Statistical learning of L2 lexical bundles through unimodal, bimodal, and multimodal stimuli

Given the fundamental contribution of formulaic sequences to the fluent, natural second/foreign language (L2) use and processing, finding effective ways of improving learners's formulaic language has increasingly engaged the attention of L2 practitioners. Focusing on lexical bundles (LBs), as the most frequent formulaic-language strings, the current quasi-experimental study examined statistical learning of recurring 3-to-6-word LBs among Iranian learners of English as a foreign language (EFL) receiving different types of distributional, narrative sensory stimuli. To this end, three intact classes received a recurring-theme narrative content in three different input modalities including only-auditory, audiovisual, and captioned audiovisual. To examine whether exposure to statistical, varied-modality input (or sensory stimuli) constituted profound knowledge of L2 lexical bundles, the three experimental groups' performances on immediate and delayed posttests were compared with those of a control group. The findings revealed that all the frequency-based modality inputs led to more sizeable LB gains. Further, multiple (captioned) audiovisual sensory stimuli more effectively fostered retention and recalling of the target LBs over time than only-auditory input. The retention differences between the bimodal and multimodal inputs were insignificant. Overall, the findings supported combined statistical, formulaic advantages for processing, retention, and recalling of L2 LBs through distributional, multimodal stimuli.