/, despite menons /mnn/; and so on. We will return
briey to this question in a discussion of alternations involving schwa.
4.5.3 4.5.3 The Phonetic Realization of Schwa

Up to this point, we have used the symbol // to represent the vowel we are
calling schwa. For many speakers, this representation of the vowel as a cen-
tral, more or less rounded, lax vowel is accurate. However, there is a strong
movement in current SF for many speakers to merge this vowel with one of
the two front rounded vowels // or //, usually //.59 Thus, for analysts such
as Dell (1973a) or Tranel (1987) (who are entirely representative concerning
this matter), schwa is pronounced [], and the following pairs are completely
homophonous:
(36) Homophonous pairs: // and //

je ne vaux rien jeune vaurien /<nvoj'/
Im worthless young good-for-nothing
je ne vois jeune voix /<nvwa/
I dont see young voice
elle se le demande elle seule demande /'lsldm#d/
she asks herself she alone asks
ample rang en pleurant /#pl#/
wide row crying
quel gent - quel jeunet /k'l<n'/
which jennet (a small Spanish saddle horse) which youngster
Schwa in word-nal open syllables, as in dites-le say it /ditl/ or quelle

honte what shame /k'lnt/, for the great majority of speakers, is pronounced
// or //, not //.60
Despite this well-advanced merger, we will continue to use a separate
symbol for schwa for two reasons, one practical or pedagogical, one more
theoretical. In practical terms, it is convenient to be able to identify a separate
schwa for pedagogical purposes and to focus on the unstable or alternating
properties of that vowel without having to confuse the issue by any reference
to a competing //. Related to this, in a more theoretically oriented way, is the
following argument. Even if we recognize the merger and work with a single
symbol // for both // and original //, we must still distinguish two separate
types of //: those that delete freely and those that do not; as well as those
that alternate with other vowels, opposed to those that are stable. We would
need, in other words, to separate the // in cela dit that being said /sladi/
/sladit/ from the // in seul dire the only one to say it /sladi/, since
only the former can delete the second is stable. Likewise, we would need to
distinguish the // in crever to burst /kve/ from that in pleurer to cry
/ple/ since the former alternates with /'/ (crevons crve) while the latter
does not (pleurons pleure). Furthermore, if word-nal schwas are also //,
then we would need to exempt them from being stressed in words like double
/dbl/ or /dbl/ (and recognize a limited exception to the exclusion of //
from nal open syllables). We would need, in other words, two distinct //
vowels, a situation reected in the decision to retain the symbol // for the
rst schwa in each pair or for the nonstressable vowels in nal position.61 For
those speakers who have merged // with //, whether regularly or sporadi-
80 Chapter 4
cally, we can deal with this independently since other elements of their speech
will continue to distinguish the two vowels. The symbol //, therefore, will be
retained here.
4.5.4 4.5.4 The Deletion of Schwa

Two principal, and not unrelated, issues arise in discussions of schwa: how
the vowel is to be represented in phonological terms, and what conditions
control its variable behaviour preservation, deletion or, (in certain cases)
insertion. Representational questions over and above the merger of schwa and
// involve a wide variety of theoretical issues beyond the boundaries of this
manual see Morin (1988) for further discussion. As for the behaviour of
schwa, any consideration of its deletion begins with the phonological phrase.
Phrasal conditions are highly relevant here, as they are in other key domains
of French phonology.62
4.5.4.1 4.5.4.1 Schwa in the Phonological Phrase

Logically, phrases may be divided into three domains: a beginning, middle,
and end, each with different effects on the behaviour of schwa. Perhaps the
simplest case is provided by phrase-nal position: here, schwa is uniformly
and consistently deleted in SF, irrespective of other phonological factors such
as the number of consonants preceding the vowel. This is illustrated in (37),
where the schwa in question is given in bold face:
(37) Schwa in phrase-nal position

une vie / ynvi || / a life
ils se marient / ilsmai || / Theyre geeting
married.
elle est petite / 'l'ptit || / Shes small.
il est trop jeune / il'to<n || / Hes too young.
jen ai quatre / <#nekat || / Ive got four of them.
voil le comble / vwalalknbl || / Thats the last straw.
monsieur le ministre / msjlminist || / Mister Minister
cest un ltre / s'tlt || / Its a lter.
il est bien dextre / il'bj'd'kst || / Hes right-handed.
Phrase-initial position governs a more complicated behaviour of schwa.63 In

general terms in SF, it is always permissible to pronounce schwa in these
syllables without creating negative stylistic impressions: cependant however

/sp#d#/, je pars Im leaving /<pa/, and so on. Nonetheless, there are con-
texts where, again without stylistic implications, it is normal to delete schwa.
These contexts are dened in phonological terms: schwa deletes after the
single fricatives /f v s z 5 </ , as in (38a) and is normally retained after other
consonants (stops, nasals, liquids), as in (38b). It is also retained after any
group of consonants, as in (38c).
(38) Schwa in initial syllables

(a) faisons cela ensemble / || fznsla#s#bl / Lets do it together.
venez nous voir / || vnenuvwa/ Come to see us.
ce que vous dites... / || skvudit / what you say
chemin faisant... / || 5m'fz# / in so doing
je taime / || 5t'm / I love you.
jetez-le-moi / || 5telmwa / Throw it to me.
(b) te voit-il64 / || tvwatil / Does he see you?
que faites-vous / || kf'tvu / What are you
doing?
demain on part / || dm'npa / Were leaving
tomorrow.
me vois-tu / || mvwaty / Do you see me?
ne mens pas / || nm#pa / Dont lie.
le voyez-vous / || lvwajevu / Do you see it?
revenez nous voir / || vnenuvwa / Come back to
see us.
(c) prenez-les / || pnele / Take them.
fredonnez-vous / || fdnnevu / Are you humming?
grenouillage / || gnuja< / shady dealings
Although these restrictions are stated in general terms, there may be very few
items that instantiate them in specic cases. There are, for example, no words in
French that begin with /z/, so any deletion in this case is moot. The same applies
to many word-initial groups that are never followed by schwa. Nonetheless, the
general principle is clear: strong phonological constraints apply to the presence
or absence of schwa in phrase-initial position. This situation is further compli-
cated by certain idiosyncratic examples that we will see in section 4.5.4.3 below.
First, however, we need to review the behaviour of schwa in medial position.
82 Chapter 4
Phrase-internal schwas present, at rst glance, the most complicated type of

behaviour, but there is a strong general pattern that simplies matters greatly.
We should perhaps rst note that the phrase-internal context also includes, in
addition to words ending in schwa, all word-internal schwas (any schwas in
neither the rst nor the last syllable of a word), since such schwas are automat-
ically in internal position in any phrase in which the word in question appears.
The key phrase-internal factor involves the number of consonants preceding
the schwa. If there is a single consonant, the schwa deletes; if there are two
or more, the schwa is pronounced. In simplied formal terms, we may sche-
matize these conditions as follows: if the sequence is VCCX (where X
represents any sequence of sounds65), the schwa deletes. If the sequence is
XCCCX, the schwa is pronounced. These two constraints reect in a more
accurate manner what has traditionally been called the loi des trois consonnes
in French, a constraint against a sequence of three consonants that would be
produced by the deletion of schwa. In fact, however, schwa deletion often
produces such triconsonantal sequences if two of the consonants follow the
schwa; it is only if two consonants precede the schwa that deletion is blocked,
so the loi is somewhat of a misnomer. To this primary constraint, we must add
an interesting further condition blocking the deletion of schwa in a sequence
VCLGV. That is, if the group of segments following the schwa consists of a
liquid-glide (LG) cluster, deletion is also blocked. These three constraints are
illustrated in (39), where the VCCX and VCCCX contexts are given in bold
face.
(39) Phrase-internal schwas

(a) VCCX (deletion of schwa)
mdecine /medsin/66 medicine
honntement /nn'tm#/ honestly
samedi /samdi/ Saturday
il arrive tt /ilaivto/ He arrives early.
en deuxime position /#dzj'mpnzisjn/ in second place
salade nioise /saladniswaz/ salade nioise
demain /adm'/ until tomorrow
nous venons /nuvnn/ Were coming.
sans demander /s#dm#de/ without asking
cette proccupation /s'tpenkypasjn/ this preoccupation
une che /yn'5/ an arrow
la dernire station /lad'nj'stasjn/ the last station
(b) VCCCX (preservation of schwa)

pret /apte/ bitterness
diablerie /djabli/ mischief
dpartement /depatm#/ department
contre coeur /akntk/ reluctantly
presque termin /p'skt'mine/ almost nished
forme principale /fnmp'sipal/ principal form
nous prenons /nupnn/ we take
tout crev /tukve/ completely worn out
sept semaines /s'tsm'n/ seven weeks
ils menaient /ilmn'/ They were leading.
(c) VCLGX (preservation of schwa; LG cluster)
atelier /atlje/ workshop
bachelier /ba5lje/ bachelor
Richelieu /i5lj/ Richelieu
vous trouveriez /tuvje/ you would nd
vous seriez /vusje/ you would be
Any of the examples in (39) indicates the phrasal nature of these constraints on
schwa deletion. Word boundaries play no role in constraining the suppression
or preservation of the vowel, since word-internal schwas (samedi), word-nal
schwas (salade nioise), or schwas in the rst syllable of a word itself pre-
ceded by another word (sans demander without asking) all delete. Moreover,
deletion can produce a sequence of three consonants, provided two of the three
follow the vowel (une che /n/). By the same token, the position of
the two blocking consonants is also independent of word structure. The conso-
nants may be word-internal (diablerie), word-nal (with the exception of the
schwa itself) (presque termin), word-initial within the phrase (tout crev), or
members of adjacent words (sept semaines). What counts is their presence, not
the forms to which they belong.
Phrase-internal schwas (usually in combination with phrase-initial partners)
are involved in a further complicated set of interactions that arise when adja-
cent syllables contain schwa. The number of words where this pattern is found
word-internally is very small (Genevois Genevan, chevelure hair, ensevelir
to bury/shroud and derivatives, words with the prex re- [redemander, reve-
nir], the highly variable papeterie stationary shop [pronounced as /pap'ti/,
/papti/ or /papti/], crevettier shrimp net/boat [with // or /'/ in either
84 Chapter 4
of the rst two syllables] and so on), but phrasal structures are exceedingly
common given the nature of many pronouns and other clitic forms that com-
bine in sequence (i.e., je, me, te, se, le, ne).
(40) Sequences of schwas

je le prends Im taking it.
je me demande I wonder.
il se levait He was getting up.
on te le propose Someone suggests it to you.
je ne le comprends pas I dont understand it.
il te le recommande He recommends it to you.
je ne te le redemanderai pas (!) I wont ask you for it again.
In SF, the general tendency in these structures is for at least one schwa to
delete, but which one (or ones, as the case may be) is often variable in partic-
ular, either the rst or the second of the sequence drops: je le prends /<lp#/
or /<lp#/.67 What is more interesting in this general context is that adjacent
schwas (schwas in adjacent syllables) may never drop: */<lp#/ is impossible.
But this constraint is directly linked to that involving the VCCCX prohibition
illustrated in (39) above, provided we use the prohibition to govern deletion in
a directional fashion starting from the beginning of the phrase. Let us consider
the sentence je me le demande, with four schwas in sequence. If the schwa of
je is deleted (cf. [38] above), then the schwa of me is preceded by the two con-
sonants /<m/ and its deletion is blocked. If the schwa of je is retained, then the
schwa of me meets the condition VCCX (/<m/) and deletes: /<mldm#d/.
In this case, the schwa of demande ask is also preceded by a single consonant
and may also delete: /<mldm#d/. Alternatively, the rst and third rather than
the second and fourth schwas could delete, giving /<mldm#d/. This alternat-
ing pattern of schwa deletion is not universally respected, in that potentially
deletable schwas often remain, but the phenomenon has generated a consider-
able amount of interest, both descriptively and theoretically.68 A further limited
but fascinating phenomenon involving schwa refers to rhythm and syllable
count, a discussion to which we now turn.
4.5.4.2 4.5.4.2 A Rhythmic Constraint

We owe to Pierre Lon (e.g., Lon 1966: 73) the discovery of this particular
behaviour of schwa. Consider the forms in (41).
(41) A rhythmic constraint

porte-monnaie porte-plume wallet pen-holder
porte-crayon porte-cl pencil-case key-case
garde-malade garde-fou nurse railing
garde-mobile garde-cte mobile guard coastguard ship
perce-oreille perce-neige earwig snowdrop
gagne-petit gagne-pain low wage earner job
vide-ordures vide-poches garbage chute glove box
ouv(re)-bouteille ouvre-boite bottle opener can opener
In the left-hand column, there is a tendency for the nal schwa of the rst ele-
ment of the compound to delete, despite the fact that this produces a violation
of the VCCCX constraint (e.g., porte-monnaie /pntmnn'/). On the right,
in contrast, the schwa is retained. It was Lon who pointed out that if the
second element of the compound is monosyllabic, the schwa is retained,
while bisyllabic second elements favour schwa deletion. In each case, the
result is a trisyllabic compound. This pattern is often violated in that the
schwa of the rst column, even when followed by bisyllabic elements, may
be retained. Moreover, the phenomenon still respects additional conditions
governing schwa (deletion before a vowel: perce-oreille; retention after a CL
cluster: souffre-douleur whipping-boy; etc.), but, as we will see below, this
rhythmic pattern is not an isolated one in French phonology. It even extends,
at least optionally, to phrasal constructions such as il parle bas He speaks
softly. /ilpalba/ versus il parle beaucoup He speaks a lot. /ilpalboku/.
4.5.4.3 4.5.4.3 Epenthesis

Up to now, we have dealt with conditions favouring or blocking the deletion of
mute-e. Additional examples show the opposite effect: the insertion of a schwa
even when there is no orthographic justication for it. Consider the forms in (42).
(42) Epenthesis of schwa

lm russe /lmys/ Russian lm
ours brun /usb/ brown bear
t-shirt jaune /ti5t<on/ yellow t-shirt
arc-boutant /akbut#/ ying buttress
match nul /mat5nyl/ tie game
Parc des Princes /pakdep's/ (a stadium)
86 Chapter 4
This epenthesis is clearly linked to the earlier VCCCX constraint, although in

this case, the presence of a word boundary is necessary for epenthesis to take
place: VCC#CX, since insertion only occurs following words ending in two
consonants, not before words beginning with two consonants (sept Franaises
seven French women */s'tf#s'z/). Furthermore, it is more frequent when
the second word is monosyllabic than when it is polysyllabic, in keeping
with the rhythmic constraint on deletion illustrated in (41): concept cl key
concept /kns'ptkle/ versus concept difcile difcult concept /kns'ptdisil/,
ours brun brown bear /usb/ versus ours polaire polar bear /uspnl'/,
for example.
A second type of schwa epenthesis also seems to be gaining ground in SF:
that following // in phrase-nal position, as in bonjour hello, bonsoir
good evening, jen avais peur I was afraid of it, tout lheur see you
soon, prends le bustake the bus, soixante-quinz seventy-ve, cest pas
croyabl thats unbelievable, despite the fact that schwa does not usually
appear phrase-nally. This epenthesis may, in fact, be more general, and
may play a role in emphatic speech: quil est snob what a snob, arrte
stop!, merde shit!, and so on, as Lon (1992: 145-46) points out. Nor is
it unknown in word-internal contexts: consider the colloquial forms lorsque
when /lnsk/, exprs on purpose /'ksp'/. Here, we see the presence
of schwa whether or not is it supported by an < e > in the orthography. Just
how widespread and stable this phenomenon will become remains to be deter-
mined at the moment, it is in the early stages of development and remains
highly sporadic.69
4.5.4.4 4.5.4.4 The Tendency towards Irregularity

The preceding discussion has perhaps left the impression that the behaviour
of schwa, while complicated, is generally regular or predictable. While there
are obvious generalizations to be extracted from the data we have seen up to
this point, it is nonetheless necessary to recognize that nothing is as orderly
as it might rst seem. There is a myriad of additional material to analyze,
some apparently irregular because of stylistic variation, some due to the inu-
ence of phonotactic constraints, some simply attributable to lexical idiosyn-
cracy. Without attempting full coverage, we will summarize some of the most
notable exceptions in this section.70
Although we will treat this topic more fully in an independent discussion in
a later chapter (see section 5.4), we should note that here, as elsewhere, aspi-
rate-h words furnish a class of systematic exceptions to schwa deletion. We
have seen a strong constraint that blocks schwa preceding vowels: *V, both
word-internally and phrase-internally. Words beginning with aspirate-h, which
are in fact vowel-initial in phonetic terms, permit schwa to be pronounced: le
hros the hero /leo/, une housse a dust cover /ynus/, cette hache this
axe /s'ta5/, quelle honte what a shame /k'lnt/, and so on.71
Secondly, despite the general tendency for schwa to delete in phrase-initial

position following fricatives, there are words where this does not occur: celui-
ci/-l/que this one/that one/the one which /sli/; ce ne sont pas they
arent /snsnpa/ but not */snsnpa/; ce quoi that at which /sakwa/, not
*/sakwa/; for example. Furthermore, there are examples, albeit much more
colloquial, where deletion may occur even after a nasal or stop consonant, as
in the forms monsieur /msj/, petit voyou! little scamp /ptivwaju/, and so
on. (See also the examples in [45] below.) Conversely, there are cases in SF
where, despite the presence of orthographic < e >, schwa is never pronounced
(e.g., samedi Saturday, hameon sh hook, caneton duckling), and the
phonological rather than the orthographic form of these words would reect
the absence of any vowel. In addition, there are cases where schwa deletes
in very specic contexts involving clitic pronouns: (re)garde-la look at her
/()gadla/, for example. In more rapid (i.e., informal) speech, schwa may
even delete following two consonants: il fera hell make /ilfa/, vous ne
serez you wont /vunse/, dpartement department /depatm#/. (Recall
that schwa deletion after CL clusters is very rare word-internally in SF, though
not uncommon regionally).
Thirdly, there are patterns that seem to override the alternating deletion pat-
tern we saw in the discussion of sequences of schwas. Certain frequent deter-
miners, such as quelle, cette, une, have feminine forms ending in schwa. If
these determiners precede a word with initial schwa (e.g., une semaine one
week), the alternating pattern would lead us to expect either /ynsm'n/ or
/ynsm'n/, quelle menace what a menace as /k'lmnas/ or /k'lmnas/, and
so on. In fact, only the second possibility is found: the schwa of the determiner
deletes obligatorily. This might lead us to question whether the phonological
representation of such forms even includes a schwa, a question that has been
highly debated in theoretical circles. Be that as it may, here, as elsewhere in
sequences of schwas, there are patterns that are xed, and the alternating pos-
sibilities outlined above are not fully realized in many cases.
Fourthly, there are what might be called distance constraints on schwa
deletion, perhaps not unrelated to the rhythmic pattern we saw above. In
general, the further removed schwa is from the stressed syllable, the more
easily it may be deleted. Thus, in a sentence like il ne reste pas Hes not stay-
ing, the schwa in reste is obligatorily pronounced in SF: /iln'stpa/. In il
ne reste pas longtemps Hes not staying long, however, it may be deleted:
/iln'stpalntC/.
Finally, the distinction between inectional and noninectional morphology
is also relevant in a discussion of the behaviour of schwa. In rst conjugation
verbs (the productive category with innitives in -er) in future and conditional
forms, a schwa may be optionally deleted before the tense marker //, even if
two consonants precede, in violation of XCCCX from (39ii) above.
88 Chapter 4
(43) Inectional conditioning (future and conditional)

parlera /pal()a/ will speak
xera /ks()a/ will fasten
formera /fnm()a/ will form
forgera /fn<()a/ will forge
ripostera /ipnst()a/ will answer back
In forms such as forgeron blacksmith /fn<n/, palmeraie palm grove

/palm'/, fumisterie fraud /fymisti/, (or even porcelet piglet /pnsl'/)
with consonant groups analogous to those of the futures/conditionals in (43),
and where the schwa is obligatorily present, we see evidence that the inec-
tional forms behave differently. Thus, although there are general patterns that
clearly emerge when schwa is considered, there remains a large residue of
forms that continue to show considerable variability. This variability is com-
pounded when we consider stylistic variation, the topic of the next section.
4.5.4.5 4.5.4.5 Stylistic Considerations

The behaviour of schwa that we have considered up to this point is, despite
its complexity and heterogeneity, fully characteristic of SF. It demonstrates, in
fact, that SF itself is not the rigid and perfectly homogeneous linguistic system
it is sometimes taken to be. At the risk of complicating matters further, we will
consider two additional areas relevant to a discussion of this phenomenon. The
rst involves the division of utterances into phonological phrases; the second
involves the formality of the speech situation.
Looked at closely, there is somewhat of a contradiction involving the inter-
action between division into phrases, degree of formality, and schwa reten-
tion. Formal speech tends to be slower and more deliberate, and phrases tend
to be shorter. But because phrases are shorter, more schwas will potentially
fall in phrase-nal position and be deleted, reducing rather than increasing
the number of schwas pronounced. The contradiction arises because formal
speech normally implies more schwas. The following examples, with the more
formal pronunciation (and shorter phrases) occurring rst, illustrate this situa-
tion.
(44) Short versus long phonological phrases
(a) / || le thtre || fermera vingt heures || / / || lteat || f'maav't || /

/ || le thtre fermera vingt heures || / / || lteatf'maav't || /
The theatre will close at 8:00 p.m.
(b) / || dans quelle chambre || dorment-ils || / / || d#k'l5#b || dnmtil || /

/ || dans quelle chambre dorment-ils || / / || d#k'l5#bdnmtil || /
In which room are they sleeping?
In (a) and (b), we see that the schwa at the end of the rst phrase has been
deleted in the slower and more formal versions. In the single longer phrase,
because the schwa in question is preceded by two consonants phrase-inter-
nally, it is now pronounced. However, in (c) we can see the opposite effect
combining shorter phrases into one longer one creates a context in which
schwa may now be deleted:
(c) / || je ne sais pas || le trouver facilement || / / || <ns'pa || ltuvefasilm# || /

/ || je ne sais pas le trouver facilement || / / || <ns'paltuvefasilm# || /
I dont know how to nd it easily.
If we ignore the implications of shorter versus longer phonological phrases,

there is a clear correlation between the number of times schwa is pronounced
and the formality of the speech situation: the more schwa is pronounced, the
more formal is the discourse. Consequently, we will look briey at two further
situations, a more formal one involving poetry, song, or formal oratory where
schwa is maximally preserved, and a popular or colloquial variety of speech
that suppresses schwa (and often other segments) as much as possible.
Poetry and related forms are perhaps the easiest to describe. In classical
verse, and in solemn public declarations and related performances, all schwas
followed by a consonant are pronounced. Rimbauds line Je ne parlerai pas,
je ne penserai rien has twelve syllables; de Gaulles declaration Cela, je ne le
ferai pas! has eight. Songs by Edith Piaf provide further testimony: Non, je ne
regrette rien, for example, has seven syllables. Needless to say, the situation
is much more complicated than this sketch implies,72 but the general tenor of
SF is appropriately summarized in the following passage by Lon (1992: 147):
Mme en labsence de toute donne scientique, on peut afrmer que dans
la conscience linguistique dun Franais, le E caduc prononc est associ avec
lide du beau langage.
If it is fair to say that the most formal varieties of French retain all possible
schwas, then colloquial French suppresses them maximally, normally in con-
junction with additional phonological changes. For example, in SF il venait
it was coming is pronounced /ilvn'/, and deletion of the schwa is blocked
by the two preceding consonants. In colloquial speech, on the other hand, il is
normally reduced to /i/, so il venait is /ivn'/. Moreover, the negative particle ne
is also usually absent, so vous ne venez pas youre not coming /vunvnepa/
would be /vuvnepa/. Deletion of word- or phrase-nal schwa following CL
clusters also exposes the liquid to deletion, so one hears table dinners
served /atab/, jen ai quatre Ive got four /<#nekat/, quatre voitures four
90 Chapter 4
cars /katvwaty/, and so on. In fact, the deletion of vowels goes far beyond
that of schwa in colloquial or popular French, as the examples in (45) illus-
trate.73
(45) Vowel deletion in colloquial French

tu as /ta/ you have
tu sais /ts'/ you know
vous tes /z't/ you are
cet homme /stnm/ this man
cet t /stete/ this summer
peut-tre /pt't/ perhaps
mais enn /m#f'/ nally
pas tout fait /pattaf'/ not completely
celui-l /sila/ that one
(less frequently: /slila/)
voil /vla/ there
cest dire /stadi/ that is to say
quest-ce que cest que a /k'ks'ksa/ Whats that?
pourquoi est-ce que /pukwask/ why
o est-ce que /usk/ where
The medial vowel in words like djeuner to eat lunch /de<ne/ also deletes on
occasion. This phenomenon, paradoxically, allows us to introduce a discussion
of stable schwas.
4.5.4.6 4.5.4.6 The Stabilization of Schwa

According to a number of recent analyses,74 SF has a large number of words
where schwas have apparently lost their status as e-caduc or e-instable and
instead fail to delete even when all conditions for their suppression are met.
Such schwas occur typically in the rst syllable of a word preceded by a single
consonant; these vowels normally delete in the phrasal context VCCX. Thus,
trois semaines three weeks /twasm'n/ but sept semaines seven weeks
/s'tsm'n/ shows the typical pattern of deletion, while trois vedettes three
stars, exclusively /twavd't/, shows a stable schwa. Examples of such stable
schwas are found in the initial syllables of the words in (46).
(46) Stable schwas

bedaine paunch
belote belote
degr degree
depuis since
gense genesis
guenon female monkey
menotte handcuffs
neveu nephew
pesant heavy
querelle quarrel
requin shark
secret secret
vedette star
Walker (1996) examines some 400 such words, plus over 800 beginning with
the prex re-. The schwa of re- remains unstable, but fully half of the remain-
ing words with initial schwa show either a stable vowel or one that is becom-
ing stable (cf. Warnants mais le se maintient souvent comment). The phe-
nomenon, then, is widespread.
Two complementary factors appear to be aiding this stabilization process.
First, as Walter and others have noted, there is a tendency in contemporary
French to expand the use of an accent dinsistance or emphatic stress, placed
on the rst syllable of words. Stress on the rst syllable would obviously hinder
any vowel deletion in that syllable, including deletion of schwa. Secondly, we
may return to the merger of schwa and // discussed above in 4.5.3. //, if we
exclude the recent invasion from schwa, is clearly a stable vowel in SF. To
the extent, then, that schwa becomes identied with //, it is not unreasonable
to expect that it also adopt the stable properties of the new vowel and begin
to resist deletion.75 The dynamism of the SF phonological system is, therefore,
made apparent in the interaction of these three innovations: the occurrence of
stress on initial syllables expands, schwa merges with the stable vowel //,
schwa stabilizes in initial position. Whether the merger and stabilization of
schwa will become permanent features of SF phonology is a question only the
future can answer. In the meantime, we must consider further complexities of
the SF phonological and morphological system linked to this enigmatic vowel.
92 Chapter 4
4.5.5 4.5.5 Alternations Involving Schwa

Several elements of French inectional and derivational morphology involve
alternations between schwa and other vowels. While it is impossible to treat
these questions fully here, some indication of the extent and the complexity
of the issues is appropriate. In inectional terms, alternations involving schwa
are most evident in the verb system, in the classic cases illustrated by mne
menons lead or appelle appelons call. Derivationally, there are also many
sets related by the presence of a schwa in unstressed position paired with vari-
ous stressed vowels: htel htelier hotel hotel keeper /ot'l/ /otlje/,
chapeau chapelier hat hatter /5apo/ /5aplje/ and a host of less frequent
and more irregular pairings.76
The most pervasive and regular of the schwa-vowel alternations occurs in
verb forms, where stressed /'/ alternates, in most forms, with unstressed //,
the latter also subject on many occasions to the normal deletion processes.
There are approximately 170 verbs in SF that fall into this category; they are
often subdivided according to their orthography, depending on whether the
stem-nal consonant is doubled or whether an accent is added to the stem
vowel.77 Examples are given in (47).
(47) /'/ // in verb stems

// or deletion /'/ // or deletion
(Innitive) (3 sg.) (1 pl.)
acheter achte achetons buy
crever crve crevons burst
haleter halte haletons pant
jeter jette jetons throw
feuilleter feuillette feuilletons leaf through
soufeter soufette soufetons slap
appeler appelle appelons call
dbosseler dbosselle dbosselons atten out
morceler morcelle morcelons divide up
In these verbs, the alternation between /'/ and // is found in all nite forms,
with the exception of the future and the conditional, which (in the normative
description of SF corresponding to the ofcial spelling) present /'/ uniformly
throughout, despite a following syllable containing schwa and despite the
absence of stress: achterai buy /a5't()e/;78 achterions /a5'tjn/; appelle-
rais call /ap'l()'/; appellerions /ap'ljn/, etc. This and similar alter-
nations have generated much theoretical debate under the name of closed
syllable adjustment, particularly since in forms like appellerions, the // alter-

nating with /'/ does not appear to be in a closed syllable.79
The situation is equally, if not more complicated when we investigate deri-
vational morphology. We will consider here only a single case: that concerning
the behaviour of stems whose nal vowel, when stressed, is /'/. When a sufx
is added to such stems, the /'/ should then reduce to schwa and be subject to
deletion if the appropriate conditions (essentially the familiar VCCX pattern
or a following vowel) are present. This is indeed the case for many types of
forms, as illustrated in (48).
(48) Schwa in derivational morphology

/'/ //
crochet crochetage hook hooking
promne promenade walk a walk
pucelle pucelage virgin virginity
chapelle chapelain chapel chaplain
duvet duveteux down downy
jumelle jumelage twin (f.) twinning
Genve genevois Geneva Genevan
vilaine vilenie nasty (f.) vileness
If the sufx begins with a consonant (a rare occurrence in SF) or with

schwa (much more frequent) this situation is again quasi-regular, since the
stem vowel will occur in a closed syllable and therefore not reduce: vilaine-
ment wickedly /vi.l'n.m#/, tincellement glitter /e.t'.s'l.m#/ (cf. tincelant
sparkling / e.t'.sl#/ /eV'UN#/), etc. The rst type of difculty arises when
we consider a set of learned sufxes, -atif, -ateur, -ation, -isme, -iste, and so
on. Despite the fact that these sufxes begin with a vowel and thus create a
preceding open syllable, reduction of /'/ to schwa does not occur, as in (49):
(49) Learned sufxes

/'-e/ //
appellatif appeler appellative to call
conglateur congeler freezer to freeze
dnivellation niveler unevenness to make uneven
rcipiendaire recevoir recipient to receive
nouvelliste renouveler short story writer to renew
94 Chapter 4
perception percevoir perception to perceive

lvation lever elevation to raise
The learned status of a word, here and elsewhere, serves to block the operation
of a number of synchronic processes and provides us with a reminder of the
phonological history of French.80
Schwa /'/ alternations are not the only ones involved in such cases. There
exists a large set of less frequent, less productive derivational relationships as
well. One type is represented in (50), where a link between /o/ (orthographi-
cally < eau > in most cases) and /l/ (< el >) is evident:
(50) /o/ /l/

bateau batelier boat boatman
cerveau cervelet brain cerebellum
chapeau chapelier hat hatter
ciseau ciseler scissors to chisel
crneau crnel crenel crenellated
jumeau jumelage twin to twin
morceau morceler piece to divide
niveau niveler level to level
Such minor cases may be extended almost indenitely, as we see in (51).
(51) Minor alternations involving schwa

tenace tnacit / e/ tenacious tenacity
cafetire caf / e/ coffee pot coffee
dangereux danger / e/ dangerous danger
mercerie mercire / j'/ notions shop
haberdasher
chenil chien / j'/ kennels dog
chasseresse chasseur / / huntress hunter
chevelure cheveu / / hair hair
peser poids / wa/ to weigh weight
champenois Champagne / a/ from Champagne
Champagne
menotte main / '/ handcuff hand
To pursue such alternations would take us too far aeld. We will return, on
occasion, to the question of learned versus nonlearned forms in French. For the
time being, however, let us consider one last role played by mute-e, a role with
dramatic phonological and morphological implications.
Two of the key elements of French structure involve differences between
masculine and feminine forms on the morphological side and the pronuncia-
tion of nal consonants in the phonological domain. Schwa plays a key role
in relating these two areas. If we restrict ourselves simply to adjectives, many
feminine forms have nal < e > in the orthography, compared to a lack of this
vowel in the masculine:81
(52) < e > in the feminine

petit petite /-t/ small
grand grande /-d/ big
long longue /-g/ long
gros grosse /-s/ fat
jaloux jalouse /-z/ jealous
gentil gentille /-j/ kind
sol sole /-l/ drunk
premier premire /-/ rst
sain saine /-n/ healthy
In phonological terms, however, the distinction in these forms is between the

presence and absence of a nal consonant. Most discussion has focused on the
best way to represent this distinction in phonological terms: (i) a representation
that mirrors the orthography and includes both a nal schwa and a nal conso-
nant (/pti+t+/); (ii) a representation without schwa but with a nal consonant,
necessitating schwa insertion on various occasions (/pti+t/); (iii) a representa-
tion lacking both nal consonants and schwa, and requiring a variety of rules
for both consonant and schwa insertion (/pti/). Whatever the solution chosen
(we will discuss some of the issues later), it is clear that any approach involves
signicant decisions regarding the functioning of schwa in French phonology.
4.5.6 4.5.6 Dialects and History

The behaviour of schwa across dialects of French is of a complexity equal to
that found in SF alone. CF, for example, appears to preserve the vowel as [],
distinct from [] and [], and popular forms of many dialects (including CF)
manifest an interesting type of metathesis: /gade/ (regarder to look at),
/lgasn/ (le garon the boy), /fevje/ (fvrier February), and so on.82 But
96 Chapter 4
perhaps the best recognized example of the distinctive behaviour of schwa is

found in MF, where the general (but not fully correct) impression is that all
mute-e are pronounced.80 Examples are found in (53).
(53) Schwa in MF
bte /b't/ animal
btise /betiz/ stupidity
tellement /t'lmC0/ so
la carte de la ville /lakatdlavil/ the map of the city
je ne vois ni Pierre /<nvwanipj'ni<ak/ I see neither Pierre
ni Jacques nor Jacques.
Needless to say, the situation in MF is not nearly this homogeneous, particu-

larly as the region falls under the inuence of northern dialects. Nonetheless,
the behaviour of schwa is highly distinctive and widely recognized as one of
the primary signals of southern speech.
Schwa in SF is the product of a variety of historical phonological processes,
and some discussion of these phenomena may help clarify the details of the
preceding sections. In the developments that led from spoken Latin through
Gallo-Roman to Old French, many unstressed vowels, particularly those in
nal syllables or in nonnal open syllables, were reduced to schwa and sub-
sequently deleted. Such deletion normally affected all Late Latin vowels
except /a/, this vowel resisting deletion and remaining as schwa. Furthermore,
because of constraints on syllable structure, certain nal consonant clusters
were proscribed in Gallo-Roman and OF, and to block their appearance in
word-nal position, a second type of schwa vowel, a voyelle dappui, devel-
oped in that context. Examples of vowel loss, of /a/ becoming //, and of nal
support vowels are given in (54).
(54) Vowel loss and development of schwa

Latin French
(a) loss of nal vowels
viginti vingt twenty
venit vient comes
habere avoir to have
scriptum crit writes
ferrum fer iron
amarum amer bitter
murus mur(s) wall

largum larc (> large) wide
grandem grant (> grand) big
(b) /a/ > //
dura dure hard
luna lune moon
alba aube dawn
pluma plume pen
amas aimes loves
(c) // as a support vowel
patrem pre (OF pedre) father
duplum double double
nostrum ntre our
inter entre between
juvenum jeune (OF juevne) young
numerum nombre number
hospitem hte (OF hoste) host
comitem comte count
Without going into detail, we may note that several of these forms show
the interaction of phonological changes. For example, the consonant cluster
requiring a support vowel in SF words like jeune or hte is no longer transpar-
ent, since changes subsequent to the development of that vowel have elimi-
nated the cluster. In words like inter or comitem, a consonant cluster (and the
consequent need for a support vowel) is present only after the syncope of the
penultimate vowel.
Subsequent to the OF period, vowel loss continues, affecting the remaining
weak vowel, namely schwa. As we have seen, this process is still going on.
The examples in (54) allow us to understand some of the synchronic details we
have seen above. First, let us consider the constraint excluding schwa (word-
internally) from closed syllables. If the diachronic process of vowel reduction
did not fully apply in closed syllables, then it is evident that schwa would not
be produced in that context, and its absence synchronically is easily under-
stood. In the same way, if schwa is the result of a weakening in unstressed
syllables or of the insertion of a nal support vowel following the stressed
syllable, the absence of stressed schwas is again not surprising. Finally, the
partial correlation between schwa (orthographic < e >) and feminine nouns
98 Chapter 4
and adjectives may be claried. In Late Latin, the antecedent of Gallo-Roman

and OF, the great majority of feminine nouns and adjectives, those belonging
to the rst declensional class, had case sufxes with /-a/ as the characteristic
vowel. As the language evolved, this noun class retained the original gender
(and attracted, on phonological grounds, additional members, specically a
number of Latin neuter plurals with a collective meaning and also ending in
/-a/: folia [feuille leaf], festa [fte holiday], labra [lvre lip], vela [voile
sail], etc.). Because the nal /a/ was a feminine marker and because OF /-/
is the direct continuation of this marker, // became as well a general signal
of feminine gender in French. The correlation is less than perfect, however,
because nal // in OF has a second source as a support vowel following a
consonant cluster, independent of the gender of the noun or adjective involved.
Hence, we see many OF masculine nouns or adjectives also ending in //, lead-
ing to much of the ambiguity in the formal expression of gender in the modern
language. Examples are found in (55).
(55) Ambiguity of nal //

(a) feminine nouns and adjectives
robe dress
salade salad
ville city
mine expression
petite (m. petit) small
grosse (m. gros) fat
grande (m. grand) big
bonne (m. bon) good
(b) masculine nouns and epicene adjectives
clotre cloister
type type
maire mayor
pre bitter
jeune young
double double
vide empty
tide luke warm
Nonetheless, a good rst approximation in attempts to determine the gender of

French substantives is provided by an examination of the nal segment of the
word: nal schwa often implies feminine gender.84
Needless to say, these historical developments are not without exception
(and are, in fact, much more complex than we have been able to outline here).
It is useful to discuss, however, one recurrent and more general type of excep-
tion that provides a contrast to vowel reduction and that is also reected in
alternations involving schwa: further evidence supplied by learned or bor-
rowed words. A number of such words, from the OF period onwards, fail to
show reduced vowels, as in the word-internal /a/ in the following OF words:
amaritude bitterness (cf. amertume); citadelle citadel, from Italian citt-
adella; medicamental medicinal (cf. medecine); paradiset small paradise;
and so on. More important for our understanding of derivational morphology,
however, are pairs of learned and nonlearned words where schwa reduction
(among other changes) is absent in the learned forms in comparison with their
popular counterparts, as in (56).
(56) Nonreduced learned vowels versus reduced popular vowels

Learned Popular
appellation appeler designation to call
conglation congeler freezing to freeze
lvation lever elevation to raise
gnuexion genou genuection knee
npotisme neveu nepotism nephew
projectile projeter projectile to plan
rbellion rebelle rebellion rebel
rception recevoir reception to receive
tnacit tenace tenacity tenacious
divination deviner foretelling to guess
satisfaction satisfaisant satisfaction satisfying
Discussion of schwa in French, in other words, puts us immediately in con-

tact with questions of phonetic detail, distributional constraints, orthography,
the lexicon, alternations between grammatically related forms, the phonologi-
cal phrase, historical considerations, stylistic and regional variation, and, no
doubt, any other area of phonological investigation one might wish to name.
It is little wonder, therefore, that schwa continues to preoccupy students and
phonologists alike. Given the dynamics of the SF vowel system, this situation
is unlikely to change.
100 Chapter 4
4.6 4.6 Semi-vowels

The SF phonological system contains the three semi-vowels (or glides, some-
times also called semi-consonants) /j w/. In articulatory terms, these semi-
vowels correspond to the three high vowels /i y u/, being front unrounded,
front rounded, and back rounded, respectively, but they lack syllabic status.
These segments are illustrated in (57) and (58).
(57) French semi-vowels

/j/ // /w/
iambe iambic huer to boo ouailles ock
iode iodine huit eight ouate cotton-
wool
ionique Ionic huile oil ouest west
iota iota huissier usher oui yes
yacht yacht cuiller to gather wallon Walloon
yaourt yogurt lui him week-end weekend
yeuse oak nuit night whisky whisky
yeux eyes puis then jouer to play
yiddish Yiddish suer to sweat louer to rent
yoga yoga aiguille needle avouer to confess
youpi yippee annuel annual panoui in full
bloom
yucca yucca essuyer to wipe secouer to shake
hier yesterday graduel gradual anchois anchovy
hirarque hierarchic habituel habitual bourgeois middle
class
bien well insinuer to insinuate foin hay
liaison liaison saluer to greet moi me
lion lion usuel usual moins less
pied foot moyen medium
scier to saw noyer to drown
tiers third roidir to stiffen
viable viable soir evening
vieux old
balayer to sweep
effeuiller to thin out the leaves
gayer to cheer up
essayer to try
veiller to awaken
lleul goddaughter
loyal loyal
payer to pay
vieillir to age
ail garlic
bail rent
mail enamel
rail rail
abeille bee
brille shines
lle daughter
maille stitch
(58) CL clusters and vowel semi-vowel alternations

/ijV/ // /[/ /w/ /u/
// /w/
bouclier shield bruine drizzle adroit adroit
brivement briey bruit noise Blois Blois
client client druide Druid croire to believe
fabliau fable fruit fruit croix cross
grief grief pluie rain gloire glory
lvrier greyhound truite trout octroyer to grant
pliable pliable /y/ trois three
prieur prior cruel cruel /u/
quatrime fourth uette slender brouette wheelbarrow
102 Chapter 4
sablier hourglass uide uid clouer to nail

triage sorting out inuer to inuence prouesse feat
triomphe triumph truand gangster trouer to pierce
vitriol vitriol truelle trowel
While their articulatory properties may be straightforward, these sounds mani-

fest a series of other characteristics involving their orthographic representation,
their distribution, and particularly their participation in a series of alternations
with the corresponding high vowels. We will deal with each in turn.
4.6.1 4.6.1 Orthographic Representations

of the Semi-vowels
Perhaps the most straightforward orthographic link is that between // and
< uV >. That is, // is uniformly represented by the letter , and the letter
 in that case must itself be prevocalic. (Not all prevocalic are //,
however, as we shall see.) Word-initially, the may be preceded by an
< h >, either aspirate (huit eight) or nonaspirate (huile oil), and // may
occur in word-initial or word-medial position. A consonant may precede // in
either of these contexts.
The back rounded semi-vowel /w/ is slightly more complicated. The normal
representation is < ouV >, initial or medial, preceded or not by a consonant
(or by < h > initially). However, certain instances of < w > are also realized as
/w/, while other instances of < w > are pronounced /v/ (whisky versus wagon).
The sequences < oi > and < oy >, moreover, require separate treatment: their
normal interpretation is /wa/ or, in nasalizing contexts, /w'/ (mois; moyen;
moins).
Finally, the status of /j/ is more complicated, both orthographically and pho-
nologically. Orthographically, /j/ is represented primarily by < y > in word-
initial position, although there are also several instances of , some pre-
ceded by < h >: yaourt yogurt, iode iodine, hier yesterday.85 Immediately
following a consonant, word initially or word medially, /j/ is written with :
bien well, dernier last. Intervocalically, /j/ is divided between < y > and
< ill > (or < ll > following /i/): essayer to try, veiller to stay up, briller to
shine, with < ill > predominating numerically. In word-nal position, /j/ is
divided between < il > (ail garlic, deuil bereavement, ventail fan, rail
rail, sommeil sleep) and < ille > (abeille bee, canaille cheap, cdille
cedilla, feuille leaf, paille straw, rouille rust).86 Here again, there are sub-
regularities: only < ille > occurs following /u/ (< ou >); < ll > is used instead of
< ill > after /i/ (), and only < lle > occurs in that context (with the excep-
tions of the loanwords drill and mandrill), and so on. Difculties arise with a
few forms where < ill(e) > is /il/, not /ij/: mille thousand (plus millsime and
other derivatives), ville city (plus village and other derivatives), tranquille
tranquil, distiller to distil, and osciller to oscillate are the most common,
but the usual interpretation of < ill > involves /j/. Let us now move to a con-
sideration of the phonology of the semi-vowels.
4.6.2 4.6.2 The Phonology of the Semi-vowels

Phonologically, the semi-vowels raise a number of interesting questions, the
rst involving their independent phonological status. Here, we must distin-
guish between /j/ and / - w/. The latter two segments are highly restricted
in their occurrence: they occur only in prevocalic position,87 and many ana-
lysts conclude that they are not independent phonemes. The basis for this con-
clusion, given the limitation of these segments to prevocalic position, is the
absence of a phonemic contrast between [y] and [] or [u] and [w] before
vowels; that is, the impossibility of a contrast between, for example, [yi] and
[i] or [ui] and [wi] in that position. From this, it is concluded that [] and
[w] are predictable positional variants (allophones) of /y/ and /u/, respectively.
The situation involving these segments is, however, slightly more compli-
cated. First, we must take into account the constraints illustrated in (58) above.
There, we see examples of [y] and [w] preceding vowels, but these sounds
are themselves preceded by CL clusters (cruel /ky'l/, uide /yid/, brouette
/bu't/, clouer /klue/). If the larger context is taken into consideration, it
could be argued that the clusters block the appearance of the glides and that
the allophonic status of [] and [w] is preserved, since their predictability
is maintained. This argument is contradicted, unfortunately, by the existence
of words like monosyllabic (and monomorphemic) bruit /bi/, pluie /pli/,
Blois /blwa/, croix /kwa/. To this, we may add the well-known examples loi
law /lwa/ versus (il) loua (he) praised /lua/, trois three /twa/ versus (elle)
troua (she) pierced /tua/, which also seem to argue for a phonemic distinc-
tion (although perhaps somewhat articially, given the place of pass simple
forms in spoken French). This being said, the forms with the glides following
CL clusters are clearly in the minority, are of marked status, and often have
an explanation based on their etymology.88 Moreover, as we will see below,
there are additional arguments involving two types of semi-vowels that sup-
port phonemic status for // and /w/. As a result, we will, for the sake of con-
venience and where warranted, continue to use the symbols // and /w/, even
if in many cases the semi-vocalic status of these segments is predictable.
The case of /j/ is different in several ways. First, unlike its round counter-
parts, /j/ may occur postvocalically (as syllable coda) and preconsonantally as
well as prevocalically: paille straw /paj/, empaillement to stuff /#pajm#/,
feuille leaf /fj/, feuilleter to leaf through /fjte/, and so on. Secondly, there
are the well-known postvocalic /i/ /j/ contrasts in pays country /pei/ versus
paye pay /p'j/ or abbaye abbey /abei/ versus abeille bee /ab'j/. Further,
unlike the marked but possible appearance of // and /w/ following CL clusters
(pluie /pli/, gloire /glwa/), no instances of CLj occur in SF. We nd, instead
104 Chapter 4
of *CLjV sequences, the outcome CLijV: /plije/, not */plje/, for plier to fold;
/tablije/, not */tablje/, for tablier apron; and so on. Nor are we nished with
the idiosyncracies of /j/. This glide is also inserted automatically as a transi-
tional segment whenever the diphthong < oi/oy > (/wa/) occurs prevocalically:
voit voyons see /vwa/ /vwajn/, croit croyez believe /kwa/ /kwaje/,
envoi envoyer shipment to send /#vwa/ /#vwaje/, joie joyeux joy
joyful /<wa/ /<waj/, loi loyal law loyal /lwa/ /lwajal/, and so on. In
fact, this transitional glide is even more widespread: balai balayer broom
to sweep /bal'/ /baleje/, ennui ennuyer boredom to bore /#ni/
/#nije/, appui appuyer support to lean on /api/ /apije/, and so on.
Finally, because of the presence of a number of sufxes in French that begin
with /j/, such as -ions, -iez, -ier, we nd frequent occurrences of geminate
/jj/, especially in imperfect or subjunctive forms of verbs when the verb stem
itself ends in /j/: cueuilliez were gathering /kjje/, fouillions were search-
ing /fujjn/, as well as aiguilliez were directing /'gijje/, although less formal
speech allows pronunciation with a single /j/. In fact, those verbs that require
a /j/ to be inserted before a vowel-initial sufx (envoie envoyer /Cvwa/
/Cvwaje/, appuie appuyer /api/ /apije/, etc.) retain this /j/ before those
imperfect or subjunctive sufxes that begin with /j/, creating a large number
of additional /jj/ geminates, as in envoyions /Cvwajjn/, appuyiez /apijje/. All
these different processes demonstrate that the case of /j/ is unlike that of the
other semi-vowels // and /w/, in that it is impossible to derive /j/ from prevo-
calic /i/. As a consequence, the independent phonemic status of /j/ seems far
less controversial, in fact necessary.
We may now turn to another major alternation in SF: the alternation between
the high vowels /i y u/ and the corresponding semi-vowels in morphologically
complex forms. This alternation occurs across morpheme boundaries and is
most easily illustrated with vowel-nal verb stems, which (other things being
equal) have the full vowel in the singular indicative and the semi-vowel in the
innitive, as in (59).
(59) Vowel and semi-vowel alternations

tudie tudier /i/ /j+e/ to study
modie modier to modify
manie manier to handle
distribue distribuer /y/ /+e/ to distribute
remue remuer to twitch
accentue accentuer to accentuate
secoue secouer /u/ /w+e/ to shake
choue chouer to fail
avoue avouer to confess
Vowel-initial derivational sufxes also condition this alternation: mari mar-

iage husband marriage /maja</, colonie colonial colony colonial
/knlnnjal/, relie relieur joins bookbinder /lj/, ignominie ignomi-
nieux ignominy ignominious /ignnminj/, and so on. As might be expected
given the constraint discussed above, the alternation is blocked by a preced-
ing CL cluster (crie crier cry /kije/, conue conuer ow together
/knye/, cloue clouer nail /klue/), and conversion to a semi-vowel is even
optional for many speakers when only a single (or no) consonant precedes, as
a glance at Martinet and Walter (1973) or Juilland (1965) will conrm: hier
yesterday /j'/ or /ij'/, ruelle alley /'l/ or /y'l/, boue buoy /bwe/ or
/bue/, and many others.89 Lastly in this context, we may also note that the alter-
nation may take place across word (or clitic) boundaries, as illustrated in (60).
This alternation, unlike the stylistically neutral gliding that occurs word-inter-
nally, is found in rapid speech, whether that speech is formal or not (although
greater rapidity often correlates with lesser formality).
(60) Gliding across word boundaries

o tes-vous all /w'tvuale/ Where did you go?
si elle venait /sj'lvn'/ if she came
ni avant, ni aprs /njav#njap'/ neither before, nor after
tu arrives tard /taivta/ Youre arriving late.
puis elle disait /pj'ldiz'/ then she said
Finally in this section we should note a phenomenon involving glides that is

reminiscent of the behaviour of aspirate-h and in fact intersects with it in cer-
tain cases. Semi-vowels in French appear to behave in two separate ways, one
analogous to vowels, one to consonants, in the same way that h-initial words
are partitioned into two classes. Some glides, that is, permit liaison and elision
of schwa just like vowel-initial words (61b); others block it in consonantal
fashion, as in (61a).
(61) Two types of glides

(a) consonantal glides (b) vocalic glides
il le jodle /illjndl/ He yodels it. liode /ljnd/ the
iodine
le yaourt /ljau/ the yogurt les yeux /lezj/ the eyes
le yoga /ljnga/ the yoga lyeuse /ljz/ the oak
la hirarchie /lajea5i/ the hierarchy lhiatus /ljatys/ the
hiatus
106 Chapter 4
la hue /lae/ the booing les hutres /lezit/ the

oysters
le huitime /litj'm/ the eighth lhuile /lil/ the oil
lhuissier /lisje/ the
hussar
le ouistiti /lwistiti/ the marmoset louest /lw'st/ the west
le western /lw'st'n/ the western loue /lwi/ hearing
There is, not surprisingly, variation in additional items: lhyne or la hyne

hyena, dhier or de hier of yesterday, louate or la ouate cotton-wool, as
well as the frequent (but nonstandard) le iambe the iamb or le hiatus the
hiatus alongside liambe and lhiatus. Here, as elsewhere, one must simply
recognize the irregularity, as must be done as well with aspirate-h in general,
as we shall see below. This does not, of course, preclude theoretical discus-
sion of how this irregularity should be described, again as we shall see. At
this stage, however, we may leave the discussion of specic aspects of the SF
vowel system, and return to a review of two more general matters.
4.7 4.7 Further Effects of the Phonological Phrase

In much of the preceding discussion, we have referred to distributional con-
straints that regulate the appearance of a number of SF sounds, normally by
excluding them from specic contexts. These constraints have been formulated
largely with respect to the notion of the word. As is well known, however, the
role of the word is often subordinated to that of the phonological phrase in
French phonology,90 and it is worthwhile to examine the status of several of
these distributional constraints in a phrasal context. The patterns with which
we are concerned are repeated in (62).
(62) Summary of distributional constraints

(a) mid vowels: /e/ does not appear in closed syllables
/n/ and // do not appear in word-nal open
syllables
(b) nasal vowels: nasal vowels may not precede vowels or nasal
consonants
(c) schwa: schwa does not appear prevocalically or in
closed syllables
When words (including clitic pronouns and other dependent morphemes) are
put together to form phonological phrases, additional processes take place that
often override word-level constraints. This is the case, in fact, with the restric-
tions in (62), each of which is violated at the phrasal level. The mechanisms
directly responsible for these violations include, in addition to the simple con-
catenation of words, the processes of schwa deletion and the resyllabication
involved with enchanement and liaison.
The simplest examples are provided by aspirate-h words. When morphemes
ending in schwa or in a nasal vowel precede such forms, we nd both //
and nasal vowels in prevocalic position: le hros the hero /leo/, cette
housse this slipcover /s'tus/, un hros /eo/, mon hros /mneo/, and
so on. Likewise, sequences of words freely bring together nasal vowels and
nasal consonants, either because of liaison following a nasal vowel (en avril
in April /#navil/, un homme a man /nnm/) or because the second word
begins with a nasal consonant (en mai in May /#m'/, un matin one morn-
ing /mat'/). Schwa deletion also leads to contradiction of word-level restric-
tions. Whenever there are sequences of schwas in which deletion occurs, the
remaining schwas are inevitably in closed syllables: je ne te le dirai pas I
wont tell you. /<n.tl.diepa/, for example. Such deletion can also yield /e/
in a closed syllable: une heure et demie an hour and a half /yned.mi/,
ces remarques these remarks /se.mak/, and so on. Finally, resyllabica-
tion across word boundaries (enchanement) within the phonological phrase
moves a word-nal consonant to the beginning of the following word, poten-
tially leaving // or /n/ in word-nal position: fort intressant very interest-
ing /fn#'te's#/, une peur abominable a terrible fear /ynp#abnminabl/,
jeune homme young man /<#nnm/, and so on.
Despite the ease with which the above constraints are violated within
phrases, they remain generally valid at the level of the word, as is shown by the
adaptation of loan words, for example. SF speakers pronounce steak as /st'k/,
not */stek/; after-shave as /aft5'v/, not */5ev/;91 camping as /k#pi0/, not
*/k#mpi0/; tango as /t#go/, not */t#ngo/; and so on.92 The persistence of these
constraints argues, pace Delattre, for the continued relevance of the word as
a phonetic and phonological unit highly relevant to the functioning of French
phonology.93
4.8 4.8 Concluding Remarks

To conclude this discussion of the SF vowel system, we will return to the ques-
tion of variation to consider what might be the state of the SF vowel system
if all of the mergers we have indicated as in progress are fully realized. The
changes in question involve // - /'/, /a/ - /#/, // - // and the complete neu-
tralization of the mid vowels based on syllable structure (the loi de position).
Implementing all of these changes would produce the vowel system in (63).
108 Chapter 4
(63) The reduced vowel system

i y u
E O 94
' n
a #
In typological terms, this system is much less complicated than that of (1)
in section 4.0 above, and is even a system instantiated in certain dialects or
sociolects of French. Valdman (1993), for example, proposes a related simpli-
ed system as an appropriate starting point for learners of French as a second
language. We have retained the maximal system in this work, however, since
the functioning of the phonology itself, independent of second-language peda-
gogical considerations, appears best illustrated if we move from more to less
rather than from less to more, from reduction of the system rather than to
its expansion. The same approach will be applied in a consideration of the
SF consonant system, although, as we will see, the consonants raise far fewer
questions than do the vowels.
Notes
1. /e/ is rare in nal closed syllables, and largely if not exclusively restricted to
loan words from English: ale /el/, attach-case /ata5ekez/, date /det/, mail /mel/,
etc. Even here, there is variation, since Rey-Debove and Gagnon (1980) list ale
and cake as /'l/ and /k'k/ respectively, along with numerous additional words
adapted with /'/ from English /e/: airdale, brain, break, claim, cornakes, grape-
fruit, maid, milk-shake, raid, skate, teen-age, trade-union, and up-to-date are all
pronounced with /'/ according to their description. No doubt a greater familiarity
with English since that work was prepared has permitted the recent adaptation of
words with /e/ rather than /'/.
2. // and /w/ do not appear post-vocalically, nor may they appear in word-nal posi-
tion (the two contexts are not mutually exclusive). Cacah(o)ute peanut appears
to provide the only common exception to the constraint blocking post-vocalic /w/.
3. Compare this to Spanish, with the following inventory: /j w i e a o u/.
4. Recall that normal stress in SF falls on the last syllable of a phonological phrase,
unless the nal syllable contains schwa, in which case stress is penultimate.
5. Of these four consonants, only // consistently provokes lengthening. There is

variation for certain speakers with the other consonants, but the generalization
holds here for the standard variety we are describing. See Walter (1977: 4445).
6. Length induced by such consonants is a more general phenomenon cross-linguis-

tically; see Chen (1970).
7. /e/ is found in nonnal closed syllables: meri emery /em.i/, mdecin doctor
/met.s'/ , cleri celery /sel.i/, etc. (all with lexicalized deletion of schwa). But
since nonnal syllables do not constitute a lengthening context, these examples
are not relevant to a discussion of vowel length. Within phrases, schwa deletion
can also produce closed syllables containing /e/, as in des melons some melons
/dem.ln/, a question to be discussed in the section dealing with schwa. We should
also note that verlan forms also systematically exclude /e/ from closed syllables:
bouger > gbou > geb /<'b/ sortir, not */<eb/, cond > dcon > dk /d'k/ cop,
not */dek/, and many others. (For a brief introduction to verlan, see chapter 7.)
8. There are rare instances of /'/ in nonnal syllables (e.g., bler [b':.le] to bleat,
ple-mle [p':l.m':l] any old way), but these are highly variable and need not
concern us further. For detailed discussion of this phenomenon, see Walter (1976),
chapter IV.
9. Passy (1892), for example, notes much more extensive vowel length than is
allowed for by the description here.
10. Catach (1995: 1129 ff.) provides much useful information.
11. The lack of consistent use of < > for /#/ is indicated by words such as phrase,
proie prey, cadre frame, gars guy, gaz gas, sabre, tas pile, and many
others, which also contain the back vowel.
12. For the /'- '/ opposition in CF, see Santerre (1974); for a general survey of col-
loquial CF pronunciation, see Walker (1984).
13. It is important to note that this constraint applies at the level of the word. Once
words are concatenated into phrases and resyllabication occurs, the constraints
are relaxed, as we will see below in 4.7. The discussion of schwa also complicates
matters, since the merger of schwa with // produces an easily described set of
exceptions to this pattern, a matter described in 4.5.3..
14. For detailed discussion, see Warnant (1996). Warnant also notes that there is a
tendency for the rst singular future sufx to be pronounced /'/, thereby merging
it with a number of conditional forms. In fact, as Tranel (1987: 5153) shows, this
distinction is far from stable, even in SF. For further discussion focusing on the
role of the distinction in the verb system (where, despite some variation, it retains
considerable importance), see Richman (1976).
15. Jene jeune provides the only true minimal pair opposing // to // in this set
(with beugle aveugle providing an identical following context). Thus, while
both vowels occur in closed syllables, the distribution is not totally free, being
affected by both random and systematic factors (such as the inuence of /z/ and
// discussed below).
110 Chapter 4
16. Since // and /o/ are intrinsically long vowels, they will be realized as [] and [o]
respectively in these words.
17. As is often the case, there are a few exceptions to this constraint. For example,
Martinet and Walter (1973) list dinosaur, maure, saur, and taure (all spelled with
< au >) where a minority of their speakers have /o/ rather than /n/ before //.
18. The apparent presence of the schwa in words such as plerin, seulement, or
effeuillement does not sufce to create an open syllable, since the vowel is never
pronounced here in SF. We should note, however, that the distributional constraint
involving // before /z/ remains in effect, requiring // rather than // in words
such as heureusement fortunately (and others with -eusement). A few additional
exceptions (lieutenant, veulerie spinelessness) may be attributed to the inuence
of the stems in // (lieu place /lj/, veule spineless /vl/), a topic to which we
will return.
19. Mute-e, also known as schwa, is discussed in detail in section 4.5.
20. Recall that VCV sequences are syllabied V.CV within the phonologi-
cal phrase. In fact, stems ending in CL clusters behave in the same way: trouble
troubler /tubl/ /tu.ble/.
21. The forms malheur misfortune /mal/ malheureuse /malz malz/,

peuple people /ppl/ peuplade /pplad pplad/, peur fear /p/ peu-
reuse /pz pz/, where derivative pronunciations with // are frequent,
provide exceptions.
22. This word is exceptional in that the vowel causing the harmony is not the stressed
vowel, but the penultimate /i/.
23. In one of the very perplexing phenomena of French phonology, the presence of
orthographic < e > following the nal consonant, whether this < e > is realized
as schwa or not, does not result in an open syllable. Thus lpre, whether /l'p/ or
/l'p/, functions as a monosyllabic closed syllable for purposes of this alterna-
tion.
24. Despite the orthography in < >, Martinet and Walter (1973) note that this form
is often pronounced /apemidi/.
25. See Buben (1935) for a detailed treatment of this matter.
26. This sound is transcribed [nA] by Martinet and Walter (1973) and Walter (1973), a
notation that indicates that a full merger with // has not been achieved, but one
that also raises the possibility of overlap with realizations of //.
27. The question must remain open, however, since Landick (1990) presents results
that indicate that the phenomenon is still very much alive in her data (obtained in
the late 1980s from students at the Ecole normale suprieure and from employees
of the Parisian metro system).
28. (26) contains examples of /a #/ in monosyllables. The results are entirely com-
parable for these vowels in polysyllables, whether in the nal syllable or not.
Walter (1976: Chapter III) presents a comprehensive analysis of this pair of
vowels, clearly demonstrating the great variability in pronunciation. Further data
is available in Warnant (1997).
29. For additional such forms, characterized by Warnant as older Parisian speech, see
Warnant (1987).
30. Unlike CF, where -ation and -ois are frequently pronounced with /#/.
31. This correlation breaks down quickly, since there are numerous words that etymo-
logically show deletion of consonants but no circumex: avocat lawyer < advo-
catus; coudre to sew < consuere; faire to make/do < facere; moudre to grind
< molere; noir black < niger; ouir to hear < audire; etc. On the other hand,
when we examine the forms in (10) or (27), or words such as ge < aetaticum
(OF eage); chsse reliquary < capsa; d must (past part.) < debutu (CL debitu,
OF deu); te ute < OF aute, ehute; jener to fast < jejunare; sr sure <
securu (OF seur); mr mature < maturu (OF meur); sol drunk < satullus; and
many others, we see the basis for a link between a circumex accent and historical
deletion. In other words, a circumex may often indicate deletion; no circumex
does not indicate no deletion.
32. See, for example, Fagan (1989), Mettas (1979).
33. E.g., Taulelle (1989), Lerond (1980). Lerond characterizes any pronunciation
with /#/ (which is subject to a separate notation) as vieilli Paris.
34. There are, in fact, additional shifts affecting the pronunciation of SF nasal vowels,
which Hansen (1998) discusses in detail. The most striking of these, discernable
in listening carefully to spontaneous speech, involves a movement of /'/ towards
/#/ and a shift of /#/ towards /n/, threatening distinctions such as blanc blond
white blond, vent vont wind they go. Neither of these innovations is as
far advanced as the /'/// merger, and they will not be discussed further here.
35. For an alternate and, to my mind, confusing approach to the transcription of nasal
vowels, see Valdman (1993: 111 ff.).
36. This merger is not taking place in other dialects, such as CF or MF.
112 Chapter 4
37. To this type of case we may also add two forms with non-: non-mtal /nnmetal/
and non-moi /nnmwa/. The word ennui boredom /#PK/ (and related forms: e.g.,
ennuyer) are also exceptional with respect to the *v constraint.
38. The situation is somewhat more complicated if nonstandard forms are consid-
ered. Nasalization of stops can yield XN sequences (lendemain the next day >
/l#PO'/), as can verlan formations (maison house > zonmai /\nOG/), a variety
of disguised speech where violation of Xv is also found (argent money> gen-ar
/<#a/). For further discussion, see Walker (1999).
39. For speakers who have merged // and /'/, this alternation is between /'/ and
/yn/.
40. This pronunciation reects the normative approach of Warnant (1987). In fact, as
Martinet and Walter (1973) demonstrate, /nn/ and /nnn/ are also found prevocali-
cally (non-tre as /nnn't/, /nn't/, or /nnn't/). The same variation may also be
found in the prex pan-: /pan/, /p#/, or /p#n/.
41. According to Warnant (1987), the two prexes circon- and circum- appear only
in preconsonantal position. In this context, one might expect the latter to end in a
nasal vowel (/sikn/), but SF contains many VNC or VN# sequences, rendering
attempts to predict the nasality of the prex vowel impossible.
42. There are very few examples where the prex en- appears before a vowel; along
with enivrer, we nd only enamourer and enamouracher to become enamoured
of.
43. We will not deal here with highly idiosyncratic cases such as divin enfant, liai-
son in compound words or xed expressions, etc. For further discussion of such
examples, see Fouch (1959) or Tranel (1981, 1987).
44. See, for example, Greenberg (1966), Dressler (1985a, b).
45. Recall the constraint from 4.4.2 above blocking nasal vowels from appearing
before other vowels or before nasal consonants. This data set presents additional
exceptions, to which we will return shortly.
46. Martinet and Walter (1973) and Warnant (1987), among others, list senamourer
(both with and without an accent aigu on the rst < e >), pronounced as /#namue/
or /enamue/. In the former case, it would be exceptional as well.
47. Most authors note the possibility of variation (presence or absence of denasaliza-
tion) for mon, ton, son, with preservation of a nasal vowel predominating. See
Fouch (1959: 436). Tranel (1987: 81-85) presents an excellent survey of this
variation.
48. The variants il- (illgal) and ir- (irrel) are not productive and will not be dis-
cussed here. Tranel (1976) remains the denitive treatment of this form (or set of
forms).
49. Standard references for this topic include Fouch (1969), Morin (1994), Nyrop
(1935), Pope (1934), Rheinfelder (1968), and Rochet (1976).
50. We may point out here two additional distributional constraints on nasal vowels
that are also historically explicable: nasal vowels are absent in SF before //
and /j/. Historically, the palatal nasal arose in OF through various processes that
required that the source of // be followed by a vowel. But if that is the case,
the following vowel removes the context needed for nasalization to occur. Hence,
only oral vowels are found preceding //. As for the constraint blocking *Xj, once
again /j/ can only arise historically through various diphthongization or palataliza-
tion processes in contexts that never involve a nasal consonant preceding those
sources. The absence of such a consonant preceding /j/ excludes, as a result, any
subsequent nasal vowel from that position.
51. The last two words belong to a different register and should probably not be con-
sidered to belong to SF. Binse means mess or hassle; clamser (also spelt cla-
mecer) means kick the bucket.
52. For details, see Walter (1982) or Carton et al. (1983).
53. In fact, CF allows for long vowels to occur in nonnal syllables much more
frequently than does SF, and these vowels may also optionally diphthongize, as
in pantoute not at all (= pas du tout) [p#wt7t], je comprends I understand
[5knwp]3], etc. Walker (1984) discusses this phenomenon in more detail.
54. For further discussion, see (from among literally hundreds of possibilities) Dauses
(1973), Dell (1973a, b), Fouch (1959), Varney Pleasants (1956) or Verluyten
(1988).
55. The same applies to dervied verbs: refaire, dfaire, contrefaire, satisfaire. The
exceptional nature of the interpretation of < ai > as schwa is shown by the fact
that certain speakers in Martinet and Walter (1973) pronounce faisan and related
forms, as well as satisfaisant, with /'/ rather than schwa.
56. Analysts are generally in agreement that the vowel in structures such as dites-le
say it, sur ce on that, parce que because, Je, soussign I, the under-
signed is // or //, not //.
57. Here, as elsewhere, loanwords may provide sporadic exceptions: edelweiss

/ed'lv's/ or /ed'lvajs/, et cetera /'ts'tea/. Note also the prex en-, as in enivrer,
enamourer, with no orthographic accent.
114 Chapter 4
58. The situation is somewhat more complicated, in reality. An accented < > is
needed in words like chelon or branler because the < ch > represents a single
segment (unlike < x >, which represents /ks/ or /gz/), and because CL clusters
syllabify with the following vowel, leaving the < e > in an open syllable and sub-
ject to pronunciation as schwa unless an acute accent on the letter is included.
Thimonnier (1970) discusses such cases in detail.
59. But see Malcot and Chollet (1977) for a study which identies schwa more
closely with //.
60. Note that the realization of schwa as // in this context furnishes an exception
to the constraint that excludes // from word-nal position. The constraint must,
accordingly, be modied to refer to original //, not to those that are an innova-
tive pronunciation of //.
61. For additional discussion of this matter, see Morin (1988) or Walker (1993).
62. Recall the discussion of the denition and role of phonological phrases from 3.3
above. We will use || to indicate phrase boundaries.
63. Since schwa can never appear in absolute word- (hence phrase-) initial position,
we use this expression as a convenient shorthand for schwa in the initial syllable
of a phrase.
64. In less formal speech, schwa may delete in the contexts of (38b): r(e)garde-moi
a, d(e)main on part, etc.
65. Inclusion of the symbol C following the schwa is not, strictly speaking, nec-
essary here we could simply refer to the number of preceding consonants. If
the segment following the schwa is a vowel (e.g., la vie humaine human life
/laviym'n/), the schwa is prohibited by the constraint in (35ii) above. If no seg-
ment follows the schwa, we are no longer dealing with a phrase-internal context,
since the schwa would be in nal position. Lastly, recall that the constraint against
schwa in closed syllables prohibits structures of the type CC#, if syllabied
CC..
66. An alternate pronunciation, /m'dsin/ (or with assimilation, /m'tsin/) reects the
constraint against /e/ in closed syllables, a constraint violated on numerous occa-
sions when schwa deletion is involved (and, as we have seen, in various other
ways as well).
67. See Malcot (1976: 99-103) for a detailed study of preferences governing which
of two schwas deletes in sequences of this type.
68. Dell (1973a) established the standard for any discussion of this phenomenon. See
also Dell (1973b, 1978).
69. See Hansen (1997), Faygal (1998) for recent discussion of this phenomenon.
70. A glance at Fouch (1959: 91-139) will conrm the complexity of the behaviour
of schwa.
71. The situation with aspirate-h is, in fact, signicantly more complicated. While
schwa is always possible here, it may also be deleted, provided that the preceding
consonant is not subsequently linked to the initial vowel of words with aspirate-h.
Thus, schwa cannot delete in || le huitime the eighth (i.e., in phrase-initial syl-
lables) but may drop in cest le huitime its the eighth /s'l.itj'm/. In the same
fashion, the schwa of une and cette can also drop in une housse, cette housse, pro-
vided no linking occurs: /yn.us/, /s't.us/; */y.nus/, */s'.tus/. In such cases, a slight
glottal constriction (glottal stop) may precede the initial vowel of the aspirate-h
word. Nonlinking also applies to nal consonants not followed by schwa: sept
haches /s't.a5/ (although even this constraint may be overridden in rapid speech:
/s'.ta5/). Cornulier (1981) provides a detailed discussion of these matters.
72. See Morier (1975: 146-168, 378-88) for an insightful discussion of the complexi-
ties that arise in poetry.
73. For further discussion, see Gadet (1989), Guiraud (1965), or the classic works of
Bauche (1920) and Frei (1929).
74. Cf. Hansen (1994), Walker (1996), Walter (1982: 221-23). Even Warnant (1987)
contains numerous word-specic indications of stability: pelade as /plad()/ with
stable schwa in the rst syllable; menu as /m()ny/ with the comment mais le
se maintient souvent but the is frequently retained (passim).
75. Although schwa emulates // in most cases, there are indications in a small
number of forms that the merger is leading // to behave as unstable schwa by
deleting in word-internal syllables: djeuner to have lunch /de<ne/, rajeunir to
rejuvenate /a<ni/, pharmaceutique pharmaceutical /famastik/, malheureux
unfortunate /mal/ are the examples most frequently cited. This is an inverse
type of evidence in support of the merger of schwa and //.
76. See Morin (1988), Tranel (1988), or Walker (1993) for further data and discus-
sion.
77. Various spelling reforms have proposed standardizing the orthography of these
sets of verbs, since no phonological differences are involved in the different
orthographic representations.
78. Spontaneous speech frequently manifests pronunciations where the /'/ is absent:
/a5t'/ (or /a5te/).
116 Chapter 4
79. Once again, Morin (1988) and Tranel (1988) provide the best analyses of this
complex set of problems. Morin, in particular, discusses nonstandard variation
in the stem vowels of these verbs ranging from the general substitution of /'/
(lever raise /l've/) to forms where schwa appears under stress (je jeute I throw
/<<"t/) or where an expected /'/ disappears, having been replaced by schwa (elle
cachte /'lka5t/ for elle cachte she seals). This latter usage seems to be expand-
ing.
80. For some discussion of the phonological consequences of learned status, see Dell
and Selkirk (1978), Zwanenburg (1983), or Walker (1975).
81. We limit ourselves to the simplest cases here.
82. As a result, these popular varieties allow schwa in absolute initial position, as well
as in closed syllables, in contrast to SF.
83. As in SF, schwa does not appear adjacent to vowels: je pars Im leaving /<pa/
versus jarrive Im arriving /<aiv/; la vie the life /lavi/, not */vi/. For dis-
cussion of schwa in MF, see Durand (1990: 27-34) or Durand, Slater and Wise
(1987).
84. Assignment of gender to French nouns rests on a complex series of phonologi-

cal, morphological, semantic, etymological, and purely arbitrary factors. Gender
assignment is less arbitrary than might rst be thought, however. For a detailed
analysis of the predictability of French gender, see Tucker, Lambert, and Rigault
(1977) or, in a more pedagogically oriented vein, McCarter (1987).
85. For some speakers, hier is pronounced /ij'/.
86. In orthographic terms, we consider the < y > in forms like paye intervocalic,
despite the pronunciation /p'j/.
87. To be completely accurate, we should note that a few (unassimilated) loanwords

have variants with postvocalic /w/: outlaw as (among others) /awtlo/ or /utlaw/,
crown-glass as /kawnglas/, etc.
88. Basically, the glides following CL clusters developed historically as elements of

diphthongs that arose in the course of phonological change. Their appearance fol-
lowing the clusters is accidental in the sense that they derive from simple vowels
in that context, and the simple vowels involved were unconstrained in their occur-
rence. Trois is from Latin tres, for example, gloire from gloria, fruit from fructus,
and several complicated series of changes led to the current diphthongs develop-
ing from simple vocalic nuclei.
89. There is also an infrequent tendency in less standard speech to insert a transitional
/j/ or /w/ (not //) glide when the conversion to a semi-vowel does not occur:
diode diode /dijnd/, boueux muddy /buw/, etc.
90. Consult in this regard Delattre (1940).
91. Recall, however, the complexities involving the appearance of /eC./ (/e/ in
closed syllables) in unassimilated loan words (e-mail /imel/) and in nonnal syl-
lables discussed earlier.
92. Samba, however, can exceptionally be pronounced both /s#mba/ and /s#ba/.
93. For detailed arguments leading to the same conclusion, see Rochet (1977) or
Lyche and Girard (1995).
94. These three symbols, a traditional archiphonemic notation in French phonology,

represent the neutralization of an opposition between the higher-mid and lower-
mid vowels /e o/ and /' n/.
119
Chapter 5
Consonants
5.0 5.0 The Consonant System

Unlike the vowel system, the SF consonant system is typologically straight-
forward. With the possible exception of //, it contains no segments of great
phonetic complexity, nor is its structure unduly complicated. Whatever dif-
culties arise are found not in the segmental inventory, but in the behaviour
of word-nal consonants, a topic of considerable descriptive and theoretical
interest. Before addressing that issue, however, there are several preliminary
matters to examine. The SF consonant inventory is given in (1); the consonants
are exemplied in (2).
(1) SF consonants
obstruents
stops
voiceless p t k
voiced b d g
fricatives
voiceless f s 5
voiced v z <
sonorants
nasals m n 0
liquids l
120 Chapter 5
(2) Examples
#__ V__V __#
p pas step appt lure cep stock
t tas pile athe atheist net clean
k cas case accord agreement sec dry
b bas low abb abbot snob snob
d dos back adieu adieu bled village
g gai gay aguets lookout bague ring
f fe fairy effet effect nef nave
s ses her assez enough bis repeat
5 chez at hacher to chop hache axe
v ville city avant before rive bank
z zone zone misre misery quinze fteen
< geai jay agir to act ge age
m mai May amont uphill me soul
n nez nose anne year ne donkey
gnle hooch agneau lamb bagne penal colony
0 swing jive
l lai lay alle lane l thread
raie line arrt stopping tir shooting
A recurrent topic of discussion in consideration of the consonant system

involves distributional constraints affecting consonants. For our purposes,
these constraints are best outlined in terms of the structure of the syllable: con-
sonants and consonant groups in syllable-initial and syllable-nal position. In
SF, all single consonants may be found syllable-initially, with the exception
of /0 /, whose marginal status we will discuss in more detail below. Likewise,
all consonants occur in syllable-nal position. The situation becomes more
interesting when we consider groups of consonants because here (as in all
languages), there are severe restrictions on both the size and composition of
consonant clusters. In SF, for example, the size of such clusters, with rare
exceptions, is limited to a maximum of two consonants in syllable-initial
position and three in syllable-nal. In syllable-initial position clusters are
normally composed of a stop or fricative followed by a liquid or by any
consonant followed by a glide.1 In syllable-nal position, conversely, we see
liquids followed by fricatives or stops, as well as CL. and some CC. groups.
Consonants 121
Moreover, not all .CL, LC., or CC. combinations are possible, and the fricative
/s/ also supplies some atypical cases. Representative examples are given in (3),
divided into normal and more marked categories.
(3) Consonant clusters in SF

(a) syllable-initial (unmarked)
/pl/ plan plan
/p/ prend takes
/t/ train train
/kl/ clair clear
/k/ crime crime
/bl/ blond blond
/b/ brun brown
/d/ drame drama
/gl/ gland acorn
/g/ grand big
// an custard tart
/f/ front forehead
/v/ vrille tendril
/sp/ sport sport
/st/ store blind
/sk/ score score
/sm/ S.M.I.C minimum wage
/sn/ snob snob
(b) syllable-initial (marked)
/spl/ splendeur splendor
/sp/ sprint sprint
/st/ strict strict
/skl/ sclrose sclerosis
/sk/ scrupule scruple
/vl/ vlan wham
/zb/ sbire henchman
122 Chapter 5
/gz, ks/ xnophobe/xylophone xenophobe/xylophone

/ps/ psychique psychological
/pn/ pneu tire
/pt/ ptrodactyle pterodactyl
(c) consonant-glide groups syllable-initially
/j/ // /w/
/p/ pied foot puits well poids weight
/t/ tient holds tuile tile toile cloth
/k/ inquiet worried acuit acuteness quoi what
/b/ biais skew buisson bush boue buoy
/d/ Dior Dior duel duel doigt nger
/g/ guier g tree arguer to deduce gouache gouache
/f/ able reliable fuir to ee foin hay
/s/ sieur Mister sueur sweat soie silk
/5/ chiot puppy chuintant hushing chouette cute
/v/ avion airplane revuiste reviewer voit sees
/z/ asiatique Asian dsuet outdated zouave Zouave
/</ plagier to juillet July joint joint
plagiarize
/m/ miette crumb muet mute mouette seagull
/n/ panier basket nuage cloud noix nut
/l/ lion lion luette uvula loin far
// rien nothing ruine ruin rouage cogwheel
2
(d) syllable-nal
/pl/ triple triple
/p/ pre bitter
/bl/ table table
/b/ sobre sober
/t/ quatre four
/d/ cidre cider
Consonants 123
/kl/ oncle uncle

/k/ vaincre to conquer
/gl/ ongle ngernail
/g/ aigre sharp
// bufe buffalo
/f/ chiffre gure
/v/ vivre to live
/lp/ poulpe octopus

/lb/ galbe curve
/lt/ halte stop
/ld/ solde balance
/lk/ calque tracing
/lg/ algue seaweed
/lf/ golfe gulf
/lv/ salve salvo
/ls/ valse waltz
/lm/ calme calm
/ln/ Elne (place name)
/p/ charpe scarf

/b/ verbe verb
/t/ porte door
/d/ absurde absurd
/k/ arc bow
/g/ orge barley
/f/ cerf stag
/v/ morve mucus
/s/ farce farce
/z/ quatorze fourteen
/5/ torche torch
124 Chapter 5
/</ a urge its urgent

/m/ charme charm
/n/ borne milestone
// hargne spite
/l/ perle pearl
/sp/ aspe reel

/st/ ouest west
/sk/ risque risk
/sm/ ralisme realism
/ft/ naphte naphtha
/mn/ hymne hymn

/0g/ pouding pudding
/ps/ clipse eclipse

/pt/ crypte crypt
/ts/ ersatz substitute
/tm/ rythme rhythm
/tl/ axolotl axolotl
/t5/ match game
/dn/ hydne hydnum
/d</ bridge bridge
/kt/ acte act
/ks/ syntaxe syntax
/km/ drachme drachma
/gm/ egme composure
/p/ pourpre purple

/b/ arbre tree
/t/ meurtre murder
Consonants 125
/d/ ordre order

/kl/ cercle circle
/sk/ lorsque when

/st/ verste verst
/sp/ aspre rocky hillside (in Roussillon)
/st/ ministre minister
/spl/ asple spore
/skl/ muscle muscle
/5t/ chtre (interjection)
/lsk/ Volsques Volsci
/lk/ spulcre sepulchre
/lt/ ltre lter
/pt/ sceptre sceptre

/kt/ spectre ghost
/kst/ texte text
/kst/ dextre dextral
What is interesting about these combinations is the fact that, as soon as we

leave the unmarked groups, the words take on a decidedly learned or foreign
avour such items are often the result of learned or foreign borrowings,
and, in many cases, are not yet fully integrated into the standard phonological
system. Examples of this type demonstrate that the phonology of French is far
from being a perfectly homogeneous and internally undifferentiated system,
either in terms of the status of the constraints that are operative or (as we have
seen in the case of the vowels) in terms of an invariant norm.
It is also striking that the number of syllable-nal clusters exceeds by a
signicant margin those found initially. There are, however, ways of explain-
ing this discrepancy. First, as the examples make clear, most of the syllable-
nal clusters are dened with respect to word-nal position, and historically
a schwa was present in these forms. This schwa would have allowed for bi-
or polysyllabic words with a different syllabication, as in pourpre purple
/pu.p/, ltre lter /l.t/, rythme rhythm /it.m/, egme composure
/'g.m/ instead of monosyllabic /pup/, /lt/, /itm/, and /'gm/ respec-
tivley, with the clusters in the latter forms being reduced to more simple (word-
internal) sequences when nal schwa is present. We see this further in the syl-
126 Chapter 5
labication of derived forms such as absurdit absurdity /ap.sy.di.te/, qua-

torzime fourteenth /ka.tn.zj'm/, clipser to eclipse /e.klip.se/, rythmique
rhythmic /it.mik/, muscl muscled /mus.kle/, ltrer to lter /l.te/,
egmatique phlegmatic /'g.ma.tik/, etc., where the complex nal clusters of
the roots (/apsyd/, /katnz/, /eklips/, /itm/, /myskl/, /lt/, and /'gm/) are
simplied (decomposed) when a vowel-initial derivational sufx is added. In
other words, the set of nal clusters has been somewhat articially expanded
and complicated by the loss of word-nal schwa, but when word-medial
sequences are considered, the situation is much less intimidating.
Nor should it be thought that any permitted nal cluster can combine with a
permitted initial cluster in the interior of words, despite the fact that either type
may occur elsewhere. To take an extreme example, nal /skl/ (muscle) may not
combine with initial /spl/ (splendeur) to form a medial sequence */VsklsplV/.
To use a more realistic case, nal /k/ (arc) may not precede initial /5/ (chez)
in French, despite the fact that both are simple and well-formed independently.
That is, no SF word contains the internal sequence */Vk5V/. As Morin (1987)
has demonstrated, signicant further restrictions, beyond the scope of this
description, apply to the combination of syllable codas and onsets word-inter-
nally.
Finally, we should return to the issue of consonant-glide clusters in word-or
syllable-initial position (#CGV, .CGV), since such groups provide an oppor-
tunity to exploit the differing syllabic representations outlined in chapter 3
(section 3.1) above. Note rst that the preceding inventory allows for #CGV
clusters, but not for structures of the form #CCGV (i.e., with two consonants
preceding the glide). But how, then, can we account for the relatively frequent
items with just that shape (cf. bruine drizzle, fruit fruit, pluie rain, croire
to believe, gloire glory, trois three, from (58) in section 4.6 above)? The
answer resides in the assignment of the glide to a complex syllable nucleus
rather than to the initial cluster, as in (4) rather than (5)
(4) Simple onset; complex nuclus

m
rhyme
onset nucleus coda
C L G V C
b i n (bruine, croire, etc.)

Consonants 127
(5) Complex onset; simple nucleus
rhyme
onset nucleus coda
C L G V C
b i n
This differential assignment of the glide in .CGV versus .CLGV onsets allows
us to preserve the unity and simplicity of the constraints on initial clusters. It
also means, in fact, that glides may be of two types in French. Some, those that
appear in initial clusters, are more consonant-like. Others, those appearing in
the nucleus, are more vowel-like.
Interestingly, this differential representation correlates directly with another
distinction we have already seen in the discussion of glides and aspiration. We
saw in the examples of (61) in section (4.6) that some instances of the glides /j
w/ permit elision and liaison while others block these processes: liode the
iodine, lhuile the oil, louest the west versus le yacht the yacht, le huitme
the eighth, le western the western, respectively. This behaviour may also be
described using distinct syllabic representations. Those glides that block elision
and liaison will be assigned to an initial consonantal position in the syllable, as
in (6), thereby identifying the words as consonant-initial:
(6)
rhyme
onset nucleus coda
G V C
j C t (yacht)
128 Chapter 5
Since the initial consonant position is lled, this class of words behaves in
a fashion similar to other words beginning with consonants: liaison does
not occur, because there is no space available into which a liaison con-
sonant could move. Exceptionally, schwa may also be maintained even
in these VCCV structures, as in jen prends le huitime Im taking the
eighth. /<#p#litj'm/ versus jen prends le neuvime Im taking the ninth.
/<#p#NPvj'm/, although deletion is normal in both cases phrase-internally.3
Here, the consonantal glides also condition behaviour in a way parallel to
aspirate-h (for a detailed discussion of the latter, see section 5.4). In words like
huitre oyster or ouest west, where both liaison and elision take place (les
huitres /lezit/; louest /lw'st/), the glide is considered as part of the syllable
nucleus, as in (7):
(7)
rhyme
onset nucleus coda
C G V C C
i t (huitre)
Since the onset is empty, liaison can occur. Since a nal schwa in any preced-
ing word is immediately followed by the nucleus rather than by a word-initial
consonant, elision can occur. Thus, we see how a type of representation that
makes explicit formal distinctions between syllable structures can account in
a unied way for processes that initially appear dissimilar. Such theoretical
developments are particulary important in a language such as French, where
the nature of syllables plays such a large part in the functioning of the phonol-
ogy.
This concludes our description of the general constraints on sequences of
consonants in SF. Before turning to more specic topics, however, we will
illustrate one case where a restriction of this type has left a signicant trace
in French morphology where an examination of the phonological history of
French illuminates certain irregularities in verb conjugations and elsewhere.
The case involves an excursus into the domain of consonantal epenthesis.
At the earliest stages of OF, nasals, liquids and fricatives appeared freely in
syllable-nal position (_.), and could be followed by a syllable-initial liquid
(_.L). However, when syllable-nal /l m n s z/ preceded /l/ or // (e.g.,
l., as in OF molre to grind), this cluster was subsequently interpreted
Consonants 129
as an unacceptable syllable contact, and an epenthetic consonant was inserted

between the two segments, as indicated in (8).
(8) Epenthetic consonants in OF

l+ > ld
m+ > mb
n+ > nd
+ > nd
s+ > st
z+ > zd
m+l > mbl
Various explanations have been proposed for this transitional consonant,

linked to the articulatory transition between the segments in question or to
their relative phonological strength.4 In any event, we note that the inserted
consonant agrees in voicing and in place of articulation with the rst conso-
nant: voiceless /t/ after /s/, voiced /d/ elsewhere; labial /b/ after /m/, apical
/t/ or /d/ elsewhere.5 In addition to its intrinsic interest, epenthesis has left a
number of traces in the morphology of SF, traces that may help to illuminate
the behaviour of irregular verbs in particular. Consider the forms in (9).
(9) Remnants of epenthesis

falloir faudra be necessary
valoir vaudra be worth
moulons moudre grind
rsolvons rsoudre resolve
cumulatif comble, combler cumulative packed, ll in
humilit humble humility humble
numro nombre number number
simuler semble simulate seem
tenir tiendra hold
venir viendra come
craignons craindre fear
(OF cremons - criembre)
incinrer cendre incinerate cinder
connaissons connatre know
130 Chapter 5
naissons natre be born

paraissons paratre appear
sera tre be
cousons coudre sew
These forms recapitulate, in a sense, several elements of the phonological his-

tory of French. The early precursors (Late Latin, effectively) of the words in
the right-hand column contained a vowel that initially separated the segments
conditioning epenthesis. When that vowel dropped, the sequences of (8) were
created, and epenthesis took place. Subsequent changes (some of which we
will explore below) involved vocalization and then loss of the /l/, nasalization
of a preceding vowel and loss of the nasal consonant, or loss of preconsonantal
/s/: compare OF faldra /falda/, nombre /nnmb/, conoistre /knnnjst/ with
modern faudra /foda/, nombre /nnb/, and connatre /knn't/, respectively.
From this perspective, the notoriously irregular verbs of (9) may be seen to be
historically regular (in that they reect the regular operation of a number of
sound changes). This data indicates, once again, that SF is the heterogeneous
result of interacting and overlapping historical processes, and that history can
often provide explanations for situations that otherwise seem inexplicable.
5.1 5.1 Geminate Consonants

All of the consonants exemplied in (2) of section 5.0 are single or simple
consonants, despite an orthographic representation that may include double
letters: abb abbot, accord agreement, effet effect, and so on. SF does
contain long or geminate consonants, however. Such geminates may arise
in a number of ways.6 Perhaps the simplest examples occur in the irregular
verbs courir to run, mourir to die, acqurir to acquire and related forms
(e.g., accourir to rush up, parcourir to travel, conqurir to conquer,
senqurir to inquire, etc.), where the future and conditional paradigms
(courrai /kue/, courrais /ku'/, mourrai /mue/, mourrait /mu'/,
acquerrai /akee/, etc.) show a geminate // in opposition to the remain-
ing verb forms (courir /kui/, mourais /mu'/, acqurons /aken/, and
so on).7 Secondly, as we might expect, the deletion of schwa may, on occa-
sion, bring identical consonants into contact, and these two consonants merge
into a single long consonant: la-dedans inside /ladd#/, honntet honesty
/nn'tte/, saintet saintliness /s'tte/ (also /s'te/), deuximement secondly
/dzj'mm#/, embaumement embalming /#bomm#/, bizarrerie strangeness
/bizai/, verrerie glassworks /v'i/, and so on. Thirdly, particularly in
more formal contexts, a number of learned words containing orthographic
geminates may also be pronouned with phonological geminates: additif addi-
tive /dd/, arrogant arrogant //, attique Attic /tt/, collgue colleague
/ll/, immdiat immediate /mm/, innombrable innumerable /nn/, and so on.
Consonants 131
When we turn to a phrasal context, many additional geminates arise through

the juxtaposition of words where the rst ends in a consonant identical to the
initial consonant of the second (with or without schwa deletion). Examples are
given in (10).
(10) Geminate consonants in phrasal contexts

sept tables /s'ttabl/ seven tables
cette table /s'ttabl/ this table
chaque consonne /5akknsnn/ each consonant
robe bleue /nbbl/ blue dress
bague grandiose /bagg#djoz/ imposing ring
chef formidable /5'ffnmidabl/ great chef
grosse salade /gossalad/ large salad
juge genevois /<y<<'nvwa/ Genevan judge
drame mystrieux /dammistej/ mysterious drama
ville lorraine /villn'n/ city in Lorraine
Finally, still in a phrasal context, there is a phenomenon in SF known as

the accent dinsistance or emphatic stress, a phenomenon that seems to be
expanding in contemporary usage. One of the manifestations of this accent
involves the reinforcement through gemination of the rst consonant of the
highlighted word, a word normally occurring near the beginning of the phono-
logical phrase, as in (11).
(11) Accent dinsistance

quel crtin /k'lkket'/ What an idiot!
cest formidable /s'ffnmidabl/ Its great.
cest terrible /s'tt'ibl/ Its awful.
cest super beau /s'ssyp'bo/ Its really nice.
magnique, son geste /mmaiksn<'st/ Magnicent, his
gesture.
Gemination is not the only manifestation of this type of accentuation. We will

return to the problem in the chapter on prosody. In the meantime, we turn to
the next consonantal domain, that dealing with nasal consonants.
132 Chapter 5
5.2 5.2 Nasal Consonants

Modern SF has the four nasal consonants /m n 0/. Only the rst two, how-
ever, are fully integrated into the system in the sense that they are free of major
distributional restrictions. /m/ and /n/ occur freely word-initially, medially, and
nally, and are found in various types of consonant clusters from which // and
/0/ are excluded:
(12) Constraints on nasal consonants

mon /mn/ my
non /nn/ no
gnon */n/8 blow
*ngon */0n/
smash /sma5/ smash
snob /snnb/ snob
*sgnob */snb/
*sngob */s0nb/
asthme /asm/ asthma
prisme /prism/ prism
The constraints on // and /0/ arise from the diverse historical processes by
which they were introduced into French. The palatal nasal //, for example,
was produced through various palatalization processes, and the contexts where
these processes took place were limited. Hence, the contexts where // may
occur are similarly limited. For example, no // occurs without a following
vowel in the orthography (vigne vine, vigneron wine grower, never *vign)
since vigne < Latin vinea [winja] with assimilation of [n] to the [j] and reten-
tion of the nal [a] as [].9 A full discussion of the nasals will involve examina-
tion of the interchange between // and /nj/, of the mechanism of introduction
of /0/ through loanwords, and of a process of nasal assimilation.
5.2.1 5.2.1 The // /nj/ interchange

The relationship between palatal // and a /nj/ sequence is still evident in SF.
For many speakers, // may be realized as /nj/, while for others /nj/ may occur
as //. Some speakers, in other words, have // in both peigner to comb and
panier basket, magnique magnicent and manier to handle, others have
/nj/ in both, and still others maintain a // /nj/ distinction. The confusion may
even extend to word-initial position, since Martinet (1945: 170 73) reports
Consonants 133
a confusion between, for example, la nielle corn-cockle and lagnelle lamb

(f.). Currently, according to Walter (1976: 392 400), the tendency appears
to be to favour /nj/ intervocalically and word-initially, with // preferred pre-
consonantally and word-nally. The distinction appears to be on the verge of
complete neutralization (with [nj] and [] in complementary distribution or
free variation). The phoneme //, in other words, appears to be facing elimina-
tion from the SF system, although conservative speakers maintain the / - nj/
opposition and Warnant (1987: LXXXIX) continues to see it as part of le bon
usage. Representative examples, taken from Walter (1976), are given in (13),
where again a majority of the informants represented in Martinet and Walter
(1973) (on which Walter [1976] is based) have /nj/ rather than // in initial and
prevocalic position, while // is favoured preconsonantally and nally.10
(13) // and /nj/

(a) /nj/ favoured (b) // favoured
agneau lamb baignerais would bathe
baigneur bather bnignement benignly
daigner to deign clignement blinking
gagnant winner cognement banging
oignon onion gagne-pain job
peigner to comb renseignement information
poignard dagger rogne-pied hoof trimmer
saigner to bleed saignement bleeding
signier to mean
gnagnan soppy cagne preparatory

class
gnocchi gnocchi charogne carrion
gnognotte useless chtaigne chestnut
gnle hooch empoigne grab
gnon bash hors-ligne road
allowance
134 Chapter 5
5.2.2 5.2.2 The Importation of /0/

If SF appears to be losing a nasal consonant on the one hand, it appears to be
gaining one on the other. The velar nasal /0/ is establishing itself as part of the
system, albeit with certain restrictions on its distribution due to the mechanism
of its importation. The primary source of /0/ is through English loan words
containing the sufx -ing, as seen in (14).
(14) /0/ in SF
brieng brieng
camping camping
doping doping
footing jogging
forcing pressure
jogging jogging/sweat suit
karting go-carting
lifting face lift
living living room
parking parking lot
shopping shopping
smoking evening suit
training training
yachting yachting
zapping channel surng
dinghy dinghy
dring ding-a-ling
gong gong
junker type of falcon
lemming lemming
ring boxing ring
shilling shilling
swing jive
tanker tanker
Consonants 135
The majority of occurrences of /0/ are found in word-nal position for an obvi-
ous reason: their source is the English sufx -ing. We do nd /0/ in monomor-
phemic forms as well, although again virtually exclusively in loan words, as
well as a few instances of the segment in preconsonantal position (camping-
car /k#pi0ka/, chewing-gum /5wi0gnm/). The fact that -ing may be used pro-
ductively in SF (with the meaning of the French derived forms not being pre-
dictably related to the source, as in parking parking lot or garage, caravaning
camping in a trailer, lifting face lift, living living room) indicates an inde-
pendence for the sufx and with it a growing independence of the phoneme
/0/. This independence is reinforced by a process we will examine in the next
section, the assimilation of stops to a following nasal (e.g., diagnostique diag-
nostic /dja0nnstik/), which introduces additional instances of the segment and
which expands the contexts in which /0/ may occur. This expansion may well
provide for, or at least contribute to, full integration of /0/ into the phonology
of SF, particularly since the parallelism of the bilabial and dental series /p b
m/, /t d n/ reinforces the position of /0/ among the velars: /k g 0/. Finally in
this context we should note that certain speakers pronounce word-nal < ing
> as /i/, not /i0/. The former pronunciation appears to be restricted to older,
more conservative speakers, and is being gradually replaced (Warnant [1987],
for example, uses only /0/), a replacement no doubt aided by the increasing
knowledge of English in France and perhaps linked as well to the instability of
// in general, as we saw in the preceding section.
5.2.3 5.2.3 Nasal Assimilation

We have already seen that /0/ may appear as a variant of /g/ in certain contexts.
This appearance is part of a more general process whereby oral stops become
nasal in nasal contexts, such contexts provided by either a preceding nasal
vowel or a following nasal consonant. Examples of the phenomenon are given
in (15).
(15) Nasal assimilation

lendemain next day /l#nm'/ < /l#dm'/
vingt-deux twenty-two /v'nd/ < /v'td/
diagnostique diagnostic /dja0nnstik/ < /djagnnstik/
subjonctif subjunctive /syb<n0tif/ < /syb<nktif/
bombe atomique atomic bomb /bnmatnmik/ < /bnbatnmik/
et demie and a half /enmi/ < /edmi/
grande ville big city /g#nvil/ < /g#dvil/
langue populaire working-class /l#0pnpyl'/ < /l#gpnpyl'/
speech
136 Chapter 5
sans demander without asking /s#nm#de/ < /s#dm#de/
bande de voyous bunch of crooks /b#ndvwajo/ < /b#ddvwajo/
longue marche long march /ln0ma5/ < /lngma5/
chambre de bonne maids room /5#mdbnn/ < /5#bdbnn/
on va prendre were taking /nvap#nlmeto/
le mtro the metro </nvap#drlmeto/
une pingle tie clip /ynep'0dkavat/
de cravate </ynep'gldkavat/
en novembre in November /#nnv#m/ < /#nnv#b/
These examples show that nasal assimilation may occur both word-internally
and between words. Such assimilation, perhaps surprisingly, is widespread in
both standard and less formal speech (especially in the pronunciation of the
numbers e.g., vingt twenty, trente thirty which allow it freely). The nal
set of four examples illustrates the interaction of schwa and liquid deletion
with nasal assimilation, since the simplication of word-nal CL# clusters is
also closely associated with the process.11 Be that as it may, when the velar
consonant /g/ participates in the process, we see that /0/ is again introduced
into the phonology of SF, this time in preconsonantal position. We may also
note that nasal assimilation (as in grande ville large city /g#nvil/) provides a
further set of circumscribed exceptions to the constraint blocking nasal vowels
from appearing before nasal consonants. As a result, it serves to illustrate the
complex interaction between synchronic phonotactic constraints, stylistic vari-
ation and historical phonological change.
5.3 5.3 Voicing Assimilation

Nasal assimilation is only one of the general assimilatory processes affecting
French consonants. Another affects voicing and calls into play notions of con-
sonantal strength and syllable structure. Once again, this phenomenon occurs
both within and between words, and is exemplied in (16), where a superscript
or subscript indicates a normally voiced consonant that has devoiced, while

a subscript indicates the converse, a normally voiceless consonant which has
become voiced.
Consonants 137
(16) Voicing assimilation

(a) syllable-initial
pied [RLG] foot
tiens [VL'] hold
quiet [ML'] calm
puis [pi] then
tuile [til] tile
cuir [ki] leather
poids [RYC] weight
toit [VYC] roof
coiffe [MYnf] headdress
plan [RN#] plan
clair [MN'] clear
prince [p's] prince
trou [tu] hole
crpe [k'p] crepe
an [HN#] custard tart
pneu [pP] tire
aplatir [CRNCVK] to atten
attrait [at'] appeal
ancrer [#ke] to anchor
je pars [5pa] Im leaving
je te vois [5tvwa] I see you
je fais [5f'] Im doing
chemin faisant [5O'fz#] in so doing
la cheville [la5Xij] the ankle
un cheval [5Xal] 12
a horse
(b) word-nal
pre [CR] bitter
tratre [t't] traitor
138 Chapter 5
encre [#k] ink

triple [tipN] triple
socle [snkl] plinth
soufe [su] breath
soufre [suf] sulphur
sarcasme [sakasO] sarcasm
13
ralisme [ealisO] realism
rythme [itO] rhythm
(c) syllable-nal obstruents
obtenir [nDVni] obtain
absurde [CDU[d] absurd
mdecin [OGFU'] doctor
ci-dessus [UKFU[] above
robe chic [nD5KM] chic dress
tas de problmes [VCFRnbl'm] lots of problems
tout de suite [VWFUKV] right away
blague stupide [blaIstypid] stupid joke
brave type [bCXVKR] good guy
onze parties [n\pati] eleven parts
village corse [vila<kns] Corsican village
jeter [a<te] to throw away
jupe bleue [<ypDN] blue skirt
cette dame [s'VFCO] this woman
chaque bote [5CMDYCV] each box
chef de le [5GHFl] leader
nous faisons [PWH\n] were making
neuf bouteilles [nfDWV'L] nine bottles
tasse de caf [VCUFMCHG] cup of coffee
dix secondes [diUgnd] ten seconds
il se marie [KUOCi] hes getting married
tche difcile [V#5FKHKUKN] difcult job
Consonants 139
(d) syllable-nal sonorants (no assimilation)

valser [valse] to waltz
altitude [altityd] altitude
partir [pati] to leave
fertile [f'til] fertile
quel cretin [k'lket'] What an idiot!
par qui [paki] by whom
caneton [kantn] duckling
mannequin [mank'] model
hameon [amsn] sh hook
drame typique [damtipik] typical drama
bonne chose [bnn5oz] good thing
feuillet [fjte] foliated
paillet [pajte] sequined
paille tasse [pajtase] packed down straw
To begin with, we may note that voicing assimilation is a realization of the

general cross-linguistic tendency of consonant clusters to agree in voicing.
Thus, SF words may begin with /sp, st, sk/, but not with /*sb, *sd, *sg/, and
end in /pt, kt, gd/ (concept, acte act, Agde [place name]) but not /*bt, *kd,
*gt/, and so on.14 Moreover, it is the weaker sounds, the liquids, nasals,
and semi-vowels that most frequently undergo voicing assimilation, becom-
ing devoiced in the context of a voiceless consonant, particularly when they
follow that consonant (as in [16a-b]). We must, however, indicate two sig-
nicant exceptions to this ordering: (i) the pronoun je (following loss of the
schwa) assimilates fully to the voicelessness of any following obstruent, as
in je tiens I hold. /5tj'/, je pense que oui I think so. /5p#skwi/, quest-ce
que je fais What am I doing? /k'sk5f'/, etc.; and (ii) the classic example of
word-internal /5v/ (achever /a5fe/, not */a<ve/), where the second consonant
assimilates to the rst. In syllable-initial cases, moreover, devoicing is often
considered to be partial; that is, the consonant is voiceless during the rst part
of its articulation, but becomes voiced part way through.
Syllable-nal obstruents (as distinct from the syllable-nal sonorants illus-
trated in [16b] and [16d]) provide the most complexity, since assimilation can
occur to either a voiceless or a voiced following consonant. In either case, we
are dealing with a VC.CV syllabication, and the syllable-nal consonant, as
the one occupying an inherently weak position, is the one that assimilates, as
the examples of (16c) indicate. Thus, voiceless consonants become (partially)
140 Chapter 5
voiced; voiced consonants become voiceless.15 Syllable-nal sonorants (the

liquids, nasals, and glides) resist this assimilation and retain full voicing, as in
(16d). Here, as elsewhere, there is much individual variation in the degree of
assimilation, which (as in the case of other assimilatory processes) is greater in
rapid speech, itself an indication of the lesser degree of formality of the speech
situation.16
5.4 5.4 Aspirate-h

Aspirate-h refers, somewhat contradictorily, not to a current phonetic prob-
lem in SF, but rather to a division of the lexicon that raises both phonological
and morphological issues. There is, in other words, no sound [h] in SF.17 The
term is used to indicate a distinction among words containing, usually begin-
ning with, the letter < h >. Nonaspirate-h words, the productive and larger
class, behave transparently. There is no phonetic realization of the letter < h >
and the words function, with respect to several processes that we will discuss
below, exactly as if they were vowel-initial or as if there were an empty syl-
lable onset preceding the vowel. Aspirate-h words, by contrast, also show no
phonetic traces indicating < h >, but they behave as if they were consonant-
initial or as if the syllable onset were lled by a consonant of some type.18
Despite the potentially confusing nature of the term, we will continue the tradi-
tion of using aspirate to refer to this latter, irregular class. Before discussing
the processes involved, let us exemplify aspirate and nonaspirate-h words.19
(17) Aspirate-h words (partial list) Nonaspirate-h words (small sample)

hache axe habile skilful
haddock haddock habitude habit
haie hedge haleine breath
haine hatred harmonie harmony
har to hate herbe grass
haler to haul in hrsie heresy
halte stop hertz hertz
hameau hamlet hsiter to hesitate
hanche hip heure hour
handicap handicap heureux happy
hanter to haunt hexagone hexagon
harceler to harass hiver winter
hardi bold homme man
Consonants 141
harem harem honnte honest

hareng herring honneur honour
haricot bean hpital hospital
harmattan harmattan horreur horror
harnais harness hte host
hasard stroke of luck htel hotel
hte haste huis door
haubert coat of mail huile oil
hausser to raise hutre oyster
haut high humain human
heaume helmet humble humble
hennir to neigh humeur mood
hraut herald
hros hero
htre beech tree
heurt collision
hideux hideous
Hitler Hitler
hocher to nod (ones head)
Hollande Holland
homard lobster
honnir to hold in contempt
honte shame
hors apart from
houblon hops
housse cover
huit eight
hurler to scream
hussard hussar
142 Chapter 5
(18) Variable words (with or without aspirate-h) 20

hatien Haitian
hameon sh hook
harpaille herd
21
htelle small piece of meat
htif precocious
heimatlos stateless
hellne Hellenic
hernie, herni hernia, herniated
hiatus hiatus
hittite Hittite
H.L.M.22 public housing
These lists, despite their apparently straightforward nature, hide a certain

number of complexities and require further comment. First, the aspirate-h
words, as do most others, participate in a number of derivational paradigms.
Complex forms derived from aspirate-h words also normally contain aspi-
rate-h, in conformity with the behaviour of the base. Thus, derived hachoir
cleaver, hache-viande meat grinder, hardiesse boldness, htivement
hurriedly, hautain haughty, hauteur height, haut-parleur loudspeaker,
heurter to strike, hideusement hideously, hollandais Dutch, hors-jeu off-
side, hors-la-loi outlaw, housser to cover, huitime eighth, rehausser to
heighten, dehors outside, and so on, are also aspirate-h words. There are,
however, several items where the situation is less regular. First, among the
base forms that vary between aspirate and nonaspirate-h, we note two items
where the derivatives are uniformly nonaspirate: hameon and hellne vary,
but hameonner to hook, hellnique Hellenic, plus the additional sufxed
forms hellnisant, -isation, -iser, -isme, and -iste are exclusively nonaspirate.
Hte has several aspirate derivatives (e.g., htivement), but htif precocious
varies and may be nonaspirate. Aspirate hraut, hros, and Hitler have the
exclusively nonaspirate derivatives hraldique, hraldiste, hroine, hroique,
hroisme, and hitlrien respectively. On the reasonable assumption that the
aspirate forms are irregular (a proposal to be justied below when we consider
the phonological implications of aspiration), the behaviour of the variable and
derived forms may be seen as a step towards regularization: the tendency for
words to pass from the marked aspirate to the unmarked nonaspirate class.
This same tendency is more general cross-linguistically, affecting the relation-
ship between basic and derived or neological forms, where the latter often lose
irregularities present in the base (e.g., leaf leaves but Toronto Maple Leafs;
life lives but Miller High Lifes; mouse mice but two computer mouses; two
Consonants 143
lice, but metaphorically two individuals who are louses, etc.).

What, then, are the phonological consequences of aspiration? Fundamentally,
although aspirate-h words (or syllables) begin phonetically with a vowel, they
behave as if they begin with a consonant. There are several phonological pro-
cess in French that are sensitive to the distinction between vowel-initial and
consonant-initial words: liaison and enchanement, elision and the general
behaviour of schwa, and the selection of certain special variants of adjectives
and determiners. These processes are illustrated in (19).
(19) Effects of the word-initial segment

#C #V
(a) elision in le, la:
le garon the boy lautomne autumn
/lgasn/ /lotnn/
la lle the girl lglise the church
/laj/ /legliz/
(b) elision of schwa:
lorsque nous when we lorsquil when he
/lnsknu/ /lnskil/
double jeu double play double chec double failure
/dubl</ /duble5'k/
(c) enchanement:
vite parti left quickly vite arriv arrived
quickly
/vit.pati/ /vi.taive/
chaque personne each person chaque enfant each child
/5ak.p'snn/ /5a.k#f#/
(d) liaison:
les personnes the people les enfants the children
/lep'snn/ /le.z#f#/
petit livre small book petit enfant small child
/ptiliv/ /pti.t#f#/
(e) special forms:
beau livre nice book bel emblme nice emblem
144 Chapter 5
/boliv/ /b'l#bl'm/

nouveau livre new book nouvel objectif new objective
/nuvoliv/ /nuv'lnb<'ktif/
ce livre this book cet emblme this emblem
/sliv/ /s't#bl'm/
du livre of the book de lemblme of the emblem
/dyliv/ /dl#bl'm/
vieux livre old book vieil objet old object
/vjliv/ /vj'jnb<'/
ma maison my house mon admission my admission
mamezn/ /mnnadmisjn/
#aspirate #nonaspirate
(a) elision in le, la:
le hros the hero lhomme the man
/leo/ /lnm/
la housse the cover lheure the hour
/laus/ /l/
(b) elision of schwa:
quelle honte what a shame quelle honneur what an
honour
/k'lnt/ /k'lnn/
quatre hros four heros quatre honneurs four honours
/kateo/ /katnn/
(c) enchanement:
sept htres23 seven beeches sept htes seven hosts
/s't.'t/ /s'.tot/
quel hraut which herald quel homme which man
/k'l.eo/ /k'.lnm/
(d) liaison:
dix hrauts ten heralds dix hommes ten men
Consonants 145
/dieo/ /di.znm/
nous hassons we hate nous honorons we honour
/nuaisn/ /nu.znnnn/
(e) special forms:
beau heaume nice helmet bel htel nice hotel
/bo.om/ /b'lot'l/
nouveau hasard new hazard nouvel habit new outt
/nuvoaza/ /nuv'labi/
ce haut this top cet htel this hotel
/so/ /s'tot'l/
du haut of the top de lhtel of the hotel
/dyo/ /dlot'l/
vieux hros old hero vieil homme old man
/vjeQ/ /vj'jnm/
ma hache my ax mon honneur my honour
/ma.a5/ /mnnnn/
To this data we may add certain minimal pairs that also demonstrate the dis-
tinctive role played by aspiration: leau the water le haut the top, laine
the groin la haine the hatred, lauteur the author la hauteur the
height, lair the air le hre the wretch, ltre the being le htre the
beech tree, lheure the hour le heurt the blow, lle the island le hile
the hilum, dors sleeps dehors outside, and so on.
As the examples show, aspirate-h words parallel the behaviour of consonant-
initial words, despite the fact that phonetically they begin with vowels.24 This,
then, is their phonologically irregular behaviour, and it is little wonder that we
see steps, albeit minor ones, leading towards their regularization. In fact, infor-
mal and popular speech goes further in this direction, since many commenta-
tors note errors that involve liaison with putatively aspirate-h words: les hari-
cots the beans /lezaiko/, les handicaps the handicapped /lez#dikape/, les
homards the lobsters /leznma/, ils harclent they harrass /ilzas'l/, and so
on. Interestingly in such cases, the errors in liaison do not appear to extend
to elision in the articles. While one might well say /lezaiko/, lharicot for le
haricot has not yet appeared. There are also cases showing liaison in the plural,
but not the singular: un hors-doeuvre /ndv/ versus des hors-doeuvre
/dezndv/, or un(e) hernie /yn.'ni/ (with no enchanement) versus des
hernies /dez'ni/.25 General schwa elision also appears to be expanding, con-
tinuing a trend noted by Damourette and Pichon at the beginning of the
146 Chapter 5
twentieth century, since in rapid speech schwa may sometimes fail to appear
before aspirate-h, although enchanement does not occur: donne-moi une
housse [yn.us], une bonne hache [ynbnn.a5], and so on.26 Any full regu-
larization in the direction of nonaspirate status, therefore, is likely to spread
slowly and to be linked to specic contexts or types of behaviour, and is at best
in its very early stages. This process will no doubt be further complicated by
the appearance of recent loan words, largely from English, that are spelled with
< h > and that, for many speakers, contain an initial phonetic [h]: half-track,
halva, hard-top, herdbook, highlander, hobby, holding, home, and so on.27
5.4.1 5.4.1 Other Types of Aspiration

We have been using the term aspirate to indicate a type of phonological
behaviour where certain forms spelt with < h >, despite a pronunciation begin-
ning with a vowel, behave as if they were consonant-initial. There are, in other
words, two types of vowel-initial words: aspirate and nonaspirate. In fact, it is
not only words spelt with < h > that exhibit this behaviour. We have already
seen one additional indication of such bipartite structures in the section on
semi-vowels; the data is repeated here for information.
(20) Two types of glides

Consonantal Vocalic
le iambe /lj#b/ iamb liode /ljnd/ iodine
le yaourt /ljau/ yogurt les yeux /lezj/ eyes
le yoga /ljnga/ yoga lyeuse /ljz/ oak
la hirarchie /lajea5i/ heirarchy lhiatus /ljatys/ hiatus
la hue /lae/ booing les huitres /lezit/ oysters
le huitime /litj'm/ eighth lhuile /lil/ oil
lhuissier /lisje/ usher
le ouistiti /lwistiti/ marmoset louest /lw'st/ west
le western /lw'st'n/ western louie /lwi/ hearing
Here again, the two types of glides (which we might also call aspirate and
nonaspirate) act in a way that correlates exactly with that shown in (19), to
which we may add the expected variation: lhyne or la hyne hyena, dhier
or de hier yesterday, louate or la ouate cotton-wool, and so on. Finally,
to complete the picture, we should note certain additional structures, such as
acronyms or metalinguistic usages (where forms are cited as linguistic objects)
in which, despite the presence of an initial vowel, neither liaison nor elision
occurs: les SDF rclament the homeless ask for /le'sde'feklam/; il a dit
Consonants 147
envers non envoi He said envers, not envoi. /iladi#v'nn#vwa/, and so
on. This behaviour affects the use of numbers and letters in particular, a situ-
ation that serves to conrm the complexity of aspiration in general: le onze
eleven /lnz/, not */lnz/; son onzime eleventh /snnzj'm/, not */snnnzj'm/;
le h letter h /la5/; du y letter y /dyig'k/, not de ly */dlig'k/, for exam-
ple. As might be expected, the situation here is still far from stable. We nd
page onze page eleven /pa<nz/ with elision, not */pa<nz/; il est onze heures
Its eleven oclock. /il'tnz/ with liaison; lh aspir aspirate-h /la5aspie/
with elision; le huit eight with no elision but dix-huit eighteen /di.zit/ with
liaison; and so on.
5.4.2 5.4.2 Historical Comments Regarding < h >

An examination of the lists in (17 18) indicates that many of the aspirate-h
words have a non-French or borrowed avour, while the nonaspirates (with
the obvious exceptions of hertz and heimatlos) appear more Latinate. This
impression is correct: nonaspirate words are largely derived from Latin and
in the general history of the language lost the [h] very early (i.e., during the
late Empire, prior to the diversication of the Romance languages, much less
the beginnings of French). Since the [h] had disappeared, these words were
consequently vowel-initial and free to participate in the historical processes
conditioned by the presence of a vowel that led to the phenomena of liaison
and elision so characteristic of French. On the other hand, the majority of the
aspirate forms arise from Germanic or other loan words and were introduced
into the language after the loss of the original Latin [h]. As a consequence, this
segment was not deleted by the initial historical process, and was present pho-
netically when liaison and elision were implemented. Aspirate-h words, that
is, were in fact consonant-initial at this early stage, and the [h], just like other
consonants, blocked liaison and elision. When this [h] was itself subsequently
lost, liaison and elision had hardened (i.e., were no longer fully productive)
so the newly appearing vowel-initial words resulting from the second loss of
[h] failed to be affected and preserved their idiosyncratic status. As we have
noted, this marked status leads to some uctuation in their current behaviour,
but not (yet) to full regularization.
In this and other respects, the current lexicon of SF is a mixture of forms
reecting different historical stages and processes. This mixture, particularly
the aberrant status of the aspirate-h forms, has led to various theoretical pro-
posals, the most abstract of which seek to assimilate these forms to consonant-
initial words through the introduction of a phantom consonant. While there
is no room to debate these issues here, one might argue that the best solution
is simply to recognize aspirate words as exceptional (a status conrmed by the
changes they are undergoing) and to mark them as lexically distinct, a solution
adopted by lexicographers but resisted by phonologists, who prefer, in many
cases, to seek generalizations while minimizing or ignoring sets of lexical (and
other) exceptions.
148 Chapter 5
5.5 5.5 Final Consonants

With the possible exception of schwa, no topic has more preoccupied students
of French than the behaviour of nal consonants. Most attention has gone to
liaison, and we will devote ample time to that matter in due course. First, how-
ever, we must review a somewhat more prosaic issue word-nal consonants
in general: their stability, role in the morphology, and relation to the orthogra-
phy. As (2) in section 5.0 above makes clear, all SF consonants may appear
in word-nal position and three of them (/ t l/) are very frequent, whether
considered lexically or textually. Statistics gleaned from Juilland (1965) also
allow the conclusion that SF has no major sanctions on closed syllables,
despite claims that it is primarily an open syllable language. (As might be
expected on typological grounds, open syllables are more frequent, but there
also exist productive processes, such as schwa deletion and the formation of
acronyms and abbreviations, that produce closed syllables.) The situation is
complicated by the role of orthographic < e >, which, in the great majority of
cases, does not result in an open syllable phonologically but is instead an indi-
cation that the nal (orthographic) consonant is to be pronounced. The issue
is not, therefore, one of phonotactics, but rather one of lexical representation:
which specic words end in which consonants?
After the basic division into vowel-nal28 and consonant-nal morphemes
(e.g., agenda /a<'da/, zro /zero/ versus arc bow /ak/, fat smug /fat/), we
must divide consonant-nal morphemes into two fundamental groups dened
on the basis of their phonological behaviour: those that end in a permanent
consonant (vite quickly, dame woman, ville city, pre father, net clean,
islam Islam, l thread, hiver winter, etc.) and those whose nal conso-
nant is latent, unstable or, in modern parlance, oating (we will use the term
latent). The former type of consonant is present in all realizations of the mor-
pheme in question;29 the latter varies present in certain phonological, mor-
phosyntactic or lexical contexts, absent in others.
5.5.1 5.5.1 Stable Final Consonants

As the name implies, stable consonants are not subject to variation they are
permanent parts of the morphemes or words in question. Among these conso-
nants, there is a further division to be made, linked to the orthography. The
most straightforward group of words with stable nal consonants ends not in
the consonant itself, but in orthographic < e >.30 A second set contains bare
consonants in the orthography, and the difculty here is to distinguish between
bare consonants that are stable and those that are latent. The examples in (21)
illustrate the two types of stable consonants.
Consonants 149
(21) Stable consonants 31
(a) word-nal < e > (b) bare consonants

/p/ crpe crepe cap cape
/t/ gte shelter mat dull
/k/ banque bank cognac cognac
/b/ robe dress snob snob
/d/ vide empty bled village
/g/ bague ring zinc /z'g/, erg zinc, erg
/f/ gaffe blunder soif thirst
/s/ bouillabaisse sh soup atlas atlas
/5/ riche rich Foch (surname)
/v/ rive shore Kiev Kiev
/z/ trapze trapezoid gaz gas
/</ juge judge hadj (/ad</) haj
/m/ femme woman islam Islam
/n/ baleine whale lichen lichen
// cagne preparatory
course
/l/ rafale gust bal dance
// mre mother mer sea
As indicated, one major difculty in dealing with the group in (21b) involves
the relationship between the orthography and the phonology. How does one
know when to pronounce bare nal consonants? While ultimately there will
be some arbitrariness in the phonological patterning, there are also generaliza-
tions to be extracted from this material. In what follows, we refer, in summary
form, to Tranel (1987: 154 67 and to Moody (1978).32 Those interested in
further details should consult the original sources. Much of this data will iden-
tify latent rather than stable consonants; we will deal with the former in the
next section. As for stable consonants, the lexical entries for these words pose
no problem: the consonant in question is directly included in the phonological
representation without need for further specication as to its status.
150 Chapter 5
(22) Realization of bare orthographic consonants

B: rare; normally pronounced (exception: plomb lead /pln/)
C: relatively rare; pronunciation varies (parc park /pak/ versus porc
pig /pn/); silent after nasal vowels (banc bank /b#/), with the
exception of donc therefore, usually /dnk/
D: relatively rare; normally silent (exceptions: sud south, week-end,
other loan words, and liaison [grand homme big man, prend-il is he
taking] where it is pronounced /t/)
F: relatively rare; normally pronounced (exceptions: clef key[also
cl], nerf nerve)
G: relatively frequent; pronounced after vowels; silent after < r > (with
the exception of erg /'g/) and < n >; (sufx -ing is /i0/)
K: rare; always pronounced (e.g., bifteck steak)
L: relatively frequent; normally pronounced (exceptions: cul ass,
sol drunk, and several words in < -il >: fusil rie, outil tool, gentil
kind, etc.)
M: rare; pronunciation varies (maximum /maksimnm/ versus parfum
perfume /paf/); use of nal < m > to indicate nasalization of
preceding vowel (faim hunger /f'/)
N: frequent; normally indicates nasalization of the preceding vowel
(bon good /bn/, n end /f'/, an year /#/), except polysyllabic words
in < Cen > are /C'n/ (abdomen /abdnm'n/, pollen /pnl'n/; exception:
examen examination /egzam'); normally pronounced in English loans
(gin /d<in/, blue-jean /blud<in/); frequent realization in liaison contexts
P: rare; normally silent (exceptions: cap cape, handicap, cep vine
stock, slip underwear, stop, etc.)
Q: rare; pronounced in cinq ve and coq rooster
R: very frequent; normally pronounced (exception: monsieur Mister
/msj/), except in words ending in < er >, which yield the possibilities
/e/, /'/, //: /e/ in innitives in < -er > (parler to speak /pale/);
the adjectives dernier last /d'nje/, entier whole /#tje/, leger light
/le<e/, premier rst /pmje/; and the sufxes -ier and -er (croupier
/kupje/, boucher butcher /bu5e/); /'/ elsewhere (cher expensive
/5'/, er proud /fj'/, enfer Hell /#f'/, hiver winter /iv'/,
gangster /g#gst'/, revolver /evnlv'/), except for some English
loans in < -er >, were it is // (leader /lid/, speaker /spik/)
Consonants 151
S: very frequent and variable; situation complicated by the use of

< s > as an inectional marker for plurals and second person singular
verb forms; extensive liaison; silent (except for liaison) in plural
endings, verb endings, adverbs, and prepositions (e.g., alors then,
toujours always, dans in, vers towards; exception: jadis formerly
/<adis/), nouns ending in < Vis > (e.g., Franais Frenchman, bois
wood), and adjectives (e.g., bas low, pais thick, gros fat,
confus confused); pronounced in words ending in < -ss, -ps > (e.g.,
express fast train, laps lapse); variable elsewhere (ours bear /us/
secours help /sku/, virus /viys/ abus abuse /aby/, vis screw
/vis/ avis opinion /avi/, pataqus mistaken liaison /patak's/
progrs progress /png'/, and many others), but more frequently
silent in these cases
T: like < s >, very frequent and variable; extensive use as third
person inectional marker (il peut, ils peuvent he/they can), where
it is always silent (except in liaison); normally silent in words ending
in < ait, aut, uit, oit, et, ot, at > (souhait wish, haut high, fruit,
droit right, guichet counter, sot silly, chat cat), after nasal
vowels (gant glove, vent wind), and after < r > (art, vert green);
normally pronounced following < c, l, p, s > (correct, contact, volt,
abrupt, concept, ouest west, test; exceptions: distinct /dist'/, succinct
/syks'/, exact /'gza(kt)/, suspect /sysp'(kt)/, il est he is /il'/); variable
elsewhere (dcit /desit/ esprit mind /'spi/, knock-out /knnkut/
debout standing /dbu/, brut /byt/ dbut /deby/, etc.)
X: relatively frequent; pronounced /s/ in six six /sis/, dix ten /dis/,
pronounced /ks/ in learned forms (thorax /tnaks/, sphinx /sf'ks/,
larynx /la'ks/, phnix phoenix /feniks/); normally silent elsewhere,
including as a plural or verb ending (choux cabbage, je veux I want,
prix price, paix peace, voix voice, faux false, doux soft, deux
two)
Z: rare; pronounced in fez /f'z/, gaz gas /gaz/, jazz /d<az/; silent
elsewhere, including the verbal suffx < -ez >, assez enough, chez
at (except in liaison), rez-de-chausse ground oor, raz de mare
tidal wave
(< h > is rare in isolation (e.g., Allah), and never pronounced; in combination
with < c > it varies between /k/ (varech seaweed), /5/ (Foch), and /t5/ (match);
in combination with < t > it is always /t/ (znith); < j > and < w > do not occur
word-nally; < v > is very rare, limited to loan words (e.g., Kiev, Tel-Aviv) and
always pronounced.)
A subsidiary issue that arises in this context involves consonants whose
pronunciation is optional in nal position. A list illustrating this phenomenon
appears in (23).
152 Chapter 5
(23) Optional nal consonants33

almanach /almana(k)/ almanac
ananas /anana(s)/ pineapple
aot /u(t)/ August
but /by(t)/ goal
Cassis /kasi(s)/ (city name)
cerf /s'(f)/ stag
fait /f'(t)/ event
fat /fa(t)/ smug
joug /<u(g)/ yoke
legs /l'(g)/ legacy
nombril /nnbi(l)/ navel
os (plural) /o/ or /ns/ bones
radoub /adu(b)/ retting
serf /s'(f)/ serf
sourcil /susi(l)/ eyebrow
yaourt /jau(t)/ yogurt
Tranel (1987: 15467) provides a brief and useful summary of the principal
generalizations involving bare nal consonants, useful in particular for learn-
ers of French as a second language. He notes, for example, that (with the
exception of liaison) grammatical markers are usually silent; that nal conso-
nants are usually silent in adjectives and are usually pronounced in borrowed
or learned words; that < s > and < t > present the greatest difculty; and that,
in statistical terms, the consonants most likely to be pronounced are < b, c,
f, k, l, m, q, r >. These remarks apply to bare word-nal consonants. When
we examine morphologically more complex structures, however, we nd that
many of the silent consonants can, in fact, make an appearance.
5.5.2 5.5.2 Latent Final Consonants

Latent consonants in French raise a number of fascinating theoretical prob-
lems. Simply put, are they present in lexical representations and deleted in
those contexts where they are not pronounced? Are they absent from lexical
representations and inserted when needed? Is there a mixed solution parallel-
ing both inserted and deleted schwas? Does the grammar list both forms (or all
Consonants 153
forms, if more than two are involved) and establish some more abstract link
between them, including a specication of the contexts in which they appear?
What type of generalizations are we seeking to identify? Consider, for exam-
ple, the pairs petit petite small or vert verte green. The nal consonants
in petit and vert clearly qualify as latent, but what about the /t/ in petite, fol-
lowed by < e >? Since this form (the feminine) is exclusively /ptit/, is the
nal consonant stable or latent? As to vert verte, what does the form verdure
greenery imply? These questions, about which opinion still seems divided,
will nd no answer in the descriptively oriented material presented here, but
they may help to illuminate the patterns and complexities involved. In particu-
lar, the frequency, range of distribution, and importance of latent consonants
should become evident.
Latent consonants are realized in a variety of contexts in French, contexts
involving both inectional and derivational morphology as well as syntax.
For the latter case, which involves liaison, we will postpone the discussion to
the next section. Here, we will examine latent consonants in inectional and
derivational processes. Historically, latent consonants became latent because
consonants were deleted in preconsonantal position and at the end of phono-
logical phrases, but retained before vowels. When these deletion processes
developed, nal < e > was still realized as a vowel (i.e., schwa), and schwa
served to block deletion. It is with the subsequent deletion of schwa that many
of the complexities arise, particularly when the relationship between orthogra-
phy and pronunciation is considered: petit petite but /pti/ - /ptit/, where in
the rst pair a nal orthographic consonant is present in both cases. Be that
as it may, three major inectional processes in French involve the presence or
absence of a latent consonant, a C alternation (or, in orthographic terms, a
< C# > < Ce > or < C# > < CV > alternation). They include the formation
of (a) feminine adjectives (and nouns), and of (b) plural and subjunctive forms
of verbs.34 These inectional contexts are illustrated in (24).
(24) Latent consonants in inection

(a) adjectives
/t/ /z/
haut high creux hollow
dissout dissolved acquis acquired
plat at confus confused
court short jaloux jealous
fort strong clos closed
vert green /s/
cuit cooked doux soft
petit small gros fat
154 Chapter 5
abstrait abstract bas low

absent absent pais thick
saint holy pars sparse
dfunt deceased divers varied
/o/ /nV/ gras fatty
sot foolish tiers third
idiot idiotic
/d/ M
chaud hot franc Frankish
froid cold 5
picard Picard franc frank
laid ugly blanc white
blond blond frais fresh
grand big /g/
' /'P long long
vain vain /j/
urbain urban gentil kind
/'/ /in/ /l/
n thin sol drunk
alpin alpine
malin shrewd /e/ '
// /yn/ lger light
un one premier rst
chacun each gaucher left-handed
/n/ /nn/ /o/ /'N/
bon good beau handsome
breton Breton nouveau new
/#/ /an/ /u/ /nl/
catalan Catalan fou foolish
paysan rustic mou soft
Consonants 155
/-kt/
abject abject
suspect suspect
exact exact
distinct distinct
Sufxes:
-ier/-ire prisonier /-je/ /-j'/ prisoner
-ain/-aine hautain /-'/ /-'n/ haughty
-en/-enne europen /-e'-/ /-e'n/ European
-ien/-ienne canadien /-j'/ /-j'n/ Canadian
-eux/-euse amoureux /-/ /-z/ in love with
-ais/-aise libanais /-'/ /-'z/ Lebanese
-ant/-ante agissant /-#/ /-#t/ active
-in/-ine blondin /-'/ /-in/ fair-haired35
(b) plural and subjunctive verb forms
rompt rompent rompe /p/ break
met mettent mette /t/ put
part partent parte /t/ leave
vainc vainquent vainque /k/ conquer
rpand rpandent rpande /d/ spread
mord mordent morde /d/ bite
crit crivent crive /v/ write
sert servent serve /v/ serve
nit nissent nisse /s/ nish
connat connaissent connaisse /s/ know
lit lisent lise /z/ read
dors dorment dorme /m/ sleep
prend prennent prenne /n/ take
vient viennent vienne /n/ come
feint feignent feigne // pretend
bout bouillent bouille /j/ boil
sait savent sache /v 5/ know36
156 Chapter 5
Latent consonants also surface in many derived forms, particularly given that
the great majority of French derivational sufxes, both learned and popular,
begin with a vowel. The number of latent consonants that appear in this way
is surprisingly large, so much so that many vowel-nal items as well as many
of those with supposedly silent orthographic consonants in (20) above actually
manifest a consonant in a variety of forms, not all of which contain the pho-
nological equivalent of the letter involved (e.g., nu nudit nude nudity;
vie vital life vital; chaos chaotique chaos chaotic; tabac tabagie
tabatire tobacco smoke shop snuff box; etc.). In the same vein, the
consonant appearing in derivation will, in normal cases, be identical to that
appearing in inection (and liaison), but not always (vert verte verdure; jus
juse juteux juice liquid used in tanning leather juicy). Derivational
examples are given in (25).
(25) Latent consonants in derivation

abandon abandonner desertion to abandon
abricot abricotier apricot apricot tree
absolu absolutisme absolute absolutism
abus abuser abuse to abuse
an annuel year annual
bnin bnignit benign mildness
biseau biseauter bevel to bevel
blanc blanchir white to whiten
blond blondin blond fair-haired
bois bois wood wooded
bras brassard arm armband
caoutchouc caoutchouter rubber to rubberize
chocolat chocolatier chocolate chocolate maker
clef clavier key keyboard
clin cligner wink to blink
cul culotte ass shorts
dos adosser back to lean against
crit crivain writes writer
tain tamer pewter to tin-plate
fusil fusiller rie to shoot
Consonants 157
gai gayer gay to cheer up

gentil gentillesse kind kindness
humain humanit human humanity
instinct instinctif instinct instinctive
jonc jonchaie rush reedbed
jour journe day day
jus juteux juice juicy
loup louveteau wolf wolf cub
nu nudit nude nudity
parfum parfumerie perfume perfume shop
plomb plombage lead lling
piano pianoter piano to tinkle at the piano
pied pitiner foot to stamp ones foot
sang sanglant blood bloody
sol soler drunk to intoxicate
tabac tabagie tobacco smoke shop
un unime one rst
vie vitalit life vitality
voisin voisinage neighbour neighbourhood
zro zrotage zero zero setting
In most cases, the relationship between the latent consonant and the orthogra-
phy is clear, and the orthography provides a direct indication of the specic con-
sonant to appear in derived forms. In several cases, however, either a consonant
different from the orthography of the base form appears (caoutchouc-caou-
tchouter, clin-cligner, tain-tamer,37 jonc-jonchaie, jour-journe, jus-juteux,
tabac-tabagie-tabatire, etc.), or no consonant is present in the base (absolu-
absolutisme, biseau-biseauter, nu-nudit, zro-zroter, etc.). Here, again, learn-
ers are confronted with arbitrary lexical occurrences reecting the history of the
items in question. Before turning to the last topic, enchanement and liaison,
one further discussion related to nal consonants and their history is in order.
This discussion involves and alternation of /al 'l l/ with /o/ or //.
Strictly speaking, this phenomenon involves instances of /l/ that are (or
were historically) in syllable codas either at the end of words or in precon-
sonantal position, rather than /l/ in strictly word-nal position. SF presents a
number of words where such preconsonantal /l/s and the vowel that precedes
158 Chapter 5
them often show the effects of a historical change leading to an alternation

with /o/ or //. Synchronically, we nd here the historical remnants of a pro-
cess, no longer productive, where in OF the phoneme /l/ had two variants, a
light or apical [l] and a dark or heavily velarized [N] (comparable to English
pill [p+N] or milk [m+Nk]), the latter occurring before consonants (in OF, nal
< s > [subsequently < x > in many cases], was pronounced /s/; deletion of
this /s/ is a later occurrence). In the course of time, the velarized allophone
weakened to [w], a closely related sound acoustically and articulatorily, and
this new glide combined with a preceding vowel to form a diphthong. These
new diphthongs were often identied with existing diphthongs and followed
the same evolutionary path, notably in monophthongizing to /o/ or // as the
case may be. Thus, the OF plural of animal was animals, with the [N] vocaliz-
ing to [w]. The resultant diphthong [aw] monophthongized to [o] (as did addi-
tional diphthongs from other sources: e.g., Latin causa, aurum > French chose
thing, or gold). Likewise Gallo-Roman els (< illos) > OF [ews] els > eus >
SF [] eux. The orthographic < x > in each of these cases is a medieval render-
ing of the < s > (usually a case marker) following and was reinterpreted
as the letter < x >. Various types of examples are given in (26).
(26) Alternations involving /l/

(a) singular plural alternations (nouns and adjectives)
animal animaux animal
capital capitaux capital
cheval chevaux horse
ciel cieux sky heavens
journal journaux newspaper
national nationaux national
oral oraux oral
original originaux original
signal signaux signal
social sociaux social
(b) derivationally related forms
altitude haut altitude high
chtelain chteaux lord of the manor castles
ciseler ciseaux to chisel scissors
coutellerie couteaux cutlery knives
crnel crneaux crenelated crenels
Consonants 159
falsier faux to falsify false

fuseler fuseaux to taper spindles
hraldique hraut heraldry herald
jumeler jumeaux to twin twins
loyal loyaut loyal loyalty
morceler morceaux to divide up pieces
niveler niveaux to level levels
sceller sceaux to seal seals
chevelure cheveu head of hair hair
(c) additional inectionally related forms
celle ceux this one these
elles eux them (f.) them (m.)
falloir il faut to be necessary its
necessary
meilleur mieux better best
valoir il vaut to be worth its worth
voulons veut we want want
Exceptions to the vocalization exemplied in (26) are seen in two areas:

learned forms such as altitude or falsier, which entered the language sub-
sequent to the process in question, and analogical reformations like cheveu,
niveau, sceau, where the singular was remodeled on the basis of the plural.
In any event, we have here another example of apparent irregularities in SF
that, in a historical context, reect regular phonological change, analogical
restructuring, and learned loan words. The synchronic structure of SF contains
many such mixtures and provides many internal clues to the history of the lan-
guage.38
Finally, after this brief excursus, we arrive at liaison, a complex domain
with links to virtually every problematic aspect of French phonological struc-
ture. Since liaison is closely allied with enchanement (in terms of issues of
syllabication, for example) we will treat the two topics together.
160 Chapter 5
5.6 5.6 Linking Phenomena: Enchanement

and Liaison
Enchanement and liaison reect the difference between stable and latent nal
consonants. Enchanement involves stable consonants, while liaison involves
latent consonants, but the contexts that determine their behaviour are identical:
both types of word-nal consonant are linked or resyllabied to the following
vowel, within but not between phonological phrases.39 The difference between
enchanement and liaison is illustrated in (27).
(27) Enchanement versus liaison

(a) enchanement
jeune enfant /<.n#f#/ young child
vite gar /vi.tegae/ quickly led astray
seize enfants /s'.z#f#/ sixteen children
petite orange /pti.tn#</ small orange
(b) liaison
bon enfant /bn.n#f#/ good child
petit lphant /pti.telef#/ small elephant
six enfants /si.z#f#/ six children
petit orage /pti.tna</ small storm
In the present context, enchanement raises few issues we have not already
seen in earlier sections, particularly those dealing with (re)syllabication and
the domains within which it operates. As a consequence, we now turn our
attention to a more detailed consideration of liaison.
5.7 5.7 Liaison

Liaison in French is clearly one of the most extensively studied of external
sandhi phenomena, in part because of its fascinating complexity, in part
because of its theoretical implications. Any attempt to deal with liaison must
immediately confront semantic, syntactic, morphological, phonological, lex-
ical, orthographic, historical, and stylistic questions, even in a variety of
the language as highly circumscribed (and thoroughly analyzed) as Standard
French. Liaison is normally described as follows:
Consonants 161
Liaison involves the phonetic realization of a word-nal consonant

that is normally silent. When such a latent consonant precedes a
vowel-initial word within the same phonological phrase, it is pro-
nounced. When the consonant is pronounced, it is syllabied with the
following vowel as the syllable onset, not with the preceding vowel
as part of the coda, despite the intervening morphological or syntactic
boundaries.40
Already in this descriptively oriented statement, we see the emergence of

a number of complex issues, not all of which we can address here. What pre-
cisely, for example, does latent mean in this context? Since many of the
major participants in liaison are cliticized articles and pronouns, how do they
relate to the notion word-nal? How are phonological phrases dened? How
do syllabication and resyllabication take place? The situation is further
complicated by the traditional tripartite classication of liaison into oblig-
atory, optional, and prohibited. It might seem that the apparent categori-
cal nature of the rst and last types renders them more straightforward, but
this is not the case, since an important lexical or item-specic component
(i.e., exceptions) enters into play in both domains, as do questions of the de-
nition of the phonological phrase. In optional liaison, moreover, stylistic fac-
tors (level of usage, rapidity of speech, etc.), as well as the variable constitu-
tion of phrases (the decision to combine two or more minimal phrases into
one longer one) complicate matters still more. Such questions serve to indi-
cate why phonological theorists have been so preoccupied with this topic. For
our purposes, however, it will sufce to provide a description of the main con-
straints involved in a description of liaison phenomena.
(28) Central properties of liaison 41

(a) Liaison is a prototypical phrasal phenomenon: it takes place across
word boundaries within phonological phrases but never occurs across
phrase boundaries:42
|| Sans ces_accords || | || aucune solution nest possible.||
/s#.se.za.kn. || | || o.kyn.sn.ly.sjn.n'.pn.sibl./
Without these agreements, no solution is possible.
|| les recherches realises || | || en_attendant le rsultat de lenqute ||
/le.5'5.e.a.li.ze. || | || #.na.t#.d#l.e.zyl.tad.l#.k't./
the tests done while waiting for the results of the inquiry
|| Je nai pas de temps || | || perdre. ||
/<ne.pad.t#. || | || a.p'd./
I have no time to lose.
162 Chapter 5
(b) Some potential instances of liaison (within phrases) are obligatorily

realized, some are prohibited, and some are optionally realized (with
varying degrees of frequency):
(i) obligatory:
determiner (+ adjective) + noun
aux_enfants /o.z#f#/ to the children
ses_lves /se.zel'v/ his students
un grand_ami /g#.tami/ great friend
les petits_amis /lepti.zami/ the small friends
(subject pronoun) (+ object pronoun) + verb (+ object pronoun)
nous_arrivons /nu.zaivn/ we arrive
elles_insistent /'l.z'sist/ they insist
on_appelle /n.nap'l/ Someones calling.
je les_ai vus /<le.zevy/ I saw them.
nous_en_avons /nu.z#.navn/ We have some.
il nous_embte /ilnu.z#b't/ Hes bothering us.
vivent_elles /viv.t'l/ do they live
allez_y /ale.zi/ go ahead
prenez_en /pne.z#/ take some
monosyllabic preposition + following word
dans_un_instant /d#.z.n'st#/ in an instant
sous_un_arbre /su.z.nab/ under a tree
en_Espagne /#.n'spa/ in Spain
sans_hsitation /s#.zezitasjn/ without hesitation
(ii) prohibited:
between a singular noun + adjective
un bois | immense /bwaim#s/ an enormous wood
une voix | leve /ynvwaelve/ a raised voice
ce banc | ombrag /sb#nba<e/ this shady bench
un temps | idal /t#ideal ideal weather
le drap | efl /ldaele/ the frayed cloth
Consonants 163
in plurals of compound words

des arcs | en-ciel /dezak#sj'l/ rainbows
les porcs | pics /lepnkepik/ porcupines
les salles | manger /lesalam#<e/ dining rooms
les moulins | vent /lemul'av#/ windmills
with proper nouns, acronyms, citations, metalinguistic usage, etc.
Paris | est une grande ville /pai'tyng#dvil/
Paris is a large city.
les | SDF rclament... /le'sde'feklam/
the homeless ask for
ses | Ah, je le vois ne me convainquent plus
/sea<lvwa/
His oh, I see doesnt convince me any more.
il a dit | envers non | envoi /iladi#v'nn#vwa/
He said envers, not envoi.
when the word ends in // + consonant
vers Angers /v'.#<e/ towards Angers
bourg anim /bu.anime/ lively city
rapport exact /apn.'gza/ exact connection
(Note: when the nal consonant is the plural maker < s >, liaison
occurs: rapports exacts /apn.z'gza/; in the absence of liaison,
the // resyllabies.)
(iii) optional:
after verb forms
je suis (|) en colre /<5K(z)#MnN' Im angry.
ils vivent (|) Paris /ilviv(.t)apai/ They live in
Paris.
elles_taient (|) alles /'lzet'(.t)ale/ they had gone
after polysyllabic prepositions
pendant (|) un repas /p#d#(.t)pa/ during a meal
durant (|) une heure /dy#(.t)yn/ for an hour
164 Chapter 5
after specic words

long (|) hiver /ln(.g)iv'/ long winter
il ma beaucoup (|) impressionn /ilmaboku.(p)'R'ULnPG
He impressed me greatly.
(c) certain structures, classes of words, or individual words provide
systematic or variable exceptions to liaison:
adjective + noun versus noun + adjective
un petit[Adj]_Irlandais[N] un petit[N] | irlandais[Adj]
a small Irishman a small person who is Irish
/pti.til#d'/ /ptiil#d'/
compound words
aspirate-h words
polysyllabic prepositions
et, trop, oui, vers, avant, selon, etc.:
groupes cologistes | et | associations /gupeknln<isteasnsjasjn/
ecological groups and associations
elle sort | et | il sen va /'lsne'ls#va/
she goes out and he leaves
des | ouis catgoriques /dewikategnik/
categorical yesses
(compare louie /lwi/)
les | oui-dire /lewidi/
hearsay
(d) liaison is associated with item-specic allomorphy:
ma/ta/sa + V, ce, beau, nouveau, vieux, etc.
mon_affaire (f.) /mn.naf'/ my business
cet_lment /s'.telem#/ this element
bel_enfant /b'.l#f#/ handsome child
nouvel_objectif /nuv'.lnb<'ktif/ new objective
vieil_homme /vj'.jnm/ old man
Consonants 165
(e) the realization of a liaison consonant may occur without associated

resyllabication (liaison sans enchanement)43:
javais /z/ un rve /<av'z.'v/ I had a dream.
nous voulons /z/ entendre /nuvulnz.Ct#dC/ We want to hear.
il est /t/ inscrit /il't.'ski/ Hes registered.
ont /t/ abouti /nt.abuti/ have succeeded
il faut /t/ encourager /ilfot.#kua<e/ one must encourage
qui sont /t/ en vrit /kisnV#veite/ which are, in truth
dans son /n/ interprtation /d#snn.'t'petasjn/ in his
interpretation
(f) widespread erroneous overapplication of liaison (so-called fausse
liaison, cuirs, velours, pataqus, plus extension to aspirate-h forms)
occurs. This behaviour frequently appears in the speech of those who
have not fully mastered the formal variety of the language when they
try to speak more correctly than they normally do:44
je ne sais pas /t/ qui est-ce /<ns'pa.taki's/

I dont know whose it is.
cest pas /t/ moi /s'pa.tamwa/
Its not mine.
elle devra /t/ tre prsente /'ldva.t'tpez#/
She should be present.
ils taient /z/ amis /ilzet'zami/
They were friends.
lorsque jy ai /z/ t /lnsk<ie.zete/
when I was there
sous une forme ou /z/ une autre /suzynfnmu.zynot/
in one form or another
un n [t] esprit /f'.t'spi/
a keen mind
les /z/ Hollandais /le.znl#d'/
the Dutch
166 Chapter 5
les allocations aux /z/ handicapps /o.z#dikape/

the benets to the handicapped
ils ont /t/ harcel /lizn.tasle/
they harassed
le palais tout /t/ hriss de drapeaux /lpal'tu.teiseddapo/
the palace all covered in ags
(g) a variety of restricted phonological adjustments is associated
with liaison, involving nasal vowels, nal consonant changes, vowel
adjustments, and so on:
mon_ami /mn.nami/ my friend
un_instant /.n'st#/ an instant
grand_homme /g#.tnm/ big man
six_hommes /si.znm/ six men
dernier_instant /d'nje.'st#/ /d'nj'.'st#/ last instant
premier_tage /pmje.eta</ /pmj'.eta</ rst oor
As we have seen, the relationship between the consonant appearing in liai-

son and that appearing in derived forms is usually one of identity, a link that
simplies descriptive (and learning) problems. On occasion, however, there
are discrepancies. Both types of behaviour are illustrated in (29).
(29) Correspondences between liaison and derivational consonants

(a) identical consonants
petit petitesse petit ami /ptitami/
small smallness small friend
bon bonier bon ami /bnnami/
good to improve good friend
deux deuxime deux amies /dzami/
two second two friends
(b) different consonants45
grand grande grand homme /g#tnm/
big (m.) big (f.) big man
fond fonder de fond en comble /dfnt#knbl/
Consonants 167
bottom to found from top to bottom

gros grosse un gros inconvnient /goz'knvenj#/
big (m.) big (f.) a big inconvenience
Against this heterogeneous background, what are the most salient properties
of liaison? First, liaison is not the unied phenomenon that the denition at the
beginning of section 5.7.1 might lead us to believe. Although there are clearly
systematic aspects to the behaviour of liaison consonants aspects involving
the presence of a following vowel, sensitivity to phonological phrases, syllabi-
cation, and so on each of these properties is also subject to greater or lesser
degrees of exceptional behaviour. Recall, for example, the fact that aspirate-h
words, despite being phonetically vowel-initial, normally block liaison, or the
fact that liaison sans enchanement results in the realization of a latent conso-
nant without that consonant being resyllabied. Nonetheless, the central prop-
erties in (28) above, those that occur consistently, appear to characterize liai-
son most simply. (Alternatively, learners of French would be well advised to
concentrate on the following simplied review of the basic features of liai-
son.)
(30) Basic features of liaison

(a) contexts (relatively few lexical items are involved: determiners,
prenominal adjectives, numerals, third person verb forms,
monosyllabic adverbs and prepositions, idioms):
determiners and following elements (nouns, adjectives)
les_arbitres the referees
un_autre jardin an other garden
tout_autre problme every other problem
mes_ennemis my enemies
adjectives and following nouns
petit_oiseau small bird
gros_ennui big problem
grand_homme big man
de graves_incidents serious incidents
six_lves six students
pronouns plus verbs
vous_allez Youre going.
elles_arrivent Theyre arriving.
168 Chapter 5
on_est l Were there.

vient_elle Is she coming?
je les_ai vus I saw them.
following monosyllabic prepositions
dans_un_instant in an instant
en_hiver in winter
sans_y rchir without thinking about it
sous_un_autre angle from another angle
following monosyllabic adverbs
trs_intressant very interesting
trop_important too important
moins_habile less skilful
plus_exigeant more demanding
tant_aime so loved
bien_aimable very likeable
the verb form est
elle est_arrive she arrived
il est_en colre Hes angry.
in numerous xed expressions (including most compound words)
rien_ faire nothing to do
comment_allez-vous How are you?
Etats-Unis United States
pied--terre temporary lodging
pot-au-feu stew
vis--vis opposite
(b) morphological role (SF structure dictates that the most frequent
occurrences of liaison will signal [or be correlated with] either plural
noun or pronominal forms, or third person verb forms):
les_lments the elements
mes_enfants my children
ces_indications these indications
Consonants 169
elles_arrivent (cf. elle arrive) They arrive.

ils_indiquent (cf. il indique) They indicate.
vient_il Is he coming?
parle-t_elle Is she speaking?
aux jeunes_lves (cf. au jeune lve) to the young students
petites_oranges (cf. petite orange) small oranges
These morphological functions have the further consequence that /z/
and /t/ are by far the most frequent consonantal participants in liaison,
whether it be obligatory or optional.46
(c) inuence of the phonological phrase: although the correlation is
far from perfect, phonological phrases are closely related to syntactic
phrases. Syntactic phrases may, needless to say, be longer or shorter; in
rapid or informal speech, minimal phrases may be conjoined to form
longer phrases, with consequences for liaison: liaison never occurs
at the end of phrases, only between words within phrases. Hence,
if minimal phrases are joined into one longer phrase, conditions for
liaison may be created, as in the examples below.
|| songez || au lendemain || || songez_au lendemain ||
think about tomorrow
|| ils travaillent || ensemble || || ils travaillent_ensemble ||
they work together
|| jen suis || extrmement er || || jen suis_extrmement er ||
Im very proud of it
(d) tendencies: there exists a general tendency in SF, except in the
most formal style, to reduce the number of optional liaisons. This
tendency even extends, in less formal registers, to the elimination
of formerly obligatory liaisons (e.g., en | un jour). Against this
background, it is less likely to cause comment if one errs by failing
to make obligatory liaisons than if one over-extends the process. This
being said, the cases indicated in (30a) remain central to the phonology
of French, and their absence in the language of learners will inevitably
have sociolinguistic consequences.
This concludes the discussion of liaison and, in fact, of consonants in general.

The next topic to be considered involves prosody a discussion of intonation
in particular, as well as a return to elements of stress assignment.
170 Chapter 5
Notes
1. The two nasal consonants / 0/ may not be followed by glides, for reasons linked
to the way these segments were introduced into the language. There are also gaps
in the set of vowels following these glides (the sequences /j', j', wa, w'/ predomi-
nate, again for historical reasons), but the absences appear accidental rather than
systematic.
2. As might be expected, some further observations are relevant to this list. For
example, many clusters (e.g., hydne hydnum (type of mushroom) or axolotl
axolotl) are represented by a single anomalous example. /z/ is uniquely found
in quatorze fourteen, but the frequency of this word makes the cluster appear
much less strange than /dn/ or /tl/, to name but two. /0g/ varies with /0/ in
words like pouding pudding, with the latter pronunciation predominating, in
fact. Because of the syntactic constraints on its use, lorsque when rarely if ever
appears phrase-nally. As a consequence, it will either be followed by a vowel,
and pronounced /ln.skV/, or by a consonant, in which case a schwa is required
and the structure will be /ln.skC/. The cluster /sk/, in other words, is articial
in nal position.
3. Phrase-initially, schwa is obligatorily retained in the huitime example: le huitime

est difcile The eighth is difcult. /litj'm'disil/.
4. These proposals are not mutually exclusive. For preliminary discussion of the rst
possibility, see Walker (1978); for a theory of the second, Vennemann (1988).
5. // represents a special case, since it assimilates to /n/ upon contact either with the
/4/ (the precursor of Modern French //) or epenthetic /d/.
6. By geminate is meant a single articulation whose duration is prolonged, not a

repeated or rearticulated consonant. Functionally and phonetically, geminate con-
sonants in French would normally be expected to provide both a syllable-nal and
a syllable-initial component, by straddling the syllable boundary. This appears to
be the case except for the geminates arising from the accent dinsistance (see [11]
below), which may occur word- and hence syllable-initially (super /ssyp'/) and
which, when intervocalic, may also be syllable-initial (affreux hideous /a.ff/).
7. To such forms we might add the geminate /jj/ produced in the imperfect forms
of verbs whose stem ends in /j/ (fouillions search, travaillions work, etc.)
although this geminate may simplify, in informal speech, to a single consonant. In
fact, the geminate // in these verb forms is also often simplied in less formal
contexts.
8. // (< gn >) does occur word-initially, but only in a relatively small number
of words of peripheral status (slang, onomatopoeia, child language, loans):
gnaf cobbler, gnangnan soppy, gnard brat, gnle hooch, gnouf the clink,
gnocchi gnocchi, gnognote useless, gnon bash, gnaf-gnaf bow-wow, etc.
Consonants 171
9. The situation is, as usual, somewhat more complicated. On occasion, // did

occur historically in nal position following apocope, but it subsequently merged
with /n/ in that position (cf. bain bath [OF baign] baigner, malin maligne
shrewd, soin soigner care, gain gagner wages, etc). This merger, along
with the other factors leading to the development of //, accounts for its absence
word-nally.
10. Note the presence of < e > (normally deleted) in the case of // in preconsonantal
and nal position. For completeness, we should also note the frequent pronuncia-
tion of < gn > as /gn/ in a set of learned words (although // and /nj/ occur here
too): cognitif /kngnitif/ cognitive, diagnostic /djagnnstik/ diagnosis, gnostique
/gnnstik/ gnostic, magnitude /magnityd/, rcognition /rekngnisjn/, etc.
11. Malecot (1972) and Malcot and Metz (1972) provide a detailed description of
schwa deletion and nasal assimilation in SF as they relate to this data, including
the social conditioning of both phenomena.
12. In CF, tradition has it that the assimilation works in the opposite direction /5/
assimilates to the /v/, which itself weakens to /w/, yielding the notorious /<wal/
joual, a term for both horse and for the often stigmatized CF vernacular.
13. There is much variation, both regional and individual, in the pronunciation of the
sufx -isme: [isO] or [izm].
14. The words sbire henchman, Sganarelle, sgrafte sgratto normally begin with
/zb/ or/zg/.
15. There is a certain debate in the phonetic and phonological literature concerning
the precise realization of these assimilated consonants and their phonological
interpretation. According to some (e.g., Valdman [1993: 213]), they remain dis-
tinct from their voiced or voiceless correlates respectively, while Carton (1974:
83 86, following Martinet) argues for the phonological identity of assimilated
and nonassimilated pairs (e.g. /p/ = /D/; /b/ = /R/). In the former view, devoiced
/b/ would remain distinct from /p/ because it remains lax, while /p/ is tense. The
other voiceless-voiced pairs behave in a similar fashion. This residual tense-lax
distinction explains why certain speakers may distinguish jeter from acheter:
the rst contains voiceless but lax /</, while the second has voiceless tense /5/.
There has been some study of this matter (Kohler et al. [1979, 1981], as well as
Gather [1998], the latter using notions of syllable structure) but further instrumen-
tal work would be of benet.
16. Walter (1976: 410-23), who provides a very detailed description of voicing assim-
ilation, notes that assimilation is more frequent when the consonants in question
are not separated by a schwa: greater frequency of assimilation in disgrce dis-
grace than in tte-bche head to tail, for example.
172 Chapter 5
17. We exclude certain marginal forms (largely interjections or onomatopoetic items)

such as hop!, h!, hein!, houp-l! which may occur with [h].
18. On occasion, one may hear, in deliberate speech, a glottal stop preceding aspi-
rate-h words. However, as Malecot (1975) demonstrates, this is part of a more
general phenomenon in French, and the presence of glottal stop is clearly not cor-
related with < h >.
19. Or at least those where orthographic < h > is involved; recall that certain glide-
initial words (yacht, watt) also behave as if there were an aspirate consonant
present.
20. Examples are from Martinet and Walter (1973). Current observers note that many
additional words (hollandais, haricot, etc., as well as words with no < h >: iambe,
ouatte) also vary. A related type of variation is seen in verbs like houer /we/ or
/ue/, huer /e/ or /ye/ (both consistently aspirate) where the variation is between
an initial vowel or semi-vowel. We omit from this discussion the interesting cases
of proper nouns with aspirate-h, cases that bring their own complexities (e.g.,
variation between aspirate and nonaspirate in several forms such as Hugo, aspira-
tion in hrisson but not in le comte dHrisson, and so on). For discussion of the
latter items (as well as the phenomenon of < h > in general), see Fouch (1959:
251 65, especially 258 ff).
21. From hte spit, leading to the meaning small piece of roast meat.
22. This acronym, pronounced [a5'l'm], stands for habitation loyer modr public
housing.
23. One frequent exception should be noted here: hasard stroke of luck is an aspi-
rate-h word (le hasard), but it does show enchanement with par: /pa.aza/.
24. This parallelism is not perfect, as there is perhaps no need to say. For example,
deletion of the schwa in le is possible before consonant-initial words (cest le
garon /s'lgasn/), but may not occur before aspirate-h words (cest le hros
/s'leo/). Thus, aspirate-h words do not always behave as if they were conso-
nant-initial.
25. See Catach (1995: 1148).
26. See Tranel (1981: 286 88) for further examples and discussion.
27. Data from Martinet and Walter (1973).
28. For reasons that will become clear shortly, this group is smaller than one might
believe, since many putative vowel-nal morphemes have derivational variants
where the stem ends in a consonant: vie vital life vital, nu nudit naked
nudity, or even noeud nodosit knot node, for example.
Consonants 173
29. To be fully accurate, this statement needs a slight modication: the consonant
itself may vary, but some consonant is present in all versions of the morpheme,
as in sec dry: sche, schement, schoir, scheresse; loup wolf: loupiot, louve,
louvet, louveteau, louveter, louveterie, etc.
30. Whether this < e > represents schwa i.e., a phonological segment present in the
representation of the forms in question remains a hotly debated topic. If it does,
we need to refer to the conditions under which it is maintained or deleted (as in
the preceding chapter). If it does not, we will need to specify the contexts in which
it is inserted where needed (e.g., preceding aspirate-h, in VCC#C sequences, and
so on). In all likelihood, the best solution is for there to be a mixture of underlying
and inserted schwas, not all of which are linked to the presence of < e >.
31. Innumerable comments could be made concerning these forms, some of which
will occur in (22) below. Perhaps the simplest observation involves the fact that,
with the exception of < r >, bare stable consonants are relatively rare in French
compared to those protected by < e >. Furthermore, because of a variety of his-
torical processes, certain restrictions apply. For example, < se > does not occur
in this list representing /s/, since < VsV > is realized as /Vz(V)/ (hence, bouil-
labaisse with double < ss >); < c > and < g > do not occur followed by < e > if
they are to be interpreted as /k/ or /g/, since < ce > and < ge > represent /s/ and
/</, respectively (hence, banque, bague); < m > and < n > are rare in nal posi-
tion since unprotected nasal consonants in this context usually indicate nasaliza-
tion of the preceding vowel; < gn > does not occur as a representative of a bare
consonant, since the conditions producing // require a following (orthographic)
vowel at all times; and so on.
32. Moody is based on a circumscribed data set drawn largely from Savard and
Richards (1970), so is less intimidating than Tranel, whose more complete data
set inevitably contains more exceptions.
33. See Walter (1976: 45158) for many more examples. Malcot and Richman 1972
provide details of usage, correlated with sociolinguistic factors, for several of
these forms.
34. Historically, once again, many adjectives form the feminine by adding an ortho-
graphic < e >, and the plural subjunctive verbal sufxes are also vowel initial.
(Many additional verb forms [e.g., the imperfect] use vowel-initial sufxes and
also provoke the appearance of the latent consonant.)
35. Needless to say, this does not exhaust the problem of latent consonants in noun
or adjective formation, but it does give some idea of the nature and extent of the
issue. We have not, for example, mentioned the processes involved in the forma-
tion of names (Fernand-Fernande, Clment-Clmence, etc.) or the interchange of
nal consonants (sec-sche dry, actif-active active, among others). For much
more detail, see Durand (1936).
174 Chapter 5
36. Once again, we are far from exhausting the implications of latent consonants in
the verb conjugation, and give these examples for illustration only. In particular,
the latent consonant appears much more widely in the conjugation in innitives,
in the imperfect, etc. Although the latent consonant is usually constant across all
verb forms, this is not always the case: dit-dites-disent say, sait-sachent-savent
know, peut-puisse-peuvent be able, etc. Encyclopedic catalogues of the details
of the SF conjugation abound; for more general discussion, consult Le Gofc
(1997), Martinet (1958b) or Fouch (1967). For initial discussion of the role of
latent consonants in the CF vernacular, see Walker (1995).
37. This and the preceding example illustrate the historical process merging word-
nal nasal consonants to /n/, and reect the phonotactic constraint on // dis-
cussed above in section 5.2.
38. Here, as elsewhere, we have only scratched the surface of a complicated set of
problems. For example, the circumscribed domain of animal names, where the
forms for offspring or female animals are often derivationally related to the mas-
culine, provides much additional data: chameau chamelle camel, corbeau
corbillat crow, crapaud crapelet frog, hirondeau hirondelle swallow,
maquereau maquerelle mackerel, moineau moinelle sparrow, oiseau
oiselle bird, etc. Certain forms ending in < ail > (= /aj/), such as bail baux
lease, corail coraux coral, travail travaux work, and vitrail vitraux
stained glass window, demonstrate that earlier palatalized /l/ also vocalized to
[w] under similar coda conditions, while becoming /j/ in others. Finally, the wide
existence of plurals ending in < als > rather than < aux > (carnavals carnivals,
chacals jackals, festivals festivals, nals nal, nasals nasal, etc.), as well
as many forms that vary (vals vaux valleys, idals idaux ideals), further
indicate that the process of vocalization is no longer productive.
39. As expected, aspiration (in the general sense, involving semi-vowels as well)
blocks both enchanement and liaison: sept homards seven lobsters /s't.nma/
and ces homards these lobsters /senma/.
40. For denitions of liaison, consult Encrev (1988), Fouch (1959), Tranel (1987),
or Valdman (1993), among innumerable others. The description to follow owes
much to Klausenburger (1984). Several terminological alternates to phonologi-
cal phrase are also in widespread use, among which the most common are groupe
rythmique and groupe accentuel. As we will see, the realization of the liaison
consonant must be separated from its enchanement i.e., its (re)syllabication
with the following vowel since there exist instances where the former but not
the latter occurs.
41. In what follows, _ , (|), and | are used to indicate obligatory, optional, and prohib-
ited liaisons respectively, and || to indicate phonological phrase boundaries. This
brief survey is illustrative only, intended to indicate certain properties relevant
to our general discussion, not to present a complete survey of the phenomenon.
Consonants 175
For well-organized discussion of the phonological, morphological, syntactic, and

semantic constraints on liaison, see Esposito (1994), Tranel (1987, chapter 11) or
Valdman (1993, chapters 10 and 11).
42. While this formulation is generally true, we should note two types of case where
a liaison consonant may be pronounced at a phrase boundary: the celebrated
cases of liaison sans enchanement: je ne lavais pas entendu I hadnt heard it
/<nlav'paz || #t#dy/ [Encrev 1988], and the realization of a liaison consonant
in parenthetical insertions where, by denition, the liaison consonant is in phrase-
initial position (un robust, mais petit, enfant a strong, but small, child /'nbyst ||
m'pti || t#f#/ [Pichon 1938]).
43. Examples from Encrev (1988, chapters I and II).
44. In other words, speakers who normally use popular language produce false liai-
sons (a type of hypercorrection) when they try to use a formal style that they do
not fully control. False liaisons, then, are not characteristic of spontaneous popu-
lar usage, but of efforts by speakers of the vernacular to use a formal register that
is not a full part of their repertoire. Interestingly, Encrev (1988) also notes false
liaisons in the formal speech of politicians who produce liaison sans enchane-
ment as well.
45. The sometimes claimed lack of correspondence affecting long (long longue /g/
un long hiver a long winter /lnkiv'/), though still sometimes found in text-
books (e.g., Fouch [1959: 436]), has been obsolete for close to a century (cf.
Pichon and Damourette 1936). Other apparent mismatches (de pied /t/ en cap
from head to foot, sang /k/ impur tainted blood) are also archaic or part of
xed expressions. Finally, the realization of < d > as /t/ in a number of inverted
verb constructions (perd perdons perd-il /p'til/ does he lose, etc.) should
probably be seen as a separate morphological adjustment rather than as liaison,
since the /t/ appears following vowels (ira-t-on will we go) as well as with verbs
whose (sometimes abstract) stem ends in a consonant other than /t/, e.g. craque-
t-il /kaktil/ does it creak, rampe-t-il /#ptil/ does it crawl, mord-il /mntil/
does it bite, me convainc-t-il /mknv'til/ does it convince me, vainc-t-il /v'til/
does he defeat, and so on).
46. Both the frequency and the morphological role no doubt account for the appear-
ance of these consonants in many instances of false liaison: quatres enfants four
children /katz#f#/, il va-t et vient he comes and goes /ilvatevj'/, etc., as well as
neologisms such as zyeuter /zjte/ to eye (< les yeux eyes) or Zanimo /zanimo/
(< les animaux), a trade name for animal crackers.
177
Chapter 6
Prosody

Traditionally applied to the study of the structure of meter and verse, prosody
is now used more generally within phonology to refer to variations in pitch,
loudness, duration, rhythm, rate of speaking, and occasionally syllabication.1
Without identifying them explicitly as such, we have already discussed several
prosodic phenomena in SF: syllabication in section 3.1, vowel length in 4.1,
and long or geminate consonants in 5.1. In this chapter, we will be primarily
concerned with stress, rhythm, and intonation.
Prosodic phenomena, intonation in particular, have often been neglected in
phonological studies. This reduced attention no doubt arises in part because
of the complexity of the physical properties through which prosody is mani-
fested (and the need for complex instrumentation and techniques to study these
properties), because of the variety of syntactic structures involved (and the
concomitant need for large bodies of data), because prosody involves virtually
every component of the grammar from phonetic detail to semantic and prag-
matic considerations, and because of the complex intermingling of general lin-
guistic structures with highly variable individual expressivity. In what follows,
we will concentrate more on the general, systematic structures of French than
on those elements reecting individual attitudes or emotions.2
6.1 6.1 Stress and Rhythm

Stress involves the assignment of greater prominence to one syllable in com-
parison with its neighbours. Cross-linguistically, the physical correlates of
stress may include one or more of greater loudness, longer duration, or higher
pitch. In SF, stress is usually indicated by the greater length assigned to stressed
syllables3 (rather than by reliance on tonal prominence, for example, although
variations in tone or increased loudness may also be present). Assuming a
difference between prominent and nonprominent syllables, rhythm refers in
general terms to a perceived pattern of prominent syllables. Rhythmically,
it is common to divide languages into those with stress-timed rhythmic pat-
terns (where stressed syllables occur at approximately equal intervals of time,
178 Chapter 6
interspersed with varying numbers of unstressed syllables) versus those with

syllable-timed patterns (where each syllable takes roughly the same amount of
time to produce and has roughly the same importance).4 French, in contrast to
English, is described as a syllable-timed language, with all syllables within a
phonological phrase given equal prominence as they lead up to the phrase-nal
stressed syllable. The contrast between stress and syllable timing may be seen
in (1), where stressed syllables are in bold face.
(1) Stress versus syllable timing

(a) stress timing
Let me not to the marriage of true minds
Admit impediments. Love is not love
Which alters when it alteration nds
Or bends with the remover to remove. (Shakespeare, Sonnets)
(b) syllable timing
Si je la hassais, je ne la fuirais pas. (Racine, Phdre)
Jose dire pourtant que je nai mrit
Ni cet excs dhonneur, ni cette indignit (Racine, Britannicus)
Jai perdu ma force et ma vie
Et mes amis et ma gaiet (Musset, Tristesse)
Against this general background, we are in a position to discuss stress assign-

ment in French in more detail.
6.1.1 6.1.1 Phrasal Stress

Stress in French is considered to be a phrasal phenomenon: the nal syllable
of phonological phrases is stressed (with the exception of syllables consisting
of schwa): larbitre the referee, jamais never, je le vois I see it, je ne le
vois jamais I never see it. Nonnal syllables, even those that are word-nal
within the phrase, are unstressed. French stress is, then, predictable and xed,
once phrase boundaries have been determined. (Consult section 3.3 for discus-
sion and exemplication of phonological phrases.) Because French stress is
predictable and consequently nondistinctive, it is unlike stress in Spanish or
English, where stress may fall on different syllables within words (often cor-
related with morphological or syntactic categories) and distinguish one word
from another: permit, compress (nouns) versus permit, compress (verbs); pho-
tograph versus photography; Spanish trmino end, termino I nish, ter-
min he nished; and so on. French stress is also unlike stress in a language
Prosody 179
like Latin, where the position of stress may vary within the word, but where
(in longer words) its placement is determined by the weight of the penultimate
syllable. The SF phrasal stress pattern is illustrated in (2), where the stressed
syllables are again in bold face.
(2) Phrasal stress

oui Yes.
voil there is/are
le voil There it is.
il y en a There are some.
il y en a trop There are too many.
il y en a trente-quatre There are 34 of them.
elle revient Shes coming back.
son lve || revient Her students coming back.
son ancien lve || revient demain Her former students coming
back tomorrow.
son ancien lve || revient || Her former students coming
sans le lui dire back tomorrow without
telling her about it.
Unstressed syllables are given equal weight within the phrase; they are pro-
nounced, that is, with a regularity and equilibrium that block any vowel reduc-
tion and lead toward the prominence of the syllable at phrase end.5 In this case,
the contrast between English and French is instructive, since English shows a
widespread weakening of unstressed vowels which reduce to schwa. Consider
the contrasts in (3), particularly (in English) the variation in pronunciation
between stressed and unstressed versions of the same syllable.
(3) French versus English: vowel reduction

F Canada /kanada/
E Canada /knd/
F canadien /kanadj'/
E Canadian /knedjn/
F photographe /fntngaf/
E photograph /fodgrf/
F photographie /fntnga/
E photography /ftngr/
180 Chapter 6
F communiquer /knmynike/
E communicate /kmjunket/
F cest une question difcile /s'tynk'stjndisil/
E thats a difcult question /&tsdIfkltkw'st5n/
F son attitude || ne me surprend pas /snnatityd nmsyp#pa/
E his attitude || doesnt surprise me /h+zdtjud dzntsprajzmi/
Since stress is phrase-nal in French, decisions concerning the combination of

minimal phrases into one longer phrase (see chapter 3, section 3.3) will affect
the stress and rhythmic patterns of longer utterances. This contrast is seen in
(4).
(4) Minimal versus longer phrases

je ne lavais pas || entendu je ne lavais pas entendu
I hadnt heard it.
tu devrais || lui en parler tu devrais lui en parler
You should speak to her about it.
le match || sest bien termin le match sest bien termin
The game ended well.
elle a promis || de nous suivre elle a promis de nous suivre
She promised to follow us.
un journal || hebdomadaire un journal hebdomadaire
a weekly newspaper
lautobus || de la ville lautobus de la ville
the city bus
un appartement || louer un appartement louer
an appartment to rent
elle me le dit || assez souvent elle me le dit assez souvent
She tells me often enough.
il faut y aller || tout seul il faut y aller tout seul
You have to go there alone.
Prosody 181
Finally, recall that vowel length (chapter 4, section 4.1) is also conditioned by
stress: long vowels occur only in stressed closed syllables. As a consequence,
if a minimal phrase containing a long vowel is combined into a longer phrase,
the loss of stress on its nal syllable will result in the loss of vowel length as
well: il espre || nous y rencontrer He hopes to meet us there. (['sp'] )
versus il espre nous y rencontrer (['sp']), il compte || y aller Hes counting
on going there. [knt] versus il compte y aller [knt].
6.1.2 6.1.2 Emphatic Stress

In addition to its normal phrase-nal emphasis, SF also uses stress to signal a
variety of semantic or affective contrasts. Examples are given in (5).
(5) Emphatic stress

jai dit intention, pas attention I said INtention, not Attention.
un effet fondamental a fundamental effect
une erreur impardonnable an unforgivable mistake
par contre, tu ne comprends pas On the other hand, you dont
understand.
des visiteurs trs nombreux very many visitors
rien de plus scandaleux nothing more scandalous
il a parfaitement raison Hes perfectly correct.
cest si amusant Its so funny.
cest tout fait impossible Its completely impossible.
Encore de la morue! Voil huit Cod again! Weve eaten nothing
jours que nous ne mangeons que but cod all week! Its
de la morue; cest dgotant! disgusting! (from Grammont
[1914: 144])
Phoneticians and phonologists begin to comment more extensively on the

appearance of the accent dinsistance or emphatic stress around the begin-
ning of the twentieth century (e.g., Nyrop 1934, rst edition 1902; Grammont
1914), and indications are that the phenomenon is expanding. Most generally,
this emphasis falls on the rst syllable of the word or phrase, either by empha-
sizing the rst vowel (intention, pas attention) or, more frequently, by length-
ening the rst consonant of the word (cest essentiel [s't'ss#sj'l]; for more
examples cf. section 5.1). Although emphatic stress normally affects full
words (nouns, verbs, adjectives), it is not unknown for clitic forms to be
accented as well, as the examples above make clear. Further, because emphatic
182 Chapter 6
stress does not replace phrase-nal stress but is in addition to it, it is easy to
see how it may contribute to alternating rhythmic patterns within phonological
phrases.
As indicated, most analysts agree (cf. Lucci 1983: 69) that use of emphatic
stress is increasing in frequency, and that this stress pattern is characteristic of
a certain type of discourse or discourse situation, particularly more formal situ-
ations involved with speaking in public: interviews, conferences, oratory. It is
also widely heard in news broadcasts.6 The increasing currency of emphatic
stress is also showing effects on other components of the phonological system.
Walter (1977: 51), for example, explicitly attributes the increasing stability of
schwa in initial syllables (discussed in section 4.5.4.6) to the increased fre-
quency of emphatic stress affecting those same syllables: Ce serait alors par
le biais de la prosodie, grce laccent dinsistance dans ce cas, plus directe-
ment sensible aux effets du sens, que serait favoris une modication du statut
dun lment phonique, par phonologisation de cet lment dans une position
donne. We see here, in other words, a classic example of the desire to produce
stylistic effects having an impact on the frequency of certain prosodic struc-
tures, with these prosodic changes in turn affecting the segmental phonology.
Intonation, as we will see in the next section, is also directly linked to stylistic
effects, as well as to more concrete syntactic patterns, in a variety of ways.
6.2 6.2 Intonation

Intonation involves variation in pitch over stretches of speech usually larger
than the word (except, evidently, in cases where the word and the phonological
phrase are co-extensive).7 Since no utterance can take place without the simul-
taneous realization of both a segmental component (consonants and vowels)
and an intonational contour, the central role (or roles) of intonation in com-
municative acts should be clear. These roles may include the delimitation of
grammatical structure (phrase or clause boundaries), the distinguishing of sen-
tence types (assertions, questions, etc.), the highlighting of new information
(topic/comment distinctions), or the communication of expressive information
reecting the speakers attitude or state of mind (agreement, doubt, anger, sar-
casm, surprise, and a host of other possibilities). Although there exist very
widespread cross-linguistic tendencies linking specic intonational patterns
with specic grammatical functions (e.g., rising intonation in yes-no ques-
tions, falling intonation for declarative sentences), the amount of dialectal or
individual variation in intonation contours makes the topic dauntingly com-
plex. In what follows, we will outline only the rudiments of French intona-
tional structures, focusing on grammatical and pragmatic rather than expres-
sive features.
This outline will take place using several descriptive parameters. First,
the domain of intonational contours will be described in tems of the phono-
logical phrases previously discussed (Chapter 3, Section 3.3).8 Secondly, as
the term contour itself implies, intonational patterns will be described using
Prosody 183
continuous lines representing relative pitch or pitch changes within phono-

logical phrases. Finally, these pitch changes occur within a range, and may
begin or end at different heights within the range. Schematically, this is dia-
grammed in (6), using the basic reference point for the discussion to follow: a
neutral declarative sentence.
(6) A simple intonation contour (one phrase)

high
range
mid
low
syllables m m m m m m
phrase || Michelle est arrive. ||
Michelle arrived.
After an initial presentation of stress, rhythm, and prosodic units, Di Cristo

(1998: 200-16) surveys the principal intonation patterns of SF in terms of
the following categories: a basic nonemphatic pattern (simple declarative
sentences), modality and expressivity (questions, imperatives and vocatives,
expressivity), focalization and contextual effects (focus and topic/comment
organization), and intonational clichs. We list the principal contours in (7)
and then exemplify each of them in turn.
(7) Principal SF intonational contours

(a) basic non-emphatic contour
(b) questions
(c) imperative and vocative
(d) expressivity
(e) focalization
(f) topic/comment
(g) intonation clichs
The basic nonemphatic contour is often described as a contour tone with a

rising falling pattern; in shorter phrases (as in elliptical answers to questions),
the rising component may be minimal.
184 Chapter 6
cest un ami
Hes a friend.
je lui en ai parl
I spoke to him about it.
sans exception
without exception
voil
there
Contour patterns are particularly evident in long utterances consisting of more

than one phonological phrase, where the end of the rise corresponds to the end
of the rst phrase and indicates noncompletion of the utterance.
je lui ai parl || de ses devoirs

I spoke to him about his homework.
ils sont tous venus || sans exception
They all came, without exception.
Questions may be classied into several types, the most frequent being yes/no
questions and partial or wh-questions. Syntactically, yes/no questions can be
marked either by inversion (Jean vient-il? Is John coming?) or by the use
of the phrase est-ce que (Est-ce que Jeanne vient? Is Jean coming?). More
frequent in the current spoken language, however, are questions with no syn-
tactic marking and solely a characteristic nal rising intonation pattern (Jean
vient?).9
ta soeur a-t-elle termin ses tudes?

Has your sister nished her studies?
veux-tu du vin?
Do you want some wine?
est-ce que le patron est libre?
Is the manager free?
vous partez demain?
Are you leaving tomorrow?
Prosody 185
Di Cristo notes a striking parallelism between interrogative and continuative

intonation, the former occurring in utterance-nal position, the latter in non-
nal position. As a result, he nds reasonable the generalization that there is a
single overall intonation pattern for both continuative and interrogative utter-
ances in French (202), both being indicated by a phrase-nal rise in intona-
tion, the difference residing in the nal or prenal position of the phrase in
question.
The second major type of question, the so-called wh-question, involves
interrogative morphemes such as qui who, o where, quand when, com-
ment how, pourquoi why.10 In these structures, the interrogative morpheme
normally represents the high point of the intonation countour, which then falls
to the end of the sentence. As a result, these questions manifest a pattern simi-
lar to the basic contour of simple declaratives.
qui est-l?
Whos there?
o tes-vous alls?
Where did you go?
comment peut-on y accder?
How can we get there?
pourquoi le dit-il ainsi?
Why does he say it like that?
lequel prfrez-vous?
Which one do you prefer?
Finally, to be complete, let us mention that two further question types occur:
alternative questions (Partez-vous demain ou aprs demain? Are you leav-
ing tomorrow or the day after?) and elliptical questions (Et ta lle? And your
daughter?; Quoi ma voiture? What, my car?). Each type uses the rising-
falling or falling contours with which we are already familiar.
In the imperative and vocative domain, imperative sentences are said to be
characterized by a rapid drop in pitch from high to low (Delattre 1966b, Lon
1974). While this general pattern is widely accepted, analysts disagree about
certain details, such as whether this pattern is exclusive to imperatives. Given
the admixture of expressive possibilities with imperative structures, this varia-
tion is not surprising. Vocatives, in turn (the last two examples), show a rising-
falling pattern, but with a rise and a fall on the last syllable, including a high
starting point and consequent large fall at phrase end. Examples follow:
186 Chapter 6
venez demain Come tomorrow.

ne faites pas a Dont do that.
donne-le moi Give it to me.
Anne-Marie Anne-Marie
excusez-moi, monsieur Excuse me, sir.
Expressivity is no doubt one of the most complicated domains of phonetic

study. It combines intonation, nonverbal communication, and a range of addi-
tional linguistic factors (pitch range, register, complex pitch movement, loud-
ness, tempo, and voice quality; cf. Di Cristo [209]) to communicate a variety
of attitudes. One very frequent pattern involves a large rise-fall movement on
the last syllable of the utterance coupled with deaccentuation of the preceding
stressed syllables. This pattern, however, may convey subtly different mean-
ings depending on the associated syntactic structures: exasperation with partial
questions, a polite invitation with imperatives, and so on, so it is difcult to
isolate specic independent expressive features (Di Cristo 208). Lon (1971,
1993) and van den Berghe (1976) are basic references in this area.
Focalization, involving extra pitch prominence, is a process used to inten-
sify or contrast certain words and is closely related to the accent dinsistance
discussed earlier. Examples, in addition to those from 6.1.2 above, include the
following (Di Cristo 209 10):
il ne dit jamais rien

He never says anything.
qui va te rencontrer
Whos meeting you?
le professeur a la clpas ltudiant
The professor has the key, not the student.
With respect to topic/comment, speakers often, in context, choose to indicate

a distinction between information presupposed to be known by the listener
(the topic) and that which is new (the comment). Normally, this difference is
signalled both by syntactic means (dislocation, resumptive pronouns, etc.) and
by intonational contours, as in the following, where voisin neighbour repre-
sents the topic and toujours malade always sick the comment, the latter
characterized by regular contours falling for statements, rising for yes-no
questions:
mon voisin, il est toujours malade
My neighbour, hes always sick.
Prosody 187
mon voisin? il est toujours malade

My neighbour? Hes always sick.
mon voisin, le connais-tu?
My neighbour, do you know him.
Further discussion of these types of pattern occurs in Di Cristo (210 14).

Clichs in general are xed, frequent, and stereotypical expressions. In the
prosodic domain, intonation clichs11 are frequent in certain situations, highly
melodic, and consequently often compared to singing. A typical example is the
la-la-lre or na-na-nre nyah, nyah expression used by children to make fun
of someone. Other examples include hop l! off you go/oops with a high-
low pitch pattern; Tu as perdu! You lost! or Elle est partie! shes gone
with a rise on the penultimate and a sustained mid-level tone on the last syl-
lable; or Bonjour! hello or coucou! peek-a-boo with high-mid tonal patterns
announcing someones arrival. In an interesting article, Fnagy et al. (1983)
survey a wide variety of intonation clichs in SF, linking them to a variety of
expressive functions as well.
This concludes the survey of SF intonation patterns, a survey that cannot do
justice to the richness and complexity of the domain. As the examples make
clear, syntactic and discourse function, expressivity, variable syntactic struc-
tures (some more characteristic of spoken and occasionally familiar language),
stress, and rhythmic patterns all combine (or collide) to render a discussion of
intonation more than challenging. To close this chapter, we will present a few
additional examples from colloquial speech, examples that illustrate some of
the syntactic variability that is possible in that register, and that have intona-
tional consequences.
6.3 6.3 Colloquial Constructions

The structures to follow, drawn from le franais informel are most appropri-
ately restricted to certain less formal contexts and may give some indication
of the direction in which French syntax will evolve. A brief, nontechnical
description of the innovation precedes each example set.12
(8) Sample colloquial constructions

(a) interrogative morpheme at end rather than beginning of sentence
tu las vu o? You saw it where?
vous partez quand? Youre leaving when?
tu las pay combien? You paid how much?
188 Chapter 6
(b) lack of inversion after interrogative morpheme

o tu vas? Where are you going?
de quoi tu parles? What are you talking about?
quoi a sert? Whats that for?
(c) suppression of pronominal subject (often with falloir)
faut pas faire a Dont do that.
faudrait lui tlphoner You should call her.
a, pas question tu sais Dont even think about it.
Rue Mouffetard?, connais pas Rue Mouffetard, I dont know it.
(d) nominal subject or object accompanied by resumptive pronoun
son frre il est malin His brother, hes tricky.
ses voisins ils sont pas gentils His neighbours arent nice.
les artichauts je les aime pas Artichokes, I dont like them.
il la rat son examen He failed his exam.
enn, il la eu, son bac? So did he get it, his degree?
(e) pronouns repeated with corresponding disjunctive forms
tu veux y aller, toi? Do you want to go there?
jai lair dun ic, moi? Do I look like a cop, me?
toi, tu me fatigues Youre bothering me.
moi, lui, je laime pas Him I dont like.
(f) prepositions stranded in nal position
combien tas pay pour? You paid how much?
on peut se laver sans We can wash without it.
tu viens avec? You coming with us?
la lle que je sors avec the girl Im dating
These examples give only the briefest indication of a very rich set of morpho-
logical and syntactic innovations in contemporary spoken French.13 Further
data and discussion, indicating the frequency and extent of these changes, may
be found in Bauche (1920), Frei (1929), Gadet (1989), Guiraud (1965), or
Mougeon (1998). Rather than dwell on an area that has at best only indirect
implications for our current phonological discussions, let us turn, in the next
chapter, to certain phonological phenomena that occur at what one might call
the periphery of the phonological system.
Prosody 189
Notes
1. For studies of SF addressing these topics, see Lucci (1983) or Lacheret-Dujour
and Beaugendre (1999). There are additional, theory-specic uses of the term
prosody in Firthian Prosodic Analysis, or in approaches known as Prosodic
Phonology and Morphology, which need not concern us here.
2. For the latter, Lon (1993) represents an excellent introduction.
3. As was observed, for example, by Delattre (1938).
4. This distinction is an approximation only. For some discussion of the complexi-

ties involved, see Wenk and Wioland (1982).
5. To be complete, we should note that there is by now an extensive literature deal-

ing with nonnal or secondary stress in French, seen in terms of rhythmic pat-
terning. This topic is beyond the bounds of our current discussion. For further
information, consult (in addition to Di Cristo [1998: 19699, 1999, 2000]),
Dell (1984), Fnagy (1979), or Lacheret-Dujour and Beaugendre (1999).
The existence of these phrase-internal rhythmic patterns does not lead syn-
chronically to the reduction of rhythmically weak vowels. It does mean,
however, that the approximate equality attributed above to all pretonic
syllables is an idealization for expository and pedagogical purposes.
The terms emphatic and phrasal stress are used in the sense of Di Cristo
(1988). Lucci (1983), who presents one of the most substantial studies of emphatic
stress in SF, also notes the terminological richness in this area. Phrasal stress is
also known as laccent dintensit, dynamique, tonique, normal, nal, among
others. Emphatic stress is variously called laccent dinsistance, intellectuel, dis-
tinctif, oratoire, affectif, motionnel (with certain authors e.g., Marouzeau
[1924] distinguishing, within emphatic stress, between affectif and intellec-
tuel). Lucci uses the term accent didactique for what we are calling emphatic
stress. See also Sguinot (1977).
6. Because the reading of the words lists is also a formal exercise, the speakers heard
on the accompanying CD-ROM also manifest a tendency to accent the initial syl-
lable of the words being read.
7. Pitch, in turn, is the perceptual correlate of the frequency of a sound, such fre-
quency determined in speech by the rate of vibration of the vocal cords: the more
rapid the rate of vibration, the higher the pitch. Languages in which pitch varia-
tion occurs within words rather than over longer sequences, where such changes
produce different meanings when associated with identical sequences of conso-
nants and vowels, are known as tone languages. In simplied terms, therefore,
tone refers to pitch changes within words while intonation refers to pitch
changes within phrases and longer sequences. As noted by Trask (1996: 184),
studies of intonation may additionally involve factors such as tempo, loudness,
190 Chapter 6
and pauses; see Fougeron and Jun (1998) for an analysis of the effect of tempo on
French intonation.
8. It is possible to distinguish intonational phrases from phonological phrases, as do

Nespor and Vogel (1986, chapter 7). This distinction is not necessary for our pur-
poses here. The discussion to follow relies heavily on Di Cristo (1998: 200 - 17).
For much more detailed information, see Rossi (1999).
9. There is a second type of yes/no question, also frequent in spoken language, called
tag questions. This type ends in markers such as oui, non, si, hein, daccord,
nest-ce pas, and shows the normal declarative contour followed by a rapid rise on
the tag marker: Elle est malade, non? Shes sick, isnt she?; On part, daccord?
Lets go, okay?
10. Actually, as Di Cristo describes (1998: 205), there are two main types of partial
questions: the neutral partial questions illustrated below and echo partial ques-
tions, in which the speaker asks for a repetition
. or reformulation of a response not
fully understood. Echo questions show a contour similar to that of yes/no ques-
tions: Qui est tomb? = Who did you say fell?
11. Also called clichs mlodiques; cf. Fnagy et al. (1983).
12. Note, as a general rule, that the negative particle ne is virtually

. absent from cur-
rent spoken French, except in the most formal registers.
13. Lexical choice is also obviously correlated with stylistic level consider a series
such as vhicule, voiture, bagnole, char, vehicle, automobile, car, jalopy for
example. A separate work would be needed to consider this matter fully.
191
Chapter 7
Around the Phonological Periphery:
Playing with Language.

Virtually all languages allow manipulation of their phonological structure
for what we might call recreational or expressive purposes. Punning, ono-
matopoeia, nicknames, and diminutives provide four widespread and familiar
examples of the type of activity involved and indicate further the primary
sociolinguistic connotations of such phonological manipulations. Word play
is, at least at the outset, more characteristic of casual or relaxed situations
where the participants know each other well than it is of formal contexts. This
being said, however, one of the intriguing aspects of French word play is the
degree to which acronyms or the use of verlan, to take the two most obvious
examples, have extended into standard usage. In the following sections, we
will illustrate four types of processes with which all speakers of French are
familiar and which have been widely discussed in both the technical and the
popular literature. While each of these processes involves obvious phonologi-
cal manipulations, we should also note that they may be considered, along
with derivation and compounding, as word-formation mechanisms (albeit of a
minor kind) that serve to expand the contents of the lexicon. 1
7.1 7.1 Abbreviations

Abbreviation, also often called clipping, is a word-formation process by means
of which a new item is formed by the removal of part of a larger word or
phrase. The new formation has the same meaning and grammatical category
membership as its parent. A variety of straightforward abbreviations is given
in (1).2
(1) Simple abbreviations

Abbreviation Full form
abrv abrviation abbreviation
agreg agrgation competitive degree
appart appartement apartment
192 Chapter 7
alu aluminium aluminum

bnf bnce benet
calva calvados Calvados
dg dgueulasse disgusting
dgueu dgueulasse disgusting
diapo diapositive slide
estome estomac stomach
fac facult faculty
fana fanatique fan
impec impeccable perfect
imper impermable raincoat
Lib Libration Libration newspaper
manif manifestation demonstration
maths mathmatiques math
perpte perptuit life sentence
prof professeur professor
pub publicit commercial
rcr rcration recess
scu scurit (sociale) Social Security
sensa sensationnel terric
sensass sensationnel terric
sympa sympathique nice
tiche t-shirt t-shirt
Simple in these cases refers to the fact that the abbreviations are identical
to part of the original forms and show no further phonological changes (other
than the regular adjustments of the mid vowels). The examples here all trun-
cate the nal part of the word, and this appears to be the most frequent pat-
tern, despite the existence of forms like ricain for amricain American, pit-
aine for capitaine captain, or bib for toubib doctor. Otherwise, we nd
some variation (e.g. sensa and sensass) and both monosyllabic and polysyl-
labic examples, as well as both vowel and consonant nal forms.3
More complicated types of abbreviation also exist, of which we will
illustrate only two: abbreviation of phrases or compound words and the
Around the Phonological Periphery: Playing with Language 193
generalization of the sufx -o, both illustrated in (2). Note that -o in some
forms may not actually be a sufx, but rather part of the word itself. These
forms are included here because of their general similarity to sufxed forms in
form and in impression conveyed.
(2) Complex abbreviations

Abbreviation Full form
(a) phrases
ctap cet aprs-midi until this afternoon
plute plus tard see you later
aprm aprs-midi afternoon
tout tout lheure see you shortly
beau-f beau-frre brother-in-law
cit-u cit universitaire student residences
comif comme il faut properly
dacc daccord okay
deuxch Deux Chevaux (type of car)
doccase doccasion second hand
sans dec sans dconner no kidding
sciences po sciences politiques political science
tag ta gueule shut up
Boul Mich Boulevard Saint Michel (street)
Nouvel Obs Nouvel Observateur (magazine)
(b) sufxes
ado adolescent adolescent
amerlo amerloque (amricain) American
anarcho anarchiste anarchist
apro apritif aperitif
aristo aristocrate aristocrat
bolcho bolchviste Bolshevist
collabo collaborateur collaborator
dbilos dbile moron
194 Chapter 7
dmago dmagogue demagogue

diapo diapositive slide
colo cologiste ecologist
gaucho gauchiste leftist
hebdo hebdomadaire weekly paper
hystro hystrique hysterical
intello intellectuel intellectual
interro interrogation interrogation
labo laboratoire laboratory
Macdo McDonald McDonalds restaurant
mto mtorologie weather
mollo mollement take it easy
Montparno Montparnasse Montparnasse quarter
moto motocyclette motorcycle
parano paranoaque paranoid
rapido rapidement quickly
resto restaurant restaurant
socio sociologie sociology
toxico toxicomane junkie
In many of these forms (e.g., aristo, labo, toxico), the nal /o/ can be simply
the segment preceding the truncated part of the abbreviation. In others, how-
ever, there is no trace of the /o/ in the original, and the vowel must be inter-
preted as a sufxal addition to the base word (apro, intello, rapido, and so
on). Many of these creations have a jocular or sarcastic tone, as indicated in
the observation of Prigniel (1966: 63): Il [le sufx -o] a une allure bon enfant,
il est familier et dsinvolte, vulgaire, ironique. Sa bonne humeur, souvent,
donne une nuance enjoue aux propos.4 The formation of words in -o is
very productive (cf. Macdo), a productivity reinforced by the ability of -o
to combine with other segments in an expanded series: classicos (classique),
dbilos (dbile), vulgos (vulgaire), cinoche (cinma), floche (flicitations),
and others.
7.2 7.2 Acronyms

Acronyms, not unrelated to abbreviations, are words formed by combining
into a new word the initial letters of the constituents of a phrase: UFO
(unidentied ying object); AIDS (acquired immune deciency syndrome).
Sometimes the acronym is spelled out (the letters are read individually), as in
UFO; sometimes the combination of letters is pronounced or read as a new
word (AIDS). There are even mixed forms, such as CD-ROM (compact disk,
read-only memory). It takes very little contact with contemporary France to
be convinced that acronyms (acronymes or sigles in French) now play a sig-
nicant communicative role, reect a very large set of mutually understood
cultural concepts, and are freely used at all levels of communication. (It is dif-
cult to read the current press without a grasp of a large number of acronyms.)
The examples in (3) are a tiny proportion of those that may be found in SF. 5
(3) Acronyms
(a) spelled acronyms
AJ Auberges de jeunesse youth hostels
ANPE Agence nationale pour national employment
lemploi agency
BCBG bon chic, bon genre preppy
BD bande dessine comic strip
BNP Banque nationale de Paris (name of bank)
CCP compte chque postal post ofce bank account
CEE Communaut conomique EEC
europenne
CES collge denseignement junior high school
secondaire
CGT Confdration gnrale (trade union)
du travail
CNPF Conseil national du (employers federation)
patronat franais
CNRS Centre national de recherche (research council)
scientique
CPFH collier de perles, snooty
foulard Herms
CRS Compagnies rpublicaines state security police
de scurit
196 Chapter 7
DST Direction de la surveillance (comparable to CIA)

du territoire
EDF Electricit de France (national electric company)
GDB gueule de bois hangover
HLM habitation loyer modr public housing
IVG interruption voluntaire abortion
de grossesse
JO Jeux olympiques Olympic games
MLF Mouvement de libration Womens Lib
des femmes
ORTF Ofce de la radiodiffusion (French national
et tlvision franaises broadcasting service)
PCI Parti communiste International Communist
international Party
PDG prsident-directeur gnral chairman and CEO
PME petites et moyennes small and medium
entreprises business
PNB produit national brut gross national product
PPH passera pas lhiver on deaths door
PS Parti socialiste Socialist Party
PTT Postes, tlgraphe, (national post ofce and
tlphone telephone service)
PV procs-verbal parking/speeding ticket
RAS rien signaler nothing new
RATP Rgie autonome des (Paris city transport
transports parisiens authority)
RER Rseau express rgional (high speed train service in
Paris region)
RPR Rassemblement pour la (political party)
Rpublique
SDF sans domicile xe homeless person
SNCF Socit nationale des (French national railway
chemins de fer franais company)
TGV train grande vitesse high speed train
TP travaux pratiques lab work
TVA taxe la valeur ajoute value added tax

UDF Union pour la (political party)
dmocratie franaise
VO version originale original version
VTT vlo tout terrain mountain bike
(b) read acronyms
CAPES Certicat daptitude au high school teachers
professorat denseignement diploma
secondaire
CHRU Centre hospitalier rgional regional university
universitaire hospital centre
DEUG diplme dtudes diploma received after two
universitaires gnrales years at university
DOMTOM Dpartements dOutre-Mer French overseas
et Territoires dOutre-Mer departments and
territories
ENA Ecole nationale college for senior civil
dadministration servants
FNAC Fdration nationale (national chain of stores)
dachat pour les cadres
INSEE Institut national de la national institute of
statistique et des economic and statistical
tudes conomiques information
ONU Organisation des UN
Nations Unies
OPEP Organisation des pays OPEC
exportateurs de ptrole
OTAN Organisation du Trait de NATO
lAtlantique du Nord
OVNI objet volant non-identi UFO
RIB relev didentit bancaire particulars of ones bank
account
SAMU Service daide mdicale mobile emergency
durgenceservice medical service
SDECE Service de documentation (comparable to CIA)
extrieure et de
contre-espionnage
198 Chapter 7
SIDA syndrome immuno- AIDS

dcitaire acquis
SMIC salaire minimum minimum wage
interprofessionnel
de croissance
TUC travail dutilit collective community work/
employment training
ZUP zone durbanisation urban development zone
prioritaire
The productive role of acronym formation is further indicated in two ways: by

the possibility that their innovative pronunciation violates regular phonotac-
tic constraints (CHRU [5y], SDECE [sd's], FNAC [fnak] are phonotactically
irregular) and by the appearance of acronyms as the base for new derivational
formations, as exemplied in (4).
(4) Acronyms in derived forms

ajisme, ajiste youth hostelling, hosteller
bcbgisme preppiness
bdphile comic strip fan
capsien student preparing the CAPES
cgtiste member of the CGT
narque student or former student of the ENA
hachelmiser to develop public housing
onusien UN ofcial
otanesque NATO-like
siden, sidatique with AIDS, AIDS sufferer
smicard minimum wage earner
tucard, tuciste community worker, employment trainee
ufologie ufology
vtteux mountain biker
zupage, zuper rapid urbanization
Initially characteristic of the spoken language and of certain limited domains

(names of countries, political parties, businesses, etc., as in USA, URSS
USSR, PC, PTT, SNCF), use of acronyms has exploded since the Second
World War. Now, virtually every semantic domain is replete with its own
siglaisons, and the frequency of many forms has led to their replacing the
original phrases (e.g., le smic). Rather than being limited to the shortening of
the names of organizations or ofcial structures, acronyms based on slang or
common phrases (TP, VO, BCBG, CPFH) now demonstrate the continually
expanding importance of this means of lexical enrichment.
7.3 7.3 Reduplication

Reduplication is a process of word formation whereby all or part of a word
is repeated for grammatical or semantic purposes. Certain languages use redu-
plication productively in their inectional systems (e.g., for plural formation).
Others, like English, make limited use of it for expressive purposes (really,
really big; teeny weeny problem; etc.). Reduplication can be total (where a
whole word is repeated) or partial (where only a part of the word is involved).
The process has interesting theoretical implications, particularly involving
syllable structure, regarding the identication of that part of the word to be
repeated and the places where the repeated parts may appear. The semantic
effects of the phenomenon are also intriguing. While reduplication need not
be considered a major aspect of French morphology, it does occur widely and
productively, as demonstrated convincingly by Morin (1972).6
Morin considers reduplication to have two primary functions in French: the
production of diminutives (chouchou favourite) and of onomatopoeic words
(ron-ron cats purring). The use of such forms occurs primarily in informal
situations, where the speakers know each other well, or in child language.
Morin also established a basic phonological description for French redupli-
cation. In partially reduplicated forms, the rst or reduplicated syllable is a
partial copy of the second or stem syllable, which must itself begin with a con-
sonant.7 Examples of reduplicated forms and their proposed bases are given in
(5); note that a further characteristic of onomatopoeic forms is that they have
no obvious base.
(5) Reduplication in French

(a) partial base form
bbte bte silly
coco communiste, communist
coco cocane cocaine
cracra crasseux dirty
dodo dormir sleep
fanfan enfant child
fofolle folle mad, foolish
200 Chapter 7
jojo joli pretty

joujou jouet toy
mm grand-mre grandma
mmre grand-pre grandpa
mimi minet cat
nounours un ours teddy bear
zonzon maison, prison house, prison
(b) full base form
blabla chatter, babble (cf. blablater)
bobo hurt
bonbon bon candy
chaud chaud chaud hot
coin-coin sound of duck
durdur dur quite difcult
foufou fou mad, foolish
gaigai gai happy
glouglou sound of running liquid or
of turkey (cf. glouglouter)
menumenu menu tiny
miamiam yum yum
ronron purr (cf. ronronner)
tata tante aunt
tonton tata+oncle uncle
vroumvroum sound of engine
Fully in line with Morins proposals, these examples should sufce to convey
both the informal nature and general semantic character of reduplication in
French.8
7.4 7.4 Word Games: Verlan

Language games, secret languages, or disguised speech, such as the Pig Latin
familiar to many English speakers, are very widespread throughout the worlds
speech communities. Such phenomena give rise to a number of interesting

theoretical and descriptive questions involving both the phonological struc-
tures resulting from the formal manipulations and the conditions of their use.9
Structurally, the processes normally involve the movement of a sound or syl-
lable from one point in the word to another and/or the insertion of sounds at
a speciable point, as well as potential further adjustments depending on the
nature of the sequences resulting from these operations. Use of such devices
is sociolinguistically complex, not unrelated to the functions of slang, profes-
sional jargon, or other indications of social class membership.
In French, the disguised language currently garnering the most attention
is verlan (French spoken lenvers).10 Linked to the vernacular of the north-
ern suburbs of Paris and initially used as a symbol of group membership by
younger and marginalized speakers of North African immigrant communi-
ties,11 verlan has now become increasingly integrated into the linguistic main-
stream. Verlan forms appear as the title of lms (Les Ripoux [= Les Pourris,
crooked policemen], in songs (Laisse bton [= laisse tomber drop it] by
Renaud Schan) and les bds [= bandes dessines comics], in publicity (Le
rap est dans la cepla [= place place], in a Virgin Records advertisement), in
book titles (Seguin, Boris: Les Cfrans [= les Franais the French] parlent
au Franais, Seuil) and even in the speech of the President of the Republic
(Mittrands famous Mais on ne dit plus chbran [= branch with it] on dit
cbl ou mme blca [= cbl with it], in response to an interviewer). The
co-opting of such expressions by mainstream media also reects a normal
sociolinguistic trajectory and contributes to the constant renewal of slang:
once the forms become too well known, they no longer serve their original
purpose of in-group identication and are replaced.
How, then, is verlan spoken? The name itself gives a good initial indica-
tion: in bisyllabic words, the order of the syllables is reversed (lenvers wrong
way around > verlan, caf coffee > fca, mtro subway > trom, bonjour
hello > jourbon, paquet dose of cocaine > kpa, etc.). Needless to say, how-
ever, the situation is more complicated. Inversion may also apply to longer
words, but differentially. The syllable order 1-2-3 may change to 3-2-1, as in
portugais Portuguese > gutupor, cigarette > rettegaci, or to 1-3-2, as in
enculer to sodomize > enlcu, couter to listen to > tcou.12 Monosyllabic
words may simply have the order of sounds reversed (fou crazy > ouf, chaud
hot > auche). More frequently, though, such words, especially when they are
consonant-nal (whether or not they end in mute-e in the orthography), are
treated is if they were pronounced with nal schwa (realized as [] or []).
Inversion then takes place, with any residual word-nal vowel deleted: mec
guy > keum (from /m'k/), femme woman > meuf (from /fam/). Finally,
once inversion has taken place, further truncation of syllables can occur, ren-
dering words still more difcult to recognize: couter > tcou > tcou; jobard
looney > barjo > barge; Arabe > Beura > Beur; bidon phoney > dombi
/dnbi/ > dombe /dnb/; etc. The following set of examples illustrates these pro-
cesses more completely. (The social domains where this type of vocabulary is
primarily used will be evident from the translations.)
202 Chapter 7
(6) Verlan
(a) vowel-nal monosyllables
bien > iemb well
bon > ombe good
a > a this
chier > iche to be very bored
feu > euf a light (for a cigarette)
moi > wam me
nez > zen nose
(b) consonant-nal monosyllables (often with addition of nal < e >
[= [] or [], subsequently spelled < eu >] to provide the new base
vowel), plus optional truncation of the nal vowel.
bac > keuba baccalaureate degree
Beur > Rebeu < Arabe (a reverlanized verlan form)
cher > reuch expensive
cul > luc ass
femme > meuf woman, girl
fesse > seffe buttock
fte > teuf party
ic > keuf cop
juif > feuj Jew, Jewish
mec > keum guy
mre > reum mother
punk > keupon punk, punk rocker
skin > neusk skinhead
soeur > reuss sister
thunes > nuts dough
tronche > chetron face (pejorative)
(c) bisyllables
poil > oilp, oilp naked
arnaque > karna swindle
bagnole > gnolba car

baskets > sketba sneakers
bizarre > zarbi weird (also > zarb)
blouson > zomblou jacket
bouffe > fbou food
braquer > kbra to stick up
cam > mca stoned
choper > pcho to catch
clochard > charclo hobo
drogu > gudro drug addict
gonzesse > zessegon chick
hachisch > chicha hashish
Paris > Ripa Paris
pdale > dallep homosexual
pinard > narpi wine
ptard > tarp joint
salope > lopsa bitch
voler > lvo to swipe
voyou > youvoi hooligan
(d) longer forms
dfonc > foncd drunk, drugged
dgueulasse > lasdgueu dirty, repulsive
fatigu > gutifa tired
frangine > ngifran sister
ma parole > marolepe you dont say!
oublier > bliou forget
partouze > zetoupar orgy
rigoler > golerie to kid around
(e) truncated forms
bouffer > fbou > feb eat
bouger > gbou > geb move
204 Chapter 7
braquer > kbra > keb steal

cond > dcon > dec cop
copain > painco > painc pal
dguis > kisd plainclothes policeman
mongol(ien) > golmon > gol stupid, moron
(for)niquer > kni > ken to copulate
parents > rempas > remps parents
pd > dp > dep homosexual
ticket > keuti > keut ticket
A closer look at certain complex forms reveals additional details that dem-
onstrate the relevance of language games in confronting a number of ques-
tions of phonological theory, such as the behaviour of certain latent consonants
(the appearance of /z/ in nez nose > zen), of consonant clusters (/bl/ remain
together in bliou < oublier to forget, while /rt/ are separated in zetoupar
< partouze orgy), of nasal vowels (no latent nasal consonant appears: bon
good > ombe, not *nombe), of schwa (schwa is identied with one of the
vowels // or //, and accordingly written with < eu >), and so on. While these
issues are very interesting, they take us beyond our current descriptive con-
cerns. Much further detail concerning the theoretical implications of verlan,
as well as large amounts of additional data, is available in the references of
footnote 9.
This concludes our discussion of language play. The phenomena consid-
ered here have counterparts in most languages. They reect, in a further way,
the adaptation of language to its context of use, and implement in addition a
very widespread tendency in informal speech: the shortening of words by vari-
ous means. More generally, language play shows yet again how phonological
processes are intimately involved in word formation in relating the varying
word forms of a single lexical item. This theme, a reection of the discussions
throughout previous chapters, underlines the close connections between pho-
nology and morphology and reinforces the conclusion that one cannot study
French sounds in isolation but only as part of an integrated set of linguistic
systems, systems used in a variety of speech situations. Hence the title with
which we began: French Sound Structure.
Notes
1. See Valdman (2000b) for an excellent survey of this type of speech behaviour.
2. This data is taken from Scullen (1997), the most comprehensive general analysis
of the various phenomena involved in this chapter. Scullen worked with a corpus
of approximately one thousand abbreviated items from SF.
3. The ease with which consonant-nal abbreviations are formed contradicts the
claim that French is a strongly open-syllable language.
4. For many additional examples and further analysis, see Offord (1989).
5. The terminology spelled and read acronyms is that of Scullen (1997).

Dictionaries of acronyms now run to hundreds of pages.
6. In Morins approach, reduplicated forms are called echo words. For further dis-
cussion, in addition to Scullen (1997), see Thiele (1987: 107 08).
7. Scullen (1997: 135) recognizes this description as largely correct. Subsequent

analyses reect theoretical developments in phonology since Morins ground-
breaking work.
8. We cannot discuss here one additional complicated and very productive area
of reduplication (and, in fact, of abbreviation as well): the formation of hypocoris-
tics or nicknames. Scullen (1997: 148 70) divides nicknames into four classes:
reduplicated (Kiki < Christophe, Bbert < Albert, Coco < Colette, Vvette <
Yvette, Popol < Jean-Paul), abbreviated (Dom < Dominique, Isa < Isabelle, Gus <
Auguste), compounded (Mah < Marie-Hlene, Marie-T < Marie-Thrse), and
sufxed (Thomassou < Thomas, Alexo < Alexandre, Toinon < Antoine). Variants
of many nicknames abound, often combining different abbreviatory or reduplica-
tive processes. Further discussion these processes may be found in the work of
Marc Plnat (e.g., Plnat 1984).
9. For an insightful review of this question in the French context, see Lefkowitz
(1991), especially chapters 2 and 3, as well as Antoine (1998). Villon, in his
Ballades en jargon from the mid-1400s, presents one early and well-known
example of a secret language in the French literary domain. Further information
may be found in Azra and Cheneau (1994), Plnat (1995), Scullen (1997), or vari-
ous recent issues of Le Nouvel Observateur (e.g., No. 1771 of October 15, 1998).
10. Other possibilities, far less actively used at the moment, include javanais (formed
by inserting -av-, -va-, or -ag-: chatte cat > chagatte, jeudi Thursday > javeu-
davi), largonji (formed by replacing the rst letter of the word by l and moving
the rst letter to the end of the word and adding a vowel: jargon > largonji, caf
> lafk), or loucherbem, a creation of Paris butchers (formed by replacing the
206 Chapter 7
rst consonant or consonant cluster by l, moving the initial segments to word-

nal position and adding -em: boucher butcher > loucherbem, truc thing, trick
> louctrm). For further details, cf. Plnat (1985, 1991).
11. The verlan form Beur (= Arab) is widely used as a designation for second genera-
tion North African immigrants, and one can now listen to Radio Beur.
12. In a number of polysyllabic verlan forms, the order of syllables actually appears
random: cigarette > rettegaci, garettci, or retciga; dfonc smashed > foncde,
fondc, or cfond; etc.
207
AppendixAppendix
Text: Le Premier Ministre ira-t-il Beaulieu

Diagnostic Word List: Standard French
Diagnostic Word List: Canadian French
Speaker 1: Georges Blary, Pontoise, Val-d'Oise
Speaker 2: Jean-Bernard Gauthier, Oucques, Loir-et-Cher
Speaker 3: Sverine Lamontre, Chaumont, Haute-Marne
Speaker 4: Etienne Grang, Pau, Pyrnes-Atlantiques
Speaker 5: Fleur Larocque, Montral, Qubec
The Appendix contains the text of recordings of the following passage and
word lists, read by ve speakers and included on the CD-ROM. (The speaker
from Montral reads the additional list designed to illustrate certain phonolog-
ical properties of the French language as spoken in Canada.) The purpose of
these recordings is to give examples of connected speech and to illustrate three
different varieties of French, a standard northern accent, a southern accent and
a Canadian accent.
Northern speakers:
Georges Blary, Pontoise, Val-d'Oise
Jean-Bernard Gauthier, Oucques, Loir-et-Cher
Sverine Lamontre, Chaumont, Haute-Marne
Southern speaker:
Etienne Grang, Pau, Pyrnes-Atlantiques
Canadian speaker:
Fleur Larocque, Montral, Qubec
208 Appendix
Note: as has been made abundantly clear in the text itself, there is considerable
variation in the pronunciation of Standard French. This means that the speak-
ers reading the texts or lists of data will themselves manifest elements of this
variation, and that the examples read may not always correspond exactly to
the point being discussed in the text. This lack of correspondence primarily
affects vowel length, the mid vowels /e ', 1 ,o n/, the nasal vowels /' /, the
low vowels /a #/, the retention or deletion of e-muet, certain nal consonants,
and the use of an accent dinsistance (inital stress on words). It is, however, a
useful exercise for listeners to be made aware of these minor discrepancies.
Le Premier Ministre ira-t-il Beaulieu?

Le village de Beaulieu est en grand moi. Le Premier Ministre a, en effet,
dcid de faire tape dans cette commune au cours de sa tourne de la rgion
en n d'anne. Jusqu'ici les seuls titres de gloire de Beaulieu taient son vin
blanc sec, un champion local de course pied (Louis Garret), quatrime aux
jeux olympiques de Berlin en 1936, et plus rcemment son usine de ptes ital-
iennes. Qu'est-ce qui a donc valu Beaulieu ce grand honneur? Le hasard, tout
btement, car le Premier Ministre lass des circuits habituels qui tournaient
toujours autour des mmes villes veut dcouvrir ce qu'il appelle "la campagne
profonde".
Le maire de Beaulieu Claude Bousquet est en revanche trs inquiet. La
cote du Premier Ministre ne cesse de baisser depuis les lections. Comment,
en plus, viter les manifestations qui ont eu tendance se multiplier lors des
visites ofcielles ? La cote escarpe du Mont Saint Pierre qui mne au village
connat des barrages chaque fois que les opposants de tous les bords mani-
festent leur colre. D'un autre ct, chaque voyage du Premier Ministre, le
gouvernement prend contact avec la prfecture la plus proche et s'assure que
tout est fait pour protger le Premier Ministre. Or, un gros dtachement de
police comme on en a vu Jonquire et des vrications didentit risquent
de provoquer une explosion. Un jeune membre de l'opposition aurait dclar:
"Dans le coin, on est jaloux de notre libert. S'il faut montrer patte blanche
pour circuler, nous ne rpondons pas de la raction des gens du pays."
De plus, quelques articles parus dans La Dpche du Centre, L'Express,
Le Nouvel Observateur et Ouest-Libert indiqueraient que des activistes
des communes voisines prparent une journe chaude au Premier Ministre.
Quelques fanatiques auraient mme entam un jene prolong dans l'glise de
St Martinville.
Le sympathique maire de Beaulieu ne sait plus quel saint se vouer. Il
s'est, en dsespoir de cause, dcid crire au Premier Ministre pour vrier
si son village tait vraiment une tape ncessaire dans la tourne prvue.
Beaulieu prfre tre inconnue et tranquille plutt que de se trouver au centre
d'une bataille politique dont, par la tlvision, seraient tmoins des millions
dlecteurs.
Appendix 209
Diagnostic Word List: Standard French

1. roc 31. djeuner
2. rat 32. ex-femme
3. jeune 33. lige
4. mal 34. baignoire
5. ras 35. pcheur
6. fou lier 36. socialisme
7. des jeunets 37. relier
8. intact 38. aspect
9. nous prendrions 39. niais
10. ftard 40. pais
11. nice 41. des gents
12. pte 42. blond
13. piquet 43. creux
14. pe 44. reliure
15. compagnie 45. piqu
16. fte 46 malle
17. islamique 47. gnle
18. agneau 48. bouleverser
19. pcheur 49. million
20. mdecin 50. explosion
21. paume 51. inuence
22. infect 52. mle
23. dgeler 53. ex-mari
24. btement 54. pomme
25. pier 55. trier
26. millionnaire 56. chemise
27. brun 57. brin
28. scier 58. lierre
29. fter 59. blanc
30. mouette 60. petit
210 Appendix
61. jene 78. quatrime

62. rhinocros 79. muette
63. miette 80. piquais
64. slip 81. trouer
65. compagne 82. piquer
66. peuple 83. creuse
67. rauque 84. beaut
68. cinquime 85. patte
69. nier 86. pte
70. extraordinaire 87. pais
71. meurtre 88. pe
72. vous prendriez 89. jeune
73. bott 90. jene
74. patte 91. beaut
75. triller 92. bott
76. faites 93. brun
77. feutre 94. brin
Diagnostic Word List Canadian French

1. mettre 13. vire
2. matre 14. juge
3. coeurer 15. court
4. vque 16. courte
5. sable 17. boulevard
6. vite 18. ltrer
7. libre 19. abusif
8. juste 20. ministre
9. plume 21. pilule
10. couple 22. touriste
11. pitoune 23. cuisine
12. ville 24. cuisiner
Appendix 211
25. pur 56. toi

26. rouge 57. bois
27. neige 58. boivent
28. neutre 59. noir
29. chaude 60. boisson
30. pre 61. voyons
31. beurre 62. soire
32. port 63. doigt
33. part 64. avoir
34. pte 65. boire
35. crainte 66. froid
36. emprunte 67. crois
37. honte 68. poign
38. lente 69. bain
39. quiper 70. quinze
40. dput 71. un
41. couter 72. jungle
42. professuer 73. crayon
43. piscine 74. honte
44. malle 75. absent
45. mle 76. il vente
46. Jacques 77. reculer
47. cadenas 78. brouette
48. clater 79. fvrier
49. clat 80. tabernacle
50. il est l, l 81. coutume
51. voyage 82. rendu
52. voyager 83. dire
53. mauvais 84. dite
54. jamais 85. duel
55. parfaite 86. tube
212 Appendix
87. tuile 102. champagne

88. pas dide 103. dehors
89. vote immense 104. hler
90. le prtre 105. chercher
91. aveugle 106. arbre
92. convaincre 107. plutt
93. vinaigre 108. il en a
94. orchestre 109. je lai vu
95. anglicisme 110. cest-tu ttu
96. debout 111. cent piastres
97. pourrie 112. sur la table
98. ombre 113. dans la maison
99. pingle 114. je les ai vus
100. signe 115. sans les voir
101. enseigner
213
References
References
Further information relevant to the use of this book and to the study of French
phonology in general may be found on the following www site:
http://www.ucalgary.ca/~dcwalker/
Ager, Dennis. 1990. Sociolinguistics and contemporary French. Cambridge: Cambridge

University Press.
gren, John. 1973. Etude sur quelques liaisons facultatives dans le franais de
conversation radiophonique: Frquence et facteurs. Uppsala, Sweden: Acta
Universitatis Upsaliensis.
Antoine, Fabrice. 1998. Des mots et des oms. Verlan, troncation et recyclage formel
dans largot contemporain. Cahiers de lexicologie 72: 4170.
Arteaga, Deborah, and Julia Herschensohn. 1998. Diachronic perspectives on the
teaching of New French. The French Review 71.4: 60720.
Azra, Jean-Luc, and Vronique Cheneau. 1994. Jeux de langage et thorie pho-
nologique: Verlan et structure syllabique du franais. Journal of French Language
Studies 4.2: 14770.
Ball, Rodney. 1997. The French-speaking world: A practical introduction to sociolin-
guistic issues. London: Routledge.
Baraduc, Jean, Gabriel Bergounioux, Vronique Castellotti, Cline Dumont, and
Marie-Hlne Lansari. 1989. Le statut linguistique des voyelles moyennes.
Langage et socit 49: 524.
Batchelor, Ronald, and Malcolm Offord. 1993a. Using French: A guide to contempo-
rary usage. Cambridge: Cambridge University Press.
. 1993b. Using French synonyms. Cambridge: Cambridge University Press.
Battye, Adrian, and Marie-Anne Hintze. 1992. The French language today. London:
Routledge.
Bauche, Henri. 1920. Le langage populaire. Paris: Payot.
Bchade, Herv-D. 1992. Phontique et morphologie du franais moderne et contem-
porain. Paris: Presses Universitaires de France.
Blanche-Benveniste, Claire, and Andr Chervel. 1969. Lorthographe. Paris: Franois
Maspero.
214 References
Bonnemason, Jean. 1993. Histoire des langues de France. Mouans-Sartoux, France:

Publications de lEcole Moderne Franaise.
Borrell, Andr, and Michel Billires. 1989. Lvolution de la norme phontique en
franais contemporain. La linguistique 25.2: 4562.
Buben, Vladimir. 1935. Inuence de lorthographe sur la prononciation du franais
moderne. Bratislava: Faculty of Philosophy, University of Bratislava.
Bullock, Barbara. 1995. Prosodic constraints and morphological alignment in French.
Lingua 96: 95117.
Carton, Fernand. 1974. Introduction la phontique du franais. Paris: Bordas.
, Mario Rossi, Denis Autesserre, and Pierre Lon. 1983. Les accents des
Franais. Paris: Hachette.
Catach, Nina. 1995. Dictionnaire historique de lorthographe franaise. Paris:
Larousse.
Chen, Matthew. 1970. Vowel length variation as a function of the voicing of the con-
sonant environment. Phonetica 22: 12959.
Clark, John, and Colin Yallop. 1990. An introduction to phonetics and phonology.
Oxford: Blackwell.
Clements, G. N., and Elizabeth Hume. 1995. The internal organization of speech
sounds. In The handbook of phonological theory, ed. John Goldsmith, 245306.
Oxford: Blackwell.
Cornulier, Benot de. 1981. H-aspire et la syllabation: Expressions disjonctives. In
Phonology in the 80s, ed. Didier Goyvaerts, 183230. Ghent: Story-Scientia.
Coustenoble, Hlne, and Lilias Armstrong. 1934. Studies in French intonation.
Cambridge, England: Heffer.
Cruttenden, Alan. 1986. Intonation. Cambridge: Cambridge University Press.
Damourette, Jacques, and Edouard Pichon. 19111940. Des mots la pense: essai de
grammaire de la langue francaise. 7 vols. Paris: DArtrey.
Dauses, August. 1973. Etudes sur le instable dans le franais familier. Tbingen:
Niemeyer.
Davenport, Mike, and Stephen Hannahs. 1998. Introducing phonetics and phonology.
London: Edward Arnold.
Delattre, Pierre. 1938. Laccent nal en franais: accent dintensit, accent de hau-
teur, accent de dure. The French Review 12: 37. Reprinted in Delattre (1966a:
6568).
. 1940. Le mot est-il une entit phontique en franais? Le franais moderne
8.1: 4756. Reprinted in Delattre (1966a: 14149).
References 215
. 1966a. Studies in French and comparative phonetics. The Hague: Mouton.

. 1966b. Les dix intonations de base en franais. The French Review 40: 114.
Dell, Franois. 1973a, Les rgles et les sons: Introduction la phonologie gnrative.
Paris: Hermann.
. 1973b. e muet, ction graphique ou ralit linguistique? In A Festschrift for
Morris Halle, eds. Stephen Anderson and Paul Kiparsky, 26-50. New York: Holt,
Rinehart and Winston.
. 1978. Epenthse et effacement de schwa dans les syllabes contigus en fran-
ais. In Etudes de phonologie franaise, eds. Benot de Cornulier and Franois
Dell, 7581. Paris: Editions du CNRS.
. 1984. Laccentuation dans les phrases en franais. In Forme sonore du lan-
gage: Structure des reprsentations en phonologie, eds. Franois Dell, Daniel
Hirst, and Jean-Roger Vergnaud, 65122. Paris: Hermann.
, Daniel Hirst, and Jean-Roger Vergnaud, eds. 1984. Forme sonore du langage:
Structure des reprsentations en phonologie. Paris: Hermann.
and Elisabeth Selkirk. 1978. On a morphologically governed vowel alterna-
tion in French. In Recent transformational studies in European languages, ed.
Samuel Keyser, 151. Cambridge: MIT Press.
Dsirat, Claude, and Tristan Hord. 1976. La langue franaise au 20e sicle. Paris:
Bordas.
Deyhime, Guiti. 1967. Enqute sur la phonologie du franais. Parts 1 and 2. La lin-
guistique 1.1: 97108; 1.2: 5784.
Di Cristo, Albert. 1998. Intonation in French. In Intonation systems: A survey of
twenty languages, eds. Daniel Hirst and Albert Di Cristo, 195218. Cambridge:
Cambridge University Press.
. 1999. Vers une modlisation de laccentuation du franais: premire partie.
Journal of French Language Studies 9.2: 143179.
. 2000. Vers une modlisation de laccentuation du franais (seconde partie).
Journal of French Language Studies 10.1: 2744.
Dressler, Wolfgang. 1985a. Typological aspects of natural morphology. Acta Linguistica
Academiae Scientiarum Hungaricae 35.1-2: 5170.
. 1985b. Morphonology: The dynamics of derivation. Ann Arbor: Karoma.
Durand, Jacques. 1990. Generative and non-linear phonology. London: Longman.
, Catherine Slater, and Hilary Wise. 1987. Observations on schwa in Southern
French. Linguistics 25.2: 9831004.
, and Marie-Anne Hintze. 1995. French phonology: Morae, syllables, words.
Lingua 95. (A special issue of Lingua devoted to French phonology.)
216 References
Durand, Marguerite. 1936. Le genre grammatical en franais parl Paris et dans la

rgion parisienne. Paris: DArtrey.
Encrev, Pierre. 1988. La liaison avec et sans enchanement: Phonologie tridimensi-
onnelle et usages du franais. Paris: Seuil.
Eschmann, Jrgen. 1976. Die Numerusmarkierung des Substantivs im gesprochenen
Franzsisch. Tbingen: Niemeyer.
Esposito, Florence. 1994. Les liaisons dangereuses: Pour ne plus lier les mots tort
et travers. Paris: Editions du Centre de formation et de perfectionnement des
journalistes.
Fagan, David. 1989. New directions in the merger of Parisian French /#/ and /a/. Acta
Linguistica Hafniensia 21.2: 6578.
Faygal, Zsuzsanna. 1998. Le retour du e nal en franais parisien: Changement
phontique conditionn par la prosodie. Paper presented at the XIIe Congrs
International de Linguistique et Philologie Romanes, Bruxelles. Available on the
Congress Web site: http://www.ulb.ac.be/philo/serlifra/cilpr98/cilpr98.html.
Fnagy, Ivn. 1979. Laccent franais: accent probabilitaire. In Laccent en franais
contemporain (Studia Phonetica 15), eds. Ivan Fnagy and Pierre Lon, 123233.
Ottawa: Didier.
. 1989. Le franais change de visage. Revue romane 24.2: 22554.
, Eva Brard, and Judith Fnagy. 1983. Les clichs mlodiques du franais
parisien. Folia Linguistica 17: 15385.
Fouch, Pierre. 1936. Les diverses sortes de franais au point de vue phontique. Le
franais moderne 4: 199216.
. 1959. Trait de prononciation franaise. Paris: Klincksieck.
. 1967. Morphologie historique du franais: Le verbe. Paris: Klincksieck.
. 1969. Phontique historique du franais. Vol. 2. Les voyelles. Vol. 3. Les con-
sonnes. 2nd edition. Paris: Klincksieck.
Fougeron, Ccile, and Sun-Ah Jun. 1998. Rate effects on French intonation, prosodic
organization and phonetic realization. Journal of Phonetics 26: 4569.
Frei, Henri. 1929. La grammaire des fautes. Geneva: Slatkine.
Gadet, Franoise. 1989. Le franais ordinaire. Paris: Armand Colin.
Gather, Andreas. 1997. Die Genusmorphologie des Franzsischen und sekundre
Phnomene. In Semiotische Prozesse und natrliche Sprache. Festschrift fr Udo
L. Figge zum 60. Geburtstag, eds. Andreas Gather and Heinz Werner, 15582.
Stuttgart: Steiner.
. 1998. Sonorittsassimilationen im Franzsischen. Unpublished manuscript,
Ruhr-Universitt-Bochum.
References 217
Girard, Francine, and Chantal Lyche. 1997. Phontique et phonologie du franais.

Oslo: Universitetsforlaget.
Goldsmith, John, ed. 1995. The handbook of phonological theory. Oxford:
Blackwell.
Gougenheim, Georges. 1935. Elements de phonologie franaise. Strasbourg:
Publications de la Facult de lettres de lUniversit de Strasbourg.
Grammont, Maurice. 1914. Trait pratique de prononciation franaise. Paris:
Delagrave.
Greenberg, Joseph, ed. 1966. Universals of language. 2nd edition. Cambridge: MIT
Press.
Guiraud, Pierre. 1965. Le franais populaire. Paris: Presses Universitaires de France.
. 1969. Franais populaire ou franais relch. Le franais dans le monde 69:
2327.
Hannahs, Stephen. 1995a. Prosodic structure and French morphophonology. Tbingen:
Niemeyer.
. 1995b. The phonological word in French. Linguistics 33: 112544.
Hansen, Anita. 1994. Etude du E caduc: stabilisation en cours et variations lexicales.
Journal of French Language Studies 4: 2554.
. 1997. Le nouveau [] prpausal dans le franais parl Paris. In Polyphonie
pour Ivn Fnagy, ed. Jean Perrot, 17398. Paris: Harmattan.
. 1998. Les voyelles nasales du franais parisien moderne: Aspects linguis-
tiques, sociolinguistiques et perceptuels des changements en cours. Copenhagen:
Museum Tusculanem Press.
Hirst, Daniel, and Albert Di Cristo, eds. 1998. Intonation systems: A survey of twenty
languages. Cambridge: Cambridge University Press.
Juilland, Alphonse. 1965. Dictionnaire inverse de la langue franaise. The Hague:
Mouton.
Kaye, Jonathan, and Yves-Charles Morin. 1978. Il ny a pas de rgles de troncation,
voyons! In Proceedings of the Twelfth International Congress of Linguists, eds.
Wolfgang Dressler, and Wolfgang Meid, 78892. Vienna, August 28September
2, 1997. Innsbruck: Innsbrucker Beirtge zur Sprachwissenschaft.
Kenstowicz, Michael. 1994. Phonology in generative grammar. Oxford: Blackwell.
Klausenburger, Jurgen. 1984. French liaison and linguistic theory. Stuttgart: Franz
Steiner Verlag.
218 References
Kohler, K. J., and H. Knzel. 1979. Produktion und Perzeption der Plosive
und Frikative im heutigen Standard-Franzsisch: Wort- und satzphonetische
Untersuchungen. Kiel: Arbeitsberichte des Instituts fr Phonetik und digitale
Sprachverarbeitung Universitt Kiel.
, W. A. van Dommelen, and G. Timmermann. 1981. Die Merkmalpaare
stimmhaft/stimmlos und fortis/lenis in der Konsonantenproduktion und -perzep-
tion des heutigen Standard-Franzsisch. Kiel: Arbeitsberichte des Instituts fr
Phonetik und digitale Sprachverarbeitung Universitt Kiel.
Lacheret-Dujour, Anne and Frderic Beaugendre. 1999. La prosodie du franais.
Paris: CNRS Editions.
Landick, Marie. 1990. The determination of the French mid-vowels. Ph.D. diss.,
Kings College, University of London.
LeBel, Jean-Guy. 1990. Trait de correction phontique ponctuelle: Essai systma-
tique dapplication. Qubec: Centre International de Recherche en Amnagement
Linguistique.
. 1991. Fiches correctives des sons du franais. Qubec: Les Editions de la
Facult des Lettres de lUniversit Laval.
Lefkowitz, Natalie. 1991. Talking backwards, looking forwards: The French language
game verlan. Tbingen: Gunter Narr Verlag.
Le Gofc, Pierre. 1997. Les formes conjugues du verbe franais oral et crit. Paris:
Ophrys.
Lennig, Matthew. 1978. Acoustic measurement of linguistic change: The modern Paris
vowel system. Ph.D. Diss., University of Pennsylvania.
Lon, Pierre. 1966. Prononciation du franais standard. Paris: Didier.
. 1971. Essais de phonostylistique. Ottawa: Didier.
. 1973. Modle standard et systme vocalique du franais populaire de jeunes
Parisiens. In Contributions canadiennes la linguistique applique, ed. Guy
Rondeau, 5579. Montral: Centre Educatif et Culturel.
. 1974. Modalit imprative et intonation. In World papers in phonetics:
Festschrift for Dr. Onishis Kiju, 25380. Tokyo: Phonetic Society of Japan.
. 1992. Phontisme et prononciations du franais. Paris: Nathan.
. 1993. Prcis de phonostylistique: Parole et expressivit. Paris: Nathan.
Lerond, Alain. 1980. Dictionnaire de la prononciation. Paris: Larousse.
Lodge, R. Anthony. 1993. French: from dialect to standard. London: Routledge.
, Nigel Armstrong, Yvette Ellis, and Jane Shelton. 1997. Exploring the French
language. London: Arnold.
References 219
Lucci, Vincent. 1983. Etude phontique du franais travers la variation situation-

nelle. Grenoble: Publications de lUniversit des Langues et Lettres de Grenoble.
Lyche, Chantal, and Francine Girard. 1995. Le mot retrouv. In French phonology:
Morae, syllables, words, eds. Jacques Durand and Marie-Anne Hintze, 20521.
Lingua 95. (A special issue of Lingua devoted to French phonology.)
Malcot, Andre. 1974. Frequency and occurrence of French phonemes and consonant
clusters. Phonetica 29.2: 15870.
. 1975. The glottal stop in French. Phonetica 31.1: 5163.
. 1976. The effect of linguistic and paralinguistic variables on the elision of the
French mute-e. Phonetica 33.2: 93112.
, and Grard Chollet. 1977. The acoustic status of mute-e in French. Phonetica
34.1: 1930.
and Patricia Lindsay. 1976. The neutralization of /'/ - // in French. Phonetica
33.1: 4561.
and G. Metz. 1972. Progressive nasal assimilation in French. Phonetica 26.4:
193209.
and Marie Richman. 1972. Optional word-nal consonants in French.
Phonetica 26.1: 6588.
Malmberg, Bertil. 1969. Phontique franaise. Malmo, Sweden: Hermods.
Marouzeau, Jean. 1924. Accent affectif et accent intellectuel. Bulletin de la Socit de
Linguistique de Paris 25: 8086.
Martinet, Andr. 1955. Economie des changements phontiques. Berne: A. Francke.
. 1958a. Cest jeuli, le Mareuc! Romance Philology 11: 34555.
. 1958b. De lconomie des formes du verbe en franais parl. In Studia philo-
logica et litteraria in honorem L. Spitzer, 30926. Reprinted in Martinet 1969:
91120.
. 1969. Le franais sans fard. Paris: Presses Universitaires de France.
. 1971. La prononciation du franais contemporain. 2nd ed. Paris: Droz.
. 1990. Remarques sur la varit des usages dans la phonie du franais. In
Variation and change in French: Essays presented to Rebecca Posner on the
occasion of her sixtieth birthday, eds. John Green and Wendy Ayres-Bennet,
1326. London: Routledge.
, and Henriette Walter. 1973. Dictionnaire de la prononciation franaise dans
son usage rel. Paris: France Expansion.
McCarter, Robbin. 1987. The mastery of French gender. Richmond, Va.: Willowbrook
Press.
220 References
Mettas, Odette. 1979. La prononciation parisienne: Aspects phoniques dun sociolecte

parisien (du Faubourg St.-Germain la Muette). Paris: SELAF.
Moody, Marvin. 1978. Stable nal consonants in French. The French Review 51.3:
40712.
Morier, Henri. 1975. Dictionnaire de potique et de rhtorique. 2nd ed. Paris: Presses
Universitaires de France.
Morin, Yves-Charles. 1971. Computer experiments in generative phonology: Low-level
French phonology. (Natural Language Studies No. 11.) Ann Arbor: University
of Michigan, Department of Computer and Communication Sciences, Phonetics
Laboratory.
. 1972. The phonology of echo-words in French. Language 48: 97105.
. 1987. On explaining cross-syllabic constraints. In Proceedings of the
Fifth International Phonology Meeting, Eisenstadt, 25-28 June 1984, eds.
Wolfgang Dressler, Hans Luschutzky, Oskar Pfeiffer, and John Rennison, 20713.
Cambridge: Cambridge University Press.
. 1988. Lajustement du schwa en syllabe ferme dans la phonologie du
franais. In La phonologie du schwa franais, ed. S. Paul Verluyten, 13389.
Amsterdam: Benjamins.
. 1994. Quelques rexions sur la formation des voyelles nasales en franais.
Communication and Cognition 27.12: 27110.
. 1999. Le franais de rfrence et les normes de prononciation. Paper pre-
sented at the Colloque international: Le franais de rfrence, at Universit
catholique de Louvain, Louvain-la-Neuve, November, 1999.
, and Jonathan Kaye. 1982. The syntactic bases for French liaison. Journal of
Linguistics 18: 291330.
Mougeon, Franoise. 1998. Quel franais parler? Initiation au franais parl au
Canada et en France. 2nd ed. Toronto: Editions du Gref (Centre universitaire
Glendon).
Mueller, Theodore, Edgar Mayer, and Henri Niedzielski. 1968. Handbook of French
structure: A systematic review. New York: Harcourt, Brace and World.
Muller, Bodo. 1985. Le franais daujourdhui. Paris: Klincksieck.
Nespor, Marina, and Irene Vogel. 1986. Prosodic phonology. Dordrecht: Foris.
Nyrop, Kristoffer. 1934. Manuel phontique du franais parl. 5th ed. (1st ed. 1902.)
Copenhagen: Gyldendal.
Nyrop, Kristoffer. 1935. Grammaire historique de la langue franaise. Vol 1: Histoire
gnrale de la langue franaise; Phontique historique. 4th ed. Copenhagen:
Gyldendal.
References 221
Offord, Malcolm. 1989. La valeur smantique des mots terminaison en -o en fran-

ais. Cahiers de lexicologie 55.2: 3952.
. 1996. A reader in French sociolinguistics. Clevedon: Multilingual Matters.
Ostiguy, Luc, Robert Sarrasin, and Glenwood Irons. 1996. Introduction la phon-
tique compare: Les sons. Le franais et langlais nord-amricains. Sainte-Foy:
Les Presses de lUniversit Laval.
Passy, Paul. 1892. Le franais parl: Morceaux choisis lusage des trangers avec
la prononciation gure. Leipzig: Reisland.
Picard, Marc. 1987. An introduction to the comparative phonetics of English and
French in North America. Amsterdam: Benjamins.
Pichon, Edouard. 1938. Genre et questions connexes (Sur les pas de Mlle Durand). Le
franais moderne 6: 2943, 10726.
, and Jacques Damourette. 1936. Questions de liaisons. Le franais moderne 4:
34753.
Picone, Michael. 1996. Anglicisms, neologisms and dynamic French. Amsterdam:
Benjamins.
Plnat, Marc. 1984. Toto, Fanfa, Totor et mme Guiguitte sont des anars. In Forme
sonore du langage. Structure des reprsentations en phonologie, eds. Franois
Dell, Daniel Hirst, and Jean-Roger Vergnaud, 16181. Paris: Hermann.
. 1985. Morphologie du largonji des loucherbems. Langages 78: 7395.
. 1991. Le javanais: concurrence et haplologie. Langages 101: 95117.
. 1995. Une approche prosodique de la morphologie du verlan. In French pho-
nology: Morae, syllables, words, eds. Jacques Durand and Marie-Anne Hintze,
97129. Lingua 95. (A special issue of Lingua devoted to French phonology.)
Pope, Mildred. 1934. From Latin to Modern French with especial consideration of
Anglo-Norman. Manchester: Manchester University Press.
Post, Brechtje. 1999. Restructured phonological phrases in French: Evidence from
clash resolution. Linguistics 37.1: 4163.
. 2000. Pitch accents, liaison and the phonological phrase in French. Probus
12.1: 12764.
Price, Glanville. 1971. The French language: Present and past. London: Edward
Arnold.
. 1991. An introduction to French pronunciation. Oxford: Blackwell.
Prigniel, M. 1966. Le sufxe populaire -o. Le franais moderne 31.1: 4763.
Pullum, Geoffrey, and William Ladusaw. 1986. Phonetic symbol guide. Chicago:
University of Chicago Press.
222 References
Reichstein, Ruth. 1960. Etude des variations sociales et gographiques des faits lin-
guistiques. Word 16: 5599.
Rey-Debove, Josette, and Gilberte Gagnon. 1980. Dictionnaire des anglicismes: Mots
anglais et amricains en franais. Paris: Le Robert.
Rheinfelder, Hans. 1968. Altfranzsische Grammatik. 1. Teil. Lautlehre. Munich:
Hueber.
Richman, Marie. 1976. The distribution of /e/ - /'/ in French verb endings. Phonetica
33.3: 30719.
Rigault, Andr, ed. 1971. La grammaire du franais parl. Paris: Hachette.
Roca, Iggy, and Wyn Johnson. 1999. A course in phonology. Oxford: Blackwell.
Rochet, Bernard. 1976. The formation and the evolution of the French nasal vowels.
Tbingen: Niemeyer.
. 1977. On the status of the word in French phonology. International Review of
Applied Linguistics 15.3: 18796.
. 1980. The mid-vowels in Bordeaux French. Orbis 29.12: 76104.
Rossi, Mario. 1999. Lintonation, le systme du franais: Description et modlisation.
Paris: Ophrys.
Rossillon, Philippe, ed. 1996. Atlas de la langue franaise. Paris: Bordas.
Sanders, Carol, ed. 1993. French today: Language in its social context. Cambridge:
Cambridge University Press.
Santerre, Laurent. 1974. Deux E et deux A phonologiques en franais qubcois.
Cahier de linguistique 4: 11745.
Savard, Jean-Guy, and Jack Richards. 1970. Les indices dutilit du vocabulaire fon-
damental franais. Qubec: Les Presses de lUniversit Laval.
Schane, Sanford. 1968. French phonology and morphology, Cambridge: MIT Press.
Scullen, Mary Ellen. 1997. French prosodic morphology: A unied account.
Bloomington: Indiana University Linguistics Club.
Sguinot, Andr. 1977. Laccent dinsistance. Emphatic stress. (Studia Phonetica 12.)
Ottawa: Didier.
Selkirk, Elisabeth. 1978. The French foot: on the status of mute e. Studies in French
Linguistics 1: 14150.
Shelly, Sharon. 1998. Teaching the sound system(s): the case of mid vowels. The
French Review 71.4: 598606.
Straka, Georges. 1950. Systme des voyelles du franais moderne. Parts 1 to 4.
Bulletin de la Facult des Lettres de Strasbourg 26: 17279, 22033; 28: 27584,
36875.
References 223
. 1952. La prononciation parisienne, ses divers aspects et ses traits gnraux.

Parts 1 and 2. Bulletin de la Facult des Lettres de Strasbourg 29: 21225; 31:
23953.
. 1955. Remarques sur les voyelles nasales, leur origine et leur volution en
franais. Revue de linguistique romane 19: 24574.
. 195859. Tendances phontiques du franais contemporain. Parts 1 and
2. Classe de franais (Revue pour lenseignement du franais) 8: 35462; 9:
3540.
. 1990. Franzsische: Phonetik und Phonemik. In Lexikon der Romanistischen
Linguistik. Vol. 5.1: Franzsisch, eds. Gnter Holtus, Michael Metzeltin, and
Christian Schmitt, 133. Tbingen: Niemeyer.
Taulelle, Dominique. 1989. Le Robert oral-crit: Lorthographe par la phontique.
Paris: Dictionnaires Le Robert.
Thiele, Johannes. 1987. La formation des mots en franais moderne. Montral: Les
Presses de lUniversit de Montral.
Thimonnier, Ren. 1970. Code orthographique et grammatical. Paris: Hatier.
Tranel, Bernard. 1976. A generative treatment of the prex in- of Modern French.
Language 52: 34569.
. 1981. Concreteness in generative phonology: Evidence from French.
Berkeley: University of California Press.
. 1987. The sounds of French: An introduction. Cambridge: Cambridge
University Press.
. 1988. A propos de lajustement de e en franais. In La phonologie du schwa
franais, ed. S. Paul Verluyten, 89131. Amsterdam: Benjamins.
. 1995. Current issues in French phonology: liaison and position theories. In
The handbook of phonological theory, ed. John Goldsmith, 798816. Oxford:
Blackwell.
Trask, R. L. 1993. A dictionary of grammatical terms in linguistics. London:
Routledge.
. 1996. A dictionary of phonetics and phonology. London: Routledge.
. 1999. Key concepts in language and linguistics. London: Routledge.
Tucker, Richard, Wallace Lambert, and Andr Rigault. 1977. The French speakers
skill with grammatical gender: An example of rule-governed behaviour. The
Hague: Mouton.
Vaissire, Jacqueline. 1991. Rhythm, accentuation and nal lengthening in French. In
Music, language, speech and brain, eds. Johan Sundberg, Lennart Nord, and Rolf
Carlson,10820. Proceedings of an International Symposium at the Wenner-Gren
Center, Stockholm, 5-8 September 1990. London: Macmillan Press.
224 References
Valdman, Albert. 1976. Introduction to French phonology and morphology. Rowley,

Mass.: Newbury House.
. 1993. Bien entendu! Introduction la prononciation franaise. Englewood
Cliffs: Prentice Hall.
. 2000a. Comment grer la variation dans lenseignement du franais langue
trangre aux Etats-Unis. The French Review 73.4: 64866.
. 2000b. La langue des faubourgs et des banlieues: de largot au franais popu-
laire. The French Review 73.6: 117992.
van den Berghe, Christian. 1976. La phonostylistique du franais. The Hague:
Mouton.
Varney Pleasants, Jeanne. 1956. Etudes sur le muet: Timbre, dure, intensit, hauteur
musicale. Paris: Klincksieck.
Vennemann, Theo. 1988. Preference laws for syllable structure and the explanation of
sound change. Berlin: Mouton de Gruyter.
Verluyten, S. Paul. 1984. Phonetic reality of linguistic structures: The case of (sec-
ondary) stress in French. In Proceedings of the Tenth International Congress
of Phonetic Sciences, eds. Marcel Van den Broecke and A. Cohen, 52226.
Dordrecht: Foris.
. ed. 1988. La phonologie du schwa franais. Amsterdam: Benjamins.
Walker, Douglas. 1975. Lexical stratication in French phonology. Lingua 37:
17796.
. 1978. Epenthesis in Old French. Canadian Journal of Linguistics 23.1:
6683.
. 1984. The pronunciation of Canadian French. Ottawa: University of Ottawa
Press.
. 1993. Schwa and // in French. Canadian Journal of Linguistics 38.1:
4364.
. 1995. Patterns of analogy in the Canadian French verb system. Journal of
French Language Studies 5: 85107.
. 1996. The new stability of unstable-e in French. Journal of French Language
Studies 6: 21129.
. 1999. On the phonotactics of French nasal vowels. Paper presented at the
14th International Conference on Historical Linguistics, Vancouver, B.C., August
913, 1999.
Walter, Henriette. 1976. La dynamique des phonmes dans le lexique franais contem-
porain. Paris: France Expansion.
References 225
. 1977. La phonologie du franais. Paris: Presses Universitaires de France.

. 1982. Enqute phonologique et varits rgionales du franais. Paris: Presses
Universitiares de France.
, ed. 1983. Phonologie des usages du franais. Langue franaise 60.
. 1997. Laventure des mots franais venus dailleurs. Paris: Robert Lafont.
. 1998. Le franais dici, de l, de l-bas. Paris: JC Latts.
Warnant, Lon. 1987. Dictionnaire de la prononciation franaise dans sa norme actu-
elle. 4th ed. Paris-Gembloux: Duculot.
. 1996. Orthographe et prononciation en franais. Louvain-la-Neuve, Belgium:
Duculot.
. 1997. Dictionnaire des rimes orales et crites. 2nd ed. Paris: Larousse.
Wenk, Brian, and Franois Wioland. 1982. Is French really syllable-timed? Journal of
Phonetics 10: 193216.
Wioland, Franois. 1991. Prononcer les mots du franais: Des sons et des rythmes.
Paris: Hachette.
Wise, Hilary. 1997. The vocabulary of Modern French: Origins, structure and func-
tion. London: Routledge.
Zwanenburg, Wiecher. 1965. Recherches sur la prosodie de la phrase franaise.
Leiden: Universitaire Pers Leiden.
. 1983. Productivit morphologique et emprunt: Etude des drivs dverbaux
savants en franais moderne. Amsterdam: Benjamins.
227
Index Index
/a - #/, 60-62 compound, 85, 163, 168, 192

/e - '/, 48-59 conjugation, 49, 57, 87
/':/ , 45 conservative, 45, 48, 60, 63, 133
/o - n/, 48-59 constraint, 48, 55, 64, 77-78, 84, 86,
/ - /, 48-59 105, 136
/sC/, 28, 52 derivation, 14, 67, 156, 191
CL, 85, 87, 89, 101, 103, 120-121, dialect, 6
136 emphasis, 181
X - VN, 65 en-, 64-65, 73
VC.CV, 27-28, 139 enchanement, 30-31, 37, 107,
VCCCX, 82-86 143-145, 160, 165, 167
VCCX, 82-86, 90 English, 1, 5, 15, 44, 62, 134-135,
146, 158, 178-179
abbreviation, 191-194 epenthesis, 85-86, 128-130
accent d insistance, 91, 131, 181-182, exception, 50, 55, 61, 64-65, 78-79,
186 83, 92, 99, 120, 150-151
acronym, 195-199 false liaison, 165
afx, 14 familiar, 4, 187
alternation, 16, 49, 59, 71, 92, feminine, 14, 47, 51, 65, 71, 87, 95,
104-105, 153, 157-158 97-99, 153
analogy, 52-56, 61 nal consonant, 95, 148-152, 163
aspirate-h, 9, 78, 86, 105, 107, 128, formal, 2-6, 35, 54, 88-89, 104-105,
140, 147 130, 136, 165, 169, 182, 187, 201
assimilation, 135, 140 geminate, 73, 104, 130-131, 177
Canadian French, See CF gender, 13-14, 95, 98-99
CF, 5-6, 44, 48, 75-76, 95 glides, See also semi-vowels, 21-22,
circonexe, 47, 62 24-25, 31, 41, 100, 103-106,
clitic, 31, 84, 87, 105, 107, 181 127-128, 146
closed syllable, 50-52, 57, 78, 93, hiatus, 47, 78, 105-106, 146
107, 110 high vowels, 22, 31, 41, 100, 104
coda, 23-27, 30, 43, 74, 103, 126-128 inection, 14, 65, 153, 156
colloquial, 3, 86-90, 187 informal, 4, 58, 87, 145, 169,
compensatory lengthening, 44, 62 199-200, 204
228 Index
insertion, 80, 85-86, 95, 97, 201 phoneme, 13, 103, 133, 135, 158
latent, 25, 148-149, 152-157, phonological hierarchy, 23, 30
160-161, 167, 204 phonological phrase, 30-37, 80-84,
Latin, 1, 6-8, 15, 96, 98, 130, 147, 88-89, 99, 106-107, 131, 161, 169,
179, 178, 182, 184
learned, 8, 15-16, 28, 47-48, 62, phonology, 5, 13, 16-17, 22, 28,
93-95, 99, 125, 130, 151-152, 156, 30-31, 91, 95, 103, 107, 125, 128,
159 149, 177, 182, 204
lengthening consonants, 42-43, 47 phonotactic, 24-25, 29, 31, 37, 77, 86,
lexicon, 14, 140, 147, 191 106-107, 121-128, 136, 198
liquids, 22, 81, 119-120, 128, 139-140 phrase-initial, 37, 80-81, 83, 87, 181
loi de position, 51-52, 55-56, 59, 107 plural, 14, 47, 65-66, 71, 145, 151,
loi des trois consonnes, 82 158-159, 163, 168, 199
masculine, 14, 47, 65, 71, 95, 98 popular, 3-4, 6, 15, 89-90, 99, 145,
merger, 59-60, 62-63, 79-80, 91, 156, 191
MF, 48, 55-57, 75, 96, 111, 116 prex, 14, 64-65, 69, 73, 83, 91
mid vowels, 22, 29, 48-60, 106-107, productivity, 14, 194
192 proper noun, 78
Midi French, See MF prosody, 131, 169, 177-187
minimal phrase, 34-36, 88-89, rapid speech, 13, 105, 115, 140, 146
180-181 reduplication, 14, 199-200
morphology, 13-17, 53, 65, 87, 92-93, register, 3-6, 186-187
99, 128-129, 148, 153, 199, 204 resyllabication, 37, 107, 161, 165
mute-e, See also schwa, 52, 85, rhythm, 31, 84-85, 177-178, 183
95-96, 201 root, 14, 52-53, 56, 61, 65, 73, 77
nasal consonants, 22, 31, 63-64, schwa, See also mute-e, 3, 25, 30-31,
73-74, 106-107, 132, 136 37, 42, 48, 60, 75, 99, 105-107,
nasalization, 44, 71-75, 130, 150 125-126, 128, 130-131, 136, 139,
neologism, 14-15 143-146, 148, 153, 178-179, 182,
nonstandard, 3, 106, 191-194, 201, 204
199-204 semi-vowels, See also glides, 21,
nucleus, 23-27, 126-128 24-25, 41, 100, 102-105, 139, 146
number, 13-14, 158 SF, See also Standard French, 6, 21,
-o, 193-194 25, 29, 41-42, 48, 55-56, 75, 88,
OF, 6-7, 15, 74, 96-99, 128-130, 147, 107, 119, 121, 135, 140, 147
158 singular, 14, 47, 65-66, 71, 145, 151,
Old French, See OF 158-159, 162
onset, 23-27, 30, 126-128, 140, 161 slang, 2, 4, 75, 199, 201
open syllable, 23, 27, 44, 48-50, stable, 51, 56, 60, 79, 86, 90-91,
52-57, 93, 148 147-149, 153, 160
optional, 26, 35, 105, 151-152, 161, standard, 2-6, 18, 48, 136, 191
163-164, 169, 202 Standard French, See also SF, 2-3, 5,
orthography, 7, 11-12, 50, 57, 59, 8, 160
73-75, 86, 92, 95, 99, 132, 148-149, strength, 27, 30, 129, 136
153, 157, 201 stress, 7, 23, 30-32, 42, 46, 53, 91-92,
person, 14, 151, 167-168 108, 116, 131, 169, 177-182, 187
Index 229
style, 5, 169
sufx, 14, 51, 53, 65, 93, 104, 126,
134-135, 193-194
syllabication, 27-31, 52, 125, 139,
159-161, 167, 177
syllable-nal, 74, 120, 125, 128,
138-140
syllable-initial, 120-121, 126, 128,
137, 139
variation, 2-3, 5, 8, 12, 16, 52, 74, 76,
86, 88, 99, 106-107, 133, 136, 140,
146, 148, 179, 182
variety, 2, 5-6, 62, 76, 89, 95-96, 160,
165, 182, 191
verlan, 191, 200-202, 204
vowel harmony, 52, 54-56
vowel system, 41-42, 57, 60, 62, 64,
106-108
word, 13-16, 30-31, 37, 48, 67, 73,
77-78, 82-83, 86-87, 90, 99, 102,
105-107, 126, 128, 131, 161-163,
178-179, 181-182, 191-195,
199-201, 204
word-nal, 30-31, 43-44, 46, 49-50,
59, 72, 75, 79, 83, 96, 102,
106-107, 119, 125-126, 135, 137,
148-149, 152, 157, 160-161, 178,
201
French Sound Structure provides a comprehensive, detailed
and well-illustrated description of the pronunciation of Modern Standard
French, incorporating comments on regional and social variation, on
abbreviatory processes and word play, and on certain historical
phonological changes which continue to be reected in the contemporary
language. It is written in a way that presupposes little or no formal training
in linguistics proper (other than some familiarity with phonetic notation, to
which students of linguistics are normally exposed independently).
This work will be of interest to university students studying French and

to students of linguistics in general. Others wishing to know more about
the nature of the French language will also nd the material useful, since
pronunciation is rarely considered in any detail in general handbooks
of French. The accompanying CD-ROM is an additional valuable asset,
providing oral examples relevant to the linguistic material under study.
[a]well-organized and systematic

presentation of the major questions
of modern French phonology in a
clearly updated fashion.
Jurgen Klausenburger, University of Washington, Seattle
D ouglas Walker, a Professor of French and Linguistics in the Department of

French, Italian and Spanish at the University of Calgary, teaches French
and general linguistics. He has held academic positions in both Canada and the
United States and has authored An Introduction to Old French Morphophonology,
The Pronunciation of Canadian French, and numerous articles and reviews.
ISBN 1-55238-033-5
xHSLFPCy3803 8z

French Sound Structre PDF

Uploaded by

Copyright:

Available Formats

French Sound Structre PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

French Sound Structre PDF

Uploaded by

Copyright:

Available Formats

French Sound Structure

University of Calgary Press

National Library of Canada Cataloguing in Publication Data

Includes bibliographical references and index.

1. French language Phonetics. 2. French language Pronunciation.

We greatfully acknowledge the International Phonetic Association for allowing us to repro-

Printed and bound in Canada by Hignell Book Printing.

Page and cover design and typesetting by Kristina Schuring.

1. The Object of Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1

2. Key Descriptive and Theoretical Concepts . . . . . . . . . . . . . . . . . . . . . 11

3. Basic Descriptive Units and Domains . . . . . . . . . . . . . . . . . . . . . . . . . .21

4. Vowels and Semi-vowels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .41

4.4.4 History and Orthography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .73

7. Around the Phonological Periphery: Playing with Language . . . . .191

Standard French Vowels and Semi-vowels

Standard French Consonants

labial apical palatal velar uvular

*In International Phonetic Association (IPA) notation, [] represents a voiced

Abbreviations and Symbols

1.0 1.0 Preliminaries

1.1 1.1 The French Language

The wide geographic distribution of the French language is understandably

Gographiquement, nous avons choisi de consigner la prononciation

Socialement, nous avons choisi de recueillir la prononciation des

Nous avons aussi choisi de ne dcrire que le parler dune gnra-

Despite Warnants efforts to circumscribe the object of his description, his

(1) Levels of usage (registers) 5

innovative (although highly formal speech can also be innovative

The following graded examples, moving from formal to familiar or popular,

(2) (a) O demeurez-vous? Where do you reside?

(e) importuner to importune

Minor lexical or phonetic differences aside, in fact, Standard Canadian French

1.2 1.2 A Brief Historical Review

2. Despite what may be thought, France itself is not linguistically homogeneous:

4. Warnant, interestingly enough in this context, is Belgian. The quotation is taken

2.0 2.0 Introduction

2.1 2.1 Orthography and Pronunciation

(1) Orthographic puzzles

(b) one spelling several sounds

These examples demonstrate the ambiguity of certain aspects of French

namely the symbols of the International Phonetic Association (IPA), which we

2.2 2.2 Morphological and Lexical Notions5

as syntactic and phonological context). Savoir to know is even more com-

-ique (informatique computer science, robotique robotics), for example,

double dupliquer double to duplicate

However we decide to approach these examples in a discussion of French pho-

(3) Phonological alternations

These alternations may be regular, systematic, predictable, phonologically

6. See Bchade (1992: 136139) for a convenient summary of conversion pro-

7. Inectional processes may also be distinguished as to their productivity. For

3.0 3.0 Introduction: The Segment

(1) SF vowels and semi-vowels

labial apical palatal velar uvular

3.1 3.1 The Syllable

(3) A simple syllable structure

onset nucleus coda

It is likely, however, that syllables have a more complicated, hierarchical

(4) A hierarchical syllable

onset nucleus coda

# C VV #