G. Cameron Marean

Acoustic analyses and perceptual data on anticipatory labial coarticulation in adults and children.

The present study investigated anticipatory labial coarticulation in the speech of adults and children. CV syllables, composed of [s], [t], and [d] before [i] and [u], were produced by four adult speakers and eight child speakers aged 3-7 years. Each stimulus was computer edited to include only the aperiodic portion of fricative-vowel and stop-vowel syllables. LPC spectra were then computed for each excised segment. Analyses of the effect of the following vowel on the spectral peak associated with the second formant frequency and on the characteristic spectral prominence for each consonant were performed. Perceptual data were obtained by presenting the aperiodic consonantal segments to subjects who were instructed to identify the following vowel as [i] or [u]. Both the acoustic and the perceptual data show strong coarticulatory effects for the adults and comparable, although less consistent, coarticulation in the speech stimuli of the children. The results are discussed in terms of the articulatory and perceptual aspects of coarticulation in language learning.

Infant psychometric functions for detection: Mechanisms of immature sensitivity

Psychometric functions are described for individual 6- to 9-month-old infants and for individual adults for auditory detection of repeated, long- and short-duration tone bursts in quiet and for single, long-duration tone bursts in quiet and in noise. In general, infant psychometric functions have reduced upper asymptotes, shallower slopes, and poorer thresholds than adult psychometric functions. Infant-adult differences in slope and threshold are greater for short-duration tones than for other stimuli. Infant upper asymptotes are around 0.85 correct for all stimuli. One explantation for these findings is that infants are inattentive a certain proportion of time during the detection task. This model cannot account for the very shallow short-duration stimulus slope, nor can it account for infant-adult threshold differences for any stimulus. Other models of immature attention, or listening strategies, may be able to account for the slope and upper asymptote as well as the threshold of infant psychometric functions. Some combination of inattentiveness and primary neural immaturity may also account for the data. Although immaturities exist, some aspects of the detection process appear to be quantitatively similar in infants and adults.

Speech perception in children with histories of recurrent otitis media

The present study investigated the ability of 5-year-old children to perceive differences in voice onset time (VOT) in naturally produced speech. Three groups of children whose hearing was within normal limits at the time of the experiment were tested on identification and discrimination tasks: (1) group C, in which the children had normal language abilities and no history of severe, recurrent otitis media (OM), (2) group OM, in which the children had histories of severe, recurrent OM but normal language abilities, and (3) group OM/DL, in which the children had histories of severe OM and delays in the acquisition of linguistic competence. Compared to group C, group OM/DL showed marked differences in their ability to identify and discriminate speech patterns. Their perception was less categorical, as well, in that there was less of a peak in the discrimination function at the region of the phoneme boundary. The performance of group OM fell between the other two groups, with deficits being more pronounced in the discrimination task than in the identification task. The results supported the idea that episodes of OM can produce periods of sensory deprivation that alter perceptual abilities. The relation of a history of OM to later language and academic difficulties was also considered.

Vowel categorization by very young infants

Six-month-old infants are known to categorize vowels despite variation in talker voice and pitch contour. Using the observer-based psychoacoustic procedure, this study asked whether 2- and 3-month-olds could categorize similarly. Infants were trained to respond whenever the vowel category alternated from /a/ to /i/ and to refrain from responding when the vowel category remained the same, despite variation in spectral cues associated with pitch and talker changes. Eighty percent of 2-, 3-, and 6-month-olds did not respond the first time a talker change occurred in the absence of a vowel change, suggesting that even the younger infants recognize these spectrally different sounds as perceptually equivalent

Regenerated hair cells in the European starling: are they more resistant to kanamycin ototoxicity than original hair cells?

Previous work from our laboratory [Marean et al. (1993) Hear. Res. 71, 125-136] has shown that a 10 day dose of 200 mg/kg/day kanamycin produced damage to the basal 34% of the starling basilar papilla. We also observed that repeating the dosing schedule following a 4 month survival period resulted in significantly less damage to the regenerated auditory epithelium. The present study investigated whether or not this apparent resistance was the result of a tendency for regenerated hair cells to be less susceptible to kanamycin ototoxicity, or if other, systemic factors may be involved. Eight European starlings were given subcutaneous injections of 200 mg/kg/day kanamycin for 10 days. Serum levels of kanamycin were measured at the time of sacrifice for all birds, and the basilar papillae of all birds were examined by scanning electron microscopy (SEM). Two of these birds (Group 1) were sacrificed immediately following the dosing period. Two of the birds were allowed to survive for 60 days (Group 2). Two of the birds were redosed with 200 mg/kg/day for 10 days after 60 days survival (Group 3). Finally, two birds were redosed with 250 mg/kg/day until serum levels of kanamycin were the same as Group 1 when sacrificed (> 9 micrograms/ml). The SEM results showed that the regenerated auditory epithelium of the birds dosed a second time sustained less damage compared to previously untreated ears, even though the dosing regimen was the same (Group 3 versus Group 1). The regenerated auditory epithelium of birds dosed a second time sustained the same damage as previously untreated animals when the dosage was increased to attain similar serum levels (Group 4 versus Group 1). These results suggest metabolic changes occur in the starling in response to the initial dose of kanamycin which do not necessarily involve changes in hair cell resistance to ototoxicity.

Auditory perception following hair cell regeneration in European starling (Sturnus vulgaris): Frequency and temporal resolution

Behavioral detection thresholds, auditory filter widths, and temporal modulation transfer functions were obtained from four starlings before, during and after 11 days of subcutaneous injections of kanamycin, an aminoglycoside antibiotic. Birds were operantly conditioned to respond to pure tones and amplitude modulated noises ranging in frequency from 0.25 to 7 kHz using adaptive staircase procedures and were tested daily for 92 days after the first injection of aminoglycoside. All birds had threshold shifts of at least -60 dB at frequencies above 4 kHz. Lower frequencies were affected in some birds, although none of the birds had hearing loss below 3 kHz. All four birds had wider auditory filters at 5 kHz immediately after the aminoglycoside series. Any changes in frequency resolution at frequencies below 5 kHz were slight, transitory, and rarely observed. Two of the four birds had permanently wider auditory filters at 5 kHz. Temporal modulation transfer functions were briefly affected in two birds during the time of greatest threshold shift. Recovery of detection thresholds began soon after the injections ceased and continued for approximately 60 days. Recovery in frequency resolution lagged behind auditory threshold by about 10 days. Normal temporal resolution was observed in the context of impaired intensity and frequency resolution. Changes in auditory threshold and frequency resolution were closely associated for all birds at 5 kHz, but were correlated with statistical significance in only two birds. Scanning electron microscopy was performed on all four birds after 90 days of recovery and confirmed that the extent of initial damage was consistent with the pattern of observed hearing loss.

Infant thresholds for short duration tone bursts

Thresholds for short duration tone bursts were measured for three‐and six‐month‐old infants and for adults. An experimenter, blind to trial type, observed the infants behavior. On each trial, the experimenter judged whether or not a tone had been presented and received feedback. The stimuli were 1000‐Hz pure tones, with 16‐ms rise/fall times and no steady‐state duration. Eight tone bursts were presented at a rate of 2/s during a signal trial. Stimuli were presented monaurally via insert earphone. Thresholds were estimated using an adaptive procedure. Adults were tested using the same stimuli and procedure. Thresholds for six‐month‐olds averaged about 13 dB higher, while those of three‐month‐olds averaged about 24 dB higher than those of the adults. The size of the infant‐adult difference for these short duration tones is nearly the same as that reported for longer duration stimuli [Olsho et al., J. Acoust. Soc. Am. Suppl. 1 80, S123 (1986)]. That the effects of duration on detection threshold are quite s...

The resolution of frequency and time following hair cell regeneration in birds

Detection thresholds, auditory filter widths, and temporal modulation transfer functions were measured in European starlings before, during, and after a 10‐day injection series of kanamycin, an ototoxic aminoglycoside antibiotic. Hearing loss in response to the drug exposure was observed in frequencies at or above 3 kHz; threshold shifts above 3 kHz were ≳60 dB. The auditory filter broadened at 5 kHz from 9% to 20% of center frequency (CF). After 60 days of recovery, detection thresholds improved to within 15 dB of predose estimates, and auditory filters at 5 kHz recovered to 11% of CF and were permanently changed in the skew toward lower frequencies. At 3 and 1 kHz, changes in filter bandwidths were small and transient. At 3 kHz, no permanent change in filter width occurred, but filters were permanently skewed toward lower frequencies. At 1 kHz, changes in skew occurred, but only during the period of greatest threshold shift at higher frequencies. Temporal resolution was largely unaffected by hearing los...

Infant psychometric functions: Asymptotic performance.

Psychometric functions have been obtained for detection of repeated, short (16‐ms) tone bursts [L. A. Werner and G. C. Marean, Assoc. Res. Otolaryngol. (1991)] and for single, long (300‐ms) tone bursts [J. Y. Bargones and L. A. Werner, Assoc. Res. Otolaryngol. (1992)] from infants and adults. Infant functions for these stimuli are similar: they tend to have shallower slopes and lower upper asymptotes than those of adults. Best infant performance for both stimuli is about 85% correct. Infant thresholds are also worse than those of adults, but this may result partially from the fact that thresholds were estimated using methods that assume an upper asymptote of 100%. However, most studies of infant threshold present a series of long duration, repeated stimuli. In the current study, upper asymptotic performance for this more optimal stimulus was examined. The observer‐based psychoacoustic procedure [Olsho et al., Devel. Psychol. 23, 627–640 (1987)] was used to test 6‐ to 9‐month‐old infants. The stimulus cons...

Speech perception: The effects of recurrent conductive hearing loss in young children

The ability of five‐year‐old children with histories of recurrent otitis media (OM) and associated conductive hearing loss to perceive syllable initial stop consonants was investigated. Naturally produced, computer‐edited stimuli were created to form continua that ranged in voice onset time (VOT) from 7 to 60 ms. The ability of children to assign these stimuli to adult phonetic categories was assessed by identification and discrimination tasks. Three subject groups were identified: One group (group C) had no history of OM and was linguistically age appropriate; another group (group OM) had considerable involvement with OM and was also linguistically normal; the third group (group OMDL) had an equal involvement as group OM with otitis media, but was linguistically delayed. All groups were matched for age, IQ, socioeconomic class, and educational experience. Results indicate that OM was associated with poorer discrimination scores, regardless of the linguistic status of the individual.