UC San Diego

What does it mean for a language to have flexible constituent order, and what are the sources of variation in this domain, both within and across languages? This thesis addresses these questions by combining methods from traditional linguistic fieldwork with those from psycholinguistics, focusing on Malayalam (Dravidian), in which all six permutations of subject, object, and verb are grammatical and have the same truth-conditional meaning. I propose a novel operational measure of flexibility in constituent order which uses formal acceptability judgment experiments: a greater preference for canonical versus non-canonical orders is associated with decreased flexibility. I demonstrate the cross-linguistic validity of this measure by comparing English and Malayalam (Experiments 1 and 2). After considering in more detail the relationship between information structure and constituent order in Malayalam (Experiment 3), I show that formal acceptability experiments can measure variation in flexibility within Malayalam, with older participants exhibiting greater flexibility than younger participants (Experiment 4). I consider language contact as a potential source of this variation, and conclude with a discussion of the role of flexibility in contact-induced syntactic change. The approach proposed in this thesis not only allows for an enriched typology of constituent order, it also has implications for our understanding of how languages interact and change.