Converging to local and supralocal norms: stylistic variation in migrants’ speech

Resumo

While stylistic variation has been the focus of several sociolinguistic studies, addressing prestige (e.g. Labov 2001), interlocutors (e.g. Bell 1984), identities and social practices (e.g. Eckert 2004), little is known about stylistic variation in internal migrants’ speech. Hora; Wetzels (2010) found migrants to São Paulo from NE Brazil used more of the northeastern [h] variant of coda /R/ in formal style, despite its local stigmatization. However, migrants in dialect contact situations potentially face multiple competing linguistic norms, not only between prestigious vs. stigmatized variants, but also between local and supralocal norms (Guy 2016). This paper analyzes the effect of style, gender, age of arrival, and length of residence in the host community, in a sample of 24 Northeastern migrants from Alagoas living in Campinas (SE Brazil), balanced for the three social predictors. Style is defined as in Labov’s (2001) Decision Tree model (careful: response, language, soapbox, residual; casual: narrative, group, kids, tangent). The main questions are whether there is change over time and to which norms speakers converge in different styles. We analyzed two phonetic variables: coda /R/ (e.g. porta ‘door’) as fricative or tap/retroflex, and coda /S/ before /t, d/ (e.g. pasta ‘folder’), as palatal or alveolar, in mixed-effects logistic regression models in R, including external and internal predictors (preceding and following phonological context, syllable stress, position in the word, word class), along with Speaker and Lexical Item as random effects. While the migrants’ native /R/ realization coincides with the national norm (fricative) and differs from the host community norm (tap/retroflex), their /S/ realization (palatal) differs from the national and local norm (alveolar). Results from the analysis of 4,211 tokens of /R/ show that speakers tend to accommodate to the local variant over time, with significant correlation with length of residence (logodds 0.08, p<.001), gender (local variant disfavored by women: logodds –1.90, p<.001), and style (local variant favored in the “soapbox” style, compared to “narrative”: logodds 0.50, p=.004; disfavored by “response” and “language”: logodds –.41 and –1.21 respectively, p<.001). The analysis of 1,247 tokens of /S/ before /t, d/ show no correlation with length of residence (p=.60) and gender (p=.91), near significance for age of arrival (logodds: –.05, p=.08) and significant correlation with style (local/national variant disfavored by “language”, compared to “narrative”: logodds –.50, p=0.03). Although we observe accommodation to the local tap/retroflex variants of /R/ over time, migrants prefer their native variant in “response”, “language” (Labov’s “careful”) topics for /R/ and in “language” for /S/. In the “soapbox” style, where speakers talk about general problems such as crime and corruption, the preference for the local host community variant suggest a more nuanced categorization than simply “careful” and “casual” styles, with local and supralocal norms at play. The gender difference for /R/ indicate that women attend to supralocal while men to local norms, a result also found in other studies (e.g. Oushiro 2020). Stylistic variation in migrants’ speech thus sheds light on competing factors driving accommodation over time.