Effects of semantic integration on subject–verb agreement: evidence from Dutch

Abstract

The generation of subject–verb agreement is a central component of grammatical encoding. It is sensitive to conceptual and grammatical influences, but the interplay between these factors is still not fully understood. We investigate how semantic integration of the subject noun phrase (‘the secretary of/with the governor’) and the Local Noun Number (‘the secretary with the governor/governors’) affect the ease of selecting the verb form. Two hypotheses are assessed: according to the notional hypothesis, integration encourages the assignment of the singular notional number to the noun phrase and facilitates the choice of the singular verb form. According to the lexical interference hypothesis, integration strengthens the competition between nouns within the subject phrase, making it harder to select the verb form when the nouns mismatch in number. In two experiments, adult speakers of Dutch completed spoken preambles (Experiment 1) or selected appropriate verb forms (Experiment 2). Results showed facilitatory effects of semantic integration (fewer errors and faster responses with increasing integration). These effects did not interact with the effects of the Local Noun Number (slower response times and higher error rates for mismatching than for matching noun numbers). The findings thus support the notional hypothesis and a model of agreement where conceptual and lexical factors independently contribute to the determination of the number of the subject noun phrase and, ultimately, the verb.

Keywords:

• subject–verb agreement
• language production
• semantic integration
• notional number

Notes

1. Solomon and Pearlmutter carried out five experiments and a meta-analysis; Brehm and Bock carried out two experiments using all items from Solomon and Pearlmutter's experiments.

2. We thank N. Pearlmutter for making the text available.

3. As for Experiment 1, absolute t-values above 2 were considered significant.

4. In an additional experiment, 12 participants completed the same task as in Experiment 2 but, similar to Experiment 1, were required to repeat and complete the preamble verbally on 24 catch trials. The results were very similar to those seen in Experiment 2: there were significantly more errors and longer response times for preambles with plural local nouns (relative to singular local nouns) and for weakly integrated preambles (relative to tightly integrated preambles). Again, the two variables did not interact.