This study examined the effect of the interaction approach on young EFL learners' lexical depth in a private language institute in Talesh, Iran. To achieve this, a Michigan test was administered to 40 EFL learners. Those language learners who met the selection criterion, i.e., performed one standard deviation above and below the mean on the test were divided into two classes (N = 28): group A (N = 14) and group B (N = 14). In both groups, the interaction approach was incorporated into vocabulary instruction. However, the design of the interaction was different in each group. In group A, interactions were mostly between learners. On the other hand, in group B, learners mainly interacted with the teacher. After 18 weeks of instruction, a test of lexical depth, word association tasks (WAT), was administered to both groups. The results of statistical analysis showed that the standard deviation (SD) was almost the same in both groups. However, the mean in group B significantly outperformed group A. A "t" test was used to compare means of test scores between groups. The findings of this study suggest that learner-teacher interaction is a more effective design in the improvement of learners' knowledge of vocabulary depth in comparison with learner-learner interaction.