Effect of age on silent gap discrimination in synthetic speech stimuli

The difficulty that older listeners experience understanding conversational speech may be related to their limited ability to use information present in the silent intervals (i.e., temporal gaps) between dynamic speech sounds. When temporal gaps are present between nonspeech stimuli that are spectrally invariant (e.g., noise bands or sinusoids), older listeners are less able to resolve temporal gaps than are younger listeners. It has also been demonstrated that temporal gap perception deteriorates as the frequency difference between the sounds bordering the silent gap increases, and this effect becomes more pronounced with age (J. Lister, J. Besing, and J. Koehnke, 2002; J. Lister, J. Koehnke, and J. Besing, 2000). In this study, the effect of age on the ability to discriminate temporal gaps in dynamic stimuli (i.e., changing in frequency and duration over time) was measured in a gap duration discrimination (GDD) task. The participants were two groups of listeners with normal hearing sensitivity through at least 4000 Hz: (a) ages 21-38 years and (b) ages 50-72 years. Stimuli simulated the frequency characteristics of 1 consonant (/s/), a steady-state vowel (/a/), a weak diphthong (/eI/), and a voiced bilabial plosive (/burst/) in 6 combinations: (a) /s-a/, (b) /s-eI/, (c) /a-a/, (d) /eI-eI/, (e) /burst-a/, and (f) /burst-eI/. For each of the 6 phoneme combinations, 2 conditions of stimulus duration were used: (a) fixed, meaning that the durations of leading and trailing noises were fixed at values typical for the speech sounds being simulated, and (b) random, meaning that duration varied randomly within a range acceptable for accurate perception of the speech contrasts. Gap duration difference limens (GDDLs) were significantly larger for the older listeners than for younger listeners. For both groups, GDDLs were poorer for the spectrally dynamic marker pairs (e.g., /burst-eI/) than for the marker pairs that were relatively stable in frequency over time (/a-a/). GDD performance also was poorer for the random conditionthan for the fixed condition. Listener age and hearing sensitivity were significantly correlated, and age was not significantly correlated with GDD when controlling for hearing sensitivity. The authors conclude that temporal resolution (as measured by gap discrimination) is affected by age, stimulus complexity, and, perhaps, by hearing sensitivity in a speech context.