With an effort to verify the involvement load hypothesis, this study investigates the effects of three different vocabulary learning tasks - reading, filling in blanks, and sentence writing - on learners' short and long term memory when they are provided with the theoretically identical level of involvement tasks. It is also to examine whether there are any differences of learning effects between input and output vocabulary learning tasks. The experiments were conducted with 86 Korean middle school students grouped by three different task types. Pre-/post-tests and delayed tests were administered and analyzed. It was found that three vocabulary tasks with theoretically identical levels of involvement do not have the same learning effects: the involvement load hypothesis was not completely proved in the case of Korean students' vocabulary learning. The findings implied that (i) the difference of vocabulary learning tasks should be considered even though they have the same involvement load; (ii) given that EFL students learn new words in a limited amount of time in class, the reading task using a dictionary is found to be very helpful; and (iii) it is important to study target vocabulary repeatedly during class and to expose the vocabulary with different tasks to enhance students' retention of the newly learned vocabulary.