Improving Incidental L2 Vocabulary Learning With Latent Semantic Analysis

Abstract

This study examined whether word-context semantic similarity computed by Latent Semantic Analysis (LSA) predicts the performance of incidental L2 vocabulary learning. LSA is relevant to the usage-based model of language learning, which premises that learners obtain lexical knowledge from the information about how words are used in context. In the experiment, 153 Japanese undergraduates were given 20 target words with contexts whose proposition had higher or lower semantic similarity to those words (HSS vs. LSS context) in the lexical inference and multiple-choice glosses tasks. A Vocabulary Knowledge Scale (VKS) test was used to assess incidental gains in word meaning and usage knowledge. The analyses of the task performances showed that the learners inferred the target word meanings from the HSS contexts more successfully than from the LSS ones, but obtained similar scores between the two context conditions in the multiple-choice glosses task. Nevertheless, the VKS results demonstrated that the HSS contexts greatly contributed to the incidental gains in word meaning and usage knowledge. Thus, LSA predicted the outcomes of incidental vocabulary learning, which indicates that the learners acquired lexical knowledge from usage-based contextual information.