Complement selection bias in L2 English: The case of infinitives and gerunds

Abstract

This study investigates the source of a complement selection bias observed in the choice between infinitives and gerunds. It has long been noted that second language (L2) learners show a stronger preference for infinitives over gerunds, but the source of this bias has remained unclear. Building on previous theoretical and empirical work, this study hypothesizes that L2 learners resort to the syntactically default option (i.e. infinitive) when they lack the relevant lexical knowledge. To test this hypothesis, we conducted three binary-choice experiments: (1) Real verbs: This experiment confirmed the previously reported bias toward the infinitive; (2) Nonce verbs without meanings assigned: This experiment examined how participants behaved when encountering entirely novel verbs; (3) Nonce verbs with meanings prototypically associated with either the infinitive or the gerund: This experiment tested whether the patterns observed in Experiment 2 shifted in accordance with the assigned prototypical meanings. A total of 91 native speakers of English (NSEs) and 98 Japanese-speaking learners of English at the intermediate level (JLEs) participated across the three experiments. Taken together, the results revealed a clear divergence between NSEs and JLEs, especially when presented with nonce verbs: (1) NSEs selected complements randomly, whereas JLEs showed a strong preference for the infinitive; and (2) while NSEs continued to respond randomly even when nonce verbs were assigned meanings, JLEs’ choices became more aligned with the complement selection patterns associated with the assigned meanings. The findings are discussed in terms of derivational economy.