Making predictions by means of verbs’ implicit causality (IC) bias is an essential skill for efficient discourse comprehension. Although English as a foreign language (EFL) learners are shown to reactively use IC bias for pronoun interpretation, whether they can proactively use IC bias for making predictions remains unknown. This study aimed to reveal whether and to what extent Japanese EFL learners predict the next-mentioned entity with IC bias. Forty Japanese university students produced continuations of sentence fragments including an IC verb in L1 (Japanese) and L2 (English). The IC verbs are assumed to bias sentence continuations to refer to the first noun phrase (NP1; e.g., “Mary annoyed Lisa because ”) or the second noun phrase (NP2; e.g., “Mary respected Lisa because ”) in the fragment. The results confirmed that the participants used IC bias for predicting the next-mentioned entity, as reflected by more NP1 continuations in the NP1- than in NP2-biasing context. However, they used NP1 bias less accurately in the L2 than in the L1 condition. This reduced accuracy with English-NP1 bias was found to lie in learners’ underspecified lexical quality of English-NP1-biasing verbs, rather than in L1–L2 morphemic difference of verbs. These findings are discussed with reference to the theoretical model of L2 comprehension and translated into educational practice to support learners’ predictive processing.
