This study explores quantitative and qualitative differences between context and word morphology as determinants of students' guesses of the meanings of novel kanji compounds (i.e., words consisting of two or more Chinese characters). In this study, 74 Japanese language learners interpreted unknown compounds consisting of familiar characters under three conditions (i.e., words in isolation, contextual clues only, and both). When the students' guesses were compared against the actual meanings, context and word morphology were equally effective. However, the combined effect of both sources was the best, suggesting that the 2 sources provide information that does not overlap. Students expressed higher confidence in their context-based guesses than in their words-in-isolation guesses. Furthermore, contextual clues generated more syntactically related guesses than did kanji clues, whereas kanji clues yielded more semantically related guesses than did context. In general, the strength of contextual support and morphological transparency worked additively when combined but sometimes affected each other negatively. These findings suggest that context and word morphology play different roles in word inference.IN THE PAST 2 DECADES, A GROWING number of studies have examined the strategies that language learners use to learn new words. When second language (L2) readers encounter an unfamiliar word while reading, they often infer its meaning using available information and knowledge without referring to a dictionary (Hulstijn, Hollander, & Greidanus, 1996;Schmitt, 1997). Contextual information and word morphology (e.g., word roots, affixes, and inflectional suffixes) are two major sources that readers use to interpret novel words (De Bot,