<p>This paper investigates how input modality and second-language (L2) vocabulary proficiency affect the processing of different types of Japanese compound verbs. In this study, 50 graduate students from Japanese language programs at a Chinese university were placed in the visual group (<i>N</i> = 25) or auditory group (<i>N</i> = 25), and were asked to judge whether the presented Japanese compound verbs and pseudo-words were actual Japanese words or not. The compound verbs and pseudo-words were randomly presented visually or auditorily. Our findings highlighted several key points. First, the headedness effect in the processing of Japanese compound verbs emphasised the importance of the second verb in the compound, and this effect was modulated by both input modality and L2 vocabulary proficiency. Second, when compound verbs were presented in isolation, input modality and vocabulary level independently affected the processing of different types of compound verbs. Third, upon visual presentation, the processing of different types of compound verbs tended to shift from decomposed to holistic processing depending on the absence of grammatical and semantic functions in the first or second verbs; conversely, upon auditory presentation, a tendency toward decomposed processing prevailed across all types of compound verbs. Fourth, lower L2 vocabulary proficiency resulted in more pronounced differences between decomposed and holistic processing across various compound verb types; in contrast, higher L2 vocabulary proficiency favoured holistic processing. These observations contribute to a deeper understanding of the cognitive processing involved in L2 acquisition and have implications for both teaching and further research in this area.</p>