Class-like speech audiometry for the clinical evaluation of school aged CI users
Increasingly more children with cochlear implants (CI's) are being educated in mainstream schools. In Belgium, already 45–74% of the deaf students with a CI are enrolled in regular elementary and secondary schools. Speech perception in secondary schools is particularly difficult for CI users due to poor room acoustics, high levels of background noise, complex language and the fact that different courses are taught by different teachers. Despite these challenges affecting CI users in school, clinical evaluation of speech perception is currently administered with easy speech materials in a sound treated room and will therefore not accurately predict the speech performance in class. For this reason a class-like speech audiometry test was developed.
23 children with cochlear implants (mean age =13,7; SD = 1,8) and a control group of 28 normal hearing children participated in this study. A test situation was created in a reverberant room with 5 signal speakers and 6 noise speakers. The signal speakers were positioned as detailed in Valente (2012) to mimick the presence of students and teacher in class (1 meter from the listener, 5 different angels). The noise speakers produced a diffuse multitalker noise field of 65 dBSPL. In a randomized procedure the Linguistically controlled Sentences (LiCoS) (Coene et al., 2016) were administered from the 5 signal speakers with an adaptive procedure to obtain the 50% SRT of each speaker. An additional trial was performed in which speech was presented from all 5 speakers randomly.
Mean SRT’s were calculated for each signal speaker (L1-L5) and for the randomized trial using all speakers (Lrandom). Children with CI scored significantly worse in all conditions compared to their normal hearing peers. A pairwise within subject comparison between the six conditions (one-way repeated measures ANOVA) showed that two conditions were perceived more difficult by the control group (F(2.37, 63.96) = 16.12, p<.05, Bonferroni corrected): L1 (Speaker in front, SRT+3,6 dB) and Lrandom (SRT+5,6 dB). This effect was not found in the CI group. A class-like speech audiometry model was created that accurately mimicked the listening difficulties occurring in typical classrooms. This ecological model can be used in addition to current existing clinical evaluation of speech perception. Further research is necessary to get more insight into the directivity effects and binaural benefits found in this model.