Wagers Colloquium

February 25
3:30 - 5:00 pm
Rosenwald 011
University of California, Santa Cruz
Constituent order and parser control processes in Chamorro
In this talk, I examine the question of whether prenominal relative clauses and postnominal relative clauses exhibit the same processing complexity and intepretive biases. And I will use the answer(s) to this question to articulate an account of why and when parsers favor the analytic choices they do.
Postnominal relative clauses show a preference for subject extraction, a fact overwhelmingly supported by decades of observation and experimentation. The status of this preference in languages with prenominal relative clauses – like Cantonese, or Basque – is more controversial; and some evidence suggests that object extraction is preferred in those languages. But, overall, we know much less about how prenominal relative clauses are parsed. In part, this is because the languages which allow prenominal relative clauses are less well-studied psycholinguistically. 
I will bring data to bear from comprehension and production experiments in the Austronesian language Chamorro. Chamorro provides an excellent grammatical setting to make a clean comparison between the two structures: it permits both prenominal and postnominal relative clauses and there are no difference in within-RC word order. Our experiments uphold the idea that prenominal relative clauses do indeed more easily allow object extractions; but they also reveal that the parser attempts the subject extraction as a first resort.