Slide Slam F1
Predictability facilitates comprehension of but not adaptation to degraded speech in a graded manner
Pratik Bhandari1, Vera Demberg1, Jutta Kray1; 1Saarland University, Germany
Previous studies have shown a facilitatory effect of semantic predictability on language comprehension at moderate levels of spectral degradation (Obleser & Kotz, Cereb Cortex. 2010). It is argued that when speech is degraded, listeners have ‘narrowed expectations’ about the sentence endings, i.e., semantic prediction may be limited to only most highly predictable sentence completions (Strauß et al., J Cogn Neurosci. 2013). It has also been shown that listeners adapt to the speech presented at the same level of degradation when the semantic predictability of the sentences is kept constant (e.g., only low predictability sentences), showing performance improvements over the course of the study (Erb et al., J Neurosci. 2013). The main objectives of this study were (i) to examine whether listeners form narrowed expectations, or they form predictions across a wide range of probable sentence endings, and (ii) to assess the relationship between perceptual adaptation and the facilitatory effect of sentence predictability. For this, we created 360 German Subject-Verb-Object sentences that varied in semantic predictability of a sentence-final target word in a graded manner (high, medium, low) and levels of spectral degradation (1, 4, 6 and 8 channels noise-vocoding). These sentences were presented auditorily to two groups: One group (n=48) performed a listening task in an unpredictable channel context in which the degraded speech levels were randomized, while the other group (n=50) performed the task in a predictable channel context in which the degraded speech levels were blocked. The results showed that at 4-channels noise-vocoding, response accuracy was higher in high-predictability sentences than in the medium-predictability sentences (β=1.14, SE=0.37, z (1608)=3.10, p=.002), which in turn was higher than in the low-predictability sentences (β=1.01, SE=0.24, z (1608)=4.20, p<.001). This suggests that in contrast to the narrowed expectation view, comprehension of moderately degraded speech ranging from low to high including medium predictability sentence conditions are facilitated in a graded manner; listeners probabilistically preactivate upcoming words from a wide range of semantic space, not limiting only to highly probable sentence endings. Additionally, in both channel contexts, we did not observe learning effects, i.e., response accuracy did not increase over the course of experiment (β=-0.0001, SE=0.01, z (6917)=-0.02, p=0.985), and response accuracy was higher in the predictable than in the unpredictable channel context (β=-0.27, SE=0.14, z (6917)=-2.02, p=.043). We speculate from these observations that when there is no trial-by-trial variation of the levels of spectral degradation, listeners adapt to speech quality at a long timescale (i.e., across the block); however, when there is a trial-by-trial variation of the high-level semantic feature (e.g., sentence predictability), listeners do not adapt to low level perceptual property (e.g., speech quality) at a short timescale (i.e., trial-by-trial learning). Taken together, these findings support the probabilistic prediction account of language processing adding a novel insight in the case of degraded speech comprehension, and also shows the probable effect of high-level semantic feature of language in adaptation to low-level sensory characteristic of speech. and also shows the probable effect of high-level semantic feature on adaptation to low-level sensory characteristic of speech.