Influence of speech sample on perceptual judgement of hypernasality

Author(s)

Maria Natália Leite de Medeiros

• Document type: Master's Dissertation
• Press: Bauru.
• Institution: Universidade de São Paulo (USP). Faculdade de Odontologia de Bauru (FOB/SDB)
• Defense date: 2015-02-27

Examining board members

Renata Paciello Yamashita; Adriana de Oliveira Camargo Gomes; Rosana Prado de Oliveira

• Advisor: Renata Paciello Yamashita

Abstract

Purpose: To investigate the influence of speech material, spontaneous conversation or sentences repetition, on intra and inter-rater reliability. Material and Methods: 120 audio recorded speech samples were selected and edited of individuals with repaired cleft palate±lip, both genders, aged 6 to 52 years old (mean 21±10), 60 samples containing spontaneous conversation and 60 samples containing sentences repetition. Recordings were analyzed by three experienced speech and language therapists, which rated hypernasality according to their own criteria using 4-point scale: 1=absence of hypernasality, 2=mild hypernasality, 3=moderate hypernasality and 4=severe hypernasality. The analyses were performed in two steps: first spontaneous speech and 30 days after, sentences repetition. Intra and inter-rater agreements were calculated for both speech samples and were statistically compared by the Z test. Differences between two speech samples were considered statistically significant at the 5% level. Results: Comparison of intra-rater agreements between both speech samples showed an increase of the coefficients obtained in the analysis of sentences repetition compared to those obtained in spontaneous conversation from 0.45 (moderate) to 1.00 (almost perfect) regarding rater 1; from 0.60 (substantial) to 0.74 (substantial) to rater 2 and from 0.44 (moderate) to 0.92 (substantial) to rater 3. Statistically significant differences were verified to raters 1 (p<0.001) and 3 (p=0.006). Comparison between inter rater agreement showed that the agreement coefficients obtained among the three raters ranged from 0.34 (fair) to 0.48 (moderate) for spontaneous conversation and from 0.31 (fair) to 0.43 (moderate) for sentences repetition with no statistically significant difference between two speech samples (p=0.970). Conclusion: Sentences repetition improved intra-rater reliability of hypernasality. However, the speech material had no influence on reliability among different raters. (AU)

• FAPESP's process: 13/14769-4 - Influence of speech sample on the perceptual judgment of hypernasality
• Grantee: Maria Natália Leite de Medeiros
• Support Opportunities: Scholarships in Brazil - Master