| Popis |
One of the primary tensions in second language (L2) research, and language acquisition generally, is the relationship between input frequency and mechanisms of innate linguistic competence for acquiring the target grammar (e.g., Rothman & Slabakova, 2018). One particularly intriguing approach is the Variational Learning Model (Yang, 2002) and its application to L2 acquisition (Yang & Montrul, 2017), stating that more frequent, unambiguous input will lead to faster acquisition of properties, but that the hypothesis space is constrained by innate mechanisms. We adopt this approach in examining the interaction of input frequency and implicational universals associated with modal existential wh-constructions (MECs). MECs are a sub-type of possibility clause and are closely related to the predicate available (Šimík, 2011). MECs consist of three core properties: (a) modality realized through the use of infinitive or subjunctive mood, (b) existential quantification derived from one of a few distinct classes of selecting verb containing an existential component, and (c) a fronted wh-word. Based on cross-linguistic evidence, as well as semantic decomposition of individual verbs, Šimík proposes implicational universals for b and c (1 and 2, respectively). While most Romance languages, including Spanish (3) and French (4), display MECs, English does not (5). We tested 60 L2 English learners (L1 Spanish: n = 30; L1 French: n = 30) and 22 L1 English controls using a speeded 88-item bimodally presented, contextualized acceptability judgement task with a 4-point Likert scale. Items consisted of combinations of the 5 verb types and 6 questions words (n = 29), combinations of the verb types with indefinites (n = 29), and fillers testing bi-clausal sentences with indefinites (n = 24). A sample test item is provided in Figure 1. Participants also completed an English proficiency test, a second version of the acceptability judgment task in their L1, and a language background questionnaire. Both L2 English groups exhibited significant restructuring across verb types and wh-question words, though we still observed residual effects of transfer (Figures 2 and 3). For the L1 Spanish speakers, we see greater acceptance of MECs with have- and find-type embedding verbs. For the L1 French speakers, we only see greater acceptance of MECs with find-type verbs. Increased acceptance of MECs with find-type verbs may be due to analysis of the sentences as containing infinitival relative clauses, even though contexts were designed to make this analysis less appropriate. Increased acceptance of MECs with have-type verbs suggests a tendency to restructure peripheral properties before core properties, following the implicational hierarchy. Both groups are also more accepting of MECs with where and less accepting for MECs with why, with the other wh-words accepted similarly. The low acceptance of MECs with why aligns with the findings for verb type, i.e., more peripheral elements are restructured first. Additionally, Šimík’s (2008) corpus results suggest MECs with what and who occur more frequently than MECs with the other wh-words. We argue that the high frequency of suitable contexts for what and who leads to faster restructuring. Consistent with Yang’s (2002) model, because they are lower on the implicational hierarchy, their restructuring forces restructuring of wh-words higher in the hierarchy, despite lower frequency; however, as where is at the same level of the hierarchy, its frequency-based restructuring is not affected by the restructuring of what and who. Differences between the two L2 groups may be due to English proficiency: the L1 French group had higher proficiency than the L1 Spanish group. As such, we interpret the results as two stages in the acquisition process, with the L1 Spanish group’s performance a precursor to the L1 French group’s performance.
|
