Is Syntax Semantically Constrained Evidence From a Grammaticality Judgment Study of Indonesian

A central debate in the cognitive sciences surrounds the nature of adult speakers’ linguistic rep-resentations: Are they purely syntactic (a traditional and widely held view; e.g., Branigan & Pick-ering, 2017a), or are they semantically structured? A recent study (Ambridge, Bidgood, Pine,Rowland, & Freudenthal, 2016) found support for the latter view, showing that adults’ acceptabil-ity judgments of passive sentences were significantly predicted by independent semantic “affected-ness” ratings designed to capture the putative semantics of the construction (e.g.,Bob was pushed by Wendy is rated as more acceptable than Bob was liked by Wendy, as Bob is more affected in the former). However, because English lacks a separate topicalization construction which provides an alternative means of highlighting the patient (e.g.,BOB, Wendy kicked), these findings have a possible alternative explanation: that highly affected entities are more likely to be topicalized, rather than passivized per se. Here we show that, in fact, Ambridge et al.’s (2016) finding repli-cates in Indonesian, a language with a topicalization construction. The present study therefore pro-vides particularly compelling evidence that grammatical representations have semantic structure. Keywords: Indonesian; Linguistic representations; Abstract syntax; Passive; Verb; Semantics; Autonomy of syntax