Age Effects in Processing Bilinguals’ Accented Speech

Older adults often complain about difficulty understanding distorted speech, such as synthetic voices on telephones, answering-machine messages that are “too-fast”, and movies with actors speaking dialects of English different from their own. Younger adults (college-students) have been reported to complain about difficulties understanding teaching assistants with “foreign” accented speech (e.g., Brown, 1992; Norris, 1991). Speech spoken by bilingual speakers with non-native accents is prevalent in many settings and likely to be experienced by the young and old listeners alike. We ask whether older adult listeners are disproportionately affected, relative to younger adults, by bilinguals’ accented speech. One might expect that older adults have more difficulty than younger adults in processing accented speech because older listeners have been documented to have more difficulty with other forms of distorted speech, namely speech-in-noise, reverberant speech, and time-compressed speech, as will be discussed further (see also Wingfield & Stine-Morrow, 2000 and Schneider & PichoraFuller, 2000). Obler et al. (1985) showed that older monolingual adults (in their 70s) perform significantly worse than younger monolingual adults (aged 30-69) when reporting the final word of Speech Perception in Noise test (Kalikow et al., 1977) sentences under low stimulus-to-noise ratios. Pichora-Fuller et al. (1995) reported a similar finding. They demonstrated that older participants found it significantly more difficult than younger adults to filter the signal from the noise. When the noise level was low, older adults performed at ceiling on word-recognition tasks, but when the noise level was high, even participants with near-normal hearing sensitivity experienced difficulty with the task. Tun (1998) added temporal distortion to the speech-in-noise component to test the effects of speech rate, level of noise, and predictability of the material. Speech rate, level of noise, and predictability of the material were varied while comparing across two groups of listeners: young (1721 years old) and old (64-78 years old) in a sentence-repetition task. She found that for all participants, faster speech and lower signal-to-noise ratios yielded poorer performance. Moreover, there was a significant interaction between the two variables such that performance was most affected at fastest presentation at the highest noise level. While the older adults performed less well than the younger group overall, smaller differences were detected at the extreme ends of the difficulty continuum, i.e., at high noise levels (when performance was generally poor) and at low noise levels (when performance was usually good). Most differences between the two groups were found at rate and noise levels along the mid-range of the continuum. Note that neither the processing speed, nor the hearing sensitivity of the subjects, appeared to account for the results observed. Other studies of speeded speech have demonstrated differential effects on older as compared to younger presumably monolingual adults. Using a time-compression technique, Gordon-Salant and Fitzgibbons found that normal-hearing older adults showed poorer performance on the Speech-innoise (SPIN) test (Gordon-Salant and Fitzgibbons, 1993) and poorer recognition scores for sentences and phrases (Gordon-Salant and Fitzgibbons, 2001), when the material was presented at a fast (timecompressed) rate compared to normal (non-compressed) rate. Vaughan and Letowski (1997) also used a time-compression paradigm but found that their older participants did not differ from the younger participants on a sentence-repetition task, even at the fastest presentation condition. However, the older adults did experience greater difficulty than the younger participants at the fastest speech rate when they were required to make inferences from the spoken paragraph-level material. Age-related