This study shed light on the effects of four modes of vocabulary instruction, i.e., extended audio glossing, lexical inferencing, lexical translation, and frequency manipulation of input on the learning of lexical collocations by Iranian intermediate EFL learners. In so doing, 80 L1 Persian EFL students were divided into four 20-participant comparison groups: Lexical Inferencing (LI), Extended Audio Glossing (EAG), Frequency Manipulation of Input (FM), and the Lexical Translation group (LT). LI, EAG, FM, and LT were treated through lexical inferencing, extended audio glossing, skewed frequency of input, and lexical translation techniques, respectively. The participants were pretested and posttested through a piloted multiple-choice lexical collocation test and instructed for ten instructional sessions. The results of the data, analyzed through repeated measures ANCOVA, showed that all the techniques examined in this study were effective on learners' achievement in lexical collocations. Comparatively, FM treated through frequency manipulation of input, significantly outperformed the other groups in lexical collocation improvement. The ANCOVA results and paired comparisons also indicated that EAG had the least achievement in lexical collocation compared to the other three groups. These results can hopefully inform language teachers, learners, and syllabus designers.