A Quantitative and Network Approach to Alignment Effects in L2 Continuation Tasks

Abstract The present study employed a quantitative and network approach to detect alignment effects in second language (L2) continuation tasks designed on the xu-argument (Wang, 2016). The materials used in this study were 6 sub-corpora consisting of two selected input stories and two groups of L2 written production based on two continuation tasks. During continuation, the participants were required to continue in English a story with its ending removed, with one group reading and continuing the Chinese version and the other group the English version, and then switching their roles in the two tasks. Results show that the alignment effect differs across the two versions of continuation. Specifically, compared with the Chinese-version continuation, L2 learners produced more use of unigrams and bigrams similar to the input story in terms of lexical items, frequency and ranking correlations in the English-version task; on the other hand, the English-version continuation can facilitate generating linguistic networks that are much closer to the native English networks. Moreover, this research corroborates that written production in L2 continuation tasks can be influenced by input content.