UC San Diego

This dissertation contains three studies that investigate whether attested patterns of constituent order distribution and change in the world's languages can be attributed, in part, to cognitive preferences for some constituent orders over others. To assess these preferences, seven experiments employed an elicited pantomime task. Communication in gesture without speech may reveal cognitive pressures that otherwise remain hidden. Study 1 manipulates the whether the to-be- described events are semantically non-reversible (e.g. a woman lifting a box) or reversible (e.g. a woman lifting a boy). Across three experiments, participants reliably preferred SOV for describing non-reversible events in pantomime, but avoided using SOV to describe reversible events. Although previous accounts have assumed similar behavior to be due to constraints on comprehension, the distribution of alternative orders (mainly SVO, OSV, & SOSV) suggests a greater role for constraints on production in explaining these patterns. Whereas Study 1 focuses on explaining drifts away from SOV, Study 2 tests the hypothesis that languages drift toward SVO because natural languages are subject to additional constraints that pantomime may not engage, such as pressures on information structure and efficiency. These additional pressures might work against OSV and SOSV, leaving SVO as a preferred solution for reversible events. Experiment 1 (English speakers) created more language-like conditions by simulating the emergence of a lexicon and of shared lexical knowledge. Under these conditions, OSV and SOSV decreased, and SVO increased. Experiment 2 found similar patterns with native speakers of an SOV language (Turkish), suggesting that the increase in SVO among English speakers was not simply a result of influence from their native language. Whereas Studies 1 and 2 focus on constituent order preferences in pantomime production, Study 3 focuses on pantomime comprehension. Speculation about constituent order preferences in comprehension has figured largely in linguistic and psycholinguistic theory, but little empirical evidence is currently available. Using a novel paradigm, Study 3 finds that constituent order preferences in comprehension are distinctly different from those in production, for both speakers of English (Experiment 1) and Turkish (Experiment 2). Results suggest that constraints on production may play a greater role than previously thought