This study investigated the effects of text modification and task on incidental learning of foreign language vocabulary by reading. The focus of the study was on how different cue types (appositives and single and multiple-choice marginal glosses) and a task (translation) would affect processing of input, initial learning, and retention of the meaning of target words in a text. Two hundred thirty-one students in Japanese universities were randomly assigned to 10 condition groups. After a pretest, a reading task, and a cloze test, they were given a surprise vocabulary test. A week later, two unexpected delayed posttests were also given. The results indicated that (a) both single and multiple-choice marginal gloss conditions performed significantly better than the no cue and the appositive conditions, (b) no statistically significant difference existed between the single and the multiple-choice gloss conditions nor between the appositive and the no cue conditions, and (c) the translation task had no effect.