This study examined the impact on speech processing of regional phonetic/phonological variation in the listener’s native language. The perception of the /e/-/ε/ and /o/-/ɔ/ contrasts, produced by standard but not southern French native speakers, was investigated in these two populations. A repetition priming experiment showed that the latter but not the former perceived words such as /epe/ and /epε/ as homophones. In contrast, both groups perceived the two words of /o/-/ɔ/ minimal pairs (/pom/-/pɔm/) as being distinct. Thus, standard-French words can be perceived differently depending on the listener’s regional accent.

Adult listeners find it difficult to discriminate speech sounds not present in their native language. Models such as Best’s “Perceptual Assimilation Model” Best et al. (1988) and Flege’s “Speech Learning Model” (1995) propose that non-native speech perception abilities can best be explained by making reference to the native phonetic space. For example, adult Japanese listeners have difficulties discriminating between American English [l] and [ɹ] (Miyawaki, Strange, Verbrugge, Liberman, Jenkins, and Fujimura, 1975), presumably because Japanese has only one liquid phoneme (/r/), to which both [l] and [ɹ] are assimilated by Japanese listeners. Difficulties in perceiving non-native speech sounds are not only observed for a foreign language, but also for the second language of highly fluent bilinguals. In a study on the perception of the /e/-/ε/ contrast found in Catalan but not in Spanish, Pallier et al. (2001) showed that bilinguals who acquired Spanish first but learned Catalan before 6 years of age, merged this contrast and processed Catalan words like /perə/ and /pεrə/ as homophones. Hence, even for fluent bilinguals, speech perception depends upon the phonemic categories of the first language.

While considerable work has accumulated on the perception of non-native phonemic contrasts, the impact on speech perception of within-language phonetic/phonological variation has only recently been investigated. Moreover, conflicting results have emerged about how phonemic contrasts specific to one language’s regional variety are perceived by listeners of another regional variety. In a recent study of the perception of vowels from two different British English accents, Evans and Iverson (2004) showed that listeners can adjust their vowel categorization decisions according to the accent of the carrier sentence. Also, Cutler et al. (2005) had speakers of Australian English identify vowels in CV and VC syllables produced by an American English talker and found that the Australian speakers’ performance was similar to that of speakers of American English. In contrast, Conrey et al. (2005) found that speakers of an American-English dialect with a /i/-/ε/ vowel merger were less accurate than speakers of an unmerged dialect in discriminating between minimal pairs of words (e.g., pin/pen) contrasting in these vowels (see also Labov et al. 1991, and Janson and Schulman, 1983 for Swedish dialects). Further evidence that regional phonetic/phonological variation affects speech perception has been obtained by Floccia et al. (2006), who showed that French words were recognized more slowly when spoken in an unfamiliar accent.

To gain a better understanding of the effect of regional accent on speech perception, we have examined the word identification performance by listeners of two varieties of French, standard French and southern French which have different phonemic inventories. Standard French has three mid vowel pairs, /e/-/ε/, /ø/-/œ/ and /o/-/ɔ/ which are all contrastive, as is the case of the minimal pairs épée /epe/ “sword” vs épais /epε/ “thick” or côte /kot/ “hill” vs cote /kɔt/ “rating”. Conversely, there is no contrastive distinction between /e/-/ε/, /ø/-/œ/, and /o/-/ɔ/ in southern French, which is viewed as having three mid vowel phonemes only, namely, /e/, /ø/, and /o/ (Durand, 1990). [ε], [œ], and [ɔ] do appear at the phonetic level, but they are in complementary distribution with respect to the corresponding mid-high variants: mid-high vowels occur in open syllables and mid-low vowels in closed syllables and whenever the next syllable contains schwa /ə/ (Durand, 1990). Thus, épée and épais will both be pronounced [epe], while côte and cote will both be pronounced [kɔtə].

How do listeners of southern French perceive standard French forms such as [epε], compared with [epe], or [kot], compared with [kɔt]? To address this question, a long lag repetition priming paradigm was used as in Pallier et al. (2001). The repetition priming effect refers to the fact that participants respond more rapidly when a word is encountered a second time. Of particular interest is the amount of priming obtained between members of minimal pairs such as /epe/-/epε/ or /kot/-/kɔt/. Standard-French listeners should clearly perceive these words as being different and show no facilitation on the processing of the second word of the pair. In contrast, since [e]-[ε] as well as [o]-[ɔ] are conditioned variants associated with a single phonemic category in southern French, these listeners should map the two words of a minimal pair onto the same lexical representation, regardless of the height (mid-high vs mid-low) of the critical vowel. Hence, unlike standard-French listeners, southern-French listeners should take the second word of the minimal pair to be a repetition of the first and show facilitation on the second word. We used a lexical decision task in which participants had to discriminate between words and nonwords.

Sixty-nine native speakers of French participated in the experiment for course credit. Thirty-two were standard-French listeners from the University of Geneva. The other 37 were southern-French listeners from the Aix-en-Provence area. All participants reported no hearing disorders.

Sixty-four (C)V.CV bisyllabic words making up 32 minimal pairs based on the /e/-/ε/ phonemic contrast were selected. This vowel contrast occurred in word-final position. These 32 pairs were split into two sets according to their morphological/semantic relation. For one set, the words forming a pair were morphologically/semantically related (e.g., /pike/ to prick–/pikε/ stake). For the other, no morphological or semantic relation existed between the words (e.g., /epe/ sword–/epε/ thick). The results of related and unrelated Pairs will be analyzed separately, since it has been shown that a morphological/semantic relation between prime and target can facilitate target word processing (Meunier and Segui, 2002). Sixteen CVC monosyllabic words forming eight minimal pairs containing the /o/-/ɔ/ phonemic contrast were also selected (e.g., /pom/ palm–/pɔm/ apple). Finally, we included 16 CV monosyllabic words forming eight minimal pairs based on the /ø/-/y/ phonemic contrast (e.g., /fø/ fire–/fy/ barrel) which exists in both regional varieties. Only eight minimal pairs were found for the /o/-/ɔ/ and /ø/-/y/ phonemic contrasts, because pairs based on these contrasts are rare in French.

For the purpose of the lexical decision task, 96 nonwords were created. Sixty-four were bisyllabic and formed 32 minimal pairs based on the /e/-/ε/ contrast. The nonwords were created by replacing the first syllable of test words by the first syllable of another word present in the list to avoid a strategy for word responses based on the repetition of the first syllable. The remaining 32 nonwords were monosyllabic, forming eight minimal pairs based on the /o/-/ɔ/ contrast and eight minimal pairs based on the /ø/-/y/ contrast. Monosyllabic nonwords were created by replacing the first phoneme of the test words. Finally, an additional 108 words and 108 nonwords were included as filler items in the experimental lists.

Four counterbalanced lists of 408 items were created such that each member of a minimal pair was repeated or followed by the other member of the minimal pair. Primes and targets were separated by 8 to 17 items. In each list, the words forming a pair appeared in the same positions.

The stimuli were recorded by a female native speaker of standard French with a Parisian accent. Each minimal pair was produced twice, and the pair for which the acoustic differences in the critical vowel were the largest between the two words was presented to participants. Figure 1 shows the frequency of the second formant (F2) as a function of the first formant (F1) for the critical vowel in each of the /e/-/ε/ and /o/-/ɔ/ words. As can be seen, both /e/-/ε/ and /o/-/ɔ/ contrasts were produced distinctly: /ɔ/ was systematically associated with higher frequencies of both the first and second formants than /o/, and /ε/ was characterized by both a higher first formant frequency and a lower second formant frequency than /e/. Although the regions associated with /e/ and /ε/ in the F2 vs F1 acoustic plane were close to each other, there was little overlap between the two.

FIG. 1.

F1 and F2 frequencies for the critical vowel in each of the /e/-/ε/ and /o/-/ɔ/ words.

FIG. 1.

F1 and F2 frequencies for the critical vowel in each of the /e/-/ε/ and /o/-/ɔ/ words.

Close modal

Participants were asked to make a lexical decision as quickly and accurately as possible and to give “word” responses using their dominant hand. Response times (RTs) were measured from the onset of the test item. An interval of 2500ms elapsed between the participant’s response and the presentation of the next stimulus. If participants failed to respond within 1800ms of a stimulus, no response was recorded and a next item was presented. The participants were tested on only one experimental list and began the experiment with 12 practice trials.

Mean correct response times (RTs) in each condition were computed and are shown in Fig. 2. The data from one participant from Aix-en-Provence were not included in the statistical analyses given the high percentage of errors on the task (more than 25%). Four word pairs (two, one, and one, for the /e/-/ε/ related, /e/-/ε/ unrelated, and /ø/-/y/ contrast, respectively) giving rise to more than 30% of errors were also excluded from the analyses. For each contrast, planned comparisons on the RTs assessed the priming effect (first vs second occurrence) within each pair (same, minimal) and for each group (Geneva, Aix-en-Provence). F values are reported by participants (F1) and items (F2).

FIG. 2.

Mean reaction times (in ms) for the first (black bars) and the second (white bars) occurrence as a function of pair type and population for each contrast.

FIG. 2.

Mean reaction times (in ms) for the first (black bars) and the second (white bars) occurrence as a function of pair type and population for each contrast.

Close modal

/e/-/ε/ related contrast: A priming effect was observed for the same pairs in both Geneva [F1(1,66)=17.81, p<.001; F2(1,40)=16.34, p<.001] and Aix-en-Provence [F1(1,66)=14.14, p<.001; F2(1,40)=10.46, p<.01]. RTs were shorter for target words encountered for the second time compared with the first time. A priming effect was observed for the minimal pairs, with RTs being faster for the second than for the first occurrence in Aix-en-Provence [F1(1,66)=14.47, p<.001; F2(1,40)=8.58, p<.01] but not in Geneva (Fs<1). Note that the lack of a priming effect for the minimal pairs in Geneva suggests that the morphological/semantic relation between words had no impact.

/e/-/ε/ unrelated contrast: A priming effect was observed for the same pairs in both Geneva [F1(1,66)=21.53, p<.001; F2(1,40)=26.18, p<.001] and Aix-en-Provence [F1(1,66)=40.16, p<.001; F2(1,40)=29.61, p<.001]. RTs were shorter for target words encountered for the second time compared with the first time. A priming effect was observed for the minimal pairs, with RTs being shorter for the second than for the first occurrence in Aix-en-Provence [F1(1,66)=10.31, p<.01; F2(1,40)=5.57, p<.05] but not in Geneva [F1<1; F2(1,40)=1.96, p=.17].

/o/-/ɔ/ contrast: A priming effect was observed for the same pairs in both Geneva [F1(1,66)=28.59, p<.001; F2(1,40)=27.34, p<.001] and Aix-en-Provence [F1(1,66)=11.67, p<.01; F2(1,40)=9.24; p<.01]. RTs were shorter when the target words were encountered for the second time relative to the first time. No priming effect was observed for the minimal pairs in either Geneva (Fs<1) or Aix-en-Provence (Fs<1).

Finally, neither group showed a priming effect for the minimal pairs based on the /ø/-/y/ contrast which was common to both regional French varieties [in Geneva (F1<1; F2(1,40)=1.32; p>.20); in Aix-en-Provence (Fs<1)].

This study has shown differences in how French listeners perceive words contrasting in only one phonetic feature depending on their regional accent. When presented with French minimal pairs containing either a /e/-/ε/ or a /o/-/ɔ/ standard phonemic contrast, standard French listeners showed no priming between members of a minimal pair. Hence, they perceived the two members of a minimal pair as being different. In contrast, southern-French listeners showed a priming effect on members of the minimal pair with the /e/-/ε/ contrast. This suggests that they treated the second member as a repetition of the first. These findings are consistent with the Best et al. (1988) model and suggest that listeners are insensitive to phonemic contrasts when the speech sounds can be assimilated perceptually to a single category in their phonemic inventory.

Another significant outcome of this study was that southern-French listeners did not respond in the same way to the /e/-/ε/ and /o/-/ɔ/ contrasts, since a priming effect was found for the former but not for the latter. The absence of a priming effect for the /o/-/ɔ/ minimal pairs suggests that the two words were perceived as being distinct by southern-French listeners. One potential explanation for this discrepancy lies in how both contrasts were produced. It may be the case, as Fig. 1 suggests, that /e/ differed from /ε/ to a lesser extent than /o/ from /ɔ/ at the phonetic level, hence allowing the first word to facilitate the recognition of the second word in the /e/-/ε/ but not in the /o/-/ɔ/ pairs. However, the lack of a priming effect for the /e/-/ε/ minimal pairs for the standard French speakers confirms that differences between /e/ and /ε/ were clearly perceivable.

The differential response pattern for the two contrasts in southern French listeners could also be related to the phonological status of these contrasts in today’s standard French. The /o/-/ɔ/ contrast is a particularly well-established phonological feature of standard French which is, as such, known to the listeners of southern French, even if this contrast is neutralized in their own productions. Indeed it probably stands as one of the shibboleths allowing southern listeners to recognize standard-French speakers. The explicit knowledge that southern French listeners have of the /o/-/ɔ/ contrast could explain why they perceive the two members of /o/-/ɔ/ minimal pairs differently.

By comparison, the /e/-/ε/ contrast, as it occurs in word-final position in standard French, is characterized by greater complexity both across and within speakers (e.g., Coveney, 2001; Fagyal et al. 2006; Tranel, 1987). For example, Tranel (1987) points out that the distribution of /e/ and /ε/ is partly speaker-dependent (e.g., quai “platform” may be pronounced [ke] by some speakers and [kε] by others), and that the pronunciation of the vowel in grammatical words such as les “the,” plural form, or est “is” commonly varies between [e] and [ε] even for a given speaker. In addition, there is experimental evidence showing that /e/ and /ε/ in word-final position are in the process of merging in Parisian French (e.g., Fagyal et al. 2002). For example, young Parisian speakers tend to no longer maintain a distinction in the pronunciation of épée vs épais. Both the complexity of the pronunciation of the /e/ and /ε/ vowels and the merging process of the two vowels in standard French make it less likely that this opposition is represented in the southern listeners’ receptive phonological knowledge of standard French. As a consequence, southern French listeners tend to assimilate the [e] and [ε] vowels to their [e] native category and perceived the two members of /e/-/ε/ minimal pairs as homophones. It thus appears that the listener’s linguistic experience influences speech processing, even within dialects of the same language.

This research was partly supported by the TCAN interdisciplinary CNRS program. Thanks to Cécile Fougeron for recording the speech material. We are also grateful to Robert Espesser for sharing his statistical expertise and to Jacques Durand, Zsuzsanna Fagyal, and Bernard Laks for helpful comments. We also thank the reviewers for their helpful comments.

1.
Best
,
C. T.
,
McRoberts
,
G. W.
, and
Sithole
,
N. M.
(
1988
). “
Examination of perceptual reorganization for nonnative speech contrasts: Zulu click discrimination by English-speaking adults and infants
,”
J. Exp. Psychol.
14
,
345
360
.
2.
Conrey
,
B.
,
Potts
,
G. F.
, and
Niedzielski
,
N. A.
(
2005
). “
Effects of dialect on merger perception: ERP and behavioral correlates
,”
Brain Lang
95
,
435
449
.
3.
Coveney
,
A.
(
2001
).
The Sounds of Contemporary French: Articulation and Diversity
(
Elm Bank Publications
, Exeter, UK).
4.
Cutler
,
A.
,
Smits
,
R.
, and
Cooper
,
N.
(
2005
). “
Vowel perception: Effects of non-native language vs non-native dialect
,”
Speech Commun.
47
,
32
42
.
5.
Durand
,
J.
(
1990
).
Generaive and Non-Linear Phonology
(
Longman
, London, UK).
6.
Evans
,
B. G.
, and
Iverson
,
P.
(
2004
). “
Vowel normalization for accent: An investigation of best exemplar locations in northern and southern British English sentences
,”
J. Acoust. Soc. Am.
115
,
352
361
.
7.
Fagyal
,
Z.
,
Hassa
,
S.
, and
Ngom
,
F.
(
2002
). “
L’opposition [e]-[ε] en syllabes ouvertes en fin de mot en français parisien: étude acoustique préliminaire (“The [e]-[ε] opposition in open syllables at word offsets in Parisian French: A preliminary acoustic study”)
,”
Proceedings of the XXIVémes Journèes d’Études sur la Parole
(
Nancy
, France, 24–27 June), pp.
165
168
.
8.
Fagyal
,
Z.
,
Jenkins
,
F.
, and
Kibbee
,
D.
(
2006
). “
French: A Linguistic Introduction
,” in
Phonetics and phonology
(
Cambridge University Press
, Cambridge, UK).
9.
Flege
,
J.
(
1995
). “
Second language speech learning: Theory, findings, and problems
,” in
Speech perception and linguistic experience: Issues in cross-language research
, edited by
W.
Strange
and
J.
Jenkins
(
York Press
, Timonium, MD), pp.
233
277
.
10.
Floccia
,
C.
,
Goslin
,
J.
,
Girard
,
F.
, and
Konopczynski
,
G.
(
2006
). “
Does a regional accent perturb speech processing?
,”
J. Exp. Psychol. Hum. Percept. Perform.
32
,
1276
1293
.
11.
Janson
,
T.
, and
Schulman
,
R.
(
1983
). “
Non-distinctive features and their use
,”
Linguistics
19
,
321
336
.
12.
Labov
,
W.
,
Karan
,
M.
, and
Miller
,
C.
(
1991
). “
Near-mergers and the suspension of phonemic contrast
,”
Lang. Var. Change
3
,
33
74
.
13.
Meunier
,
F.
, and
Segui
,
J.
(
2002
). “
Cross-modal morphological priming in French
,”
Brain Lang
81
,
89
102
.
14.
Miyawaki
,
K.
,
Strange
,
W.
,
Verbrugge
,
R.
,
Liberman
,
A. M.
,
Jenkins
,
J. J.
, and
Fujimura
,
O.
(
1975
). “
An effect of linguistic experience: The discrimination of /r/ and /l/ by native listeners of Japanese and English
,”
Percept. Psychophys.
18
,
331
340
.
15.
Pallier
,
C
,
Colomé
,
A.
, and
Sebastián-Gallés
,
N.
(
2001
). “
The influence of native-language phonology on lexical access: Exemplar-based versus abstract lexical entries
,”
Psychol. Sci.
12
,
445
449
.
16.
Tranel
,
B.
(
1987
).
The Sounds of French
(
Cambridge University Press
, Cambridge, UK).