Age Differences in the Effects of Semantic Context on Speech Perception: The Role of Uncertainty.

摘要

Recent experiments have shown that semantic context effects on spoken word recognition differ between young and older adults (Rogers, Jacoby & Sommers, 2012; Sommers et. al, in prep). These experiments have shown that older adults provide significantly more incorrect responses than young adults when context predicts a phonological competitor that is semantically more likely than the actual target item. Incorrectly responding with the semantically predicted item rather than the true target item, with high confidence, is referred to as false hearing. The general design used in the studies reported for this dissertation is to compare identification of sentence-final items (presented in noise) in three different conditions: (1) a condition in which target words were congruent with the preceding context; (2) a baseline condition in which no semantic context was presented; (3) an incongruent condition in which the target item was a semantically acceptable phonological competitor of the congruent trial target word.;The current work was conducted to see if reducing uncertainty would decrease age differences in false hearing. Uncertainty was manipulated in three different ways. Experiment 1 compared open to closed-set response formats to investigate if age differences would still emerge when uncertainty in responses was reduced by only having two responses to choose from relative to an open-set design. Results indicated that reducing uncertainty by limiting the number of responses did not reduce age differences in false hearing rates. Experiment 2 compared young and older adults' false hearing when the signal-to-noise ratio (SNR) was manipulated to either increase or decrease level of stimulus degradation. Results of Experiment 2 demonstrated that regardless of how much uncertainty was increased (by making the SNR harder), older adults' performance when context was congruent did not change. Experiment 3 manipulated uncertainty by presenting a validity cue ("likely predictive" or "likely misleading") prior to sentence presentation. The results showed that even when told an incongruent trial was likely to be misleading older adults' performance was no different than when presented a neutral cue. These results are discussed within the context of age differences in cognitive control as a contributing proof to differences in false hearing.