8th Speech in Noise Workshop, 7-8 January 2016, Groningen

Exploring the roles of auditory and cognitive factors and lexical difficulty in individual differences in word-in-noise perception by older adults

Sarah Knight(a)
MRC Institute of Hearing Research, Nottingham, GB

Antje Heinrich(b)
MRC Institute of Hearing Research, Nottingham, GB

(a) Presenting
(b) Attending

Spoken word recognition relies on the ability to match representations derived from acoustic information to an existing item in the mental lexicon. Listeners’ ability to do this depends upon a number of factors, including overall audibility and factors intrinsic to the target word. In particular, lexical characteristics of the target word, such as word frequency, neighbourhood density (number of phonemically similar words) and neighbourhood frequency (word frequency of phonemically similar words), have been shown to affect word recognition: lexically easy targets (words with high frequency and/or low density and/or low-frequency neighbourhoods) are recognised more easily than lexically hard targets (words with low frequency and/or high density and/or high-frequency neighbourhoods), in normal-hearing adults (Sommers & Danielson, 1999), children (Eisenberg et al, 2002), cochlear implant users (Kaiser et al, 2003) and native and non-native speakers (Bradlow & Pisoni, 1999). Moreover, one study investigating factors contributing to individual differences in young and old listeners’ ability to recognise lexically hard versus easy words suggested a role for cognitive processes such as inhibition (Sommers & Danielson, 1999).

This study further investigated the relationship between lexical difficulty and individual differences in speech-in-noise intelligibility in older listeners. Fifty older adults (ages = 61-86; mean = 70, age-normal hearing) listened to 200 monosyllabic words whose lexical characteristics varied in terms of word frequency (WF; high vs. low) and neighbourhood density (ND; high vs. low), presented in a background of speech-modulated noise at two signal-to-noise ratios (SNR; +1dB and -2dB). Individual measures of hearing (PTA0.25-8kHz and temporal processing) and cognition (inhibition, working memory and linguistic skill) were also obtained.

Initial analyses show that hearing thresholds, but not age, influenced overall intelligibility scores. When accounting for hearing thresholds, main effects of SNR, WF and ND were as predicted, with participants performing better in the higher SNR, for high-frequency words and for low-density words. A significant interaction of SNR x ND showed that the effect of ND was larger in the more challenging SNR, while a significant interaction of WF x ND showed that participants performed better for high-frequency, low-density words than any other combination of WF and ND. Intelligibility scores were also significantly affected by inhibition as measured by a visual Stroop task, with better inhibitory abilities associated with higher intelligibility; inhibitory abilities may also account for the larger ND effect in the low SNR. Further analyses will examine the relationship between individual differences in intelligibility and cognitive abilities in more detail.

Last modified 2016-05-12 14:22:09