Francesco Romano
Francesco Romano


One well-known finding in studies of generative SLA is that ultimate attainment in heritage speakers differs in important ways from L1 speakers. One prominent area of divergence is the morphosyntactic domain of Romance languages, where heritage speakers’ competence has been found to differ from L1 competence, particularly for inflectional morphology (for Spanish gender see Montrul, Foote, & Perpiñán, 2008, mood morphology in Montrul & Perpiñán, 2011 and Silva-Corvalán 1994, 2014; for Italian article use see Schmidt & Wörner, 2012 in Kupisch & Rothman, 2015). Heritage speakers are simultaneous or sequential bilinguals, usually speakers of a stronger, dominant, environment language and a weaker, family language. An open question is whether the end-state grammar of the weaker language is more similar to the incomplete grammars of L2 acquisition than the complete and rather uniform end-state of L1 acquisition. To the extent that L2 acquisition is incomplete and fundamentally different from L1 (Bley-Vroman, 1990; Meisel, 1997; Clahsen & Muysken, 1986), UG-access can be questioned, by extension, in the weaker, heritage language. Such a view vies with one that endorses recourse to UG and maintains that its architecture constrains L2 acquisition (and by extension, heritage languages) in ways comparable to L1 grammars (Lardiere, 2015; Vainikka & Young-Scholten, 2011; Schwartz & Sprouse, 2007).

In this talk, I explore the plausibility of these views in a structural priming study on the oral production of L2 Italian accusative clitics with restructuring verbs in the presence of clitic-left dislocation. The structures tested are exemplified in (1)a-c. [For a correct disposition of the example, se Abstract pdf below.]

(1) a.  I pesci Aurora li cucina all’aperto               (lexical verb)
            OBJ                 cl V

      b. I pesci Aurora li  fa cucinare da Raniero     (causative verb)
            OBJ                 cl V  V

      c. I pesci Aurora li vuole  cucinare all’aperto  (modal verb)
            OBJ                cl MOD   V

Highly proficient first-generation heritage speakers with Italian as the weaker and Swedish as the stronger language were tested alongside a group of proficiency-matched adult L2 speakers exposed to Italian after age 13 and a group of L1 Italian speakers recruited in Italy. Morphological and syntactic competence were examined as accuracy in use of clitic morphology and strength of priming effects, respectively. A fundamental difference view predicts similar gender assignment errors and clitic omission phenomena across the L2 and heritage groups. It also predicts more similar priming effects between the L2 and heritage groups than either of these groups and the L1 speakers. In contrast, a full access view predicts significantly less variability in use of clitic morphology as well as more similar priming effects between the heritage and L1 groups compared to the L2 speakers.

The clitic morphology analysis finds both the L2 and heritage group differ significantly from the L1 speakers largely due to omission but not gender assignment errors, partially consistent with a fundamental difference view. When morphology is used by the heritage and L2 speakers, however, it is accurate, suggesting both groups are qualitatively similar to the native speakers and have access to UG. An analysis of priming effects finds overall stronger priming in the native group for all types of verbs, consistent with a fundamental difference. When the verb condition is factored in, however, the native and heritage groups are qualitatively similar, showing stronger priming in the causative and lexical than the modal verb condition, while the L2 group shows stronger representations for the lexical, followed by the causative, and modal verbs, consistent with full access. On the whole, I argue these results to show that heritage speakers have recourse to UG and that successful ultimate attainment is not inevitable for both heritage and L2 speakers.