Abstract
Why do some popular songs become queer anthems while others, including many by openly queer artists, do not? We tested musical affordance and historical contingency accounts in a purposive corpus of 100 songs (69 artists) designed to dissociate queer authorship, explicit queer lyrical content, and community adoption. Adoption was measured as an evidence-coded anthemization score from URL-verified documentation of Pride, drag, ballroom, club, activist, and queer-media use. Anthemization was unrelated to whether artists were publicly queer (rs = −.02) and only weakly related to explicit queer lyrics (rs = .22). Among 16 language-model-coded lyrical themes, camp potential predicted anthemization after family-wise correction and controls for era, genre, authorship, and explicit content (partial r = .40, n = 55). The association replicated with masked lyrics, among songs the coding model could not identify, with a second coding model, and with two independent outcome coders (partial r = .45–.54, n = 46), although it was not robust when the model's own anthem judgements were also partialled out. Full-mix acoustic descriptors related to formant energy and voicing predicted anthemization (|rs| = .38–.43, family-wise p < .05) but vanished when computed on Demucs-separated vocal stems (n = 80), locating the acoustic signal in production texture rather than the singer's voice; spectral brightness was confounded by genre and era. Outcome reliability was moderate (weighted κ = .57). The results support reception-centred accounts of musical queerness. They implicate lyrical camp and production as affordances for queer adoption, not a characteristic queer voice.