Speech Identification Task Research Articles

Word frequency effects for a closed-set word identification task.

Previous work demonstrated that adults achieved higher performance scores than young children on a 4-alternative forced-choice (4 AFC), picture-pointing speech identification task even though all stimuli (words and pictures) had been developed to be familiar to 3-year-old inner-city children. It was hypothesized that these results reflected 'word frequency effects' in the sense that, even though children knew the words, adults had much greater experience with the stimuli and, consequently, required less acoustic information for correct performance. In the current study, this hypothesis of word frequency effects was tested by creating new materials in which the stimulus items (recorded words and pictures) remained the same but new picture foils representing less frequently occurring words replaced the incorrect response alternatives. According to Broadbent's word frequency, response bias views, it was hypothesized that adults would show improved performance on the modified test materials, compared to the original ones. Data supported this hypothesis and also showed that, in children, the magnitude of word frequency effects is related to receptive vocabulary skills. This phenomenon needs to be considered in developing clinical tests of speech or language that have a closed-set format.

Prescription of gain-setting and prognosis for use and benefit of post-aural hearing aids.

Degree of hearing aid use, aid gain setting and the improvement given by the aid in a speech identification task were determined for a sample of post-aural hearing aid users. Measures of the spectral effects of gain settings received by individuals suggested that the poor frequency responses of the aids examined severely restricted the overall gain and hence the frequency range of useful gain. The outcome variables of use, gain and improvement in identification were each examined in relation to a selection of audiological predictor variables that would be easily obtained clinically. Multiple-regression analysis furnished equations accounting for a considerable proportion of the variance in gain setting and in use and for a moderate proportion of the variance in the identification improvement score. These results suggest that it is valid to predict use on an individual basis when determining the emphasis of rehabilitative programmes and it is helpful to predict gain for individuals when advising them on the gain setting to be attempted. The patterns of correlations obtained justify determining the speech detection threshold and the uncomfortable loudness level in order to predict a most comfortable level and when assessing the benefit and use of an aid.

Dichotic listening in a patient with a presumed lesion in the region of the medial geniculate

Dichotic and monotic performance data were obtained from a patient with a lesion involving thalamic and parathalamic structures. Information gained from the patient's neurological examination, arteriograms and ventriculograms suggest the lesion is confined to the area of the medial geniculate and/or the brachium of the inferior colliculus on the right side. Data will be presented showing normal monaural and monotic performance in “target-challenge” speech identification tasks [Berlin et al., IEEE Trans. Audio Electroacoust, AU-21, 189 (1973)], but differential effects in dichotic target-challenge identification tasks, depending on whether the target was presented to the right ear or left ear. [Supported by National institues of Health, USPHS Grant No. NH 11647-02.]

Selectivity versus summation in multiple observation tasks: Evidence with spectrum parameter noise in speech

The effect of stimulus repetition was investigated in a speech identification task where intelligibility was lowered not by white noise added to the audio waveform but by ‘structural’ noise added to the spectrum parameters. Several aspects of the results argue for the improvement in intelligibility between one and two presentations being due not to statistical averaging over internal or external noise, but to increased perceptual selectivity under the influence of the first presentation's stimulus properties.

Earmold Influence on Aided Speech Identification Tasks

The physical characteristics of hearing aids are modified when the instrument is coupled with an earmold and fitted to an ear canal. This study was undertaken to compare the electroacoustic characteristics of four specific earmold variations with behavioral speech audiometric tasks in five normal-hearing and seven sensorineural-hearing-loss subjects. Speech-Békésy thresholds were obtained under each earmold condition with continuous discourse for detectability, intelligibility, most comfortable loudness, and tolerance. Speech discrimination ability was evaluated with the Modified Rhyme Test in three signal-to-noise ratio conditions. Significant differences in test scores attributable to earmold modification were difficult to demonstrate, in spite of the fact that substantial variations in the ear inserts were readily apparent in both structural and electroacoustics analysis. Clinical implications of earmold effect on hearing and performance are discussed.

Temporal ordering and speech identification abilities.

The relationship between auditory temporal ordering ability and speech identification ability was investigated in 30 normal subjects. Two tones, differing in frequency, served as signals for the temporal ordering task. Speech identification ability was measured at various presentation rates by replaying sentences at more rapid speeds than recorded. The speech identification task and the temporal ordering task showed a correlation of 0.08. These results demonstrate no significant relationship between temporal ordering ability, as measured by a two-signal procedure, and speech comprehension in normals.