Digit Speech Reception Threshold Testing in Tswana/English Speakers

Abstract

The purpose of this study was to establish whether digit stimuli offer a more accurate measure for Speech Reception Threshold (SRT) testing when assessing first-language Tswana (or Setswana), second-language English speakers, as compared to an English word list (CID W-1) and a Tswana word list. Forty Tswana first language speaking participants (17 males and 23 females) aged between 18 and 25 years, participated in this study. All participants were undergraduate students at a tertiary institution in Johannesburg, Gauteng. This study utilized a quantitative single group correlation design which allowed for a comparison between three SRT scores (CID-SRT, T-SRT, and D-SRT). Participants underwent basic audiological assessment procedures comprising of otoscopy, tympanometry, conventional pure tone audiometry and SRT testing. SRT measures were established using monitored live voice testing. Basic audiometric data were descriptively analyzed to ensure that hearing function was with in normal limits, and PTA-SRT averages and means were calculated. Furthermore, analysis of the SRT-PTA correlation data was conducted through the use of the non-parametric Spearman's correlation co efficient and linear regression. Results from this study were statistically significant (p <.05) and indicated that digit-pairs were not the most effective stimuli for establishing SRT, compared to the CIDW-1 and Tswana word lists. On the contrary, findings of the current study revealed that PTA-SRT comparison was best in Tswana (r= 0 .62), followed very closely by CID W-1 (r = 0.61), and lastly digit- pairs (r = 0.60). The results however, confirm the efficacy of using digit pairs as alternative stimuli when more appropriate speech stimuli for the establishment of SRT are unavailable, as the correlation between SRT for digit pairs and PTA was also a strong one (r= 0.60). Linear regression analyses indicated that all three lists were acceptable speech stimuli for the population under investigation with the standard error of estimate being significantly smaller than the 5dB-stepused to collect the data (1.62 for Tswana, 3.56 for CID W-1, and 3.80 for digit-pairs).