Toward Clinical Measures of Speech Comprehension: Cortical Speech Processing in Normal-Hearing Listeners, Hearing Aid Users, and Cochlear Implantees

Intact speech perception (SP) is essential for effective communication. SP includes the detection, discrimination, and comprehension of speech sounds. Hearing loss (auditory deprivation) and hearing device treatment (compensatory auditory stimulation) are known to cause cortical changes that can affect SP. Measures that reflect these changes could further our understanding of these processes and could also inspire new clinical test methods. Electroencephalography (EEG) is an established method for studying the neurophysiological processes underlying SP. In audiological practice, the assessment of SP focuses on detection and discrimination, whereas comprehension-related abilities are ignored. The primary purpose of this PhD project was to develop clinically applicable measures for assessing speech comprehension and to evaluate them with normal-hearing individuals, hearing aid (HA) users, and cochlear implant (CI) users. Another aim was to explore the influence of hearing device treatment on these measures. To that end, three studies were conducted.

The first study developed EEG measures for assessing comprehension-related abilities based on clinically used digit materials. The materials were manipulated to obtain sequences of congruent and incongruent digit triplets. These sequences provided the basis for N400 and Late Positive Complex (LPC) measurements. The N400 reflects the brain’s response to an erroneous stimulus, while the LPC reflects the associated repair mechanism. Here, they served as neurophysiological proxies of comprehension abilities. The digits were presented either acoustically (‘A’) or first visually (digits 1-2) and then acoustically (digit 3; ‘VA’). VA presentation helps to ensure that the first two digits can be detected and discriminated, which could facilitate the assessment of comprehension related abilities in hard-of-hearing individuals. The digits were presented against stationary speech-shaped noise 10 dB above individually measured SRTs. Thirty young normal-hearing adults were tested. The targeted EEG responses could be evoked with the devised stimulus conditions. No differences in the EEG responses obtained with the two presentation modes were found. However, behavioral response times reflecting digit triplet congruency were significantly longer with A than with VA presentation, possibly because of a faster recall mechanism when the first two digits are presented visually.

The second study used the developed measures to explore the influence of HA use on comprehension-related abilities. Three age-matched groups of older individuals (N = 3 × 15) with (1) age-appropriate normal hearing, (2) mild-to-moderate sensorineural hearing loss (SNHL) but no prior HA experience, and (3) mild-to-moderate SNHL and >2 years of HA experience participated. The participants in groups 2 and 3 were fitted with HAs that ensured good speech audibility. All measurements were performed in the free field with stationary speech-shaped noise. As in study 1, N400 and LPC responses were evoked 10 dB above individually measured SRTs with A and VA presentation. While the EEG responses of interest could be evoked, no group differences were observed. The same was true for the SRTs and behavioral response times. As in study 1, the behavioral response times were significantly longer with A than VA presentation. Overall, these results could suggest that when speech audibility is taken care of, cortical responses reflecting comprehension abilities remain intact in individuals with SNHL, regardless of HA experience.

In the third study, the developed measures were applied to CI users who had used their devices for at least 12 months as well as to age-matched normal-hearing controls (N = 2 × 13). In addition to A and VA presentation, a visual-only condition was included. As in study 2, the analyses revealed no group differences in the evoked EEG responses. Regarding the behavioral response times, however, the CI users were slower and showed a different response pattern to the three presentation modes compared to the controls. These results suggest that experienced CI users need longer to comprehend speech and that response times can serve as a proxy for speech comprehension in such listeners.

In summary, this PhD project developed and tested measures for assessing comprehension-related abilities based on digit materials, which are in widespread clinical use. The measures were found to be usable with normal-hearing listeners, HA users, and CI users. The lack of group differences at the neurophysiological level could have been caused by good speech audibility, compensatory neural changes following hearing device treatment, or insufficient statistical power. The behavioral response times appear sensitive to some group differences and can potentially serve as a subjective measure of speech comprehension. Follow-up research with larger sample sizes is needed to better understand the roles of stimulus audibility and compensatory brain mechanisms for speech comprehension and to enable the transfer of the new measures to clinical settings.