Retention of Word Meanings Inferred from Context and Sentence‐Level Translations: Implications for the Design of Beginning‐Level CALL Software

Abstract

This article attempts to reconcile two positions on second language (L2) vocabulary learning: (a) inferring word meanings produces greater retention because it promotes deeper processing, and (b) incorrect meanings are retained if wrong inferences are made. In beginning‐level computer‐assisted language learning (CALL) software, native language translations at the sentence level may satisfy both positions. When deriving meaning from translations of this type, users must make associations (i.e., search for semantic equivalents) and focus their attention on structural differences between the two languages. Moreover, these translations increase the likelihood of making correct associations by more readily triggering appropriate schemata and, hence, the likelihood of committing the correct meaning to memory. In the present study, a MANOVA indicated that, given bilingual multiple‐choice tests, learners with access to sentence‐level translations demonstrated significantly greater short‐term (ST) and long‐term (LT) retention (p > .001) of correct word meanings than learners without sentence‐level translations. Results suggest a possible synergy between deep processing and verifications of meaning.