This study aims to ascertain the relative importance of certain aspects of task design for the acquisition of vocabulary by high-proficiency second language speakers. Previous research suggests that vocabulary learning is most effective when "task-induced involvement" is high and that involvement can be measured as a sum of three components: need for the vocabulary in the task, searching for a meaning of a word or a word to fit a meaning, and evaluation of how the word meaning compares to that of other words or how the word fits in with surrounding words. We attempt to corroborate those premises and, further, to examine the hypothetical components of involvement separately and investigate whether they interact with each other and/or differ in impact on the learning process. To address this question, we administered a different task to each of four groups of high-proficiency learners of English, varying the search and evaluation components among the tasks. We measured the vocabulary gains on immediate and delayed tests and compared the results among the four groups. The results on the immediate test accord with earlier studies in showing greater gains for the groups with more task-induced involvement than for the control group. Concerning search and evaluation, the two components appear to compensate for each other, suggesting that the presence of either one is as effective as the presence of both. The delayed post-test results followed the same patterns, although the results are not statistically significant.