Is the overt pronoun constraint learnable?

Abstract

Studies on implicit learning have provided evidence for L2 acquisition of syntactic features, yet limited effort has been made to gauge the applicability of the implicit learning paradigm on syntactic structures that are posited by nativists as innate and need not to be learned. This thesis investigates the implicit learning of the Overt Pronoun Constraint(OPC), a claimed UG-derived constraint (White, 2003a,b; Hawkins, 2008)that prevents overt pronouns from taking quantified NPs as antecedents in null-argument languages(Montalbetti, 1983), and seeks alternative explanations to such knowledge from the usage-based perspective in SLA. In Experiment 1, participants’L1 prior knowledge of the binding constraint of the overt pronoun he in Mandarin Chinese and English was investigated respectively. Results show that Chinese participants accepted the bound variable interpretation of the pronoun他 (he) when the matrix subject (the subject of the main clause) was 有人someone, suggesting that the OPC may not be fully applicable in Chinese, and that the OPC may not be a universal phenomenon in all null-argument languages as claimed by nativists (e.g. Kanno, 1997). In terms of English participants, they rejected bound variable interpretations more often when the matrix subject of the sentence was a quantified NP than when it was a referring NP, indicating some biases of the interpretation towards the reference of the overt pronoun. Potential explanations for these cross-linguistic differences include the popularized use of singular they in English (Bhat, 2004) to refer to gender-ambiguous antecedents, and the degree of consistency in definiteness between the matrix subject and the pronoun as the sub-clause subject. In Experiment 2, Chinese L1 speakers were exposed to a semi-artificial language system that combined the binding constraint of the Japanese pronoun “kare” with Chinese to see whether they could acquire the OPC implicitly. The learning was measured by a timed Grammatical Judgment Test (GJT), and awareness was assessed by confidence ratings, source attributions and verbal reports. Results show that learning effect (both implicit and explicit) was observed in the Chinese group. In Experiment 3, Chinese participants were exposed to a semi-artificial language system that combined the overt pronoun binding constraint with their L2 English, and no learning effect was observed in this group, indicating that implicit learning could be affected by participants’L2 proficiency. In Experiment 4, the implicit learning of the pronoun constraint by English native speakers was investigated and L1 transfer effect was found in this experiment. To sum up, results show that the interpretation bias of the overt pronoun might be learned implicitly, although the learning process could be affected by participants’ prior linguistic knowledge. It also suggests that this bias might be learned without the assumption of UG existence, which show support for the usage-based approach in SLA.