[Clinical study of speech understanding in noise].

The goal of this study was to determine the extent to which the difficulty frequently experienced by hearing-impaired listeners in understanding noisy speech can be clinically described in terms of the auditory site of lesion. To this end, shifts of speech reception threshold in noise, using standardized franco-québécois bisyllabic word lists, were obtained from noise-exposed workers and compared to Plomp's model to ascertain content validity. Shifts resulted from monaural listening to a broad-band speech spectrum noise held constant at 0 dB signal-to-noise ratio. In this manner, increasing the presentation levels of both speech and noise was expected to progressively minimize the contribution of Plomp's class A attenuation component to speech hearing loss and eventually show an unequivocal manifestation of its distortion component (class D speech hearing loss). Listeners were 457 noise-exposed workers giving a total of 914 observations. Rejection of those ears with suspected middle-ear pathology led to a reduction to a total of 709 observations representing various degrees of permanent hearing impairment. An additional 98 observations showed normal hearing sensitivity in spite of noise exposure averaging 9.6 years. Results indicated the presence of five categories of speech threshold shifts varying independently of hearing sensitivity. 59.8% of those observations with permanent hearing impairment (424/709) simultaneously showed a problem understanding speech in noise. On 78 occasions, listening in noise was so disturbed that the 50% performance level corresponding to the speech threshold could no longer be established within reasonable limits, that is up to 65 dB above the speech threshold in a quiet environment. Conversely, 26.5% of the observations with essentially normal hearing sensitivity (26/98) showed reduced speech reception in noise. Fitting Plomp's model confirmed our expectations for two of the five categories of scores. A principal component analysis identified similar behaviour of speech threshold shifts in noise compared with Plomp's class D speech hearing loss. Hence, these speech threshold shifts in noise were interpreted as predominantly reflecting peripheral auditory disorders, that is defective frequency selectivity and temporal resolution. The remaining categories were found to be under the predominant influence of two other mechanisms: (1) band width reduction resulting from impaired hearing; (2) disorders associated with certain non-auditory cognitive factors, namely selective attention and the ability to make use of the redundancy of the message.