Typical research in second language (L2) implicit knowledge acquisition has centred on the role of learning conditions (e.g., explicit learning, focus on form, and oral output) in stimulating automatic processing in the L2 developing system. However, the process of automating L2 grammatical knowledge can be affected not only by learning mechanisms but also by the modality of stimulus presentation and working memory (WM) capacity. The present investigation aimed i) to examine whether encoding modality (auditory vs. visual) affects L2 implicit grammatical knowledge acquisition differently, and ii) to explore the modulating effects of WM capacity on the rate of emergence of L2 implicit knowledge in each encoding modality. In each of the two experiments, Experiment 1 with Chinese first-language speakers (n = 77) and Experiment 2 with Arabic first-language speakers (n = 37), participants were split into two groups, one of which was trained on three syntactic structures (tag questions, negative adverbs, and counterfactual conditional) under auditory exposure, whereas the other group was trained on the same structures but under visual exposure. The development of implicit grammatical knowledge, accrued from each input modality, was assessed via a pretest/posttest design using the measures of implicit knowledge (i.e., a timed grammaticality judgement task [timed GJT] and an elicited imitation task [EIT]). The WM capacity was measured using an operation-word-span (OSPAN) task. The results of both experiments demonstrated that auditory and visual exposure to L2 English syntax were found to result in successful emergence of implicit grammatical knowledge, as determined by both implicit knowledge measures: timed GJT and EIT. Furthermore, only in Experiment 1 was WM capacity found to correlate positively with automatized knowledge development accrued from auditory exposure on the GJT, but not on the EIT. However, the results demonstrated no significantly differentiable effects between visual and auditory interventions, but there was an observed trend towards visual modality advantage in grammatical knowledge automatization in Experiment 1 only, in both timed GJT and EIT. Overall, the study suggests that modality effects may indeed be another relevant factor in stimulating the automatic processing in the L2 developing system. I argue, from a psycholinguistic view, that computational demands exerted by auditory modality on WM capacity are greater than those exerted by visual modality, resulting in detrimental effects on the development of automatic processing.