A long-standing debate among the vocabulary researchers is the depth of processing to learn vocabulary. This paper is a quantitative research which considers a revision in the “involvement load hypothesis” proposed by Laufer and Hulstijn in 2001. It investigates the role of proficiency and evaluation in this hypothesis in order to better reveal its potential contribution to vocabulary learning. It was based on task-induced involvementthat comparesdifferent tasks in incidental vocabulary acquisition in EFL context. The participants were 66 learners fromtwo different English institutes who were classified into two major high and low proficient groups based on Nelson Proficiency Test. The participants in each group were randomly assigned to three tasks prepared to compare“moderate”, “strong”, and “no evaluation” in involvement load hypothesis.The “strong evaluation”subgroup (making original sentences) in low proficiency supported Laufer and Hulstijn’s hypothesis and yielded better retention of the target words. The study suggests that the level of proficiency and evaluation in task induced involvementneeds reconsideration. The results have implications for language teachers, materials developers, and syllabus designers.