Hearing speech sounds: Top-down influences on the interface between audition and speech perception

Davis, Matthew H.; Johnsrude, Ingrid S.

doi:10.1016/j.heares.2007.01.014

Cited by 369 publications

(312 citation statements)

References 161 publications

Supporting

Mentioning

263

Contrasting

Unclassified

Order By: Relevance

“…These findings are in line with several studies that have associated temporal cortex with speech intelligibility (Davis & Johnsrude, 2007; Hickok & Poeppel, 2007; Narain et al., 2003). In contrast, our results showed BOLD decreases in temporal cortex between the first and second presentations of the distorted stimuli (see also Blank & Davis, 2016).…”

Section: Discussionsupporting

confidence: 93%

Predictive processing increases intelligibility of acoustically distorted speech: Behavioral and neural correlates

et al. 2017

View full text Add to dashboard Cite

IntroductionWe examined which brain areas are involved in the comprehension of acoustically distorted speech using an experimental paradigm where the same distorted sentence can be perceived at different levels of intelligibility. This change in intelligibility occurs via a single intervening presentation of the intact version of the sentence, and the effect lasts at least on the order of minutes. Since the acoustic structure of the distorted stimulus is kept fixed and only intelligibility is varied, this allows one to study brain activity related to speech comprehension specifically.MethodsIn a functional magnetic resonance imaging (fMRI) experiment, a stimulus set contained a block of six distorted sentences. This was followed by the intact counterparts of the sentences, after which the sentences were presented in distorted form again. A total of 18 such sets were presented to 20 human subjects.ResultsThe blood oxygenation level dependent (BOLD)‐responses elicited by the distorted sentences which came after the disambiguating, intact sentences were contrasted with the responses to the sentences presented before disambiguation. This revealed increased activity in the bilateral frontal pole, the dorsal anterior cingulate/paracingulate cortex, and the right frontal operculum. Decreased BOLD responses were observed in the posterior insula, Heschl's gyrus, and the posterior superior temporal sulcus.ConclusionsThe brain areas that showed BOLD‐enhancement for increased sentence comprehension have been associated with executive functions and with the mapping of incoming sensory information to representations stored in episodic memory. Thus, the comprehension of acoustically distorted speech may be associated with the engagement of memory‐related subsystems. Further, activity in the primary auditory cortex was modulated by prior experience, possibly in a predictive coding framework. Our results suggest that memory biases the perception of ambiguous sensory information toward interpretations that have the highest probability to be correct based on previous experience.

show abstract

Section: Discussionsupporting

confidence: 93%

Predictive processing increases intelligibility of acoustically distorted speech: Behavioral and neural correlates

et al. 2017

View full text Add to dashboard Cite

show abstract

“…In the frame of psycholinguistic models of language processing, our results suggest that degraded speech comprehension, which requires restoration or compensation of partial linguistic information, constitutes a privileged situation in which to look at the importance of bottom-up and top-down interactions during speech processing (see also Davis and Johnsrude, 2007 Cocktail party situations may further allow testing the issue of subliminal processing in multi-talker babble as participants can not consciously process all the overlapping speech streams. Whereas subliminal processing has been extensively studied in visual word processing (Forster and Davis, 1984), it has hardly ever been tested in the auditory modality due to the lack of a suitable paradigm.…”

Section: Discussionmentioning

confidence: 93%

“…Our findings show that this lexical interference is stronger when words in the babble are highly frequent, as these words may strongly activate the mental lexicon, hence increasing lexical competition between target speech and background. As mentioned in the introduction, psycholinguistic models of language processing postulate that word recognition is the result of strong competitive mechanisms between concurrently activated lexical candidates (Davis and Johnsrude, 2007;Marslen-Wilson et al, 1996;McClelland and Elman, 1986). Since less information is required to activate highly frequent words compared to low frequent words, in a particularly challenging situation, these words may be stronger competitors during word recognition than low-frequency words.…”

Section: Discussionmentioning

confidence: 99%

“…Most psycholinguistic models of language processing postulate that speech recognition is supported by an interactive system in which bottom-up processes (going from low-level acoustic information to higher-level information such as meaning) are combined with topdown processes where high-level information may modulate lower-level processing (e.g., Davis and Johnsrude, 2007). If the representation of a word becomes active, this lexical activation might increase the activation of the phonemes that compose the word, allowing them to be recognized with less acoustic evidence than they would need without the top-down lexical influence (Samuel, 2001).…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Real-time lexical competitions during speech-in-speech comprehension

Boulenger

Hoen

Ferragne

et al. 2010

Speech Communication

View full text Add to dashboard Cite

This study aimed at characterizing the cognitive processes that come into play during speech-inspeech comprehension by examining lexical competitions between target speech and concurrent multi-talker babble. We investigated the effects of number of simultaneous talkers (2, 4, 6 or 8) and of the token frequency of the words that compose the babble (high or low) on lexical decision to target words. Results revealed a decrease in performance as measured by reaction times to targets with increasing number of concurrent talkers. Crucially, the frequency of words in the babble significantly affected performance: high-frequency babble interfered more strongly (by lengthening reaction times) with word recognition than low-frequency babble. This informational masking was particularly salient when only two talkers were present in the babble due to the availability of identifiable lexical items from the background. Our findings suggest that speech comprehension in multi-talker babble can trigger competitions at the lexical level between target and background. They further highlight the importance of investigating speechin-speech comprehension situations as they may provide crucial information on interactive and competitive mechanisms that occur in real-time during word recognition.

show abstract

“…It was hypothesized that for human listeners, native speech had advantages on the task of auditory detection, which does not require linguistic related responses, over non-native speech and non-speech sounds. Detecting native phonemes is enhanced possibly by the top-down mechanism during auditory processing (Davis and Johnsrude, 2007). Given the approach of Liu and Jin's study (2015) that compared detection thresholds of speech stimuli between Chinese and Korean listeners, this study was to measure a) Author to whom correspondence should be addressed.…”

Section: Introductionmentioning

confidence: 99%

Auditory detection of non-speech and speech stimuli in noise: Native speech advantage

Huo

Tao

Wang

et al. 2016

The Journal of the Acoustical Society of America

View full text Add to dashboard Cite

Detection thresholds of Chinese vowels, Korean vowels, and a complex tone, with harmonic and noise carriers were measured in noise for Mandarin Chinese-native listeners. The harmonic index was calculated as the difference between detection thresholds of the stimuli with harmonic carriers and those with noise carriers. The harmonic index for Chinese vowels was significantly greater than that for Korean vowels and the complex tone. Moreover, native speech sounds were rated significantly more native-like than non-native speech and non-speech sounds. The results indicate that native speech has an advantage over other sounds in simple auditory tasks like sound detection.

show abstract

Hearing speech sounds: Top-down influences on the interface between audition and speech perception

Cited by 369 publications

References 161 publications

Predictive processing increases intelligibility of acoustically distorted speech: Behavioral and neural correlates

Predictive processing increases intelligibility of acoustically distorted speech: Behavioral and neural correlates

Real-time lexical competitions during speech-in-speech comprehension

Auditory detection of non-speech and speech stimuli in noise: Native speech advantage

Contact Info

Product

Resources

About