Interplay of sociolinguistic factors in rhythmic variation in a minority French dialect Svetlana Kaminskaïa 1 1 University of Waterloo skaminsk@uwaterloo.ca Abstract A contact setting often limits speakers’ possibilities to practice the minority language in an array of contexts and thus affects its stylistic and sociolinguistic variation. This paper examines rhythmic variation in read and spontaneous speech samples from speakers of minority Ontario French (Canada). Rate, nPVI-V, VarcoV, %V and the CC model are used to examine the extent of sociolinguistic variation in the dataset and test the hypotheses of convergence to English and of sociolinguistic discontinuity. Age, gender and speaking style each appear to be significant factors, without showing interaction. Females and older speakers showed a more syllable-timed pattern than males and younger participants. In spontaneous speech, all speakers had a less syllable-timed rhythmicity, despite a faster rate. Overall, speakers did not converge to English. Index Terms: rhythm, metrics, French in a minority setting, social and stylistic variation 1. Introduction Sociolinguistic variation in a language spoken in a situation of intense contact with a dominant language often becomes reduced due to limited possibilities for the minority language to be employed. This may lead to the same forms being used, i.e. decreased variation across an array of contexts and by various populations, or to a situation where spontaneous interactions will be marked by a transfer from the dominant language. For example, [1] observed that fluent, or unrestricted, native speakers of French use vernacular forms and pronunciation, whereas more restricted speakers adhere to standard forms maintained at school. As a consequence, syntactic, lexical and prosodic structures are regularly transferred from English into French in this group [2, 3, 4]. In Ontario, linguistic restriction is often correlated with age [5] because younger generations, being more socially and geographically mobile, are more in contact with Anglophone speakers, media, and culture. Also, women, being on the one hand leaders of sociolinguistic changes [6], and on the other, tending towards a supralocal norm [7], may either join younger generation, or adhere to more standard variants. This study focuses on prosodic rhythm in French spoken in Ontario through examining the extent of sociolinguistic variation in a minority setting. The question addressed here is: will we observe variation where groups of speakers, depending on the type of production (spontaneous speech or reading task), converge to an English stress-timed pattern, or will variation observed be within the range of a syllable-timed French? Older participants, are expected to demonstrate French timing. Younger participants, however, are predicted to show a less syllable-timed pattern overall, and even more so in spontaneous speech. Women are expected to show a very syllable-timed pattern in the reading task, and to stay within the range of French rhythmicity in a less controlled interaction. As for men, they will probably converge to English: the younger generation because of the reasons specified above, and the older one, because of a greater contact with the dominant language at work (vs. women of the older generation staying at home, and thus having less interaction with English speaking environments). This is the first analysis of this kind conducted on Canadian French data in general, and on a minority French spoken in an intense contact with English in particular. Using popular methods of rhythmic analysis (nPVI-V, VarcoV, %V, and the CC model), this work allows us to compare current results with studies of other languages and dialects, and thus contributes to the general discussion about the performance of rhythm metrics. 2. Rhythm metrics French is described as a syllable-timed language [8, 9], or, in terms of [10], as a “controlling” language, where all elements in a syllable receive similar articulatory efforts. Such languages (cf. Spanish) show no vocalic reduction and prefer CV type syllables. Stress-timed, or “compensating” languages, such as English, demonstrate a more diverse syllabic typology accompanied by variably distributed articulatory efforts across syllables; these lead to co-articulations, overlaps and reductions. Such phonological properties of languages impact the relationships between vocalic and consonantal intervals. The variability of the intervals has thus been traditionally measured to assess the rhythmic pattern of a speech sample as leaning toward the stress-timed or the syllable-timed end of the rhythmic continuum. Preferred syllable structure and relative durational equality of the intervals in syllable-timed languages lead to a higher proportion of vocalic intervals (%V [11]) but to lower values of raw or rate-normalized standard deviations (ΔV, ΔC [11], VarcoV [12], VarcoC [13]) or other metrics, such as nPVI-V [14], in comparison with stress-timed languages. Studies of first and second languages, of various dialects and of languages in contact have applied rhythm metrics to test their discriminatory power. While duration-based metrics remain most exploited, other correlates of prominence and their combinations become utilized ([15, 16, 17], among others). Some analyses demonstrated unstable or contradictory results, and the metrics have been criticized for their dependence on the phonotactic composition of the material uttered, as well as for not being able to reflect the origins of the variation observed (see [18, 19] among others). Moreover, rate has proven to be an important contributor to social and dialectal variation ([20, 21, 22, 23]), whereas the best performing metrics neutralize its effect. The CC (“Control/Compensation”) model claims to solve these issues by offering a method that takes into Speech Prosody 2016 31 May - 3 Jun 2106, Boston, USA 1191 doi: 10.21437/SpeechProsody.2016-245