Item

Abstract

Traditionally, the production of agreement has been studied using preamble completion paradigms, where participants produce inflected verbs to complete provided subject phrases (Bock & Miller, 1991). This paradigm, however, also captures comprehension as participants must understand the preamble before completing it. We designed a production task (Experiment 1), in which preambles were not linguistically offered. Participants described simple scenes with singular and plural local nouns in conceptually integrated and unintegrated settings. Similar to previous findings, there were effects of local noun number: more agreement errors were made when the numbers of the head and local noun mismatched. There were also effects of integration: more errors were made for unintegrated items than for integrated items with singular head nouns (Brehm & Bock, 2011). Using the same items in a preamble completion task (Experiment 2), we only replicated the attraction effects. This suggests that during completion tasks, notional effects might sometimes not surface because speakers do not generate the message themselves. The production task we developed is a good alternative to the completion paradigm, as it replicates previous findings while requiring speakers to generate their own messages. We conclude that during production, both grammatical and conceptual number mismatches make correct agreement harder.