Emily Markessis

French native speakers in the making: from language-general to language-specific voicing boundaries.

By examining voice onset time (VOT) discrimination in 4- and 8-month-olds raised in a French-speaking environment, the current study addresses the question of the role played by linguistic experience in the reshaping of the initial perceptual abilities. Results showed that the language-general -30- and +30-ms VOT boundaries are better discriminated than the 0-ms boundary in 4-month-olds, whereas 8-month-olds better discriminate the 0-ms boundary. These data support explanations of speech development stressing the effects of both language-general boundaries and linguistic environment (attunement theory and coupling theory). Results also suggest that the acquisition of the adult voicing boundary (at 0 ms VOT in French vs. +30 ms VOT in English) is faster and more linear in French than in English. This latter aspect of the results might be related to differences in the consistency of VOT distributions of voiced and voiceless stops between languages.

Perceptual biases for rhythm : the Mismatch Negativity latency indexes the privileged status of binary vs non-binary interval ratios

OBJECTIVE Rhythm perception appears to be non-linear as human subjects are better at discriminating, categorizing and reproducing rhythms containing binary vs non-binary (e.a. 1:2 vs 1:3) as well as metrical vs non-metrical (e.a. 1:2 vs 1:2.5) interval ratios. This study examined the representation of binary and non-binary interval ratios within the sensory memory, thus yielding a truly sensory, pre-motor, attention-independent neural representation of rhythmical intervals. METHODS Five interval ratios, one binary, flanked by four non-binary ones, were compared on the basis of the MMN they evoked when contrasted against a common standard interval. RESULTS For all five intervals, the larger the contrast was, the larger the MMN amplitude was. The binary interval evoked a significantly much shorter (by at least 23 ms) MMN latency than the other intervals, whereas no latency difference was observed between the four non-binary intervals. CONCLUSIONS These results show that the privileged perceptual status of binary rhythmical intervals is already present in the sensory representations found in echoic memory at an early, automatic, pre-perceptual and pre-motor level. SIGNIFICANCE MMN latency can be used to study rhythm perception at a truly sensory level, without any contribution from the motor system.

Categorical perception of voicing, colors and facial expressions: A developmental study

The aim of the present paper was to compare the development of perceptual categorization of voicing, colors and facial expressions in French-speaking children (from 6 to 8 years) and adults. Differences in both categorical perception, i.e. the correspondence between identification and discrimination performances, and in boundary precision, indexed by the steepness of the identification slope, were investigated. Whereas there was no significant effect of age on categorical perception, boundary precision increased with age, both for voicing and facial expressions though not for colors. These results suggest that the development of boundary precision arises from a general cognitive maturation across different perceptual domains. However, this is not without domain specific effects since we found (1) a correlation between the development of voicing perception and some reading performances and (2) an earlier maturation of boundary precision for colors compared to voicing and facial expressions. These comparative data indicate that whereas general cognitive maturation has some influence on the development of perceptual categorization, this is not without domain-specific effects, the structural complexity of the categories being one of them.

N1b and Na subcomponents of the N100 long latency auditory evoked-potential: Neurophysiological correlates of voicing in French-speaking subjects

OBJECTIVE To look for the presence of neurophysiological correlates of language-general voicing boundaries in French by analyzing the morphology of two N100 subcomponents (N1b and T-complex). METHODS /d/ and/t/ syllables with a voice onset time (VOT) value varying evenly from -75 and +75 ms were presented to French-speaking adults as stimuli for scalp-recorded auditory evoked-potentials. Morphologies and peak latencies of N1b and T-complex subcomponents were assessed. RESULTS The Na subcomponent of the T-complex was double-peaked for VOT values below -30 ms and above +30 ms. N1b subcomponent revealed a double-peaked response above +30 ms VOT and a single-peaked response for all other VOT values. Whenever the response was double-peaked, there was a correlation between the VOT value and the N1b or Na supplementary peak latency. CONCLUSIONS The combined morphologies of N1b and Na yield clear neurophysiological correlates of the language-general boundaries. For negative VOT values, the differential behavior of N1b and Na subcomponents suggests that only Na possesses physiological properties indexing the two language-general boundaries. SIGNIFICANCE Rather than being lost, the universal sensitivity of human newborns to language-general boundaries remains present even if in some languages such as French, they do not separate phonological categories.

Frequency tuning curves derived from auditory steady state evoked potentials: a proof-of-concept study.

Objectives: Assess the feasibility of drawing tuning curves from the masking function of steady state potentials. Develop a noninvasive tool for research applications on cochlear frequency selectivity in sedated animals. Obtain pilot human data validating auditory steady state evoked potential-derived (ASSEP) tuning curves against psychophysical data. Design: ASSEP tuning curves were drawn in 10 Beagle puppies and six human adults using amplitude-modulated probes. Two probe frequencies (1 and 2 kHz) were used in dogs and only one (2 kHz) in humans. The modulation rates of the two probes were set to 81 and 88 Hz, respectively. Psychophysical tuning curves were obtained in 12 normal human subjects using the same maskers and either a pure-tone or an amplitude-modulated probe to verify if the latter had a specific effect on tuning curve parameters. Six of these 12 subjects participated in the electrophysiologic measurements. For each tuning curve, the intensity of the narrowband masker required just to mask the fixed probe was plotted for different masker center frequencies. Masker center frequencies extended to about half an octave above and an octave below the probe frequencies in 100-Hz steps. Tuning curve width (Q10 dB values), high- and low-frequency slopes (in dB/octave) and the masker frequency yielding the lowest masking threshold (maximal masker frequency) were computed. Canine Q10 dB values obtained were compared with those published for several species with other techniques. For humans, ASSEP and psychophysical tuning curves were directly compared in the same subjects and with published data. Results: In dogs, the ASSEP method yielded reproducible tuning curves with qualitative and quantitative parameters similar to other physiologic measures of tuning obtained in various animals. Q10 dB values were greater at 2 than at 1 kHz, reflecting the well-known correlation between sharpness of tuning and central frequency. In humans, ASSEP Q10 dB values were slightly smaller than the psychophysical ones, but were greater by a factor of 2 than those obtained with previously published electrophysiologic procedures. In both species, detuning—a shift of the tip of the curve away from the probe frequency—was frequently observed as upward shifts with a maximal value of 200 Hz. Human psychophysical tuning curves also showed a certain amount of upward detuning. The intraindividual comparison of the two types of probes performed on human subjects with the psychophysical method did not indicate a specific effect of the amplitude-modulated probe on the curve parameters. Neither did the intraindividual comparisons indicate that an amplitude-modulated probe per se promoted detuning. Detuning has been observed with several other techniques and is usually attributed to nonlinear interactions between masker and probe in simultaneous masking. Conclusions: The results demonstrate the feasibility of measuring realistic ASSEP tuning curves in sedated dogs and in sleeping human adults. The ASSEP tuning curves exhibit a series of classical features similar to those obtained with time-honored methods. These results pave the way for the development of a noninvasive electrophysiologic method for tuning curve recording and its applications in noncooperative experimental animals or clinical subjects.

Modification of the Threshold Equalising Noise (TEN) test for cochlear dead regions for use with steeply sloping high-frequency hearing loss

Steeply sloping high-frequency hearing loss is often associated with cochlear dead regions. These can be identified by measuring pure-tone thresholds in quiet and in Threshold-Equalising Noise (TEN). However, many patients cannot be adequately tested because the low frequencies in the TEN lead to uncomfortable loudness. We investigated the effect of high-pass filtering on the TEN-test results and the loudness of the TEN. Twenty-four normally hearing subjects and 35 subjects with steeply sloping high-frequency hearing loss were tested, using the standard TEN (TENs), and TEN high-pass filtered at 0.5 kHz (TEN0.5) or 1 kHz (TEN1). For both groups, masked thresholds did not differ across noise types for frequencies above 1 kHz. Over 50% of the hearing-impaired ears tested met the criteria for a dead region at 4 kHz, using all three noise types. However, masked thresholds and the prevalence of positive TEN-test results at 1 kHz were both lower with the TEN1. The TEN1 was judged the most comfortable noise by 68% of the hearing-impaired subjects. We conclude that high-pass filtering would allow testing at higher TEN levels for patients with steeply sloping hearing loss. Sumario La hipoacusia de perfil abruptamente descendente en frecuencias agudas se asocia frecuentemente con regiones cocleares muertas. Esto se puede identificar midiendo los umbrales tonales en silencio y en ruido ecualizado al umbral (TEN). Sin embargo, algunos pacientes no pueden ser estudiados adecuadamente porque las frecuencias bajas en el TEN, les produce molestia. Investigamos el efecto del filtro pasa-alto en los resultados de las pruebas TEN y la intensidad subjetiva del TEN. Se examinaron 24 sujetos normoyentes y 35 con hipoacusia de perfil abruptamente descendente en frecuencias agudas, utilizando el TEN habitual y el TEN con filtro pasa/alto a 0.5 kHz (TEN0.5) o 1 kHz (TEN1). Los umbrales de enmascaramiento no difirieron con los diferentes tipos de ruido para las frecuencias por arriba de 1 kHz en ambos grupos. Más del 50% de los oídos hipoacúsicos cumplieron con los criterios para considerar una zona muerta a 4 kHz, con los tres tipos de ruido. Sin embargo, los umbrales de enmascaramiento y la prevalencia de una prueba TEN positiva a 1 kHZ fueron menores con el TEN1. El 68% de los hipoacúsicos consideraron al TEN1 como el ruido más confortable. Concluimos que el filtro pasa-alto puede permitir realizar pruebas con mayor intensidad TEN en pacientes con hipoacusia de perfil abruptamente descendente en frecuencias agudas.

Auditory steady-state evoked potentials (ASSEPs): a study of optimal stimulation parameters for frequency-specific threshold measurement in dogs.

OBJECTIVE To define the optimal stimulation parameters (AM/FM vs AM alone and modulation rate) for frequency-specific threshold measurements using ASSEPs in dogs. Dependent variables were thresholds and recording times needed to obtain a response at threshold. To compare the ASSEP threshold results obtained with the optimal stimulation parameters to those obtained with the Tone-Burst/Auditory Brainstem Response (TB/ABR) combination. METHODS Thirty-two sedated Beagle puppies were tested at 5 audiometric frequencies (0.5-8 kHz) and 6 ASSEP modulation rates (21-199 Hz). RESULTS The ASSEP threshold-modulation rate functions had a high-pass profile with corner frequencies of 101 Hz for 0.5, 1 and 2 kHz carriers and of 151 Hz for 4 and 8 kHz carriers. AM stimuli did not yield higher thresholds than the AM/FM ones except at 1 kHz. Modulation type had no effect on testing duration. Audiometric profiles were obtained much more rapidly with ASSEPs than with TB/ABRs (mean: 50 vs 135 min). Both ASSEP and TB/ABR provided thresholds estimates characterized by low intersubject variability. CONCLUSIONS ASSEPs are a valid and rapid method for audiometric assessment in sedated dogs. SIGNIFICANCE ASSEPs offer a new, competitive tool for frequency-specific assessment of hearing in the canine species.

Effect of phonological training in French children with SLI: Perspectives on voicing identification, discrimination and categorical perception

The aim of the present study was to investigate the effect of auditory training on voicing perception in French children with specific language impairment (SLI). We used an adaptive discrimination training that was centred across the French phonological boundary (0 ms voice onset time--VOT). One group of nine children with SLI attended eighteen twenty-minute training sessions with feedback, and a control group of nine children with SLI did not receive any training. Identification, discrimination and categorical perception were evaluated before, during and after training as well as one month following the final session. Phonological awareness and vocabulary were also assessed for both groups. The results showed that children with SLI experienced strong difficulties in the identification, discrimination and categorical perception of the voicing continuum prior to training. However, as early as after the first nine training sessions, their performance in the identification and discrimination tasks increased significantly. Moreover, phonological awareness scores improved during training, whereas vocabulary scores remained stable across sessions.

Effect of presentation level on diagnosis of dead regions using the threshold equalizing noise test

The effect of the level of threshold equalizing noise (TEN) on the diagnosis of dead regions (DRs) was investigated. Participants comprised 23 adults with sensorineural hearing impairment. Masked thresholds were measured monaurally with TEN at 60, 70, 80, and 90 dB HL/ERBN. Absolute and masked thresholds (with TEN at 80 dB HL/ERBN) were retested. The diagnosis was unaffected by TEN level at any frequency for eight of the 13 participants who met the criteria for a DR. For four of the latter, increasing the TEN level changed the diagnosis from DR to no DR, mainly at 1.5 kHz, corresponding to the edge frequency, fe, of the DR. For one participant with a low-frequency DR, increasing the TEN level changed the diagnosis from no DR to DR, only at 1 kHz. The changes with level were too large to be explained in terms of test repeatability. Overall, the results of the TEN test are usually not affected by TEN level, except for test frequencies close to fe when absolute thresholds are near-normal for frequencies adjacent to fe.

Evidence for a dual versus single origin of the MMNs evoked by cued versus cueless deviants

OBJECTIVE This study was designed to separately test the effect of the cued/cueless nature of deviant stimuli and that of temporal distance between sound and deviance onsets on the mismatch negativity (MMN) as well as to look for discrepancies between behavioural discrimination performances and MMN amplitude when deviants are cueless. METHODS Ten healthy adults passively listened to stimuli that were contrasted by the presence or absence of a frequency sweep starting early or late within the sound. Discrimination performances were collected after the electrophysiological sessions. RESULTS MMNs were much larger for cued than for cueless deviants. The temporal distance between sound and deviance onsets affected MMNs evoked by both cued and cueless deviants, even to the point of abolishing the MMN when cueless deviance occurred late in the stimulus. Behavioural data were at ceiling levels for all conditions, contrasting with the absence of MMN evoked by cueless deviants with late onset. CONCLUSIONS Two mechanisms contribute to the MMN evoked by cued deviants: the memory comparison process and the adaptation/fresh-afferent one. Within the temporal window of integration, the delay at which each component disappears is different. SIGNIFICANCE Comparing waveforms evoked by cued versus cueless deviants provides a fairly simple way of isolating the MMN memory-based component.