On the interpretation of the number attraction effect: Response time evidence

Speakers frequently make subject-verb number agreement errors in the presence of a local noun with a different number from the head of the subject phrase. A series of four experiments used a two-choice response time (RT) paradigm to investigate how the latency of correct agreement decisions is modulated by the presence of a number attractor, and to investigate the relative latency of errors and correct agreement decisions. The presence of a number attractor reliably increased correct RT, and the size of this RT effect was consistently larger in conditions that also had larger effects on accuracy. Number attraction errors, however, were similar in RT to correct responses in the same experimental condition. These results are interpreted as supporting a model according to which an intervening number attractor makes the agreement computation process more difficult in general (Eberhard, Cutting, & Bock, 2005), with errors arising probabilistically. However, attraction from a non-intervening noun resulted in only mildly inflated correct RT, but dramatically inflated error RT, suggesting that non-intervening attraction errors may reflect confusion about the structure of the subject phrase.