While much research has proved the benefits of subtitled audiovisuals for foreign language learning, few studies address such practices in out-of-classroom settings or focus on Asia-based video-sharing platforms. This study bridges this gap by introducing an increasingly popular viewing-commenting system in Japan and China, known as danmu or danmaku, which displays viewers’ timeline-synchronized comments on video content. The study analyses the metalinguistic comments which entail viewers’ knowledge of the language, their comprehension issues and sociolinguistic attitudes toward its use. Adopting an inductive or data-driven methodology, the study extracted and manually coded 390 comments that are related to the Spanish language, Spanish–Chinese translation and learning Spanish. Results show that viewers are mostly interested in linguistic features that differ from Chinese or English (e.g. the complex grammar) and they use danmu to access sociolinguistic issues that are central to daily communication such as the fast speech rate, language varieties, and frequent use of vulgarisms.