Skip to navigation – Site map

Accommodation in learner corpora: A case study in phonetic convergence

Léa Burin and Nicolas Ballier

Abstracts

Phonetic convergence is the process by which a speaker adapts his speech to sound more similar to his/her interlocutor. While most studies analyzing this process have been conducted amongst speakers sharing the same language or variety, this preliminary experiment focuses on native/non-native interactions. Segmental (vowel quality) and suprasegmental features (vowel duration and speech rate) were analysed. Results suggest that there was more deviation in the F1 dimension regarding the vowels /æ/ and /ɑ:/. This corroborates Babel’s (2009) results that low vowels tend to be more subject to imitation than high vowels, and especially within the F1 dimension.
Convergence in vowel duration in order to sound more native-like and accommodation from all speakers regarding speech rate have also been observed (Zając 2013).

Conclusions drawn from this preliminary research study suggest that previous findings on intralanguage processes of accommodation appear to be validated in the case of native/non-native interactions.

Top of page

Author's notes

We wish to acknowledge Adrien Méli for granting access to parts of his annotated data (Méli & Ballier 2015) used for his PhD. We also thank the three reviewers for their comments. We take responsibility for any remaining errors.

Full text

1. Introduction

1 The process of accommodation has been defined by Giles et al. (1991) as “the ways in which we adapt our language and communication pattern toward others”. Hence, it indicates that people are able to modify their speech and that the choice to do so, although unconscious, can be strategic and motivated (Coupland & Giles 1988: 175). One can use different strategies of conformity and identification (Giles et al. 1991: 27) and thus shift in pronunciation depending on social context (Murphy 2014: 435). This idea is supported by Meyerhoff (1998: 209) whose claim is that a speaker’s choice of a particular variant “may be a response to the linguistic and social context of that turn”. Accommodation – that appears to be determined by processes occurring during the perception step (Garnier et al. 2013: 1) – can be separated into three different categories known as divergence, maintenance and convergence.

  1. One can diverge from his/her conversational partner in order to lower social approval (Giles et al. 1991: 32) and as a way of increasing social distance (Babel 2012: 179). In this case, speakers accentuate speech differences between themselves and their interlocutors.

  2. One can maintain his/her natural way of speaking in order to maintain one’s group identity (Bourhis 1979).

  3. One can adjust his/her speech to sound more similar to his/her interlocutor (Babel 2012: 177), leading to a decrease in the dissimilarities of acoustic-phonetic forms between talkers, or to “an increase in segmental and suprasegmental similarity of the speech of one talker to another” (Pardo 2006: 2384). This phenomenon has been the most frequently observed amongst conducted studies and is known as ‘phonetic convergence’.

2 Those three different processes can, however, intertwine: talkers can converge in a specific dimension while diverging in other dimensions. Divergence and convergence may also be either upward or downward, the former referring to “a shift toward a consensually prestigious variety” and the latter to “modifications toward more stigmatized or less socially valued forms in context” (Giles et al. 1991: 11)

3 In the literature on phonetic accommodation, speech rate and fundamental frequency have been found to be the two features the most subject to accommodation (Pardo 2010, Sato et al. 2013). Many other prosodic features such as utterance length, pausal phenomena (Bilous & Krauss 1998) or vowel duration (Zając 2013) are also subject to accommodation. Giles et al. (1991) depict this process as a motive of integration, as a tactic to reduce one’s cultural distinctness and as a way of achieving solidarity with a conversational partner. In addition, imitating a person’s behaviour is often used to communicate liking for this person and to create rapport. Speakers who are mimicked by their conversational partners will, in turn, like them more, resulting in a higher degree of harmony in the interaction (Nilsenovà & Swerts 2012: 88). The main intention that emerges from this concept of convergence is thus to lessen social distance, though accommodation can take place in all types of social situations. It can occur when speakers simply produce words (Goldinger 1997, 1998; Goldinger & Azuma 2003; Namy, Nygaard & Sauerteig 2002) but is mostly observable in socially rich, dyadic conversations (Pardo 2006, Pardo et al. 2010, Babel 2011).

  • 1 “The Map Task is a cooperative task involving two participants. The two speakers sit opposite one a (...)

4 Welkowitz et al. (1972) showed that pair talkers who perceived themselves as being similar converged more towards each other in terms of vocal intensity than randomly paired talkers. Giles (1973) analysed a conversation between an inspector and a traveller in a train and observed that the latter would converge towards the speech of the former. Hence, the way people speak can lead to determine which person is socially dominant in a conversation (Nilsenová & Swerts 2012: 77). It has been found that, in dyadic conversations, each person is assigned a role and people are more likely to converge towards the person who has more ‘power’ (Watzlawick et al. 1967). Following this idea, Pardo (2006) conducted the experiment of the Map Task1 (Anderson et al. 1991) to analyse the degree of convergence depending on conversational role. She tried to determine which participant – the instruction giver or the instruction follower – would phonetically converge to a higher degree towards the other. She first collected samples of speech before, during and after the conversational task. If phonetic convergence occurred, a talker’s speech would sound more similar to the interlocutor’s speech during the post task than it was before the interaction. Gender was also taken as an independent variable for women have been found to converge more than men (Babel 2012, Babel et al. 2014). Eisikovits (1987) analysed the speech of teenagers (intergroup girls/intergroup boys) and discovered that female informants converged more towards her, the female interviewer, than male informants did, as they even tended to diverge from her speech. A possible explanation to this pattern is that “women might be more sensitive to indexical features of talkers, which could lead to greater convergence” (Pardo 2006: 2388). Namy et al. (2002) carried out an experiment to assess whether gender differences in vocal accommodation took place in socially minimal situations. Participants (both male and female) were instructed to repeat isolated words produced by different talkers. Results corroborate previous findings that female accommodate more than male and that both groups converge more towards male than towards female speakers. It has, indeed, been found that women are, in general, more likely to accommodate to a conversational partner than men but also that both female and male participants tend to converge more towards male interlocutors (Bilous & Krauss 1988; Gallois & Callan 1988; Willemyns et al. 1997). The predictions for Pardo’s experiment were thus receivers converging towards givers and women converging more than men. The conclusions did not, however, follow these predictions: givers converged to receivers more than receivers converged to givers, and male speakers converged more than female speakers.

5 Most of the studies regarding the process of accommodation have been conducted amongst people sharing the same language or variety. Yet the growing interest for the investigation of second language acquisition (SLA) has led researchers to study the interlanguage processes of phonetic accommodation and to compare results with intralanguage processes. Giles and Johnson (1987) found evidence that a non-native speaker will be likely to imitate a native speaker if they both share important social identities, related to ethnicity or not. Zuengler (1982) demonstrated that L2 pronunciation can vary, by diverging or converging, if a native-English-speaking interlocutor conveys negative or positive attitude towards the ethnic group the non-native speaker belongs to. In the study he conducted, non-native speakers who perceived threat would phonetically diverge if they identified firmly as ethnic group members, or if they wanted to defend their ethnic solidarity. Zając (2013) investigated how phonetic imitation can be influenced by the model talker being a native or a non-native speaker of English by studying the duration variability of English vowels. The series of front vowels /æ, e, ɪ, iː/ were analysed in the shortening and lengthening b_t and b_d environments. The durational characteristics of this set of vowels were chosen as pre-fortis clipping which is “a feature characteristic of English pronunciation [that] may cause difficulties for Polish learners” (Zając 2013: 21). Polish learners of English were then interviewed both by a native speaker of English and by a native speaker of Polish but who would speak to them in English. She found that, in the case of pre-fortis clipping (Trask 1996: 149), the informants converged towards the native model talker and diverged from the non-native model talker. As an explanation, Zając suggested that the subjects might have been aware of the foreign accent of the non-native speaker and have thus diverged from her in order to distance themselves from other foreign-accented talkers, and converged towards the native speaker in order to sound more native-like. Conversely, Berkowitz (1986) predicted that the influence of perceived cultural empathy on L2 pronunciation could lead informants not to produce their ‘best’ L2 English. Sounding native-like appears to be tremendously difficult for language learners. Learners can speak with a fully acquired grammar or a very diversified lexicon but what tends to fail them is their pronunciation. They indeed have to be aware of the prosodic patterns (intonation, rhythm...) of the L2 and acquire them (Nilsevona & Swerts 2012: 78). However, even if factors such as learning, practising, training or motivation contribute to sound more native-like, they are found to be less predictive than what is called ‘factors of circumstances’ and luck such as length of residency in the L2-speaking country or age of acquisition (Flege et al. 1995, Piske et al. 2001, Birdsong 2006, Best & Tyler 2007). In contrast, studies have shown that L1 speakers may also adapt their speech towards L2 speakers. This is often referred as ‘foreigner talk’, an adaptation process defined by the use of a higher pitch, shorter sentences and a slower rate (Snow 1995).

6 While most of the research studies regarding phonetic accommodation have been conducted amongst people sharing the same language or variety, this preliminary research study proposes an analysis of spontaneous conversations between an advanced learner of English and two British native speakers in order to observe if phonetic accommodation towards the native speakers occurred. Two conversations were extracted from the Diderot Longdale corpus. Segmental (vowel quality) and suprasegmental features (vowel duration, speech rate) were analysed to determine whether conclusions drawn from this study correlate previous findings. Speech rate has been found to be one of the features the most subject to accommodation (Pardo 2006). We also expect to observe convergence in vowel duration in order to sound more native-like (Zając 2013) and low vowels to be more imitated than high vowels especially within the F1 dimension (Babel 2009).

2. Corpus

7 The data used for this research study was collected from the Diderot Longdale corpus (Goutéraux 2015) in which 15 advanced learners of English (12 female and 3 male) were recorded in spontaneous conversations with a native speaker over a period of two years. For this preliminary investigation, though, only one male advanced learner is analysed, resulting in 3 recordings lasting 4’8” on average (a 15-minute speech sample was, therefore, analysed). The first interview (which carries on during the first two recordings) was conducted by a female British native speaker and the second one by a male British native speaker. In the first session, the student had to describe a country while in the second he had to describe a film/play or ‘an experience which has taught you an important lesson’, reproducing the LINDSEI protocol (Hillenbrand et al. 1995).

8 All interviews were recorded in an individual stereo 16-bit resolution sound file at a sampling rate of 44100 Hz captured in an uncompressed, pulse code modulation format using an Apex435 large diaphragm studio condenser microphone with cardioid polar pattern.

  • 2 The study of durational characteristics in case of pre-fortis clipping could have been interesting (...)

9 The back vowels /ɑː, ʊ, uː/ and front vowels /ɪ, iː, e, æ/ were manually segmented but automatically extracted, resulting in the analysis of 745 phonemes. These monophthongs were selected because they were the ones which would occur most frequently through the two interactions. While researchers usually analyse vowels occurring in the same phonological environment to compare the production of two speakers, the influence of the following consonants (e.g. locus equation) has not been analysed in this preliminary study due to the small size of the corpus. However, it would have been particularly interesting for the investigation of vowel duration for the voicing of the surrounding consonants tends to impact vowel duration2 (Rossi et al. 1981: 51).

10 On top of the statistical question of the representativeness of our sample used as a corpus, the distribution of vowels, as displayed on Figure 1, calls for specific comments on the protocol. Obviously, native speakers more often only try to keep the conversation going. This can be illustrated by many examples of non-committal statements such as “ok so, can you tell me more about it, where did you go in the United States?” or “what things did you do while you were there?”. The pragmatic limitations of our native speakers in the classic learner corpus research data collection (Granger et al. 2015) may reflect on the lexical scarcity of the observable data but does not change the possibility of the phonetic accommodation and its specific effect on learners. However small our sample may be, it is illustrative of the unequal distribution of vowels in spoken data.

Figure 1: Distribution of vowels per speaker

Figure 1: Distribution of vowels per speaker

FNS correspond to the female native speaker and MNS to the male native speaker

3. Method

11 Values for F1 and F2 were extracted using a PRAAT (Boersma & Weenink 2015) script for each target vowel at different percentages of the distance from the onset to the offset (respectively 0 %, 25 %, 50 %, 75 % and 100 %) of the vowels. However, only the values at 25 %, 50 % and 75 % were kept in order to avoid too much coarticulatory influence as the phonological environment of vowels has not been taken into account for this preliminary research study. Coarticulation can be assigned to the influence of a following segment, a phenomenon known as anticipatory coarticulation, or to the influence of a preceding segment, also called ‘persevatory coarticulation’ (Thomas 2011: 149). The average F1 and F2 values were then calculated based on the three different values extracted in order to obtain a single, mean value for each vowel. All vowels were normalized using the Lobanov method (Thomas 2011: 166). For a detailed discussion of the normalisation issues of this learner corpus, see Méli & Ballier (2015).

12 A script designed by Nivja H. De Jong & Ton Wempe (2009) was used to measure speech rate by automatically detecting syllable nuclei. Speech rate was, therefore, calculated in terms of syllables per time unit. A combination of intensity and voicedness was used to detect syllable nuclei (De Jong & Wempe 2009: 386).

13 Vowel duration was extracted from the onset to the offset for each target vowel using a PRAAT script designed by Mietta Lennes (2002).

4. Results

4.1. Analysis of formant frequencies

Table 1 : Percentages of deviation in the three dimensions (F1, F2, F1/F2) per speaker

Low Vowels

Speaker

F1

F2

F1/F2

Female native speaker

67 %

33 %

0 %

Male native speaker

100 %

0 %

0 %

Learner

67 %

16,5 %

16,5 %

High Vowels

Speaker

F1

F2

F1/F2

Female native speaker

20 %

70 %

10 %

Male native speaker

20 %

80 %

0 %

Learner

20 %

53 %

27 %

Back Vowels

Speaker

F1

F2

F1/F2

Female native speaker

40 %

60 %

0 %

Male native speaker

0 %

100 %

0 %

Learner

33 %

45 %

22 %

Front Vowels

Speaker

F1

F2

F1/F2

Female native speaker

25 %

62,5 %

12,5 %

Male native speaker

30 %

50 %

0 %

Learner

33 %

42 %

25 %

14 Table 1 displays the percentages of deviation in the F1, F2 and F1/F2 dimensions per speaker: 1) F1 indicates a higher degree of deviation in the F1 dimension; 2) F2 indicates a higher degree of deviation in the F2 dimension; 3) F1/F2 indicates a similar deviation in both dimensions.

15A higher degree of deviation means that the pronunciation of a vowel has shifted to a higher degree in a specific dimension. For instance, we can see that the production of low vowels when produced by the learner is represented by 67 % of deviation in the F1 dimension. This indicates that the learner has produced differently these vowels to a higher degree in terms of height – either higher or lower – through the course of the interaction. In other words, the position of low vowels when produced by the learner has more shifted in the F1 dimension than in the F2 dimension within the phonetic space.

16 The conclusion that can be drawn from this table is that high vowels, back vowels and front vowels show a higher degree of deviation in the F2 dimension – thus more deviation regarding advancement (Thomas 2011: 145) – when produced by all interlocutors. Conversely, low vowels show a higher degree of deviation in the F1 dimension, indicating more deviation in terms of height (Thomas 2011: 145). This latter observation could correlate Babel’s (2009) conclusion that low vowels are more subject to imitation, and especially within the F1 dimension since they are subject to more deviation.

4.2. Analysis of speech rate

17 Speech rate has been proved to be one of the suprasegmental features the most prone to be imitated (Pardo 2006). Figure 2 displays the evolution of speech rate for each interlocutor according to time. FNS corresponds to the female native speaker and MNS to the male native speaker. Their productions are represented by the solid line while the production of the L2 learner is represented by the dotted line. The smooth curve indicates the conditional mean and the grey zone represents the confidence interval. It illustrates the “range” in which our predictions would be if we were to repeat the sampling over and over. In other words, it suggests how much uncertainty there is in the smoothing curve.

Figure 2 : Evolution of speech rate through time for the three recordings

Figure 2 : Evolution of speech rate through time for the three recordings

FNS corresponds to the female native speaker and MNS to the male native speaker. Their productions are represented by the solid line while the production of the L2 learner is represented by the dotted line.

18The first graph depicts the evolution of speech rate for the first part of the interaction between the female native speaker and the L2 learner. We can observe that, at the beginning of the interaction, the female native speaker speaks at a faster rate than the student which is not surprising for native speakers usually speak at a faster rate than non-native speakers. Her speech rate, however, decreases during the first half of the recording, implying that she has converged towards the L2 learner. A number of studies have, indeed, shown that L1 speakers adapt their speech to L2 speakers by, for instance, using a higher pitch or a slower rate (Snow 1995). At the same time, the L2 learner increases his speech rate, leading him to speak faster than the native speaker. The fact that the two curves intersect several times show that both speakers accommodate towards each other.

19 On the second graph – which displays the evolution of speech rate for the second part of the interaction between the female native speaker and the L2 learner – we can observe that both curves follow the same pattern. The only difference is that the student’s speech rate is always slightly slower than the female native speaker’s. We can, however, observe complete convergence at the end of the recording.

20 The third graph depicts the evolution of speech rate for the interaction between the male native speaker and the L2 learner. As during the first interaction, the L1 speaker speaks at a faster rate than the L2 learner. Moreover, he adapts his speech rate to the student’s as well during the first part of the interaction by decreasing it. Regarding the L2 learner’s production, we can observe that he keeps increasing his speech rate through the whole interaction. At around 150s, both speakers have almost reached the same rate of speech. During the first part of the interaction the L1 speaker slows down his speech rate but then increases it as the learner increases it as well. He then might have realised that he was still speaking at a higher rate for during the last part of the interaction his speech rate decreases again, leading the student to speak at a higher rate at the end of the interaction.

21 The overall conclusion that can be drawn from the analysis of speech rate is that all interlocutors have accommodated towards one another, correlating Pardo’s (2006) observation that speech rate tends to be constantly imitated.

4.3. Analysis of vowel duration

Figure 3: Boxplot displaying vowel duration for each speaker in both interactions

Figure 3: Boxplot displaying vowel duration for each speaker in both interactions

FNS corresponds to the female native speaker and MNS to the male native speaker.

  • 3 The median corresponds to the number separating the higher half of a data sample from the lower hal (...)

22Figure 3 displays the duration of all vowels for each speaker through time. The general observation is that the L2 learner tends to produce longer vowels. He seems, however, to be producing shorter vowels when speaking with the male native speaker (where duration d is between 45ms and 120s) than with the female native speaker (50ms < d > 145ms). Moreover, the female native speaker tends to produce slightly longer vowels (50ms < d > 90s) than the male native speaker (50ms < d > 85ms). The fact that the notches in Figure 3 are overlapping in each case is strong evidence that the medians3 do not differ (McGill et al. 1978: 14). In addition, the observation made that the student produces longer vowels when speaking to the female native speaker and shorter vowels when speaking to the male native speaker might be evidence that the student converged in terms of duration towards both interlocutors.

Figure 4: Boxplots displaying vowel duration for each speaker

Figure 4: Boxplots displaying vowel duration for each speaker

23Figure 4 displays the overall duration of each vowel for each speaker in both interactions. The values given by the medians for the female native speaker appear to be slightly higher than the ones for the male native speaker. However, we can observe that the first quartile is always bigger than the third quartile for the female native speaker’s production, meaning that the majority of the vowels produced is longer than the duration given by the median. The reverse pattern is observable in the case of the male native speaker’s production. This indicates that he produces a higher amount of vowels whose duration is shorter than the duration given by the median. These results corroborate our previous observation that the female native speaker produces longer vowels than the male native speaker. Moreover, the values of the medians for the student are all higher in the first interaction. Yet, in both interactions, the student produces a higher amount of vowels whose values are higher than the value given by the median. Overall though, all vowels appear to have a longer duration in the first interaction than in the second interaction. This observation is strongly correlated to the fact that, when a speaker increases his/her rate of speech, the duration of the produced vowels will automatically be reduced. This additional observation leads us to the same conclusion that convergence from the L2 learner towards both native speakers has taken place.

24 In order to validate this assumption, plots (Figure 5) showing the evolution of vowel duration through time were created, allowing a comparison between the different interlocutors’ production. Unfortunately, we have not been able to create plots for all vowels due to the lack of tokens for certain vowels.

25 The overall conclusion that can be drawn from this analysis is that every interlocutor has converged towards the other in terms of vowel duration. We can observe that the curves tend to follow the same pattern and sometimes intersect, which indicates that interlocutors have reached the same speech rate. Moreover, it appears that the L2 learner’s vowel duration is always shorter at the end of the interactions. This pattern is observable for the production of all vowels, and correlates Zając’s (2013) observation that a non-native speaker tends to converge towards a native speaker in terms of vowel duration in order to sound more native-like, or – as one of the reviewers suggested – it might be due to the fact that they feel more confident at the end than at the beginning of the interaction.

Figure 5: Duration of the FLEECE, KIT, GOOSE and DRESS vowels through time for both interactions

Figure 5: Duration of the FLEECE, KIT, GOOSE and DRESS vowels through time for both interactions

Recordings 1 and 2 correspond to the first interaction and Recording 3 corresponds to the second interaction. FNS corresponds to the female native speaker and MNS to the male native speaker. Their productions are represented by the solid line while the production of the L2 learner is represented by the dotted line.

5. Discussion and conclusion

26 Phonetic accommodation has mostly been studied amongst speakers sharing the same language or the same variety. Over the last decades, though, this process has been of major concern in the case of native/non-native interactions. The aim of this preliminary research study was to determine whether some of the conclusions drawn (mostly) from previous intralanguage studies could also be drawn in the case of native/non native interactions.

  • 4 http://www.clillac-arp.univ-paris-diderot.fr/projects/longdale

27 Two spontaneous conversations between an L2 advanced learner of English and two British native speakers (1 male and 1 female) were extracted from the Diderot - Longdale corpus.4 Vowel quality, vowel duration and speech rate were analysed.

28 Speech rate has been found to be one of the features the most subject to accommodation during dyadic conversations (Pardo 2006). The results obtained show that this also appears to be true in the case of native/non-native interactions. The L2 learner was expected to increase his speech rate, for native speakers are known to speak at a faster rate than non-native speakers. This assumption appeared to be validated. Nonetheless, the L2 learner would still, most of the time, speak at a slower rate than the native speakers.

29 Zając (2013) conducted a research study amongst Polish English learners and observed convergence in vowel duration towards the native speaker in order to sound more native-like. This appears to be also the case for a French learner of English. This observation is strongly correlated with the fact that the L2 learner accommodated in terms of speech rate. Indeed, speaking at a slower rate would result in the production of longer vowels, and vice versa. Since the L2 learner would increase his speech rate through the different interactions, the duration of his vowels would decrease through time as well.

  • 5 Non-function words repeated by the interlocutors were: you, to and me for in first interaction, and (...)

30 The analysis of formant frequencies has shown that there was more deviation in the F1 dimension regarding the production of low vowels, corroborating Babel’s (2009) results that low vowels tend to be more subject to imitation than high vowels, and especially within the F1 dimension. The corpus was, however, not appropriate for the study of accommodation. Among the issues met by this kind of analysis based on spontaneous data is “embeddedness”: the same word may be pronounced by different speakers or by the same speaker at different time of the recording and the same vowel may be contained in other words that are similarly repeated by the co-speaker. Analysing accommodation based on the same lexical template (identical phonological environment) would lead to even more complicated issues as the same word may not be in the same position in the prosodic hierarchy and, therefore, be hardly comparable. If anything, the candidates5 for echoing words could be summed up in the following table:

Table 2: Repeated words occurring in both interactions. The number of occurrences is displayed for each speaker.

First interaction

Repeated words

you

to

me

Female native speaker

17

20

3

Learner

21

17

5

Second interaction

Repeated words

have

in

do

Male native speaker

3

3

2

Learner

3

3

2

31 While phonetic convergence may be facilitated when the language distance is close (same L1 or same dialect) (Kim et al. 2011: 137), results from previous studies, and conclusions drawn from this preliminary study, show that phonetic convergence can also occur at different levels in native/non-native interactions. Because we are aware of the insufficiency of data for what is only a preliminary investigation of the research question, our conclusions try to give a fairer appreciation of the representativeness of our data. These preliminary conclusions being promising, further research is, however, needed to establish patterns of accommodation in native/non-native interactions.

  • 6 T-flapping is a phonological process found in several varieties of English, mostly in American Engl (...)

32What should also be borne in mind is that the two native speakers were British so that we could have investigated the realisation of certain consonants (e.g. the consistency of the rhoticity, specific aspiration in Voice Onset Time (alveolar VS. dental) among the non-natives, etc.). The L2 learner in this sample tends, for instance, to speak with an American accent. Several occurrences of t-flapping6 have indeed been identified, as in the realisation of the word writing (thus pronounced as [ˈraɪɾɪŋ]). It would then be interesting to analyse the realisations of /t/ in intervocalic position through the course of the interactions to observe whether accommodation took place or not regarding this specific consonant.

Top of page

Bibliography

6. References

Anderson, Anne H., Miles Bader, Ellen Gurman Bard, Elizabeth Boyle, Gwyneth Doherty, Simon Garrod, Stephen Isard, Jacqueline Kowtko, Jan McAllister, Jim Miller, Catherine Sotillo, Henry S. Thomson and Regina Weinert. “The HCRC Map Task Corpus.” Language and Speech 34.4 (1991): 351-366.

Babel, Molly. “Selective Vowel Imitation in Spontaneous Phonetic Accommodation.” UC Berkeley Phonology Lab Annual Report, 2009. DOI: https://escholarship.org/uc/item/6c95w485.

Babel, Molly. “Evidence for Phonetic and Social Selectivity in Spontaneous Phonetic Imitation.” Journal of Phonetics 40.1 (2012): 177-189.

Babel, Molly. “Imitation in Speech.” Acoustics Today 7.4 (2011): 16-22.

Babel, Molly, Grant McGuire, Sophia Walters and Alice Nicholls. “Novelty and Social Preference in Phonetic Accommodation.” Laboratory Phonology 5.1 (2014): 123-150.

Ballier, Nicolas and Adrien Méli. “CV-patterned Transfers among French Speakers of English.” Proceedings of the EPIP4 - 4th International Conference on English Pronunciation: Issues and Practices (2015): 14-18.

Berkowitz, Diana. The Effects of Perceived Cultural Empathy on Second Language Performance. Unpublished Ph.D. dissertation, Columbia University, 1986.

Best, Catherine and Michael Tyler. “Nonnative and Second-language Speech Perception: Commonalities and Complementarities.” In M. J. Munro and O. S. Bohn (eds.), Second Language Speech Learning: The Role of Language Experience in Speech Perception and Production. Amsterdam: John Benjamin, 2007. 13-34.

Bilous, Frances and Robert Krauss. “Dominance and Accommodation in the Conversational Behaviours of Same- or Mixed-gender Dyads.” Language & Communication 8 (1988): 183-194.

Birdsong, David. “Age and Second Language Acquisition and Processing: A Selective Overview.” Language Learning 56.1 (2006): 9-49.

Boersma, Paul and David Weenink. Praat: Doing Phonetics by Computer (Version 5.4.08), 2015. DOI: http://www.praat.org.

Bourhis, Richard. “Language in Ethnic Interaction: A Social Psychological Approach.” In H. Giles and B. Saint-Jacques (eds.), Language and Ethnic Relations. Oxford: Pergamon, 1979. 117-141.

Coupland, Nicolas and Howard Giles. “Introduction: The Communicative Contexts of Accommodation.” Language and Communication 8.3/4 (1988): 175-182.

De Jong, Nivja and Tom Wempe. “Praat Script to Detect Syllable Nuclei and Measure Speech Rate Automatically.” Behavior Research Methods 41.2 (2009): 385-390.

Eisikovits, Edina. “Sex Differences in Inter-group and Intra-group Interaction among Adolescents.” In A. Pauwels (ed.), Women and Language in Australian and New Zealand Society. Sydney: Australian Professional Publications, 1987. 45-58.

Flege, Munro and Ian R. A. MacKay. “Factors Affecting Strength of Perceived Foreign Accent in a Second Language.” Journal of the Acoustical Society of America 97.5 (1995): 3125-3134.

Gallois, Cynthia and Victor Callan. “Communication Accommodation and the Prototypical Speaker: Predicting Evaluations of Status and Solidarity.” Language and Communication 8 (1988): 271-283.

Garnier, Maëva, Laurent Lamalle and Marc Sato. “Neural Correlates of Phonetic Convergence and Speech Imitation.” Frontiers in Psychology 4.600 (2013): 1-15.

Giles, Howard, and Patricia Johnson. “Ethnolinguistic Identity Theory: A Social Psychological Approach to Language Maintenance.” International Journal of the Sociology of Language 68. (1987): 69-99.

Giles, Howard, Justine Coupland and Nikolas Coupland. Contexts of Accommodation: Developments in Applied Sociolinguistics. Cambridge: Cambridge University Press, 1991.

Goldinger, Stephen. “Perception and Production in an Episodic Lexicon.” In K. Johnson and J. W. Mullenix (eds.), Talker Variability in Speech Processing. San Diego: Academic Press, 1997. 33-66.

Goldinger, Stephen. “Echoes of Echoes? An Episodic Theory of Lexical Access.” Psychological Review 105.2 (1998): 251-279.

Goldinger, Stephen and Tamiko Azuma. “Puzzle-solving Science: The Quixotic Quest for Units in Speech Perception.” Journal of Phonetics 31. (2003): 305-320.

Granger, Sylviane, Gaëtanelle Guilquin and Fanny Meunier. The Cambridge Handbook of Learner Corpus Research. Cambridge: Cambridge University Press, 2015.

Hillenbrand, James, Laura A. Getty, Michael J. Clark and Kimberlee Wheeler. “Acoustic Characteristics of American English Vowels.” The Journal of the Acoustical society of America 97.5 (1995): 3099-3111.

Kim, Horton and Ann Bradlow. “Phonetic Convergence in Spontaneous Conversations as a Function of Interlocutor Language Distance.” Laboratory Phonology 2.1 (2011): 125-156.

Lennes, Mietta. “Mietta’s Praat scripts”, 2015. DOI: http://www.helsinki.fi/~lennes/praat-scripts/.

McGill, Robert, John W. Tukey and Wayne A. Larsen. “Variations of Box Plots.” The American Statistician 32.1 (1978): 12- 16.

Méli, Adrien and Nicolas Ballier. “Assessing L2 Phonemic Acquisition: A Normalisation-independent Method?” International Congress of Phonetic Sciences (ICPhS 2015), Aug 2015, Glasgow, United Kingdom. Proceedings of the 18th International Congress of Phonetic Sciences, The University of Glasgow (2015): 805-810.

Meyerhoff, Miriam. “Accommodating Your Data: The Use and Misuse of Accommodation Theory in Sociolinguistics.” Language and Communication 18 (1998): 205-225.

Murphy, Monique. “Sociophonetic Convergence in Native and Non-Native Speakers of French.” Proceedings of the International Symposium on the Acquisition of Second Language Speech. Concordia Working Papers in Applied Linguistics 5. (2014): 435-450.

Namy, Laura L., Lynne C. Nygaard and Denise Sauerteig. “Gender Differences in Vocal Accommodation: The Role of Perception.” Journal of Language and Social Psychology 21.4 (2002): 422-443.

Nilsenová, Marie and Marc Swerts. “Prosodic Adaptation in Language Learning.” In J. Romero-Trillo (ed.), Pragmatics and Prosody in English Language Teaching. Dordrecht: Springer Science+Business Media B.V., (2012): 77-94.

Pardo, Jennifer S. “On Phonetic Convergence During Conversational Interaction.” The Journal of the Acoustical Society of America 119.4 (2006): 2382-2393.

Pardo, Jennifer S. “Expressing Oneself in Conversational Interaction.” In E. Morsella (ed.), Expressing Oneself/Expressing One’s Self.. New-York: Psychology Press, (2010): 183-196.

Pardo, Jennifer S., Isabel Cajori Jay and Robert Krauss. “Conversational Role Influences Speech Imitation.” Attention, Perception, & Psychophysics 72.8 (2010): 2254-2264.

Piske, Thorsten, Ian R. A. MacKay and James Emil Flege. “Factors Affecting Degree of Foreign Accent in an L2: A Review.” Journal of Phonetics, 29.2 (2001): 191-215.

Sato, Marc, Krystyna Grabski, Maëva Garnier, Lionel Granjon, Jean-Luc Schwartz and Noël Nguyen. “Converging towards a Common Speech Code: Imitative and Perceptuo-motor Recalibration Processes in Speech Production.” Frontiers in Psychology 4.422 (2013): 1-14.

R Development Core Team. R: A Language and Environment for Statistical Computing. R Foundation. for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0. DOI: http://www.R-project.org.

Rossi, Mario, Albert Di Cristo, Daniel Hirst, Philippe Martin and Yukihiro Nishinuma. L’intonation : de l’acoustique à la sémantique. Paris : Klincksieck, 1981.

Snow, Catherine. “Issues in the Study of Input: Fine-tuning, Universality, Individual and Developmental Differences and Necessary Causes.” In P. Fletcher and B. MacWhinney (eds.), The Handbook of Child Language. Oxford: Blackwell, 1995. 180-194.

Thomas, Erik. Sociophonetics: An Introduction. Basingstoke, UK: Palgrave Macmillan, 2011.

Trask, Robert L. Historical Linguistics. London, UK: Hodder Education Publishers, 1996.

Zając, Magdalena. “Phonetic Imitation of Vowel Duration in L2 sSeech.” Research in Language 11.1 (2013): 19-29.

Watzlawick, Beavin and Don Jackson. Pragmatics of Human Communication. New York: Norton, 1967.

Welkowitz, Joan, Stanley Feldstein, Mark Finkelstein and Lawrence Aylesworth. “Changes in Vocal Intensity as a Function of Interspeaker Influence.” Perceptual and Motor Skills 35.3 (1972): 715-718.

Willemyns, Michael, Cynthia Gallois, Victor J. Callan and Jeffery Pittam.”Accent Accommodation in the Job Interview: Impact of Interviewer Accent and Gender.” Journal of Language and Social Psychology 16. (1997): 3-22.

Zuengler, Jane. “Applying Accommodation Theory to Variable Performance Data in L2.” Studies in Second Language Acquisition 4 (1982): 181-192.

Top of page

Notes

1 “The Map Task is a cooperative task involving two participants. The two speakers sit opposite one another and each has a map which the other cannot see. One speaker – designated the Instruction Giver – has a route marked on her map; the other speaker – the Instruction Follower – has no route. The speakers are told that their goal is to reproduce the Instruction Giver’s route on the Instruction Follower’s map. The maps are not identical and the speakers are told this explicitly at the beginning of their first session. It is, however, up to them to discover how the two maps differ.”, Definition retrieved from the HCRC Map Task Guide.

2 The study of durational characteristics in case of pre-fortis clipping could have been interesting but, unfortunately, many occurrences were words ending with an open syllable, especially when they were produced by the native speakers. Very few tokens would have been extracted and it would have been really difficult to compare the production of the different locutors.

3 The median corresponds to the number separating the higher half of a data sample from the lower half. It corresponds to the middle value between all observations from the lowest value to the highest.

4 http://www.clillac-arp.univ-paris-diderot.fr/projects/longdale

5 Non-function words repeated by the interlocutors were: you, to and me for in first interaction, and have, in and do in the second interaction. Again, one should take into account the position and phonological environment for strong and reduced realisations of the function words.

6 T-flapping is a phonological process found in several varieties of English, mostly in American English. The consonant /t/ is pronounced as an alveolar flap in intervocalic positions.

Top of page

List of illustrations

Title Figure 1: Distribution of vowels per speaker
Caption FNS correspond to the female native speaker and MNS to the male native speaker
URL http://journals.openedition.org/anglophonia/docannexe/image/1127/img-1.png
File image/png, 29k
Title Figure 2 : Evolution of speech rate through time for the three recordings
Caption FNS corresponds to the female native speaker and MNS to the male native speaker. Their productions are represented by the solid line while the production of the L2 learner is represented by the dotted line.
URL http://journals.openedition.org/anglophonia/docannexe/image/1127/img-2.jpg
File image/jpeg, 120k
Title Figure 3: Boxplot displaying vowel duration for each speaker in both interactions
Caption FNS corresponds to the female native speaker and MNS to the male native speaker.
URL http://journals.openedition.org/anglophonia/docannexe/image/1127/img-3.jpg
File image/jpeg, 40k
Title Figure 4: Boxplots displaying vowel duration for each speaker
URL http://journals.openedition.org/anglophonia/docannexe/image/1127/img-4.png
File image/png, 28k
Title Figure 5: Duration of the FLEECE, KIT, GOOSE and DRESS vowels through time for both interactions
Caption Recordings 1 and 2 correspond to the first interaction and Recording 3 corresponds to the second interaction. FNS corresponds to the female native speaker and MNS to the male native speaker. Their productions are represented by the solid line while the production of the L2 learner is represented by the dotted line.
URL http://journals.openedition.org/anglophonia/docannexe/image/1127/img-5.jpg
File image/jpeg, 283k
Top of page

References

Electronic reference

Léa Burin and Nicolas Ballier, « Accommodation in learner corpora: A case study in phonetic convergence », Anglophonia [Online], 24 | 2017, Online since 28 November 2017, connection on 28 May 2018. URL : http://journals.openedition.org/anglophonia/1127 ; DOI : 10.4000/anglophonia.1127

Top of page

About the authors

Léa Burin

Université Paris-Diderot,EA 3967 (CLILLAC-ARP)
lea.burin@univ-paris-diderot.fr

Nicolas Ballier

Université Paris-Diderot,EA 3967 (CLILLAC-ARP)
nicolas.ballier@univ-paris-diderot.fr

Top of page

Copyright

Licence Creative Commons
Anglophonia – French Journal of English Linguistics est mis à disposition selon les termes de la licence Creative Commons Attribution - Pas d'Utilisation Commerciale - Pas de Modification 4.0 International.

Top of page
  • Logo Presses universitaires du Midi
  • OpenEdition Journals