Conditioning variables at interfaces for Spanish pronominal subject expression : heritage language learners across proficiency levels



Journal Title

Journal ISSN

Volume Title



Contributing to a more detailed understanding of heritage language (HL) learner levels of Spanish competence, this study examines the first-person singular Spanish pronominal subject expression by 17 HL learners across three levels of proficiency in oral interviews – as rated by modified ACTFL Proficiency Guidelines: intermediate (n = 3), intermediate-advanced (n = 4), and advanced (n = 10). Using a variationist approach, I analyze five conditioning variables (including discourse function and verb class) that have been found to affect pronominal expression in native speaker Spanish (Bentivoglio, 1980; Cameron, 1992, 1994; Cacoullos & Travis, 2010; Otheguy et al, 2007; Silva-Corvalán, 1989, 2001; Travis, 2005, 2007), as well as two additional variables (correct verbal inflection and presence of hesitation phenomena) seen in second language acquisition studies.
Following theories of interface vulnerability (cf. Montrul, 2011b; Sorace, 2004, 2005; Sorace & Serratrice 1995), I hypothesized that the incremental development of sensitivity to conditioning variables would be reflective of the complexity of the interface at which the variable is conditioned. Based on previous studies of HL leaners (Montrul, 2004; Montrul & Rodríguez-Louro, 2006) and second language learners (Geeslin & Gudmestad, 2010a, Rothman, 2007a, 2007b, 2009), I predicted that the variables related to the more vulnerable and complex interface of syntax-discourse/pragmatics would be acquired only by the advanced proficiency group. My analysis indicated that HL pronominal expression is in free variation and unconstrained by any of the interface-related variables at the intermediate-advanced proficiency level. Neither the intermediate-advanced nor the advanced proficiency level HL leaners showed native-like sensitivity to variables conditioned at the syntax-discourse/pragmatic interface. Importantly, the only variable that was significant in first-person singular subject expression by HL leaners of all proficiency levels was the presence of hesitation phenomena. These results support the claim (Evans, 1985; Fehringer & Fry, 2007; Kormos, 1999; Riazantseva, 2001; van Hest 1996a, 1996b; Verhoeven, 1989) of a relationship between hesitation phenomena in oral speech and anxiety and increased processing demands while speaking the heritage language.
Comparing the results across three proficiency levels, I analyze when and to what degree HL learners show sensitivity to these variables. The findings illustrate the benefits of extending the application of sociolinguistic methodology (specifically the comparative multivariable analysis and the stepwise logistic regression procedure) to HL and second language acquisition studies because it facilitates a fine-grained examination and comparison of leaners across proficiency levels.