Processing complex sentences: A cross-linguistic study

The Competition Model is an interactive-activation framework for the study of sentence processing that is designed to handle quantitative as well as q ualitative variations in performance across natural languages. Previous stu dies within this framework have shown that adult listeners base their inter pretation of simple sentences on the most valid and reliable cues in their language (e.g. more use of word order in English and more use of subject-ve rb agreement in Italian). Critics have argued that such effects may reflect heuristics that are only applied to simple sentences. The present study sh ows that these cross-linguistic differences are maintained when participant s are asked to interpret complex sentences with an embedded relative clause . A comparison of "off-line" (untimed) and "on-line" (timed) versions of th e same experiments shows that these effects hold up under time pressure. Th e on-line versions also provide new information about cross-linguistic diff erences in timing and demands on processing. In particular, the processing costs associated with centre embedding and non-canonical order are greater in English, which may be the price that English listeners have to pay for h eavy reliance on word order information.