Visuals (e.g., figures) are prevalent in academic writing across disciplines, and they are usually accompanied by commentaries that summarize and/or interpret the information they contain. Commentaries on visuals have not yet been studied in terms of their rhetorical and phraseological features, and insights into these features could help novice academic writers integrate such commentaries more effectively. To address the gap, this study examined and reported on frequently employed rhetorical functions and phrase-frames for commentaries on visuals, based on a corpus of such commentaries extracted from 100 social science research reports. Specifically, the corpus was first manually annotated using a rhetorical function framework developed based on previous studies. Five- to seven-gram phrase-frames were also automatically extracted based on corpus statistics and then manually scrutinized to generate a pedagogically useful list. The finalized 169 phrase-frames were then linked to the rhetorical functions based on their occurrences and sorted accordingly in the final list. Under each rhetorical function, the phrase-frames were further categorized into specialized, semi-specialized, and non-specialized types, following previous studies. Pedagogical and empirical implications of the phrase-frame list for English for specific purposes programs are discussed.