Acquisition of French vowels by Japanese-speaking learners: close and close-mid rounded vowels

In Japan, one foreign language is mandatory at secondary education (12-18 years old). Since it is English that is chosen in almost all schools, the quasi-totality of Japanese students learn English first, then another language at higher education, or in a small number of high schools. French is thus learnt as a third language in the great majority of cases. While L2 (English) and L3 (French) share a lot of similarities in their word order (SVO, as opposed to SOV in Japanese), vocabulary, etc., which might favour positive transfer from L2, there are not many similarities in phonetics and phonology. Indeed, some characteristics of French are more similar to Japanese than those of English: syllable structure (more open syllables in French), syllable-timed rhythm (Japanese is considered as a mora-timed one, but opposed to stress-timed English in terms of perceived isochrony), lack of aspiration, etc. However, when it comes to vowels, L3 French (10 oral and 3 nasal vowels in present day Parisian French) presents some discrepancies both from L1 Japanese (5 vowels in Tokyo Japanese) and from L2 English (around 20 vowels). These include: 1) the presence of phonemically nasal vowels, 2) that of front rounded vowels /y/ /ø/ /œ/ (Gottfried 1984; Strange and Levy 2004; Lauret 1998 reports the realisation of /y/ as [ju] by American English speakers), 3) and the quantal nature of French /u/: F1 and F2 close together (Vaissière, in press), as opposed to a higher F2 in English (Hawkins and Midgley 2005, among others) and in Japanese (Sugito 1995, among others). Flege 1987 reports a high F2 in French /u/ pronounced by American English speakers. As far as these features are concerned, we could not expect positive transfer either from L1 or from L2. Table 1 summarises these characteristics. For this reason, it is of particular interest to investigate into the realisation of these phenomena in L3. In the present paper, we take the production and perception of the French (L3) vowels in isolation by Japanese speaking learners, with a special focus on close and close-mid rounded vowels. a) Japanese speaking learners participated in auditory perceptual experiments (an identification test with 5 listeners, and an AXB discrimination test with 14 listeners). The stimuli were pronounced by 4 native speakers of French (2 female and 2 male x 2 repetitions). The results of the identification test (Table 2) show that the confusion between /u/-/ø/ accounts for 11% of all confusions. In the AXB test (Figure 1) this same pair presents a score (85%) significantly inferior to that of /u/-/y/ (95%) and /ɛ/-/a/ (99%). These findings could be explained by the acoustic nature of the difference between French /u/ and Japanese /u/, the latter resembling acoustically French /ø/ with F2 situated halfway between F1 and F3. b) We then examined the production of isolated vowels by Japanese learners. 3 learners pronounced twice the 13 French vowels in isolation (presented in IPA and in a typical French spelling, and embedded in a carrier sentence). We found that F2 of /u/ tended to be higher (1000-1300 Hz for male, 1300-1400 Hz for female speakers) than that of native speakers, making this vowels close to /ø/. As for /y/, 2 learners produced close F2/F3, which is characteristic of this vowel (Vaissière, in press), while the other speaker pronounced an opening diphthong similar to /ju/ in Japanese, and to the production of Lauret (1998)'s American English speakers. c) The 10 oral vowels pronounced by the same 3 Japanese learners were used in an identification test with 26 French native listeners. 1) /u/ pronounced by the learners was judged as such in only 43% (39% as /ø/, 13% as /y/), 2) /y/ was perceived correctly in 69% of the cases (14% as /u/, 7% as /ø/). 3) /ø/ was judged as /y/ in 62% of the cases. Table 3 indicates that the confusion among these three vowels is remarkable, along with that among close-mid and open-mid vowels. These findings suggest the following: 1) "New" vowels (front rounded vowels, which have no corresponding ones in L1 or L2), as well as "similar" vowels (French /u/ as opposed to phonetically different close back vowel /u/ in L1 and L2) are confused both in perception and production. 2) The confusion in production is particularly related to the high F2 of /u/, the “similar vowel”, produced by the learners, which makes this vowel sound like /ø/. This finding corroborates the result of Flege 1987. We have also found in Experiment c) that /ø/ produced by the Japanese learners tended to be judged as /y/. This is probably due to the negative transfer from the L2 English spelling of “eu” /ju:/. Since Japanese speakers' /u/ tends to be perceived as /ø/, they would not have much difficulty pronouncing /ø/, but the English spelling might have influenced the pronunciation (the vowels were presented in IPA with a typical French spelling: “eu” for /ø/). The influence of the L2 spelling can also be observed in some learners' production of “er” /ɛʀ/ in French as Japanese /a:/ (coincidently resembling the phonetic realisation in North American French), since “er” /ə/ or /ɚ/ in English is interpreted as /a:/ in Japanese