Queensland’s senior secondary composition assessment makes some songwriting practices easy to recognise and others harder to name. For Kodály practitioners and researchers alike, a useful way into that problem is musical literacies: the difference between what students can do musically and what they can say about what they can do. When assessment language rewards explanation, classification and pinpointing, some songwriting moves are harder to credit because they do not translate neatly into atomised descriptors.

I address this by re-reading songwriting assessment as boundary-making work: what counts as ‘composition’, what becomes sayable about recorded work, and what teachers learn to reward. In Rita Irwin’s terms, the paper proceeds a/r/tographically, reading from inside an entangled field of practice as songwriter, teacher and researcher. Working with an assemblage of curriculum documents, task sheets, marking guides, recordings, drafts, classroom talk and reflexive practice journalling, I look for patterns of legibility: which literacies are repeatedly foregrounded in judgements and feedback, and which are quietened or made secondary. Adult co-writing is not employed as a model for students, but as a diagnostic device that makes the system easier to read.

Three outcomes follow. First, composing emerges as a naming economy that blurs tools, objects and audible results, which matters when the composing artefact is a recording rather than a score. Second, unity and cohesion are high-stakes yet under-specified, so teachers and students often lean on inherited heuristics rather than shared language for track-level coherence. Third, assessment privileges retrospective justification and atomistic pinpointing, elevating literacies of explanation over the situated labour of track-making. The paper closes by proposing more usable ways of describing composition as track-making praxis, legitimising recordings as composing artefacts, and re-orienting judgement toward holistic decisions about how tracks communicate meaning to listeners.