When the second language attracts but the first does not: A large-scale study of number agreement attraction in Czech learners of English

We investigate agreement attraction effects in the L2 English of native speakers of Czech, a language that has little-to-no evidence of attraction effects. Our experiments involve two groups of participants. The first group (N = 415) participated in an L2 English-only experiment, and the second group (N = 183) participated in both L2 English and L1 Czech versions of the experiment (in a randomized order with a two-week interval). Standard attraction effects were observed in L2 English, contrasting with the absence of such effects in L1 Czech. Our results provide unique evidence that an L2 can significantly attract, even when the L1 does not. However, our results also revealed that the attraction effect in L2 English disappeared when the L1 Czech version was completed first. These findings are discussed in relation to the Unified Competition Model and the effects of L2-induced increases in working memory demands.