Research on endangered languages generally focuses on documenting speech patterns of remaining fluent speakers (FSs). Yet in a language death context, it is common to find a final generation of non-fluent speakers called semi-speakers (Dorian 1973, 1977). Semi-speakers feature incomplete acquisition of the target language and non-targetlike speech patterns. For this reason, their linguistic practices can provide a window into ways variation is interpreted and acquired by non-fluent speakers. With this study, we apply methods from second-language-acquisition research (Andersen 1989; Geeslin & Long 2014) to the study of semi-speaker variation, adding to the sparse research on sociolinguistic variation in language-death contexts.
We examine first-person-singular subject expression in Louisiana French (LF), which is undergoing gradual language death (Picone 1997). In the variety of LF we examine, there are 12 potential ways of expressing the first-person-singular referent in subject position. This extreme variation exists in part because of phonetic alternations whereby the first-person singular clitic ‘je’ may be pronounced as variants such as /ʒ/, /h/, and Æ, and because the non-clitic pronoun ‘mon’ may be variably appended. While there is research on phonological variation in the first-person pronoun (Carmichael 2008), pro-drop (Dajko 2009), and the rise of ‘mon’ usage in LF (Rottet 1995, 1996, 2001, 2005), there has yet to be a unified explanation of the linguistic and social factors that predict the phonological and morphological variation observed with first-person-singular subject expression. We address this issue by analyzing data from 28 interviews with LF speakers, divided according to fluency: FSs and semi-speakers. To compare the linguistic systems of FSs and semi-speakers, we generated a series of multinomial logistic regressions examining the impact of eight linguistic and social factors on first-person subject expression. The results revealed subtle similarities in the semi-speakers’ use of first-person subject forms, although their linguistic system largely differs from that of the FSs. Furthermore, we demonstrate that, like non-native speakers, semi-speakers’ language patterns exhibit two types of systematic variability: that which is also present in FSs and that which is unique to semi-speakers. Thus, a second-language-acquisition approach permits us to model variation across fluency groups within this language-death context.
