Neural Representation Of Speech Research Articles

Perceptual flexibility in word learning: Preschoolers learn words with speech sound variability

Children’s language input is rife with acoustic variability. Much of this variability may facilitate learning by highlighting unvarying, criterial speech attributes. But in many cases, learners experience variation in those criterial attributes themselves, as when hearing speakers with different accents. How flexible are children in the face of this variability? The current study taught 3–5-year-olds new words containing speech-sound variability: a single picture might be labeled both deev and teev. After learning, children’s knowledge was tested by presenting two pictures and asking them to point to one. Picture-pointing accuracy and eye movements were tracked. While children pointed less accurately and looked less rapidly to dual-label than single-label words, they robustly exceeded chance. Performance was weaker when children learned two distinct labels, such as vayfe and fosh, for a single object. Findings suggest moderate learning even with speech-sound variability. One implication is that neural representations of speech contain rich gradient information.

Impaired auditory processing and neural representation of speech in noise among symptomatic post-concussion adults

ABSTRACTBackground: The purpose of the study was to examine auditory event-related potential (AERP) evidence of changes in earlier and later stages of auditory processing in individuals with long-term post-concussion problems compared to healthy controls, with a secondary aim of comparing AERPs by functional auditory behavioral outcomes.Methods: P1-N1-P2 complex and P300 components recorded to speech in quiet and background noise conditions were completed in individuals with ongoing post-concussion symptoms following mTBI and healthy controls. AERPs were also examined between sub-groups with normal or impaired auditory processing by behavioral tests.Results: Group differences were present for later stages of auditory processing (P300). Earlier components did not significantly differ by group overall but were more affected by noise in the mTBI group. P2 amplitude in noise differed between mTBI sub-groups with normal or impaired auditory processing.Conclusion: AERPs revealed differences between healthy controls and those with chronic post-concussion symptoms following mTBI at a later stage of auditory processing (P300). Neural processing at the earlier stage (P1-N1-P2) was more affected by noise in the mTBI group. Preliminary evidence suggested that it may be only the proportion of individuals with functional evidence of central auditory dysfunction with changes in AERPs at earlier stages of processing.

Neural representation of speech in pediatric cochlear implant recipients

ObjectiveCochlear implantation (CI) is an established treatment for selected individuals with bilateral severe-to-profound sensorineural hearing loss who derive limited benefits from conventional hearing aids.This work was designed to assess speech processing at the brainstem and the cortical level in children fitted with CIs after a variable duration of implantation and speech therapy compared with language acquisition.Patients and methodsThirty-one children between 4 and 5 years of age fitted with unilateral CIs of variable duration ranging from 1 to 3 years were assessed at 1 year (n=10), 2 years (n=12), and 3 years (n=9) after device activation. They underwent aided sound–field audiological evaluation, speech-evoked auditory brainstem response, and speech-evoked mismatch negativity test. The results were compared among the study groups and then correlated with language assessment and speech perception tests.ResultsBoth speech-evoked auditory brainstem response and mismatch negativity test responses were significantly different among the three groups. Moreover, language development showed a significant difference among the three groups.ConclusionChildren fitted with CI showed cortical and brainstem activation from the first year and these activity changes continue with CI use and both are highly correlated with receptive and expressive language. Thus, both electrophysiologic tests could be early and critical objective indicators of optimal speech encoding after programming of the CI device.

Native language shapes automatic neural processing of speech

The development of the phoneme inventory is driven by the acoustic-phonetic properties of one's native language. Neural representation of speech is known to be shaped by language experience, as indexed by cortical responses, and recent studies suggest that subcortical processing also exhibits this attunement to native language. However, most work to date has focused on the differences between tonal and non-tonal languages that use pitch variations to convey phonemic categories. The aim of this cross-language study is to determine whether subcortical encoding of speech sounds is sensitive to language experience by comparing native speakers of two non-tonal languages (French and English). We hypothesized that neural representations would be more robust and fine-grained for speech sounds that belong to the native phonemic inventory of the listener, and especially for the dimensions that are phonetically relevant to the listener such as high frequency components. We recorded neural responses of American English and French native speakers, listening to natural syllables of both languages. Results showed that, independently of the stimulus, American participants exhibited greater neural representation of the fundamental frequency compared to French participants, consistent with the importance of the fundamental frequency to convey stress patterns in English. Furthermore, participants showed more robust encoding and more precise spectral representations of the first formant when listening to the syllable of their native language as compared to non-native language. These results align with the hypothesis that language experience shapes sensory processing of speech and that this plasticity occurs as a function of what is meaningful to a listener.

Physiologic discrimination of stop consonants relates to phonological skills in pre-readers: a biomarker for subsequent reading ability?(†).

Reading development builds upon the accurate representation of the phonological structure of spoken language. This representation and its neural foundations have been studied extensively with respect to reading due to pervasive performance deficits on basic phonological tasks observed in children with dyslexia. The subcortical auditory system – a site of intersection for sensory and cognitive input – is exquisitely tuned to code fine timing differences between phonemes, and so likely plays a foundational role in the development of phonological processing and, eventually, reading. This temporal coding of speech varies systematically with reading ability in school age children. Little is known, however, about subcortical speech representation in pre-school age children. We measured auditory brainstem responses to the stop consonants [ba] and [ga] in a cohort of 4-year-old children and assessed their phonological skills. In a typical auditory system, brainstem responses to [ba] and [ga] are out of phase (i.e., differ in time) due to formant frequency differences in the consonant-vowel transitions of the stimuli. We found that children who performed worst on the phonological awareness task insufficiently code this difference, revealing a physiologic link between early phonological skills and the neural representation of speech. We discuss this finding in light of existing theories of the role of the auditory system in developmental dyslexia, and argue for a systems-level perspective for understanding the importance of precise temporal coding for learning to read.

Neural representations of vowels and vowel-like sounds

Neural representations of speech at every level of the neuraxis have nonlinear features that are not described by spectrograms or linear filter-banks. In this study, recent computational models for populations of cells in the auditory periphery, brainstem, and midbrain were used to explore the implications of vowel features for neural responses. Peripheral neural responses are characterized by strong periodicities, dominant frequencies, that depend upon the distribution of energy across the harmonics and vary across vowels [Fant, 1970]. Strong periodicities related to the fundamental and low-frequency harmonics are observed for peripheral neurons tuned to a wide range of frequencies [Delgutte & Kiang, 1984]. The computational model captures this feature of the physiological responses. These periodicities are interesting because many midbrain neurons are tuned to fluctuations in this frequency range. Single-formant vowels allowed systematic manipulations of the relationship between formant and harmonic frequencies. Larger envelope fluctuations occur when harmonic and formant frequencies are mismatched than when they are aligned. The auditory models suggest that these differences in fluctuation amplitude are significant for responses of higher–order auditory neurons that are tuned to fluctuation rate. The long-term goal is to understand the interrelationship of vowel space and neural responses. Support: NIH-NIDCD-R01-001641(LHC) & NSF-0853929(JMM)

Intelligibility predictors and neural representation of speech

Intelligibility predictors tell us a great deal about human speech perception, in particular which acoustic factors strongly effect human behavior, and which do not. A particular intelligibility predictor, the Articulation Index (AI), is interesting because it models human behavior in noise, and its form has implications about representation of speech in the brain. Specifically, the Articulation Index implies that a listener pre-consciously estimates the masking noise distribution and uses it to classify time/frequency samples as speech or non-speech. We classify consonants using representations of speech and noise which are consistent with this hypothesis, and determine whether their error rate and error patterns are more or less consistent with human behavior than representations typical of automatic speech recognition systems. The new representations resulted in error patterns more similar to humans in cases where the testing and training data sets do not have the same masking noise spectrum.

Abstract Coding of Audiovisual Speech: Beyond Sensory Representation

Is there a neural representation of speech that transcends its sensory properties? Using fMRI, we investigated whether there are brain areas where neural activity during observation of sublexical audiovisual input corresponds to a listener's speech percept (what is "heard") independent of the sensory properties of the input. A target audiovisual stimulus was preceded by stimuli that (1) shared the target's auditory features (auditory overlap), (2) shared the target's visual features (visual overlap), or (3) shared neither the target's auditory or visual features but were perceived as the target (perceptual overlap). In two left-hemisphere regions (pars opercularis, planum polare), the target invoked less activity when it was preceded by the perceptually overlapping stimulus than when preceded by stimuli that shared one of its sensory components. This pattern of neural facilitation indicates that these regions code sublexical speech at an abstract level corresponding to that of the speech percept.

Neural representation of spectral and temporal information in speech

Speech is the most interesting and one of the most complex sounds dealt with by the auditory system. The neural representation of speech needs to capture those features of the signal on which the brain depends in language communication. Here we describe the representation of speech in the auditory nerve and in a few sites in the central nervous system from the perspective of the neural coding of important aspects of the signal. The representation is tonotopic, meaning that the speech signal is decomposed by frequency and different frequency components are represented in different populations of neurons. Essential to the representation are the properties of frequency tuning and nonlinear suppression. Tuning creates the decomposition of the signal by frequency, and nonlinear suppression is essential for maintaining the representation across sound levels. The representation changes in central auditory neurons by becoming more robust against changes in stimulus intensity and more transient. However, it is probable that the form of the representation at the auditory cortex is fundamentally different from that at lower levels, in that stimulus features other than the distribution of energy across frequency are analysed.

Amplitude envelope onsets and developmental dyslexia: A new hypothesis.

A core difficulty in developmental dyslexia is the accurate specification and neural representation of speech. We argue that a likely perceptual cause of this difficulty is a deficit in the perceptual experience of rhythmic timing. Speech rhythm is one of the earliest cues used by infants to discriminate syllables and is determined principally by the acoustic structure of amplitude modulation at relatively low rates in the signal. We show significant differences between dyslexic and normally reading children, and between young early readers and normal developers, in amplitude envelope onset detection. We further show that individual differences in sensitivity to the shape of amplitude modulation account for 25% of the variance in reading and spelling acquisition even after controlling for individual differences in age, nonverbal IQ, and vocabulary. A possible causal explanation dependent on perceptual-center detection and the onset-rime representation of syllables is discussed.

Discriminability of vowel representations in cat auditory-nerve fibers after acoustic trauma.

This paper attempts to connect deficits seen in the neural representation of speech with perceptual deficits. Responses of auditory-nerve fibers were studied in cats exposed to acoustic trauma. Four synthetic steady-state vowels were used as test signals; these stimuli are identical, except that the second format (F2) resonator in the synthesizer was set to 1.4, 1.5, 1.7, or 2 kHz, producing four spectra that differ mainly in the vicinity of the F2 frequency. These stimuli were presented to a large population (523) of auditory-nerve fibers in four cats with sloping high-frequency threshold shifts that reached 50-70 dB at 2-4 kHz. In normal animals, May et al. [Auditory Neurosci 3, 135-162 (1996)] showed previously that the discharge rates of fibers with best frequencies near the F2 frequencies provide enough information to allow discrimination of these stimuli at the performance levels shown by cats in behavioral experiments. Here it is shown that, after acoustic trauma, there is essentially no rate information which would allow the vowels with different F2 frequencies to be discriminated. However, information that could allow discrimination remains in the temporal (phase-locked) aspects of the responses.

Effects of high sound levels on responses to the vowel /ε/ in cat auditory nerve

The vowel /ε/ was used to study auditory-nerve responses at high sound levels (60–110 dB). By changing the playback sampling rate of the stimulus, the second formant (F2) frequency was set at best frequency (BF) for fibers with BFs between 1 and 3 kHz. For vowel stimuli, auditory-nerve fibers tend to phase-lock to the formant component nearest the fiber's BF. The responses of fibers with BFs near F2 are captured by the F2 component, meaning that fibers respond as if the stimulus consisted only of the F2 component. These narrowband responses are seen up to levels of 80–100 dB, above which a response to F1 emerges. The F1 response grows, at the expense of the F2 response, and is dominant at the highest levels. The level at which the F1 response appears is BF dependent and is higher at lower BFs. This effect appears to be suppression of the F2 response by F1. At levels near 100 dB, a component 1/component 2 transition is observed. All components of the vowel undergo the transition simultaneously, as judged by the 180° phase inversion that occurs at the C2 transition. Above the C2 threshold, a broadband response to many components of the vowel is observed. These results demonstrate that the neural representation of speech in normal ears is degraded at high sound levels, such as those used in hearing aids.

Frequency-shaped amplification changes the neural representation of speech with noise-induced hearing loss

Temporal response patterns of single auditory nerve fibers were used to characterize the effects of a common hearing-aid processing scheme, frequency-shaped amplification, on the encoding of the vowel /ε/ in cats with a permanent noise-induced hearing loss. These responses were contrasted with responses to unmodified stimuli in control and impaired cats. Noise-induced hearing loss leads to a degraded representation of the formant frequencies, in which strong phase locking to the formants is not observed in fibers with best frequencies (BFs) near the formants and there is a wide spread of formant phase locking to fibers with higher BFs ( Miller et al. 1997a, b). Frequency shaping effectively limits the upward spread of locking to F1, which improves the representation of higher frequency components of the vowel. However, it also increases phase locking to harmonics in the trough between the formants, which decreases the contrast between F1 and the trough in the neural representation. Moreover, it does not prevent the spread to higher BFs of responses to the second and third formants. The results show a beneficial effect of frequency shaping, but also show that interactions between particular gain functions and particular spectral shapes can result in unwanted distortions of the neural representation of the signal.