A Comparative Study of Lexical Inferencing During Text Reading: Evidence from Chinese–English–Japanese Trilinguals

Lexical inferencing (LIF) draws on multiple knowledge sources (KSs), but little is known about how the same multilingual readers report using these sources across different reading tasks. This exploratory study examined 31 Chinese–English–Japanese trilinguals who completed think-aloud inferencing tasks in three language-and-material conditions. Nine KS Types were coded, and trial-level ordinal mixed-effects models related KS use to rater-coded LIF success and participant-selected immediate Vocabulary Knowledge Scale (VKS) categories. Sentence Meaning was the most frequently reported KS Type in all three conditions, while several word-level KS Types and co-occurrence patterns varied across conditions. KS Amount was not associated with LIF success. The joint KS Type test produced weak, threshold-sensitive evidence, and no individual KS Type contrast survived multiplicity correction. VKS categories differed across conditions but were not consistently associated with LIF success, KS Amount, KS Type, or exploratory co-occurrence measures. Because language was confounded with texts, target sets, target status, proficiency, and language experience, between-condition contrasts are descriptive and cannot isolate language or transfer effects. The findings question a simple ‘more is better’ account of reported KS use and identify descriptive patterns for testing in future matched-material studies.