UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Production and perception of phonemic vowel duration in French Duncan, Elizabeth D. 1975

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
831-UBC_1975_A6_7 D85_3.pdf [ 2.62MB ]
Metadata
JSON: 831-1.0100021.json
JSON-LD: 831-1.0100021-ld.json
RDF/XML (Pretty): 831-1.0100021-rdf.xml
RDF/JSON: 831-1.0100021-rdf.json
Turtle: 831-1.0100021-turtle.txt
N-Triples: 831-1.0100021-rdf-ntriples.txt
Original Record: 831-1.0100021-source.json
Full Text
831-1.0100021-fulltext.txt
Citation
831-1.0100021.ris

Full Text

PRODUCTION AND PERCEPTION OF PHONEMIC VOWEL DURATION IN FRENCH by Elizabeth D. Duncan B.A. , University of B r i t i s h Columbia, 1973 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE i n the Department of Paediatrics Division of Audiology and Speech Sciences We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA A p r i l , 1975 In p r e s e n t i n g t h i s t h e s i s i n p a r t i a l f u l f i l m e n t o f t h e r e q u i r e m e n t s f o r an a d v a n c e d d e g r e e a t t h e U n i v e r s i t y o f B r i t i s h C o l u m b i a , I a g r e e t h a t t h e L i b r a r y s h a l l m a k e i t f r e e l y a v a i l a b l e f o r r e f e r e n c e a n d s t u d y . I f u r t h e r a g r e e t h a t p e r m i s s i o n f o r e x t e n s i v e c o p y i n g o f t h i s t h e s i s f o r s c h o l a r l y p u r p o s e s may be g r a n t e d by t h e H e a d o f my D e p a r t m e n t o r by h i s r e p r e s e n t a t i v e s . I t i s u n d e r s t o o d t h a t c o p y i n g o r p u b l i c a t i o n o f t h i s t h e s i s f o r f i n a n c i a l g a i n s h a l l n o t be a l l o w e d w i t h o u t my w r i t t e n p e r m i s s i o n . D e p a r t m e n t ofjyLtlL (Uh7 C J - fykAjJJl/f^ kJf)s j L ±Jl-ju T §JUUL <tL SCJJLAHU^ T h e U n i v e r s i t y o f B r i t i s h C o l u m b i a V a n c o u v e r 8, C a n a d a ABSTRACT The present study consists of two experiments which ex-amine the use of Vowel duration i n French i n such word pairs as "maitrermettre" and "paterpatte". The f i r s t experiment examines the use of vowel duration from the point of view of production. Subjects recorded sentences containing the words under study both i n (unemphatically) stressed and unstressed p o s i t i o n . S i g n i f i c a n t differences i n vowel duration were found between the "long" and the "short" members of the p a i r s . No s i g n i f i c a n t differences i n vowel duration were found between the / a : / : / a / "long"."short" vowel r a t i o s and the / e : / : / e / vowel r a -t i o s . Vowel r a t i o s were s l i g h t l y smaller when words appeared i n unstressed p o s i t i o n than i n stressed p o s i t i o n but these differences were not found to be s i g n i f i c a n t . Speakers from the Midi were ob-served to make the length d i s t i n c t i o n less frequently than speakers fjEom other areas. The second experiment examines the use of vowel duration by the same group of native speakers i n d i s t i n g u i s h i n g perceptually between members of such pairs as those mentioned above. Using a small computer, vowels of the "long" and "short" members of a p a i r were shortened and lengthened r e s p e c t i v e l y i n steps of approximate-l y 16 msec. Subjects were asked to assign a l e x i c a l value to the s t i m u l i so obtained. Subjects' responses were found to be influenced by s h i f t s i n vowel duration. Lengthening the "short" vowel had a greater e f f e c t i n reversing l i s t e n e r judgement of the word than did i i i shortening the "long" vowel. When members of a pair were accompanied by a marked difference i n vowel qu a l i t y , shortening the "long" vow-e l had r e l a t i v e l y l i t t l e effect on the l i s t e n e r s ' judgement of the word. I t thus appears that for the group of speaker/listeners selected, length does play a role i n distinguishing, both produc-t i v e l y and perceptually, between such pairs as "maitrermettre" and "paterpatte". i v TABLE OF CONTENTS Page ABSTRACT . . . . . . •. . . ... - i i i TABLE OF CONTENTS. . • v LIST OF TABLES v i i LIST OF FIGURES. v i i i ACKNOWLEDGEMENT i x Chapter 1. INTRODUCTION 1 Chapter 2. LITERATURE REVIEW 2 2.1 Introduction . . . . . . . . . . . . . . . . 2 2.2 Vowel Duration: Production 2 2.3 Vowel Duration: Perception 10 Chapter 3. AIMS OF THE INVESTIGATION. . . . 16 3.1 Experiment I .16 3.2 Experiment I I 17 Chapter 4. EXPERIMENT I • 18 4/1 Experimental Apparatus and Procedures. . . . 18 4.11 Corpus . . . . . . . . . 18 4.12 Subjects 19 4.13 ^Instrumentation and'Arrangement. . . . 19 4.14 Experimental Design 20 4.2 Analysis of the Data 20 4.21 Instrumentation. 20 4.22 C r i t e r i a for Determining Vowel Boundaries . .' 21 4.23 Determination of Average Measure-ment Error 27 v Page 4.3 Results 28 4.31 Averages of "Long" and "Short" Vowel Durations .. . 28 4.32 Average "Long":"Short" Vowel Ratios . . 29 4.33 Comparison of /a:/:/a/ with /e:/:/e/. . 29 4.34 Stressed vs. Unstressed Position. . . .31 4.35 Speaker Differences 31 4.36 Speaker's I n t u i t i o n vs. Speaker's Performance . . . . . . . . . . . . . . 33 Chapter 5. EXPERIMENT I I . . . . . . . . . . . 34 5.1 Experimental Apparatus and Procedures . . . . 34 5.11 Preparation of Test Tape. 34 ^Selection of Recording .34 Selection of Word Pairs . . . . . . . . 34 Vowel Editing 35 5.12 Subjects -37 5.13 Test Procedure and Equipment 37 5.2 Results 38 5.21 Vowel Lengthening vs. Vowel Shortening. . . 49 5.22 Other Factors Influencing Listener Judgement 50 5.23 Listener's I n t u i t i o n vs. Listener's Performance . 51 Chapter 6. SUMMARY AND DISCUSSION . , 52 6.1 Vowel Duration: Production 52 6.2 Vowel Duration: Perception 54 6.3 Vowel Duration: Production and Perception 56 BIBLIOGRAPHY. .58 APPENDIX . . . . . . . . .61 v i LIST OF TABLES Table Page I. Average "Long":"Short" Vowel Ratios for Each Subject 30 -I"II. Presence or Absence of Length D i s t i n c t i o n Be-tween "Long" and JoShort" Members of Each Pair for Individual Subjects . 32 v i i LIST OF FIGURES Figure Page 1. Mingogram of the word " f a i t e " 22 2. Mingogram of the word "mettre" 23 3 Mingogram of the word " l a " ; . . . • 25 4 Mingogram of the word "tache" . . 26 5 Listeners' judgements based on vowel duration only • 39 6 Listeners' judgements of o r i g i n a l and edited versions of "Bale" and " b a l l e " 40 7«r Listeners' judgements of o r i g i n a l and edited versions of "pate" and "patte" 41 8 Listeners' Judgements of o r i g i n a l and edited versions of "male" and "malle" 42 9 Listeners' judgements of o r i g i n a l and edited versions of " l a s " and " l a " . . . . . . . 43 10 Listeners' judgements of o r i g i n a l and edited versions of "tache" and "tache" 44 11 Listeners' judgements of o r i g i n a l and edited versions of "maitre" and "mettre". . . . 45 12 Listeners' judgements of o r i g i n a l and edited versions of "bete" and "bette" 46 13 Listeners' judgements of o r i g i n a l and edited versions of "bele" and "b e l l e " 47 14 Listeners' judgements of o r i g i n a l and edited versions of " l ' e t r e " and " l e t t r e " . . . . 48 v i i i ACKNOWLEDGEMENT I would l i k e to express my gratitude to the people who ave had a part i n this thesis. My thanks go especially to Dr. A.-P. Benguerel for a l l his help and guidance throughout the course of this investigation; Dr. J. Gilbert and Dr. R. Gregg for their time spent serving on my committee; my family for a l l their support and encouragement during the past two years. i x "The French don't care what they do ac t u a l l y , so long as they pronounce i t properly." (My Fair Lady, Act I, scene, i ) X Chapter 1 INTRODUCTION Describing the phonological features of a language i s a d i f -f i c u l t and involved task. Language i s constantly undergoing change; features that were at one time considered d i s t i n c t i v e are often replaced by other d i s t i n c t i v e features, or they simply disappear. In some cases, phonological changes occur at the segmental l e v e l . For example, an-cient Greek contrasted aspirated and unaspirated voiceless stop con-sonants. In modern Greek, the aspirated stops have been replaced by f r i c a t i v e s and the feature of aspiration i s no longer d i s t i n c t i v e . Such changes can also occur at the suprasegmental l e v e l . For example, differences i n duration are considered i n certain languages to be phonemic. In the French language., a difference i n vowel duration between members of such pairs as "maitrermettre" and "paterpatte" was at one time considered to be phonemic and i n the "standard" pronuncia-tion of French, the difference i n duration i s s t i l l observed. In the sound systems of many native speakers of French, however, the length contrast has tended to disappear and for these speakers, members Of such pairs are homonymous. The present study examines phonemic vowel duration, both from the point of view of production and that of perception, i n such pairs as "maltre:mettre" and "paterpatte" i n order to gain further i n -sight into the status of the length contrast i n contemporary French. - 1 -Chapter 2 LITERATURE REVIEW 2.1 Introduction This chapter surveys the l i t e r a t u r e concerning vowel dura-tion from two points of view. Section 2.2 deals with vowel duration from the aspect of production and section 2.3 discusses vowel duration from the aspect of perception. 2.2 Vowel Duration: Production One of the most obvious factors influencing speech segment duration i s rate of speech. There are, however, several other fac-tors which influence segment duration. In the case of vowels, cer-t a i n features of the phonetic environment play an important role i n lengthening or shortening the vowel. Several investigators have found evidence to support the hypothesis that vowels preceding voiced con-sonants are generally longer than those preceding voiceless ones. Studies include those of House and Fairbanks (1953), Belasco (1958), Peterson and Lehiste (1960) for American English; Zimmerman and Sapon (1958) for Spanish; Delattre (1966) for French. In one of his early a r t i c l e s , Delattre suggests that " . . . l a duree de l a voyelle est i n -verse de l a force d ] a r t i c u l a t i o n consonantique subsequente." (Delattre, 1966, p. 132). His findings have been substantiated by the studies of House and Fairbanks (1953) and Belasco (1958). - 2 -- 3 -Stress has also been shown to have an influence on vowel duration. Fry (1955) found that the vowel of a stressed s y l l a b l e i s longer than the same vowel i n unstressed position i n such pairs as "object (noun):object (verb)". Lehiste (1970) and Nooteboom (1972), studying Estonian and Dutch respectively, demonstrated that position i n utterance as w e l l as the number of sy l l a b l e s i n the ut-terance have a marked effect on vowel duration. Nooteboom found that with an increasing number of sy l l a b l e s i n an utterance the vow-e l tends to be shortened. Other factors which have been found to influence vowel duration are the ov e r a l l phrase length and the seman-t i c importance of the word i n which the vowel occurs. Certain languages, however, make use of vowel lengthening at the phonological l e v e l . A difference i n segment duration i n these languages i s the primary cue for distinguishing pa r t i c u l a r phonemes. For some languages, e.g. Japanese (cf. F u j i s a k i ert a l . , 1973), Nor-wegian, Finnish, Czech (c'f. Nooteboom, 1972), the vocal tract con-figuration i s i d e n t i c a l for two vowel phonemes which are distinguished solely by a difference i n duration. In other languages, e.g. Swedish, Dutch, German, English (cf. Nooteboom, 1972), a difference i n vowel duration i s often accompanied by a difference i n vowel quality. A phonological d i s t i n c t i o n i n vowel length i s said to be learned while a difference i n vowelllength which i s influenced by the phonetic environment i s said to be conditioned (Delattre, 1966). House (1961), for example, contends that i n English the / i : / : / i / - length difference i s learned, while the Jj/e/: A / length difference i s conditioned. The phoneme /!:/ i s longer than A / not because i t i s less open but because of the s u r v i v a l of a former d i s t i n c t i v e feature of length - 4 -that was present i n Middle English ( / i : / : / i / ) . This length d i f f e r -ence has changed to a less central/more central d i s t i n c t i o n with the difference i n length remaining but s l i g h t l y less marked. Several studies concerning d i s t i n c t i v e length have been carried out 'in languages that use length i n a meaningful way. Pho-nemic vowel length has been examined by Hadding-Koch and Abramson (1964) and by Fant (1971) for Swedish. For German, Meyer (1904), Vietor (1923) , and Weitkrus (1932) measured vowel durations of long and short vowels i n stressed position and arrived at long.short ratios of 4:2 to 5:2 (references cited i n Delattre and Hohenberg, 1968). Delattre and Hohenberg (1968) examined phonemic lengthening i n German when the vowel occurred i n unstressed position and conclu-ded that the "tense" (long) vowel i s always longer than i t s " lax" (short) counterpart even when the vowel i s i n unstressed position. Nooteboom (1972) studied Dutch vowels i n nonsense sy l l a b l e s and ex-amined the role of duration i n distinguishing between vowel pairs that were also accompanied by a difference i n qual i t y . I t was found that V:. . .on the whole subjects seem to adhere to surprising-l y fixed temporal patterns." (Nooteboom, 1972, p. 75). The role of phonemic vowel duration i n Czech has been investigated by Chlumsky (1928) and Janota and Jancak (1970). French i s not generally known as a language that makes extensive use of vowel length i n a d i s t i n c t i v e manner. However, i f one looks Hiachronically at the development of the French sound system, i t appears that certain vowel contrasts have arisen that are distinguished by a difference i n duration. These contrasts have arisen from the loss of a consonant or consonants i n the o r i g i n a l - 5 -Latin forms, or through analogy to these forms. e.g. L a t i n : magister ->• maiestre ( l l 1 " * 1 C.) ->- maitre ttl L a t i n : pasta -* paste (12 C.) -»• pate The disappearance of the consonant(s) toward the end of the four-teenth century resulted i n compensatory lengthening of the preceding vowel. The contrasts which arose were /a:/:/a/ and 'Xe:/:/e/. The poste r i o r i z a t i o n of the /a:/ phoneme was subsequent to i t s lengthen-ing and, according to Delattre, " I I semble...que, meme aujourd'hui, ce s o i t l a duree qui conditionne l e timbre et non 1'inverse." (De-l a t t r e , 1966, p. 107). The difference i n quality was mentioned for the f i r s t time i n 1753 by Boindin i n his Remarques sur les sons de  l a langue. Michaelis and Passy (1897), i n their Dictionnaire Pho- netique de l a Langue Francaise, indicate by means of phonetic tran-scriptions that length i s a distinguishing c h a r a c t e r i s t i c i n such pairs as "maitre:mettre" and "pate:patte" ( i n the l a t t e r case they indicate a qua l i t a t i v e difference as well) . Rousselot (1927)f1ques-tions the role of length i n distinguishing such pairs as "maitre: mettre" and "pate:patte". According to him, the difference i s a qua l i t a t i v e one (open:close). Nonetheless, he does admit that " l ' a de pate est plus long que l ' a de patte." (Rousselot, 1927, p. 90). Grammont (1933) also explains the difference between>/a:/: /a/ to be one of quality (posterior:anterior). He states further that i n certain cases the open /e/ may be long. Fouche (1958) claims that " . . . l e systeme qu a n t i t a t i f du francais est...essentiellement phonetique." (Fouche, 1958, p. 89) and does not regard length as being s i g n i f i c a n t i n French. Discussing phonemic vowel length i n - 6 -French, Malmberg (1954) says: Le nombre de cas on l a quantite seule distingue deux mots (bette :bete; renne: reine) est re s t r e i n t et le Francais n'est pas habitue a attribuer beaucoup d'importance aux variations de quantite vocalique (Malmberg, 1954, p. 92). Recent manuals of French pronunciation (Valdman e_t j a l . , 1964; Leon, 1966) indicate that the length difference i s s t i l l a factor i n distinguishing certain words. Delattre (1951) points out the length d i s t i n c t i o n and states that the additional length of the long vowel i n "maitre" or "pate" i s r e l a t i v e l y unstable. Ce supplement de duree est instable parce q u ' i l est independant de 1'in-fluence psycho-physiologique de l a con-sonne qui suit...C'est l e seul cas oii le signe de duree doive s'employer en francais (Delattre, 1951, p. 16). Experimental data to support the presence or absence of a d i s t i n c t i v e feature of length i n such pairs as "maitre:mettre" and "pate:patte" are scanty. In an early study (Delattre, ,1938), Delattre recorded a native speaker of French reading sentences con-taining the minimal contrasts mentioned and found that i n such mini-mal pairs as "maitre:metre" a measurable difference i n duration was made. Comparing the r a t i o of the vowel lengths of "maitre: metre" to those of "serre:sec", where the vowel duration i s phone-t i c a l l y conditioned, he found the f i r s t r a t i o to be much smaller. Delattre points out that whereas the phonetically conditioned length - 7 -difference i n the pair "serreisec" i s due to the influence of the following consonant, the phonemic d i s t i n c t i o n i n length i n "maitre: metre" i s consciously made (since the phonetic environments are i d e n t i c a l ) . Delattre argues that the /e/ of "maitre" must be a phoneme d i s t i n c t from the /e/ of "serre" which, despite a great d i f -ference i n length, belongs to the same phoneme as the /e/ i n "sec". As Delattre points out i n a l a t e r a r t i c l e (Delattre, 1964), the use of this length d i s t i n c t i o n varies according to the area of France from which the speaker comes and " . . . i t (the length d i s t i n c -tion) i s observed only i n careful d i c t i o n and by a minority of speakers." (Delattre, 1964, p. 99). Martinet carried out a survey i n 1941 (Martinet, 1945) concerning the status of the /a:/:/a/ and /e:/:/e/ di s t i n c t i o n s i n the speech of Parisian subjects. The method .of investigation took the form of a questionnaire i n which subjects were asked whether they made a difference between pairs such as "pate:patte". The study revealed that for the most part, Parisians distinguish be-tween "un 'a' et 'e' bref" and"'uun 'a' et 'e' long". Martinet remarks , however, that': En syllabe fermee, i l s opposent, pour l a plupart un 'e' bref a un 'e' long mais i l est evident que cette opposi-tion a de l a peine a se maintenir. Les sujets nes entre 1900 et 1910 confon-deht, dans l a proportion de pres de deux t i e r s , les mots ' f a i t e ' et 'fete'. La d i s t i n c t i o n entre les deux 'a' est uni-v e r s e l l e en syllabe tonique ouverte et tres generale dans les autres positions. I c i encore, beaucoup de sujets ne per-coivent pas de difference quantitative entre 'patte' et 'pate' (Martinet, 1949, p. 36). - 8 -A follow-up of Martinet's study was done i n 1956-57 by Reichstein (1960). Subjects were asked whether they made a difference between pairs similar to those used by Martinet. Subjects were also asked to read sentences containing the study words and the investigator indicated whether a difference i n length was heard or not. The subjects used were Parisian school g i r l s ranging i n age from 13-15 years. Results showed that the /a:/:/a/ d i s t i n c t i o n that was well established i n 1941 among adult Parisians was much less marked i n the 1956-57 study. Comparing Reichstein's results to those of Martinet, we find that i n 1941, the "paterpatte" d i s t i n c t i o n was made i n 100% of the cases while i n the 1956-57 study, between 46-49% of the subjects made a difference. Similar results were quoted for the /ei:,/:/e/ vowel contrast. Deyhime (1967) carried out a further study concerning the status of the length d i s t i n c t i o n i n French. The method of investigation again took the form of a questionnaire and sentence reading. Approximately 500 subjects were used and the subjects were from a l l parts of France (unlike the two previous studies). Deyhime found a difference between the vowel systems of subjects from the Midi and those from other parts of France. In the Midi the length oppositions under study were not apparent. He found that among non-meridional speakers ( i . e . not from the Midi region), the /a:/:/a/ d i s t i n c t i o n was present but the /e:/:/e/ d i s t i n c t i o n ap-peared with less regularity (used by only 44% of the non-meridion-a l speakers). A recent study by Leon (1972) dealing with the pronuncia-tion of accented /e/ among a group of young Parisians was under-- 9 -taken i n part to examine the status of the /e:/:/e/ length d i s t i n c -tion i n the speech of young French students. A group of t h i r t y -one boys, a l l born i n or near Paris was selected. The subjects were a l l fourteen years old at the time of the study. Subjects were recorded reading a l i s t of sentences containing the study words. Leon says that an ef f o r t was made to place the words i n i d e n t i c a l contexts; i . e . when;involving members of the same minimal pa i r . The test pairs used were: "bele:belle" "bete:bette" " l ' e t r e : l e t t r e " Results support those found by Reichstein and Deyhime. Leon found that the opposition i s jiust as unstable i n the "langue populaire" as i n the "langue cultivee". The surveys of Martinet, Reichstein and Deyhime r e l i e d on the speaker's a b i l i t y to analyze his own speech and the i n v e s t i -gator's s k i l l i n transcribing as accurately as possible what he heard. No instrumental analysis of the data was done. In the study by Leon, the investigator states that an attempt was made to mini-mize the difference between the sentence frames i n which members of the same pair appeared. However, from the examples given i n his study, i t i s evident that the sentences are not i d e n t i c a l : "C'est l'etre l e plus i d i o t . " and "C'est l a l e t t r e l a plus i d i o t e . " ; "Le mouton bele." and "Betty est b e l l e . " . Based on the results of the studies mentioned e a r l i e r concerning the influence of phonetic - 10 -environment, stress, number of syl l a b l e s i n the utterance, etc., i t would seem that d i f f e r i n g phonetic contexts could be a factor i n the neutralization of the length difference. I t appears then, that an investigation to determine the status of the length contrast i n French i s needed i n which vowel durations are measured instrumentally rather than subjectively and where external factors influencing vowel duration are controlled. 2.3 Vowel Duration: Perception Relatively few studies have been done concerning percept tion of duration of speech or non-speech s t i m u l i . A study by Creel-man (1962) examined the a b i l i t y of human observers to discriminate differences i n duration between short auditory signals. The l i s -teners were presented with a 1000 Hz tone i n a white noise back-ground. The duration and int e n s i t y of the tone were varied and the listeners were asked to compare i t to a tone of fixed length and inte n s i t y . The results of Creelman's study indicate that the per-ception of time i s an independent process, not dependent on such Sectors as in t e n s i t y . Duration discrimination depends on suf-f i c i e n t i ntensity to mark the time un-ambiguously; i t depends on d e t e c t i b i l i t y but not on loudness (Creelman, 1962, p. 592). Creelman proposed a model for the perception of duration based on the results of his study. In his model, human observers use - 11 -...a separate and independent mechanism to measure short durations. This mechan-ism functions by 'counting' input pulses during the duration to be judged ( i b i d . ) . The counting mechanism would be analagous to an accumulator which could store neural pulses. Abel (1972a) examined l i s t e n e r s ' a b i l i t y to discriminate between a pair of f i l l e d intervals of different durations and i n a l a t e r study (Abel, 1972b) looked at l i s t e n e r s ' a b i l i t y to judge d i f -ferences i n duration of s i l e n t gaps between tone bursts. Abel also found discrimination of duration to be independent of the intensity of the stimulus. Her results support Creelman's neural counter model for the perception of duration. The study of perception of non-speech s t i m u l i appears to be related to temporal discrimination i n speech perception, where cues to meaning i n some languages depend on r e l a t i v e duration. Perception of duration of speech s t i m u l i has been studied i n d i r e c t l y by Denes (1955), who examined the effect of vowel dura-tion on perception of voicing of the following consonant, and by Fry (1955), who investigated the effect of vowel duration on the perception of stress i n English. In the f i r s t study, vowel lengthen-ing resulted i n a greater number of judgements of a following voiced consonant. In the second study, vowel lengthening played a major role i n judgements of stress.-Discussing the perception of vowel duration, Lehiste (1972) suggests that the phonetic image of the vowel f a l l s along a continuum and the l i s t e n e r "locates" the stimulus on the continu-- 12 -um by reference to certain a r t i c u l a t o r y target points stored i n memory. She states further that a l i s t e n e r i s not only able to compare the duration of two sti m u l i but also can match the stimu-l i with some kind of inte r n a l "durational image". This notion i s based on Lehiste's work i n 1970 on Estonian i n which she found that l i s t e n e r s agreed consistently i n assigning l i n g u i s t i c labels to stimuli- d i f f e r i n g only i n the duration of a vowel or consonant. In languages using d i s t i n c t i v e quantity, abrupt changes i n manner of a r t i c u l a t i o n serve as reference points with regard to timing judgements. This idea supports the notion of p a r a l l e l processing i n speech perception. Whatever the process by which the dura-tion of one segment i s compared with that of another (or with a stored 'durational image') i t can very w e l l take place at the same time as the cues for the point of a r t i c u l a t i o n are extracted from the same acoustic signal (Lehiste, 1972, p. 7). Nooteboom (1972) also postulates an i n t e r n a l representa-tion of vowel and s y l l a b l e duration which i s part of an in t e r n a l representation language users have of how the words of their own language should sound. In his study, subjects were asked to ad-just the vowel duration of a nonsense word according to some i n -ternally generated c r i t e r i o n , such that the word sounded as natur-a l as possible. His results suggest that duration discrimination may ;be considerably better than the duration of one cycle of the vocal folds, or approximately 8 msec. - 13 -In an adjustment experiment i n which three phonetically non-naive subjects were asked to keep an i n t e r n a l l y gen-erated c r i t e r i o n for the duration of a stressed s y l l a b l e nucleus...as con-stant as possible i n successive set-tings, i t was found that this could be done with a high degree of accuracy. The i n t e r n a l representation of a s y l -lable nucleus duration may be more ac-curate than the spectrographic measure-ment of i t s acoustic correlate (Nooteboom, 1973, p. 42). In a study by Huggins (1972), sentences containing-words with a l -tered speech segment durations were presented to l i s t e n e r s who were asked to indicate whether the vowel or consonant i n question was "normal", "short", or "long". Results indicate that a difference i n vowel length of 20 msec, could have perceptual relevance. Rossi (1972), using isolated vowels, attempted to specify the value of threshold for r e l a t i v e duration for the vowel /a/. Spectrum, fun-damental frequency and intensity were held constant. Subjects were required to compare a fixed stimulus duration to variable stimulus durations. Results indicate that the jnd (just noticeable difference) or DL (difference limen) for duration increases s i g n i f i c a n t l y as the fixed duration increases. Between 60 and 140 msec, Rossi found the DL to be close to 30 msec. Between 130 and 290 msec, the DL was found to be a constant percentage ('22.5%) of the fixed dura-tion. Delattre and Hohenberg (1968) studied the perception of vowel duration i n German, s p e c i f i c a l l y the duration of unstressed vowels. Synthetic speech was used and vowel durations were varied without varying vowel quality. The results of the investigation - 14 -show that there i s a sharp boundary between judgements of "tense" and "lax" vowels. When a difference i n vowel quality accompanied the length difference, i t was found that duration was the more im-portant cue i n distinguishing the members of the pai r . In a similar study on the perception of vowel duration i n Swedish by Hadding-Koch and Abramson (1964), three vowel pairs (longrshort) were selected for study. The vowels of the f i r s t pair were very si m i l a r i n quality and were distinguished largely by a difference i n duration. The second pair was characterized by a marked difference i n quality as wel l as the difference i n duration. The t h i r d pair was chosen to exemplify a moderate q u a l i t a t i v e d i f -ference along with the duration difference. Recorded speech was used and the long vowel of the minimal pair was shortened i n degrees by means of tape cutting and s p l i c i n g u n t i l i t equalled the length of i t s short counterpart. The results show that i n the case where a difference i n duration was the only distinguishing feature, shor-tening the long vowel of the pair showed a constant decrease i n "long" judgements. When the o r i g i n a l long vowel reached the same duration as i t s short counterpart, subjects were unanimous i n as-signing the "short" member of the minimal pair to the stimulus. In the second case where a difference i n vowel duration was accom-panied by a s t r i k i n g difference i n vowel spectrum, shortening the long vowel did decrease the number of "long" judgements but only by approximately £5%. In the th i r d case where a difference i n vowel length was accompanied by a s l i g h t difference i n vowel spectrum, results were very similar to those of the f i r s t vowel pai r . The conclusion arrived at was that i n the f i r s t and t h i r d cases, the - 15 -vowels are distinguished by a feature of length, while i n the second case the two phonemes are distinguished by a q u a l i t a t i v e difference. The perception of length as the primary cue distinguish-ing the pairs /a:/:/a/ and /ea/:/e/ has yet to be investigated i n French. I t would be interesting to examine whether length i s an important perceptual feature for French l i s t e n e r s i n distinguishing such pairs as "maxtre:mettre" and "pate.patte" or whether as Malm-berg says: . . . l e Francais n'est pas habitue a attribuer beaucoup d'importance aux variations de quantite vocalique (Malmberg, 1954, p. Chapter 3 AIMS OF THE INVESTIGATION From the l i t e r a t u r e discussed i n the previous chapter, i t i s apparent that most of the studies dealing with vowel length i n French have taken the form of surveys. L i t t l e instrumental data have been gathered since Delattre's i n i t i a l investigation i n 1938. The present study consists of two experiments which ex-amine the use of phonemic vowel length by a small group of native speakers of French i n such minimal pairs as "maitre:mettre" and "paterpatte". 3.1 Experiment I Experiment I examines vowel duration from the point of view of production to determine i f the native speakers selected make a measurable and/or s i g n i f i c a n t d i s t i n c t i o n i n length between members of such pairs as "maitrermettre" and "paterpatte" when these words appear i n the same frame sentence. I t examines vowel dura-tion when the study words appear 'dm. both stressed and unstressed positions (in this study stress refers to unemphatic stress i n French which f a l l s on the l a s t s y l l a b l e of a rhythmic group''") . A rhythmic group i s usually composed of a noun phrase or a verb phrase, more rarely of a whole sentence. - 17 -3.2 Experiment I I The second experiment examines the perception of d i f -ferences i n vowel duration and i t s use i n distinguishing between members of a minimal pair.I The s p e c i f i c questions raised are: a) . I f the vowel i n "maitre" (or "pate":)!' i s . shortened, does the l i s t e n e r perceive the word as "mettre" (or "patte")? I f so, by how much must the vowel be shortened i n order to cause this re-versal i n the listener's perception of the word? b) . I f the vowel i n "mettre" (or "patte") i s lengthened, does the l i s t e n e r perceive the word as "maitre" (or "pate")? I f so, by how much must the vowel be lengthened i n order to cause this reversal i n the listeners's perception of the word? In addition, both experiments examine the r e l a t i o n be-tween the speaker/listener's concept of his production and percep-tion of the length difference and his actual performance. Chapter 4 EXPERIMENT I 4.1 Experimental Apparatus and Procedures 4.11 Corpus Seventeen minimal pairs of words were selected for the study. Selection of the pairs was based on the phonetic transcrip-tio n given i n the Micro Robert: dictionnaire du franqais primordial. Ten of the pairs exhibited the /a:/:/a/ contrast and seven the /e:/: Izl contrast. The words chosen were monosyllabic i n a l l but one case ("hale:alle") and were closed s y l l a b l e words for a l l but three pairs ("las :1a", "mat:ma", "bas:bat"). Two l i s t s of sentences were constructed. L i s t 1 consisted of ordinary French sentences. The study words appeared i n sentence f i n a l position except i n the case of an open s y l l a b l e word pair. Both members of a minimal pair of words appeared i n the same frame sentence but the sentences varied between pairs. As i t was not possible to find suitable sentence frames for the pairs "mat:ma", "bas:bat", "bele:belle", they were omitted from the f i r s t l i s t . L i s t 2 consisted of <a single frame sentence. The frame sentence was chosen based on the results of a p i l o t study (the subject, a native speaker of French from St. Pierre and Miquelon, was recorded reading both l i s t s ) . The frame sentence selected was " I I a prononce l e mot ' 1 plus f o r t . " Each l i s t began and ended with two buffer sentences. - 18 -- 19 -4.12 Subjects Twelve native speakers of French, s i x men and s i x women were subjects for the experiment. They were a l l born and raised i n France and represented the following'regions: Subject 1: Grenoble Subject 2: A l b i Subject 3 & 9: Quimper Subject 4 & 11: Paris Subject 5: Marseille Subject 6 & 7: Toulouse Subject 8: Jura Subject 10: Bordeaux Subject 12: Limoges A l l subjects except Subjects 11 and 12 taught French or l i n g u i s t i c s . Subjects had no p r i o r knowledge of what was being examined i n the experiment. 4,13 Instrumentation and Arrangement The subject stood four to s i x inches from an Altec 681A L0 microphone which was placed on a boom i n an IAC 1204 sound-proof room. A Scully tape recorder Model 280-2, situated outside the booth, was placed such that the examiner (who was i n the sound-proof room) could observe the VU meter through the double glass window. The tape movement was controlled by the examiner by means of a Scully remote control box. - 20 -4.14 Experimental Design The subject was given written instructions i n French explaining that the recording involved two l i s t s of sentences and that the sentences should be read at a normal speaking rate and with the same normal intonation, i . e . with f a l l i n g intonation at the end of each sentence. The subject was asked to read through the l i s t s of sentences to f a m i l i a r i z e himself with them. Practice reading of the f i r s t few sentences was allowed to determine i f further instructions were needed. L i s t 1 and L i s t 2 were then recorded. L i s t 2 was recorded a second time and the subject was asked not to stress the word i n quotations or to make a pause be-tween i t and the following phrase "plus f o r t " . Again, the sub-ject was permitted to practise reading the sentences i n th i s man-ner before being recorded. After recording the l i s t s , the subject was asked what he thought was being studied and whether or not he thought he made the length d i s t i n c t i o n i n question. The subject was also asked i f he f e l t he had read the sentences i n a natural manner. 4.2 Analysis of the Data 4.21 Instrumentation Mingograms were made of L i s t 1 and the second recording o f ' L i s t 2 for each speaker. Using a Siemens Oscillomink graphic recorder, f i v e signals were displayed on the mingograms: - 21 -1. Microphone signal (channel 1) 2. Duplex oscillogram (channel 2) 3. Fundamental frequency (channel 3) 4. Log of average speech power (channel 4) 5. Average speech power (channel 5) The tape recorder used i n making the mingograms was a Revox Model A77. The duplex oscillogram and fundamental frequency tracings were produced v i a a Frcjikjer-Jensen transpitchmeter• Speech power was recorded by means of a spee'eh power detector si m i l a r to the one developed by Peterson and McKinney (1961). Some spectrograms were made on a Kay Sonagraph Model 7029A to aid i n segmentation of the utterances. 4.22 C r i t e r i a for Determining Vowel Boundaries 1. Vowel Onset a)'. When preceded by a stop consonant, the beginning of the vowel was ea s i l y determined by examining the microphone and duplex oscillogram tracings. Vowel beginning was measured immediately after theeburst following stop closure. b) . When preceded by a f r i c a t i v e , the beginning of the vowel was measured at the point where the duplex oscillogram t r a -cing associated with the f r i c a t i v e returned to the zero l i n e (see Fig. 1). c) . When preceded by a nasal, the beginning of the vowel was measured at the offset of the cha r a c t e r i s t i c microphone and duplex oscillogram tracings associated with nasals (see Fig. 2). f F i g . 1 Mingogram of the word " f a i t e " . - 24 -d) . When preceded by /l/, the beginning of the vowel was determined by examining the speech power tracing i n conjunc-tion with the microphone and duplex oscillogram tracings (see Fig. 3). e) . When preceded by a period of silence, vowel begin-ning was e a s i l y determined at the point of voice onset measured on the fundamental frequency tracing. 2. Vowel Offset a) . When followed by a stop consonant ( i n this study a l l postvocalic stop consonants i n question are v o i c e l e s s ) , the end of the vowel was easily determined at the point of voice offset measured on the fundamental frequency tracing or by examination of the microphone and duplex oscillogram tracings. b) . When followed by a f r i c a t i v e , vowel offset was mea-sured at the point where the duplex oscillogram tracing returned to the zero l i n e before proceeding negatively for the f r i c a t i v e (see Fig. 4). c) . When followed by /l/, the end of the vowel was de--' termined by a dip i n the speech power and duplex oscillogram t r a -cings. In a few d i f f i c u l t cases, spectrograms were made. d) . When followed by a nasal, vowel offset was measured at the onset of the microphone and duplex oscillogram tracings associated with nasals. In a few d i f f i c u l t cases, spectrograms were made. e) . When followed by a period of silence, the end of the vowel was measured at the point of voice offset measured on the Fig. 3 Mingogram of the word " l a " . Fig. 4 Mingogram of the word "tache". - 27 -fundamental frequency tracing. 4.23 Determination of Average Measurement Error Five utterances were selected at random from the two l i s t s . Each of these utterances was segmented and vowel durations were measured for the second time. This was done for a l l twelve subjects. The second duration measurement was compared to the f i r s t and, using the formula _1 n n E i = l ( X i " V where X. i s the value of the f i r s t measurement for x . th _ the 1 word, Y. i s the value of the second measurement for l the i^1 word, and n i s the number of words measured (in this case n=60) the average error was calculated and found to be 3.71 msec. - 28 -4.3 Results 4.31 Averages of ''Long11 arid ''Short'1 Vowel Durations Average vowel durations were calculated for the follow-ing four categories':' 1. "long" /a:/ ( L i s t 1 "short" /a/ (L i s t 1 2. "long" "short" 3. "long" "short" 4. "long" "short" /a:/ ( L i s t 2 /a/ ( L i s t 2 Iz-.l ( L i s t 1 Izl ( L i s t 1 /e:/ ( L i s t 2 /e/ ( L i s t 2 for each speaker..'.'/ A treatments-by-subjects analysis of variance was run to determine i f the difference between the "long" and "short" vowels i n each of the above categories was s i g n i f i c a n t . Levels of s i g n i f i -cance are indicated below. Category 1 p < .001 Category 2 p < .001 - 29 -Category 3 p < .005 Category 4 p < .01 For this group of speakers i t appears that there i s a s i g n i f i c a n t length difference made between the members of such pairs as "maitre: mettre" and "pate:patte" when the words appear i n either stressed of, unstressed position. 4.32 Average "long":"Short" Vowel Ratios In an early study (Delattre ,ill938) , Delattre found that when the length d i s t i n c t i o n was made, the "long":"short" vowel ra-tios were close to 1.60. In the present study, average "long":"short" ratios were calculated for each speaker and are presented i n Table 1. In a l l cases they were found to be less than Delattre's r a t i o of 1.60. 4.33 Comparison of /aa/:/a/ with / e £ ' / : / e / In the three survey studies mentioned e a r l i e r (Martinet, 1945; Reichstein, 1960; Deyhime, 1967), i t was found that the /a:/: '/a/ d i s t i n c t i o n was more frequently maintained than the / e : / : / e / d i s t i n c t i o n . A t-test was run for each subject to determine i f there was a s i g n i f i c a n t difference between the mean ratios of /a:/:/a/ and / e : / : / e / . None of the speakers showed a s i g n i f i c a n t difference be-tween the two vowel pairs. - 30 -TABLE 1 Average "Long":"Short" Vowel Ratios for Each Subject. L i s t 1 L i s t 2 s l 1.23 1.22 S2 4723 1,'15 S3 1.36 1.30 S4 1.45 1.03 S5 1.18 1.09 S6 1.08 1.01 S7 1.03 1.03 S8 1.38 1.29 S9 1.16 1.11 S10 1.27 1.12 S l l 1.07 1.12 S12 1.17 1.11 - 31 -4.34 Stressed vs. Unstressed P o s i t i o n I d e a l l y the two l i s t s would have shown the study words i n (unemphatically) stressed ( L i s t 1) and unstressed ( L i s t 2) po-s i t i o n s . However, due to the frame sentence chosen for L i s t 2 ("II a prononce l e mot ' ' plus f o r t . " ) , c e r t a i n subjects had a tendency to emphasize the word i n quotations despite having been asked not to. Examination of the data reveals that f o r 10 of the 14 word pairs common to both l i s t s , a greater percentage of speakers made a dif f e r e n c e when the pair appeared i n L i s t 1 than when i t appeared i n L i s t 2 (see Table 2). When a length d i f f e r e n c e was made i n L i s t 2, the "long":"short" r a t i o s were generally reduced (only one subject, S^^, had a greater average r a t i o f or L i s t 2 than for L i s t 1. This subject f e l t she made no d i s t i n c t i o n i n her speech between the word pairs under consideration), t - t e s t re-su l t s i n d i c a t e , however, that there i s no s i g n i f i c a n t d i f f e r e n c e between the mean r a t i o s for the two l i s t s . 4.35 Speaker Differences Comparing the mean "long" to the mean "short" vowel dura-tions, i t was found that f o r eight speakers, the "long" average dura-tion was longer than the "short" average duration i n each of the four categories mentioned i n section 4.31. Three of the four speakers who did not have greater average "long" than average "short" values i n a l l fodr categories were from the Midi region of France. This observation tends to support Deyhime's f i n d i n g (Deyhime, 1967) that the speakers from the Midi region of France make the length d i s t i n c -s„ s, 1 2 3 4 5 6 7 8 9 10 11 12 ane:Anne ++ ++ . ++ +_ — _+ ++ l a s : la. ++ +_ ++ +_ ++ ++ +_ ++ ++ — ++ paterpatte ++ ++ ++ ++ ++ -+ — ++ ++ ++ +_ ++ Bale:balle ++ ++ ++ ++- ++ +_ +_ ++ ++ ++ ++ ++ mate:mate ++ ++ ++ =+ -+ — ++ ++ +- -+ — _+ tache:tache ++ _+. +- ++ ++ .+- ~ ++ ++. +_ +_ ++ h a l e : a l l e ++ +- ++ +_ +_ _+ ++ ++ ++ _+. +-male:malle ++ ++ ++ ++ ++ +_ — +_ ++ ++„ ++ +-reine:renne ++ — ++ +- -+ — ++ ++ ++ +- +.+ f e t e : f a l t e ++ _+ +- +- +- — — ++ -+ -+ . *+ a i l e : e l l e ' +- ++ +_ +_ ++ — — ++ -+ ++ ++ ++ l ' e t r e : l e t t r e ++ ++ ++ ++ ++ ++ i+ ++ ++ ++ . ++ +* maitre:mettre ++ ++ ++ ++ ++ +_ +_ ++ +- ++ ++ ++ bete:bette ++ ++ +- +- -+ ++ +_ ++ ++ +_ +-bele:belle + + + - + - - + + + + + bas:bat + + - - - - + - - + + -mat:ma + + + + + -TABLE I I Presence or Absence of Length D i s t i n c t i o n Be--jt E c y K tween "Long" and "Short" Members of Each Pair for Individual Subjects. F i r s t entry corresponds to L i s t 1 Second entry corresponds to L i s t 2 '4v +== "long":"short" vowel ratios > 1 - = "long":"short" vowel ratios < 1 Grenpble A l b i Quimper Paris Marseille Toulouse Toulouse Jura Quimper Bordeaux Paris Limoges '10 '11 '12 ho - 33 -tion less frequently than speakers from other areas of France. No s i g n i f i c a n t difference between meridional and non-meridional subjects could be found by means of t-tests for the following four categories: 1. mean ratios ( L i s t 1) 2. mean ratios ( L i s t 2) 3. mean no. of ratios > 1 (Li s t 1) 4. mean no. of ratios > 1 ( L i s t 2) 4.36 Speaker's I n t u i t i o n vs. Speaker's Performance A t-test was run to determine i f there was a s i g n i f i c a n t difference between subjects who thought they made the length d i f -ference and those who thought they did not (subjects who were un-sure, or who f e l t they might make a difference i n some cases were excluded from the t e s t ) . The same categories as mentioned i n sec-tion 4.35 were examined and the levels of significance for each category are shown below. Category 1 p < .01 Category 2 NS 1 Category 3 p <<.05 Category 4 NS "Not s i g n i f i c a n t Chapter 5 EXPERIMENT I I 5.1 Experimental Apparatus and Procedures 5.11 Preparation of the Test Tape Selection of the Recording. L i s t 2 recorded by one of the sub-jects for the f i r s t experiment was selected for use i n the second part of this study. Selection of the recording was based on the voice quality of the speaker, r e l a t i v e n e u t r a l i t y of accent, i . e . r e l a t i v e closeness to the so-called "standard" French pronunciation', and the presence of a length difference between members of a mini-mal word pair. Selection of Word Pairs. Nine minimal pairs of words were selec-ted, f i v e exhibiting the /ac,/:/a/ contrast and four the /e:/:/e/ contrast. The pairs were selected on the basis of the results of Experiment I. Cr i t e r i o n for selection was a minimum of eight of the twelve speakers with "long"."short" vowel ratios greater than 1.00 for the minimal pair i n question. The pairs selected were: "Bale:balle" "pate:patte" "male:malle" "las :1a" '"tache:tache" - 34 -- 35 -"maitre:mettre" "bete.bette" "bele.belle" " l ' e t r e . l e t t r e " Vowel Editing. The selected words were spliced from the ca r r i e r phrase " I I a prononce l e mot ' ' plus f o r t . " from a dubbing of the o r i g i n a l tape. To ensure that the isolated words had not been clipped or distorted as a result of s p l i c i n g and dubbing, mingo-grams were; made of the words from the spliced tape and compared with mingograms of the entire sentence taken from the o r i g i n a l tape. An int e n s i t y scale was established to convert centimeters (measured on the speech power tracing on the mingogram) to deci-bels. Differences i n vowel intensity between members of the same minimal pair were measured on the mingogram and converted to deci-bels using the prepared scale. Discrepancies i n int e n s i t y were adjusted with an attenuator. Mingograms were made of the in t e n s i t y -adjusted words and differences i n intensity were found to be 1 dB or less between the members of a minimal pair of words. Using a PDP-12 computer and the program WAVES developed by Lloyd Rice, the isolated stimulus words were d i g i t i z e d and stored on computer tape. The set of programs WAVES d i g i t i z e s the speech signal and permits the speech waveform, which can be displayed on the computer's oscilloscope screen, to be altered by means of adding or removing segments of the speech wave. The computer i s able to sample the speech signal at a rate of up to 12,000 samples per se-- 36 -cond. This maximum sampling rate, however, i s effe c t i v e only for signal durations of less than approximately 1.5 seconds. For s i g -nals longer than t h i s , the computer would be unable to transfer samples fast enough, due to the slow rate of transfer to the tape, and would begin to skip samples once the core buffer had been f i l l e d . The skipping of samples would result i n a noticeable d i s t o r t i o n of the s i g n a l , therefore, the audio tape containing the nine word pairs was played at half speed. This signal was sent through a Rockland Programmable Dual Hi/Lo f i l t e r Model 1520 with the high frequency cutoff set at 3,000 Hz. The program's sampling rate was set at 6,000 samples per second, which would be equivalent to 12,000 samples per second i f the tape were played back at normal speed. Once the speech wave had been stored on computer tape i n this manner, i t was then possible to view i t on the oscilloscope screen. The display of the speech wave could be moved back and forth across the computer's screen by means of a knob which was manipulated rbyithe operator. The program allowed the operator to lengthen or shorten the vowel waveform by repeating or removing the number of cycles desired. The edited word was then stored on a separate computer tape. Vowel waveforms were lengthened or shor-tened at a point i n the waveform, steady i n terms of amplitude and shape, and the point of editing was always marked at a zero-cros-sing between two cycles of the fundamental ( i f the vowel was not spliced at the zero-crossing, noticeable d i s t o r t i o n resulted). I t was calculated that each period i n the vowel waveform corresponded to an average of 8 msec, for the speaker chosen. The - 37 -vowel waveform of each word was altered i n f i v e steps of two cycles ( i . e . 16 msec,,).-, for a t o t a l range of 64 msec, between the longest and shortest version of the edited word. Ninety stimulus words were prepared i n this manner (fi v e different versions of both mem-bers of the nine minimal pairs of words). Using the program WAVES, the d i g i t i z e d signal was converted to an analogue s i g n a l , fed back through the high-pass f i l t e r set at 6,000 Hz and recorded on audio tape using a Revox Model A77 tape recorder. The stimulus words were recorded i n random order with the condition that s t i m u l i from the same word or the same word pair were not i n sequence. Six buffer words selected from the ninety test s t i m u l i were recorded, three at the beginning and three at the end of the tape. The i n -ter-stimulus i n t e r v a l was fixed at f i v e seconds (the program WAVES allows the operator to set the desired number of seconds that are to precede and follow each output segment). 5.12 Subjects Twelve native speakers of French served as subjects for the second experiment. In a l l but one case they were the same sub-jects used i n the f i r s t experiment. 5.13 Test Procedure and Equipment The test tape was played back on a Revox Model A77 tape recorder through Sharpe HA10 headphones at a l e v e l corresponding to 50-60 dB SPL measured through TDH-39 Maico Headphones with a Briiel and Kjaer 2203 precision sound l e v e l meter, a Briiel and Kjaer 4152 3 a r t i f i c i a l ear and a 6cm coupler. The subject was seated i n a - 38 -quiet room and was asked to underline on the answer sheet provided the word he heard. For each stimulus presentation the response sheet offered a choice of the "long" and "short" version of the word. 5.2 Results In order to analyze the results of the perception test a tabulation of the l i s t e n e r s ' responses was made i n the following manner: The random l i s t of test responses was reordered so that the fi v e versions of each word were grouped together. Within each group of f i v e , the words were ordered according .to vowel duration, i. e . from longest to shortest. Each l i s t e n e r ' s response to a stimu-lus was c l a s s i f i e d as either "long" or "short" and entered opposite the corresponding word i n the reordered l i s t . A l l subjects' res-ponses for that word were then examined and converted to a percen-tage (out of twelve) based on the number of li s t e n e r s who heard that word as the o r i g i n a l unedited version. The percentages were then plotted as a function of vowel duration measured i n m i l l i s e -conds (see Fig. 6 to Eig>14). In some cases the end points of the curve for the "long" version and the curve for the "short" version of a pair are not aligned. This i s due >tQ> the fact that the differences between the vowel durations of the o r i g i n a l "long" and "short" vowels were, i n these cases, more than the 64 msec, range, e.g. the vowel durations of "pate" and "maitre" exceeded those of "patte" and "mettre" by 100 msec. - 39 -The shape of the curves i n F i g . 6 to F i g . 14 i n d i c a t e that the l i s t e n e r s ' judgements were indeed af f e c t e d by the a l t e r e d vowel length of the o r i g i n a l word. I f the l i s t e n e r were basing h i s judgement s o l e l y on the feature of vowel length, however, one would expect the two curves to be the inverse of each other as i n the f i g u r e below. vowel duration i n milliseconds F i g . 5 L i s t e n e r s ' judgements based on vowel duration only. F i g . 6 to F i g . 14 i n d i c a t e that while length does appear to be af-? f e c t i n g the l i s t e n e r ' s judgements, other factors are i n f l u e n c i n g h i s decisions as w e l l . Fig. 8 Listeners' judgements of o r i g i n a l and edited versions of "male" and "malle". 100 80 60 A 3 i>5 40 \ \ \ \ \ til — l a las 150 160 — — 170 220 230 180 190 200 210 vowel duration i n milliseconds Fig. 9 Listeners' judgements of o r i g i n a l and edited versions of " l a s " and " l a " . 240 vowel duration i n milliseconds Fig. 11v Listeners' judgements of o r i g i n a l and edited versions of "maitre" and "mettre". 100 100 - 49 -5.21 Vowel Lengthening vs. Vowel Shortening The results of the experiment indicate that the l i s t e n -er's judgement of a lengthened "short" vowel d i f f e r s from his judge-ment of a shortened "long" vowel. For a l l the word pairs i n the study (with the exception of "las :1a" and "tache:tache") a smaller s h i f t i n vowel duration was required to reverse the listener's judgement of the o r i g i n a l "short" version of the word than to re-verse his judgement of the o r i g i n a l "long" version. This s h i f t i n duration i s taken to be the difference i n milliseconds between the vowel duration of the o r i g i n a l word and the vowel duration correspon-ding to the 50% response point on the graph. The 50% response point i s considered the category boundary between the "long" and "short" vowel. The average number of additional milliseconds needed to re-verse the lis t e n e r ' s judgement of the o r i g i n a l l y "short" version of the word (to the 50% response point) was 32 msec. The average number of subtracted milliseconds needed feo reverse the li s t e n e r ' s judgement of the o r i g i n a l l y "long" version of the word was 51 msec. Comparing the effect of maximum lengthening and maximum shortening of the vowel on the li s t e n e r ' s judgement, i t was found that when the o r i g i n a l l y "short" vowel was maximally lengthened ( i . e . by 64 msec), only 19% of the li s t e n e r s s t i l l heard i t as "short". On the other hand when the originally ?-:"long" vowel was maximally shortened ( i . e . by 64 msec), 41% of the list e n e r s s t i l l heard i t as "long". - 50 -5.22 .Other Factors Influencing Listener Judgement Spectrograms were made of the o r i g i n a l word pairs ( i . e . before vowel editing) and the f i r s t three formant frequencies were measured. Differences i n formant frequency between the steady-state portion of the /e :/ :/e/ word pairs were found to be non-sig-n i f i c a n t . However, the vowels i n "maitre" and " l ' e t r e " were ac-companied by a noticeable glide (the glide was remarked on by sever-a l of the l i s t e n e r s ) . Differences i n formant frequency between the /a:/:/a/ word pairs were measured and i t was found that, with the exception of "las :1a", the second formant of the "short" version was consis-tently higher than the second formant of the "long" version. This observation i s i n keeping with the articulatory correlate of "an-t e r i o r " (corresponding to a high second formant) vs. "posterior" (corresponding to a low second formant). The greatest difference between the second formants of the "long" and "short" versions was found for the pair "ma<le :malle" (a difference of 150 Hz was mea-sured) . This larger vowel quality difference could account for the i r r e g u l a r i t y of the curve representing the l i s t e n e r s ' judgement of-the edited versions of "male" (see Fig. 8). It appears, therefore, not too su r p r i s i n g l y , that when the vowel pair exhibits a difference i n formant frequency, or when the "long" vowel i s accompanied by a glid e , shortening the o r i g i n a l -l y "long" version has a smaller reversal effect on the l i s t e n e r s ' judgement of the word than does lengthening the o r i g i n a l l y "short" version. Similar results were found by Hadding-Koch and Abramson - 51 -(1964) for Swedish. In their study, they found that when a vowel pair was differentiated only by a contrast i n duration, shortening the "long" version had a great effect i n reversing the l i s t e n e r s ' judgement of the word. When the vowel pair was accompanied by a marked difference i n vowel quality as well as a difference i n dura-ti o n , the effect of shortening the "long" version had a much small-er ^reversal effect on the l i s t e n e r s ' judgement of the word (see Chapter 2, section 2.3 of the present study). 5.23 Listener's I n t u i t i o n vs. Listener's Performance A l l of the subjects f e l t that the task required, i . e . assigning a p a r t i c u l a r l e x i c a l item to the stimulus, was a d i f -f i c u l t one and several subjects f e l t they were guessing part/,of the time. At the end of the test, subjects were asked what they f e l t they were basing th e i r judgements on. Half of the speakers f e l t their judgements were based on vowel length and the other half said they thought their judgements were based on quality or on both quality and length. Examination of the test responses re-vealed no difference between the three groups. Chapter 6 SUMMARY AND DISCUSSION 6.1 Vowel Duration: Production The f i r s t part of the study examined the use of vowel length by a small group of native speakers of French i n the pro-duction of such pairs as "maitre:mettre" and "pate:patte". Sever-a l recent a r t i c l e s (Delattre, 1966; Deyhime, 1967, Leon, 1972) sug-gest that the length d i s t i n c t i o n i s maintained infrequently i n con-temporary French. The re'sults of the present study, however, i n d i -cated that, for the population selected, the "long" vowel was s i g -n i f i c a n t l y longer than the "short" vowel when the words appeared in the same frame sentence i n both (unemphatically) stressed and unstressed positions. The "long":"short" vowel ratios were not found to be so large as those measured by Delattre (Delattre, 1938). No s i g n i f i c a n t difference was found between the "long": "short" ratios for the /a:/:/.a/ contrast and those for the /e:/: Izl contrast. This finding does not support DeyKime's suggestion (Deyhime, 1967) that the /a:/:/a/ contrast i n length i s maintained with a greater degree of consistency than the /e :/ :/e/ contrast. Another tendency observed from the experimental results i s that the speakers from the Midi region of France tended to make a difference i n length less frequently than speakers from other areas. The difference between the vowel ratios for the two groups, however, was not found to be s i g n i f i c a n t . - 52 -- 53 -A difference was observed between the vowel ratios of L i s t 1 and L i s t 2. When the words appeared i n (unemphatically) stressed position ( L i s t 1 ) , the "long":"short" ratios were more frequently larger than unity than when they appeared i n unstressed position ( L i s t 2). Also, the ,;iilong'.fe"short" vowel ratios were generally larger when the words appeared i n stressed position but these observed differences were not found to be s t a t i s t i c a l l y s i g -n i f i c a n t . A comparison was made between the speakers who claimed they made a difference and those who claimed they did not. When the study words appeared i n stressed position, a s i g n i f i c a n t d i f -ference i n "long":"short" vowel r a t i o s was found between the two groups of speakers. No s i g n i f i c a n t difference was found between the two groups when the words appeared i n unstressed position. The present study concerning the use of vowel length i n French has several l i m i t a t i o n s . As the population examined was r e l a t i v e l y small, i t i s not possible to draw any conclusions about the status of the length d i s t i n c t i o n among French speakers i n gen-e r a l . The speakers selected were a l l highly educated and thus re-presented only one s o c i a l group. In future studies i t would be interesting to examine the use of vowel length, using a si m i l a r method of investigationtj';. by a larger group of speakers including subjects from d i f f e r i n g s o c i a l and educational backgrounds. A further l i m i t a t i o n of the present study i n determining the use of vowel length by the selected group of speakers was the r e l a t i v e l y a r t i f i c i a l atmosphere of the recording session. A l -though the subjects were asked to read the sentences inlia natural - 54 -manner, many commented that they found i t d i f f i c u l t to be com-ple t e l y natural and awareness of their pronunciation was height-ened. This factor could have affected their production of the length contrast since Delattre observed that i t i s maintained only i n " c a r e f u l - d i c t i o n " (Delattre, 1964, p. 99). However, since the subjects were unaware of what was being examined i t i s unlikely that they were overly conscious of their production of the d i s -t i n c t i o n i n length between such pairs as "maitre:mettre" and "pate: patte". I f more time had been available, i t would have been i n -teresting to record each subject on several occasions i n order to examine the consistency of the length contrasts and the consistency of the "long":"short" vowel r a t i o s . In order to obtain more conclusive results concerning the status of the length d i s t i n c t i o n i n French, future studies might benefit from using a^controlled recording procedure and ob-j e c t i v e method of analysis, s i m i l a r to that u t i l i z e d i n the pre-sent study, on a larger, more varied population of native spea-kers of French. 6.2 Vowel Duration: Perception The second part of the study examined the l i s t e n e r ' s use of vowel length i n distinguishing perceptually between such pairs as "maitre:mettre" and "pate:patte". The vowel durations of these words were varied i n steps of two cycles of the fundamental f r e -quency (or approximately 16 msec.) and li s t e n e r s were asked to assign to the stimulus a l e x i c a l item (either the "long" or "short" version of the word). - 55 -Listeners' judgements were found to be affected by s h i f t s i n vowel durations. Lengthening the "short" vowel seemed to have a greater reversal effect on l i s t e n e r judgement than did shortening the "long" vowel. In the cases of "maitre:mettre" and "paterpatte", where the o r i g i n a l "long" and "short" versions were separated by a difference of 100 msec., the addition of 64 msec, to the o r i g i n a l "short" version was a greater a l t e r a t i o n of the o r i g i n a l duration from the point of view of percentage than'jthe subtraction of 64 msec, from the o r i g i n a l "long" version. For these word pairs this could explain the greater effect that leng-thening the "short" version had on the listener's judgement of the word. In some cases, e.g. "male", vowel quality was a more important cue than length i n making judgements of the edited ver-sions of the o r i g i n a l l y "long" word. Vowel quality of the o r i g i n a l -l y "short" vowel did not appear to be as important a cue as the length feature i n assigning a l e x i c a l item to the stimulus. Be-cause of the d i f f e r i n g effect of . l e n g t n e n i n g the "short" version and shortening the "long" version, i t was not possible to arrive at a c r i t i c a l unit of duration that, when added to the o r i g i n a l l y "short" version or subtracted from the o r i g i n a l l y "long" version, would reduce "short" and "long" responses respectively to the 50% response point. Using synthetic speech s t i m u l i , many of the pro-blems encountered i n analyzing the results could be avoided. For-mant frequencies for the "long" and "short" members of a pair could be averaged and held constant along with intensity and fundamental frequency while varying only vowel duration. In this way i t would be possible to determine with greater accuracy the c r i t i c a l unit of - 56 -duration required to reverse l i s t e n e r judgement. In a recent study (Leon, 1972), Leon ran a l i s t e n i n g test which consisted of the pairs "bete:bette"; "bele :belle", and " l ' e t r e r l e t t r e " re-corded by ten native speakers. I t was found that when the vowel duration of the "long" version exceeded that of the "short" ver-sion by approximately 30 msec., the words were consistently judged correctly by native l i s t e n e r s . This figure i s close to the aver-age of 32 msec.,^ > found i n this study, which was needed to reverse the l i s t e n e r s ' judgement of the o r i g i n a l l y "short" version to the 50%JJresponse l e v e l . By eliminating differences i n formant f r e -quency be means of synthetic speech, one could determine i f the c r i t i c a l duration for reversing the judgement of the o r i g i n a l "long" version to the 50% response point i s around 30 msec. In this study, the l i s t e n i n g task did not require that the subject compare the duration of one stimulus to that of an-other. Rather, the l i s t e n e r had to base his judgement on some i n -te r n a l l y generated c r i t e r i o n . The results of the test tend to sup-port Lehiste's notion of a "durational image" (Lehiste, 1970) to which the l i s t e n e r can compare the stimulus and arrive at a judge-ment about the length of the stimulus i n r e l a t i o n to this i n t e r n a l "durational image". 6.3 Vowel Duration: Production and Perception' Comparing the results of the f i r s t experiment with those of the second experiment, the question arises as to the r e l a t i o n between the production and perception of the length difference. Is perception dependent oir production or vice versa, or are the - 57 -two processes independent of each other? The results of the second experiment do not indicate a difference between the responses of the speaker/listeners who claimed they made the length difference and those who claimed they did not. This observation would suggest that perception of the length d i s t i n c t i o n i s not dependent upon i t s production. A con-sideration, however, that must not be overlooked i n comparing the results of the two experiments, i s that most of the subjects taught French to English spe'aking students. The pronunciation of "standard" French, therefore, was fam i l i a r to a l l the subjects, and i n this standard French subjects were aware that a length d i s -t i n c t i o n i s made between such pairs as "maitre:mettre" and "pate: patte". This fact could have affected the results of the l i s t e n -ing test, i.e. subjects may have learned to make and perceive the length d i s t i n c t i o n when teaching the standard French pronunciation to English speaking students even when the d i s t i n c t i o n does not occur i n their own speech. In examining the r e l a t i o n between production and percep-tion.- of the vowel length feature i n French, i t would be more en-lightening to run the l i s t e n i n g test on subjects less aware of the standard pronunciation of French. One might find i n such a study, that subjects who do not make the length d i s t i n c t i o n i n their own speech do not perceive differences i n vowel duration as consistently as those who are accustomed to making the length d i s t i n c t i o n . BIBLIOGRAPHY ABEL, S.M. (1972a). "Duration Discrimination of Noise and Tone Bursts," J. Acoust. Soc. Amer. 51, 1219-1223. ABEL, S.M. (1972b). "Discrimination of Temporal Gaps," J. Acoust. Soc. Amer. 5_2, 519-524. BELASCO, S. (1958). "Variations i n Vowel Duration: Phonemically or Phonetically Conditioned?" J. Acoust. Soc. Amer. 30, 1049-1050. BOINDIN. (1'753) . Remarques sur les sons de l a langue. CHLUMSKY, J. ;(1928). Ceska kvantita, melodie a prizvuk (Prague). CREELMAN, CD. (1962). "Human Discrimination of Auditory Dura-t i o n , " J. Acoust. Soc. Amer. 34, 583-593. DELATTRE, P. (1938). "Duree .Sons ciente et duree inconsciente des voyelles," The French Review, XII, 49-50. DELATTRE, P. (1951). Principes de Phonetique Francaise a 1'Usage  des Etudiants Anglo-Amrericains (Middlebury College, New York). JDELATTRE, P. (1964). "Comparing the Vocalic Features of English, German, Spanish and French," International Review of Applied L i n g u i s t i c s i n Language Teaching, I_I, 71-97. DELATTRE, P. I</1966). Studies i n French and Comparative Phonetics (Mouton & Co., London, The Hague, P a r i s ) . DELATTRE P. and HOHENBERG, M. (1968). "Duration i n German Unstressed Vowels," International Review of Applied L i n g u i s t i c s i n Language Teaching, IV, 183-198. DENES, P. (1955). "Effect of duration on the perception of voic-ing," J. Acoust. Soc. Amer. 27_, 761-764. DEYHIME, G. (1967). "Enquete sur l a phonologie du francais con-temporain," La Linguistique jL, 97-108, _2, 57-84. FANT, C.G.H. (1971). "Notes on the Swedish Vowel System," Form and  Substance, L. Hammerich, R. Jakobson, E. Zwirner, Ed., (Akademisk Forlag, Denmark), 259-268. - 58 -- 59 -FOUCHE, P. (1958). Phorietique Historique du Francais: Intro- duction (Klincksieck, P a r i s ) . FRY, D.B. (1955). "Duration and Intensity as Physical Correlates of L i n g u i s t i c Stress," J. Acoust. Soc. Amer. 27_, 765-768. FUJISAKI, H., NAKAMURA, K., and IMOTO, T. (1973). "Auditory Perception of Duration of Speech and Non-Speech Stim-u l i , " Annual B u l l e t i n (Research I n s t i t u t e of Logope-dics and Phoniatrics, University of Tokyo) _7, 45-64. GRAMMONT, M. (1933). Traite de Phonetique (Delagrave, P a r i s ) . HADDING-KOCH, K. and ABRAMSON, A. (1964). "Duration vs. Spec-trum i n Swedish Vowels: Some Perceptual Experiments," Studia L i n g u i s t i c a 18, 94-107. > HOUSE, A.S. (1961). "On Vowel Duration i n English," J . Acoust. Soc. Amer. 33, 1174-117-8. HOUSE, A.S. and FAIRBANKS, G. (1953). "The Influence of Conson-ant Environment upon the Secondary Acoustical Charac-t e r i s t i c s of Vowels," J. Acoust. Soc. -Kmer. 25, 105-113. HUGGINS, H.W.F. (1972). "Just Noticeable Differences i n Segment Duration i n Natural Speech," J. Acoust. Soc. Amer. 51, 1270-1278. JANOTA, P. and JANCAK, P. (1970). "An Investigation of Czech Vow-e l Quantity by Means of Listening Tests,Kl Acta Univer-s i t a t i s Carolinae, P h i l o l o g i c a 1, 31-68. LEHISTE, I. (1970). Suprasegmentals (M.I.T. Press, Cambridge,. Massachusetts, and London, England). LEHISTE, I. (1972). "The Units of Speech Perception," Working Pa-pers i n Ling u i s t i c s 12_, Ohio State University, 1-32. LEON, P. (1966). Prononciation du Franqais Standard (Didier, Paris) LEON, P. (1972). "Etude de l a prononciation du 'e' accentue chez un groupe de jeunes Parisiens," Papers i n Li n g u i s t i c s  and Phonetics i n Memory of Pierre Delattre, A/ Valdman, Ed. (Mouton, The Hague), 317-327. MALMBERG, B. (1954). La Phorietique (Presses Universitaires de France, P a r i s ) . MALMBERG, B. (1968). Manual of Phonetics (North Holland Publish-ing Co., Amsterdam). - 60 -MARTINET, A. (1945). La prononciation du. franqals contemporain (Paris). MARTINET, A. (1949). Phonology as Functional Phonetics (Oxford University Press, London). MIGHAELIS,EH. and PASSY, P. (1897). Dictiorinaire Phonetique de l a Langue Frariqaise (Charles Meyer, Hanover and B e r l i n ) . NOOTEBOOM, S.G. (1972). Production and Perception of Vowel Dura-'. •tion: A Study of Durational Properties of Vowels i n  Dutch (Utrecht). NOOTEBOOM, S.G. (1973). "Perceptual Reality of Prosodic Durations," J. of Phonetics 1, 25-45. PETERSON, G.E. and LEHISTE, I. (1960). "Duration of Syllable Nuc-l e i , " J. Acoust. Soc. Amer; 32, 693-703. PETERSON, 'G.E. and McKINNEY, N.P. (1961). "The Measurement of Speech Power," Phonetica ]_, 65-84. REICHSTEIN, R. (1960). "Etude des variations sociales et geogra-phiques des f a i t s linguistiques," Word 16_, 55-59. ROBERT, P. (1971). MicrooRobert: dictionnaire du francais primor- d i a l (Societe du Nouveau L i t t r e , P a r i s ) . ROSSI, M. (1972). "Le s e u i l d i f f e r e n t i e l de duree," Papers i n L i n - g u i s t i c s and Phonetics i n Memory of Pierre Delattre, A. Valdman, Ed. (Mouton, The Hague), 435-450. ROUSSELOT, P.-J. (1927). Precis de Prononciation Francaise (Didier, P a r i s ) . VALDMAN, A., SALAZAR, R.J., and CHARBONNEAUX, M.A. (1964). A D r i l l - book of French Pronunciation (Harper & Row, New York). ZIMMERMAN, S.A. and SAPON, S.M. (1958). "Note on Vowel Duration Seen Cross-Linguistically," J . Acoust, Soc. Amer. 30, 152-153. - 61 -APPENDIX - 62 -Instructions for Experiment I: 1. La premiere l i s t e contient trente-deux phrases francaises ordinaires. La deuxieme l i s t e contient trente-hui.t phrases toutes de l a forme " I I a prononce l e mot ' ' plus f o r t . " ou ' ' sera toujours un mot francais. 2. Lisez l e numero; de chaque phrase, puis l i s e z chaque phrase a vitesse normale et avec l a meme intonation ribrmale en faisant tomber l a voix a l a f i n de chaque phrase. 3. S i vous f a i t e s une erreur quelconque en l i s a n t ou que vous ayez une hes i t a t i o n , repetez simplement l a phrase en question. 4. Lisez les phrases une f o i s pour vous. L i s t 1 1. 11 porte une cotte de maille. 2. l i s mangent de l a glace. 3. 11 l ' a appele bel ane. 4. Pres de l a ruche i l y avait plusieurs rennes. 5. 11 est l a ce s o i r . 6. 11 f a i t reposer sa patte. 7. l i s vont t i r e r a b a l l e s . 8. Je n'ai que l a robe de fete. 9. 11 s'est envole sans e l l e . 10. 11 faut qu'on le mate. 11. C'est sa premiere journee sans tache. 12. 11 est las ce s o i r . 13. C'est un changement de l e t t r e . 14. 11 f a i t reposer sa pate. 15. 11 pourrait l e f a i r e mettre. 16. 11 l ' a appelee "Belle Anne". 17. 11 est hale. 18. 11 pourrait l e f a i r e maitre. * 19. Regardez ces betes. 20. C'est sa premiere journee sans tache. 21. E l l e adore ses malles. 22. 11 est a l l e . 23. l i s vont t i r e r a Bale. 24. 11 s'est envole sans a i l e s . - 63 -- 64 -25. C'est un changement de l ' e t r e . 26. E l l e adore ces males. 27. I I faut qu'on le mate. 28. Je n'ai que l a robe de f a i t e . 29. Pres de l a ruche , i l y avait plusieurs reines. 30. Regardez ces bettes. 31. I I a achete deux chaises. 32. Le p e t i t chat est dans l'arbre. L i s t 2 1. 11 a prononce le mot "nette" plus f o r t . 2. 11 a prononce le mot "quatre" plus f o r t . 3. 11 a prononce le mot " l a " plus f o r t . 4. 11 a prononce le mot "mettre" plus f o r t . 5. 11 a prononce le mot " l ' e t r e " plus f o r t . 6. 11 a prononce le mot "Anne" plus f o r t . ' 1 7> 11 a prononce le mot "renne" plus f o r t . 8. 11 a prononce le mot "tache" plus f o r t . 9. 11 a prononce le mo t "malle" plus f o r t . 110. 11 a prononce le mot "mate" plus f o r t . 11. 11 a prononce le mot " b a l l e " plus f o r t . 12. 11 a prononce le mot "pate" plus f o r t . 13. 11 a prononce le mot "maitre" plus f o r t . 14. 11 a prononce le mot " a l l e " plus f o r t . 15. 11 a prononce le mot "mate" plus f o r t . 16. 11 a prononce le mot "bat" plus f o r t . 17. 11 a prononce le mot " a i l e " plus f o r t . 18. 11 a prononce le mot " l a s " fplus f o r t . 19. 11 a prononce le mot , "ane" plus f o r t . 20. 11 a prononce le mot " e l l e " plus f o r t . 21. 11 a prononce le mot "ma" plus f o r t . 221 11 a prononce le mot "patte" plus f o r t . 23. ' 11 a prononce le mot " l e t t r e " plus f o r t . 24. 11 a prononce le mot " b e l l e " plus f o r t . - 65 -- 66 -25. 11 a prononce le mot "tache" plus f o r t . 26. 11 a prononce le mot "has" plus f o r t . 27. 11 a prononce le mot "bette" plus f o r t . 28. 11 a prononce le mot "hale" plus f o r t . 29. 11 a prononce le mot "belee..-plus f o r t . 30. 11 a prononce le mot "fete" plus f o r t . 31. 11 a prononce- le mot "male" plus f o r t . 32. 11 a prononce le mot "mat" plus f o r t . 33. 11 a prononce le mot "reine" plus f o r t . 34. 11 a prononce le mot "Bale" plus f o r t . 35. 11 a prononce le mot "bete" plus f o r t . 36. 11 a prononce le mot " f a i t e " plus f o r t . 37. 11 a prononce le mot "tasse" plus f o r t . 38. 11 a prononce le mot "dette" plus f o r t . 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            data-media="{[{embed.selectedMedia}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0100021/manifest

Comment

Related Items