Low-predictability Sentences Research Articles

Auditory measures in clinically depressed individuals. II. Auditory evoked potentials and behavioral speech tests

This is the second paper in a series of two papers comparing auditory measures in depressed and non-depressed individuals. In this paper, we describe the auditory brainstem responses (ABRs), auditory late responses (ALRs) and behavioral speech measures obtained from the same set of 36 individuals as in our previous paper. No changes were made to the inclusion criteria or subject classification. The results indicated a significantly larger amplitude growth with increase in intensity for ABR peak V and ALR peak N1P2 in the unmedicated group compared to the normal group. The unmedicated group performed less favorably on most behavioral speech tests administered compared to the control group, but the difference was significant only in the left ear for the Low Predictability Sentence List of the R-SPIN (Revised-Speech Perception in Noise) test. The mean test scores of the medicated group were closer to the scores of the control group.

Acoustic and linguistic factors in the perception of bandpass-filtered speech.

Speech can remain intelligible for listeners with normal hearing when processed by narrow bandpass filters that transmit only a small fraction of the audible spectrum. Two experiments investigated the basis for the high intelligibility of narrowband speech. Experiment 1 confirmed reports that everyday English sentences can be recognized accurately (82%-98% words correct) when filtered at center frequencies of 1500, 2100, and 3000 Hz. However, narrowband low predictability (LP) sentences were less accurately recognized than high predictability (HP) sentences (20% lower scores), and excised narrowband words were even less intelligible than LP sentences (a further 23% drop). While experiment 1 revealed similar levels of performance for narrowband and broadband sentences at conversational speech levels, experiment 2 showed that speech reception thresholds were substantially (>30 dB) poorer for narrowband sentences. One explanation for this increased disparity between narrowband and broadband speech at threshold (compared to conversational speech levels) is that spectral components in the sloping transition bands of the filters provide important cues for the recognition of narrowband speech, but these components become inaudible as the signal level is reduced. Experiment 2 also showed that performance was degraded by the introduction of a speech masker (a single competing talker). The elevation in threshold was similar for narrowband and broadband speech (11 dB, on average), but because the narrowband sentences required considerably higher sound levels to reach their thresholds in quiet compared to broadband sentences, their target-to-masker ratios were very different (+23 dB for narrowband sentences and -12 dB for broadband sentences). As in experiment 1, performance was better for HP than LP sentences. The LP-HP difference was larger for narrowband than broadband sentences, suggesting that context provides greater benefits when speech is distorted by narrow bandpass filtering.

Inhibitory processes and spoken word recognition in young and older adults: the interaction of lexical competition and semantic context.

Two experiments were conducted to examine the importance of inhibitory abilities and semantic context to spoken word recognition in older and young adults. In Experiment 1, identification scores were obtained in 3 contexts: single words, low-predictability sentences, and high-predictability sentences. Additionally, identification performance was examined as a function of neighborhood density (number of items phonetically similar to a target word). Older adults had greater difficulty than young adults recognizing words with many neighbors (hard words). However, older adults also exhibited greater benefits as a result of adding contextual information. Individual differences in inhibitory abilities contributed significantly to recognition performance for lexically hard words but not for lexically easy words. The roles of inhibitory abilities and linguistic knowledge in explaining age-related impairments in spoken word recognition are discussed.

Intelligibility of bandpass-filtered speech

Warren et al. [Percept. Psychophys. 57, 175–182 (1995)] reported high intelligibility for sentences processed by one-third octave bandpass filters. The present study examined the contribution of acoustic and linguistic redundancy to the intelligibility of filtered speech. Separate groups of listeners identified sentences or isolated final words from the SPIN (‘‘speech perception in noise’’) test. One group heard unfiltered speech, while others heard filtered speech with center frequencies of 750, 1100, 1500, 2100, or 3000 Hz. Gammatone filters with bandwidths of 1 ERB were used to match auditory filter widths. Consistent with Warren et al., the 1500-, 2100-, and 3000-Hz bands were more intelligible than the lower bands. Filtered words in high-predictability (HP) sentences were identified more accurately than those in low-predictability (LP) sentences (mean difference 20%). Isolated filtered words were less intelligible than LP sentences (23% difference). Unfiltered HP and LP sentences and final words were equally intelligible, with scores exceeding 94%. Filtered HP sentences were well identified for the 2100- and 3000-Hz bands, with performance similar to unfiltered sentences. These findings support the hypothesis that context is emphasized when the acoustic signal is degraded, and illustrate the interactive nature of multilevel processing.

Temporal Factors and Speech Recognition Performance in Young and Elderly Listeners

This study investigated factors that contribute to deficits of elderly listeners in recognizing speech that is degraded by temporal waveform distortion. Young and elderly listeners with normal hearing sensitivity and with mild-to-moderate, sloping sensorineural hearing losses were evaluated. Low-predictability (LP) sentences from the Revised Speech Perception in Noise test (R-SPIN) (Bilger, Nuetzel, Rabinowitz, & Rzeczkowski, 1984) were presented to subjects in undistorted form and in three forms of distortion: time compression, reverberation, and interruption. Percent-correct recognition scores indicated that age and hearing impairment contributed independently to deficits in recognizing all forms of temporally distorted speech. In addition, subjects' auditory temporal processing abilities were assessed on duration discrimination and gap detection tasks. Canonical correlation procedures showed that some of the suprathreshold temporal processing measures, especially gap duration discrimination, contributed to the ability to recognize reverberant speech. The overall conclusion is that age-related factors other than peripheral hearing loss contribute to diminished speech recognition performance of elderly listeners.

Speech recognition measures with noise suppression hearing aids using a single-subject experimental design.

Research aimed at quantifying the benefits a hearing aid user might expect from noise suppression hearing aids purported to improve hearing in the presence of background noise have yielded widely varying results. We suggest this may in part be due to the inappropriate use of experimental approaches based on group design and inferential statistical analysis. Included in this paper is our rationale for employing a single-subject experimental design to investigate subject performance with two commercially available hearing-aid noise suppression systems. Preliminary results with two subjects indicate that both the Siemens Automatic Signal Processing (ASP) and Zeta Noise Blocker (ZNB) noise suppression systems markedly improve listener scores on the low predictability sentence material of the SPIN Test. We conclude these two noise suppression systems may improve performance as the listening situation becomes contextually more difficult, and that single-subject experimental designs could be a valuable addition to applied behavioral research with hearing aids.

Mathematical treatment of context effects in phoneme and word recognition.

Percent recognition of phonemes and whole syllables, measured in both consonant-vowel-consonant (CVC) words and CVC nonsense syllables, is reported for normal young adults listening at four signal-to-noise (S/N) ratios. Similar data are reported for the recognition of words and whole sentences in three types of sentence: high predictability (HP) sentences, with both semantic and syntactic constraints; low predictability (LP) sentences, with primarily syntactic constraints; and zero predictability (ZP) sentences, with neither semantic nor syntactic constraints. The probability of recognition of speech units in context (pc) is shown to be related to the probability of recognition without context (pi) by the equation pc = 1 - (1-pi)k, where k is a constant. The factor k is interpreted as the amount by which the channels of statistically independent information are effectively multiplied when contextual constraints are added. Empirical values of k are approximately 1.3 and 2.7 for word and sentence context, respectively. In a second analysis, the probability of recognition of wholes (pw) is shown to be related to the probability of recognition of the constituent parts (pp) by the equation pw = pjp, where j represents the effective number of statistically independent parts within a whole. The empirically determined mean values of j for nonsense materials are not significantly different from the number of parts in a whole, as predicted by the underlying theory. In CVC words, the value of j is constant at approximately 2.5. In the four-word HP sentences, it falls from approximately 2.5 to approximately 1.6 as the inherent recognition probability for words falls from 100% to 0%, demonstrating an increasing tendency to perceive HP sentences either as wholes, or not at all, as S/N ratio deteriorates.

Comprehension of natural and synthetic speech: effects of predictability on the verification of sentences controlled for intelligibility

A sentence verification task (SVT) was used to study the effects of sentence predictability on comprehension of natural speech and synthetic speech that was controlled for intelligibility. Sentences generated using synthetic speech were matched on intelligibility with natural speech using results obtained from a separate sentence transcription task. In the main experiment, the sentence verification task included both true and false sentences that varied in predictability. Results showed differences in verification speed between natural and synthetic sentences, despite the fact that these materials were equated for intelligibility. This finding suggests that the differences in perception and comprehension between natural and synthetic speech go beyond segmental intelligibility as measured by transcription accuracy. The observed differences in response times appear to be related to the cognitive processes involved in understanding and verifying the truth value of short sentences. Reliable effects of predictability on error rates and response latencies were also observed. High-predictability sentences displayed lower error rates and faster response times than low-predictability sentences. However, predictability did not have differential effects on the processing of synthetic speech as expected. The results demonstrate the need to develop new measures of sentence comprehension that can be used to study speech communication at processing levels above and beyond those indexed through transcription tasks, or forced-choice intelligibility tests such as the Modified Rhyme Test (MRT) or the Diagnostic Rhyme Test (DRT).

Speech perception in noise by fluent, non-native listeners

This experiment measured the ability of fluent, non-native listeners to understand American English in the presence of babble noise. The Speech Perception in Noise (SPIN) test was administered at 75 dB SPL to 11 native and 14 non-native young, normal listeners. The non-native listeners were university students and teachers who had lived more than four years in the United States. Six noise levels were chosen in 2-dB steps to encompass the range from 20%–80% correct performance. At each noise level, 50 new sentences were presented and the percent correct for 25 low- and 25 high-predictability sentences was measured. After each sentence the listener wrote the last word in the sentence and repeated it to the experimenter, who also wrote it. Results show (1) the native listeners could obtain 50% correct performance at significantly higher noise levels (about 3 dB) than the non-native listeners, (2) both groups performed significantly better on the high-predictability sentences than on the low-predictability sentences, and (3) the difference in tolerable noise levels (levels yielding 50% correct) between the high- and low-predictability sentences was the same for the native and non-native groups. [Work supported by the Research and Scholarship Development Fund of Northeastern University.]

Quantifying syntactic and semantic redundancy effects in the recognition of words in sentences

The goal was to measure the contribution of semantic and syntactic constraints to the probability of word recognition in sentences. Normally hearing adult Ss listened to sets of four‐word sentences at four S/N ratios. There were three types of sentences: (1) zero predictability, i.e., nonsense word strings; (2) low predictability, i.e., syntactically correct but semantically anomolous sentences; (3) high predictability, i.e., syntactically and semantically correct sentences. Word recognition probabilities were measured for each condition. The contribution of syntax was computed as the ratio of the logarithms of error probability in the low and zero predictability sentences. The contribution of semantics was computed as the ratio of the logarithms of error probability in high and low predictability sentences. The resulting parameter k is the amount by which the number of independent channels of acoustical information would need to be increased in order to produce the measured increase of recognition probability. Average values of k were 1.4 and 2.0 for syntax and semantics, respectively. Statistical analysis revealed that k was independent of S/N ratio. [Work supported by PSC. CUNY award 6‐63137.]