This paper describes a scripted role-playing task used to elicit a basic inventory of intonational tune types in Nasal (ISO 639-3: nsy; glottocode nasa1239), an endangered Austronesian language of Sumatra currently spoken in three villages by around 3,000 people. This study is a subset of the ongoing prosodic description that forms part of a larger Nasal documentation project. Prosodic description is not often included in language documentation, especially for Austronesian languages; when included, these descriptions are often very limited or based on impressionistic descriptions (Himmelmann & Kaufman 2020). I make the case here that prosodic description based on carefully planned experimentation and acoustic measurement is not only achievable but a necessary part of linguistic fieldwork for language documentation. In the case of Nasal, eight participants (four men and four women) were recorded reading scripted lines from role-play dialogues in a variety of real-life scenarios. These recordings were transcribed, labeled according to the HCRC dialogue coding scheme (Anderson et al. 1991, Carletta et al. 1996), and analyzed with a cluster analysis using the Contour Clustering app (Kaland 2023). The alignment of tune patterns with different utterance types reveals a preliminary inventory of prototypical intonational tunes, including a falling tone pattern for polar questions, a pattern that is crosslinguistically uncommon.