Investigating lexical stress in Hindi–Urdu pop music

Although Hindi–Urdu is described as having lexical stress, experimental studies have failed to find consistent acoustic correlates and perceptual cues that signal stress (Ohala 1986, 1991; Dryud 2001). We present evidence consistent with proposals that Hindi–Urdu does not have lexical stress from musical text-setting. If descriptions of Hindi–Urdu as having lexical stress are accurate (e.g., Kelkar, 1968), stressed syllables in music should fall on longer beats. If there is no lexical stress, no association is expected between stressed syllables and their beat duration. We extracted 818 syllabic tokens from recent Hindi–Urdu pop songs and used Bayesian hierarchical logistic models with weakly informative priors to test the effect of syllable type on beat duration. Our results revealed no association between putatively stressed syllables and their beat duration. We did find evidence, however, for heavier syllable onsets and longer beats, after controlling for compression effects, and for polysyllabic shortening, a negative correlation between the length of a word and the duration of syllables within it, observed in many disparate languages to maintain speech rhythm (cf. White and Turk 2010). We contrast these results with findings from spoken Urdu to present a more complete description of lexical stress in this language.