ENTREVISTA A DOCENTES

Effect of stimulus choice on hearing aid processing

Objective aided outcome measures involve measurement of AEPs while stimuli are transduced through a hearing aid. In essence, the hearing aid acts as a relay station because the elicited AEP is dependent on the hearing aid output. Choice of stimuli is therefore vital to valid aided AEP testing. Hearing aid interaction is an

important factor to consider because non-linear features in hearing aids such as compression may respond differently to non-speech inputs that are not close

representations of speech in terms of spectral composition, crest factor and temporal features (e.g., Henning & Bentler, 2005; Scollie & Seewald, 2002; Stelmachowicz, Lewis, Seewald, & Hawkins, 1990).

Application of objective measures in threshold estimation has led to increased emphasis on frequency specificity of stimuli, but these stimuli may not be preferred to predict hearing aid processing for speech. Since compression achieved depends on the duration of stimuli and its temporal characteristics (Stone & Moore, 1992), the compression ratio for speech is often lower than that for static pure tones (Fortune, 1997; Henning & Bentler, 2008; Stelmachowicz, Kopun, Mace, & Lewis, 1996). The aided output of non-speech weighted pure tones could differ as much as 20 dB relative to that of real speech (Scollie & Seewald, 2002). The largest discrepancies occur at the higher frequencies and at higher input levels for non-linear hearing aids

(Henning & Bentler, 2005; Stelmachowicz et al., 1996). These discrepancies are greater for high power hearing aids, typically prescribed for profound degrees of hearing loss (Scollie & Seewald, 2002), longer release times, higher numbers of channels and high compression ratios (Henning & Bentler, 2005). Additional features such as noise reduction may differentially process speech and

non-speech-like inputs, and de-emphasize non-speech-like stimuli (Bentler & Chiou, 2006; Scollie & Seewald, 2002). In short, stimuli with closer resemblance to natural speech in temporal characteristics will likely ensure a better representation of hearing aid functioning for speech (experienced in everyday listening conditions) during the test.

Effect of current AEP stimulus protocols on hearing aid function

Brief stimuli such as clicks or tone bursts used for ABRs are deemed unsuitable for aided applications due to inaccuracies in the representation of hearing aid function, especially for stimuli-like speech (Beauchaine, Gorga, Reiland, & Larson, 1986; Brown, Klein, & Snydee, 1999; Frye, 1987; Gorga, Beauchaine, & Reiland, 1987; Stelmachowicz et al., 1990). Speech syllables used for cABRs are likely more suitable stimuli as they are longer than clicks and tone bursts (e.g., Anderson & Kraus, 2013). Phonemes or syllables extracted from running speech or a syllable/word have been used as CAEP stimuli (Golding et al., 2006; Tremblay, Billings, et al., 2006). Although CAEP protocols use natural speech, the segments of speech used as stimuli are not used in their natural form. CAEP protocols typically intersperse stimulus phonemes with 1–2 seconds of inter-stimulus interval. In contrast, in running speech, phonemes are mostly interspersed with other amplitude-varying phonemes. Therefore, input-level-dependent non-linear hearing aids may or may not process the same phoneme similarly in the isolated CAEP context versus running speech. A discrepancy in output levels of the same phoneme in the two contexts

may imply that hearing aid function during CAEP testing is not representative of hearing aid function during natural speech. The accuracy of representation of hearing aid function during CAEP testing is investigated in Chapter 4.

Additionally, hearing aid processing may alter stimulus characteristics in one or more ways that may affect the AEP recorded (Billings, Tremblay, & Miller, 2011; Billings, Tremblay, Souza, & Binns, 2007; Jenstad, Marynewich, & Stapells, 2012; Marynewich, Jenstad, & Stapells, 2012). Hearing aid noise floor leading to poorer Signal-to-Noise Ratio (SNR) in aided conditions (Billings et al., 2011, 2007), and the increase in tone burst rise-time leading to inaccurate representation of hearing aid gain at stimulus onsets (Jenstad et al., 2012; Marynewich et al., 2012; Stapells, Marynewich, & Jenstad, 2010) have been cited as barriers to using CAEPs as a valid objective aided outcome measure. With linear processing in digital hearing aids, changes in stimulus rise-time resulted in lower gain at stimulus onset relative to the gain achieved during verification using pure tones (Jenstad et al., 2012). Studies investigating the effect of hearing aid processing on stimulus onset and CAEPs are limited to linear hearing aids. The changes caused by linear hearing aid processing may differ from stimulus changes caused by non-linear hearing aid processing. Since most hearing aids used today are non-linear (Dillon, 2012; Johnson, Cox, & Alexander, 2010), recognizing stimulus changes caused by non-linear processing and the effect of such changes on CAEPs is important and warrants further investigation. The effect of non-linear hearing aid processing on tone burst CAEPs is investigated in Chapter 3.

Speech-evoked EFRs and FFRs may prove to be advantageous because the stimulus could take the form of a sentence similar to natural running speech. EFRs to vowels embedded within naturally spoken words in sentence contexts have been recorded in clinically feasible test times (Choi et al., 2013). As seen above, stimuli presented in

a sentence form with envelope modulation rates similar to that of running speech are likely favorable to represent non-linear hearing aid function for running speech encountered in everyday listening situations. However, in addition to considering hearing aid factors, we also need to consider factors in Category III (AEP

functionality) to evaluate the usefulness of AEPs for objective aided outcome evaluation. The following section includes comparison of AEPs and AEP protocols in their potential to represent changes in audibility due to level, amplification and bandwidth.

Category III: AEP functionality