The reliability and validity of an isiZulu speech reception Threshold test in quiet for hearing and hearing-impaired individuals in KwaZulu-Natal

Panday, Seema

The reliability and validity of an isiZulu speech reception Threshold test in quiet for hearing and hearing-impaired individuals in KwaZulu-Natal

Thesis / Dissertation

2025

Publisher

University of Cape Town

Department

Division of Communication Sciences and Disorders

Faculty

Faculty of Health Sciences

Abstract

Speech reception threshold testing is part of the initial audiological test battery, and should be in the first language of the individual being tested. A newly developed isiZulu Speech Reception Threshold (SRT) test in KwaZulu-Natal Province, South Africa, required further reliability and validity evidence before it can be used for clinical practice. The aim of this study was therefore twofold: to determine the reliability and the validity of the newly developed isiZulu Speech reception threshold (zSRT) test for individuals with and without hearing loss. Methods: This study utilised an observational, analytical, repeated measures research design and had two phases. Phase 1 consisted of the reliability, as well as content, convergent, and divergent validity testing components, while Phase 2 consisted of the concurrent validity aspect. These five components were identified for inclusion in a newly developed unified validation framework for this study. Of the 176 participants who were selected through non-probability sampling, 100 had normal hearing and 76 had hearing loss (mild and moderate conductive, moderately severe and severe sensorineural hearing loss). Results: For the reliability study, the zSRT test was found to have high internal and external consistency (reliability) [ICC values ranged from 0.69 to 0.79]; Bland and Altman analyses showed that the isiZulu SRT test scores differed by no more than 7.5 to 8.7 dB HL between original and repeat assessments. Regarding the content validity, more than 20% of participants rated three of the word recordings below ‘strongly agree' in the categories of pitch or tone, and for one word recording i.e. /cinga/ below strongly agree in the categories of pitch or tone, clarity or articulation and naturalness or dialect. First language raters proved useful in identifying problematic word recordings that other methods of assessment would have missed. Intraclass correlation coefficient analyses showed zSRT scores were in substantial to very high agreement with the Pure Tone Average (PTA) scores for the normal hearing and hearing loss groups (NH– right ear ICCconsistency = 0.78, left ear ICC = 0.67; HL – right ear ICCconsistency = 0.97, left ear ICCconsistency = 0.95), thereby confirming the convergent validity of the isiZulu SRT test. A mixed model analysis indicated a significant difference between the zSRT and PTA (p < 0.001), despite there being no clinical significance (2.9 to 3.0 dB for those with normal hearing, and an average of 2.0 to 3.0 dB for those with hearing loss). For divergent validity, there was a significant difference between the PTA and zSRT and eSRT (p < 0.001). However, clinically, there was a 0 – 3dB difference between the PTA and zSRT and a 5dB difference between the eSRT and PTA for both ears. The mean psychometric slope (%/dB) at 50% correct perception for all words in the zSRT test was 4.92%/dB for the mild conductive hearing loss group, 5.26%/dB for the moderate group, 2.85%/dB for the moderately severe sensorineural group, and 2.47%/dB for the severe sensorineural hearing loss group. These slopes were appropriate for the degree of hearing loss observed in each group, confirming the concurrent validity of the zSRT test. Discussion: The collective examination of each measurement-related element, according to the unified validation framework (reliability; content, convergent, divergent, and concurrent validity), was used to confirm the reliability and validity of the zSRT. The findings suggest the need for multiple levels and methods to be considered to validate speech reception threshold tests to ensure that the test can be used in clinical practice. There is value in including first language community members as raters to establish the content validity of the test, with a broader scientific lens being required when interpreting the SRT and PTA correlation for tonal languages. Contextual factors may impact the overall validation of new speech audiometry tests. Conclusion: The zSRT test was found to be reliable and valid and can be considered for clinical practice in KwaZulu-Natal Province (notwithstanding the study limitations). The study also offers a new framework for validating speech audiometry tests, which includes both reliability and validity elements, and considers the contextual factors.