Effect of listeners' linguistic background on perceptual judgements of hypernasality

Abstract

Many speech and language therapists work in a multilingual environment, making cross-linguistic studies of speech disorders clinically and theoretically important. To investigate the effect of listeners' linguistic background on their perceptual ratings of hypernasality and the reliability of the ratings. The speech samples were nine Cantonese non-nasal sentences produced by 22 speakers (20 speakers with hypernasality and two speakers with normal resonance). Twenty-four non-expert listeners (twelve Cantonese and twelve English) rated the speech samples using direct magnitude estimation. The Cantonese listeners gave significantly higher ratings to the female speech samples (mean = 76.02) than the English listeners (mean = 59.24; t = 3.189, p<0.05). The difference in direct magnitude estimation ratings between the Cantonese (78.57) and English (70.83) listeners was not significant for the male samples (t = 2.097, p>0.05). Despite the difference in numerical ratings between the two groups of listeners, the high correlations between their ratings indicated that they ranked the speech samples in terms of hypernasality severity similarly. Both groups of listeners showed high inter-judge reliability but low intra-judge reliability for rating the two sets of speech samples. There was a significant difference in intra-judge reliability between the Cantonese (r = 0.55) and English (r = 0.39) listeners for the male samples (t = 2.125, p<0.05). Generally, the non-expert Cantonese and English listeners ranked the Cantonese samples in terms of hypernasality in a similar way. The reliability of ratings by non-expert listeners was moderate. The need for further cross-linguistic studies into perceptual evaluations of speech disorders is highlighted.