UCLA

This dissertation interrogates the representation of syntactic locality constraints, specifically intervention locality, within child grammar. Our focal point is the Intervention Hypothesis, which suggests that children are governed by a strict version of featural Relativized Minimality, leading to comprehension difficulties when partial featural overlap between moved elements and interveners is present.The unique morphosyntactic properties of Mandarin bei-passives afford an ideal platform for our investigation. Syntactic investigations indicate that long bei-passives involve movement of the internal argument (IA) across the external argument (EA), which is an embedded subject in this construction, whereas the IA movement in short bei-passives does not cross an EA because the EA is not syntactically projected. As such, the IA movement in long bei-passives, demonstrated in (1a), provokes intervention, which, based on the Intervention Hypothesis, should cause more difficulty for children’s comprehension than short bei-passives (1b). Furthermore, unlike in English, Mandarin lacks verbal/adjectival passive homophones to facilitate children’s comprehension of short passives. Consequently, children must establish a dependency between the IA and its gap, necessitating crossing an intervening argument in long passives. Our corpus study shows that long bei-passives are more frequent than short bei-passives in both child spontaneous speech and their input, which is not observed in any other languages in previous literature. In line with the Intervention Hypothesis, our results from both Experiments 1 and 2 highlight that children’s comprehension of long passives is worse than short passives, despite their significantly higher frequency in child and child-direcetd Mandarin. The dissertation further investigates which features are computationally relevant for interven- tion in child grammar. By manipulating three distinct features in Mandarin – Animacy, Number, and Shape – our research illuminates the role of linguistic features in forming a syntactic dependency. Experiment 1 shows that Animacy mismatch improved children’s performance on long passives, indicating its participation in children’s computation of intervention. An alternative ex- planation exists for this effect, however. Experiment 2 shows that a mismatch in Number or Shape does not affect children’s intervention difficulty with long passives. This leads us to propose that only morphosyntactic features triggering syntactic movement play a role in calculating intervention locality. In summary, this dissertation provides evidence supporting the Intervention Hypothesis, underscoring the intervention difficulty in children’s comprehension of long, but not short, bei-passives. Our research suggests that only morphosyntactic features triggering syntactic movement participate in computing intervention locality. This discovery introduces new dimensions to our understanding of syntactic locality constraints in child grammar and exposes cross-linguistic variations in the engagement of different morphosyntactic features in this aspect.