Towards accurate measurement of incidental vocabulary acquisition: input-matched assessments of reading, listening, and reading-while-listening

Abstract

This study explored the effects of reading-only, listening-only, and reading-while-listening input modes on incidental vocabulary acquisition and retention of knowledge over time. 104 L2 English, L1 Japanese-speaking participants from two universities in Japan were divided into three input-mode groups. A PsychoPy application was employed to present a narrative and test participants using written-receptive meaning-recall and multiple-choice meaning-recognition tests. A 2716-word narrative containing nine pseudowords that replaced concrete nouns at a frequency of 10, 15, or 20 encounters in the text was used. Input mode matched testing mode in this study, whereas previous studies tested reading-while-listening using reading-based tests only as reported by Brown et al. (Reading in a Foreign Language 20:136–163, 2008), Malone (Studies in Second Language Acquisition 40:651-675, 2018), Teng (Innovation in Language Learning and Teaching 12:274–288, 2018), and The TwiLex Group (Studies in Second Language Acquisition 2024-06:1-26, 2024) and did not consider transfer appropriateness during testing, which may cause mode-specific lexical knowledge to go unmeasured. The data were analyzed using a generalized linear mixed effects model, which found that incidental vocabulary gains were statistically lower for listening-only than reading-while-listening and reading-only on both test types. No statistically significant difference was found between reading-only and reading-while-listening for meaning-recognition tests, despite an 8-min average difference in reading time for reading-only, meaning reading-while-listening appeared to be more efficient for acquisition when testing and treatment mode matched. There was a statistically significant difference in favor of reading-only vs. reading-while-listening for meaning-recall tests. No differences were found in the rate of decay for input mode or test type, but meaning-recall displayed lower gains than meaning-recognition tests. A future study that might better measure mode-specific differences is also proposed.