CONSTRUCTING SUBJECT VERB AGREEMENT IN SPEECH - THE ROLE OF SEMANTIC AND MORPHOLOGICAL FACTORS

A sentence completion task, first introduced by Bock and Miller (1991) with English speakers, is employed here with Italian to explore the i ssue of interactivity of different levels of processing in sentence pr oduction. In a series of three experiments, we tested the effects of t hree variables on the number of subject-verb agreement errors. Like Bo ck and Miller, we found that mismatching singular and plural features between the subject head noun and a local noun in a complex NP increas ed the number of errors. It was also found that ambiguous morphophonol ogical marking on the subject noun increased errors. The most striking result was that the preferred interpretation of the complex NP had a significant influence. Singular heads, normally taken to denote severa l tokens, such as the label on the bottles, tend to induce incorrect p lural verb completions. This is in contrast to the findings for Englis h speakers and raises the question as to whether the construction of s ubject-verb agreement is a purely syntactic process in all languages. A modified version of a model of syntactic encoding (Kempen & Hoenkamp , 1987) is proposed that can satisfactorily deal with the number misma tching effect and crucially the presence/absence of semantic effects o n verbal agreement in different languages. (C) 1995 Academic Press, In c.