Cognitive Science Program

Sentence perception in noise by hearing-aid users: Relations with syllable-constituent perception and the use of context (2019)
James D. Miller, Charles Watson, Marjorie R. Leek, David Wark
The Journal of the Acoustical Society of America, 145 (3), 1821

Abstract

Masked sentence perception by hearing-aid users is strongly correlated with three variables: (1) the ability to hear phonetic details as estimated by the identification of syllable constituents in quiet or in noise; (2) the ability to use situational context that is extrinsic to the speech signal; and (3) the ability to use inherent context provided by the speech signal itself. These conclusions are supported by the performance of 57 hearing-aid users in the identification of 109 syllable constituents presented in a background of 12-talker babble and the identification of words in naturally spoken sentences presented in the same babble. A mathematical model is offered that allows calculation of an individual listener's sentence scores from estimates of context utilization and the ability to identify syllable constituents. When the identification accuracy of syllable constituents is greater than about 55%, individual differences in context utilization play a minor role in determining the sentence scores. As syllable constituent scores fall below 55%, individual differences in context utilization play an increasingly greater role in determining sentence scores. When a listener's syllable constituent score is above about 71% in quiet, the listeners score in quiet will above about 55% in noise. [Watson and Miller are shareholders in Communication Disorders Technology, Inc., and may profit from sales of the software used in this study.]

Efficacy of two speech-perception training methods for aided listeners with mild-to-moderately-severe age-related hearing losses (2019)
James D. Miller, Charles Watson, Gary R. Kidd
The Journal of the Acoustical Society of America, 145 (3), 1795-1796

Abstract

Preliminary analyses of interrelations among variables that correlate with measures of speech perception by aided listeners with mild-to-moderately-severe age-related hearing loses are described. The SPATS Group of nearly 120 hearing-aid users was trained in quiet and noise to identify syllable-constituents and to identify words in simple sentences. The Listen Group of about 60 aided listeners listened for an equal amount of time to recorded narratives. All were given audiometric, working memory, intelligence tests, a battery of psycho-acoustic tests, and a battery of speech-perception tests prior to training. The audiometric and speech perception tests were repeated after training and again after a 3-month retention period. Prior to training, there were large individual differences speech perception scores that could only be partially accounted for by the severity of audiometric loss. Training appears to be more effective for the SPATS Group than for the Listen Group. While improvements with training were generally modest, those who were very good or very poor initially tended to show little or no improvement, while those with middling scores tended to show more improvement. Sentence scores are highly correlated with syllable-constituent scores and with the use of context, and the use of context is correlated with working memory. (Miller and Watson are stockholders in communication Disorders Technology, Inc., and may profit from sales of some the software used in this study.)

Syllable-constituent perception by hearing-aid users: Common factors in quiet and noise (2017)
James D. Miller, Charles Watson, Marjorie R. Leek, Judy R. Dubno, David Wark, Pamela E. Souza, Sandra Gordon-Salant, Jayne B. Ahlstrom
The Journal of the Acoustical Society of America, 141 (4), 2933-2946

Abstract

The abilities of 59 adult hearing-aid users to hear phonetic details were assessed by measuring their abilities to identify syllable constituents in quiet and in differing levels of noise (12-talker babble) while wearing their aids. The set of sounds consisted of 109 frequently occurring syllable constituents (45 onsets, 28 nuclei, and 36 codas) spoken in varied phonetic contexts by eight talkers. In nominal quiet, a speech-to-noise ratio (SNR) of 40 dB, scores of individual listeners ranged from about 23% to 85% correct. Averaged over the range of SNRs commonly encountered in noisy situations, scores of individual listeners ranged from about 10% to 71% correct. The scores in quiet and in noise were very strongly correlated, R = 0.96. This high correlation implies that common factors play primary roles in the perception of phonetic details in quiet and in noise. Otherwise said, hearing-aid users' problems perceiving phonetic details in noise appear to be tied to their problems perceiving phonetic details in quiet and vice versa.

Evaluation of Speech-Perception Training for Hearing Aid Users: A Multisite Study in Progress (2015)
James D. Miller, Charles Watson, Judy R. Dubno, Marjorie R. Leek
Seminars in Hearing, 36 (4), 273-283

Abstract

Following an overview of theoretical issues in speech-perception training and of previous efforts to enhance hearing aid use through training, a multisite study, designed to evaluate the efficacy of two types of computerized speech-perception training for adults who use hearing aids, is described. One training method focuses on the identification of 109 syllable constituents (45 onsets, 28 nuclei, and 36 codas) in quiet and in noise, and on the perception of words in sentences presented in various levels of noise. In a second type of training, participants listen to 6- to 7-minute narratives in noise and are asked several questions about each narrative. Two groups of listeners are trained, each using one of these types of training, performed in a laboratory setting. The training for both groups is preceded and followed by a series of speech-perception tests. Subjects listen in a sound field while wearing their hearing aids at their usual settings. The training continues over 15 to 20 visits, with subjects completing at least 30 hours of focused training with one of the two methods. The two types of training are described in detail, together with a summary of other perceptual and cognitive measures obtained from all participants.

Characteristics of 40,000 calls to the National Hearing Test (2015)
Charles Watson, Gary R. Kidd, Jill Preminger, James D. Miller
The Journal of the Acoustical Society of America, 138 (3), 1830

Abstract

The National Hearing Test (NHT) is a telephone-administered screen for hearing loss (Watson et al., J. Am. Acad. Audiol., 2012) that obtains thresholds for three-digit sequences in a noise background. The NHT has been validated by comparing threshold SNR values to mean pure-tone loss. During five weeks in 2014 the NHT was offered without charge to the general public. Over 40,000 calls were made after articles describing the test appeared in eight large-circulation and 19 smaller circulation newspapers, estimated to reach less than 10% of the US public. Among those who completed the test, 81% were estimated to have clinically significant loss in one or both ears. Call numbers suggested that 88% were made from landlines and the remainder from cell phones. Threshold SNRs for cell phones were about 1.0 dB higher than those for landline phones. Samples of the callers were later contacted through telephone and email surveys. Responses indicated a positive influence in terms of the likelihood of seeking further evaluation and obtaining hearing aids if advised to do so. [Research supported by the National Institute for Deafness and other Communication Disorders of the National Institutes of Health under award number 5R44DC009719.]

The U.S. National Hearing Test, a 2013–2014 progress report (2014)
Charles Watson, Gary R. Kidd, James D. Miller, Jill Preminger, Alex Crowley, Daniel P. Maki
The Journal of the Acoustical Society of America, 136 (4), 2308

Abstract

A telephone-administered screening test for sensorineural hearing loss was made publically available in the United States in September 2013. This test is similar to the digits-in-noise test developed by Smits and colleagues in the Netherlands, versions of which are now in use in most European countries and in Australia. The test was initially offered in the United States for a small fee ($8, then $4) but after a year of promotion it became clear that either the fee or the complexity of paying it was inhibiting. During the first month in which the test was subsequently offered free of charge, 31,806 calls were made to the test line, of which 26,507 were completed tests. Analyses of test performance suggest that about 81% of the test takers had at least a mild hearing loss, and 40% had a substantial loss (estimated to be in excess of 45 dB PTA). Follow-up studies are being conducted to determine whether those who failed the test sought a full-scale hearing assessment, and whether those advised to obtain hearing aids did so. [Work funded by Grant No. 5R44DC009719 from the National Institute for Deafness and other Communication Disorders.]

Distributions of confusions for the 109 syllable constituents that make up the majority of spoken English (2014)
James D. Miller, Charles Watson, Roy Sillings
The Journal of the Acoustical Society of America, 136 (4), 2212

Abstract

Among the interests of Kewley-Port have been the perception and production of English Speech Sounds by native speakers of other languages. ESL students from four language backgrounds (Arabic, Chinese, Korean, and Spanish) were enrolled in a speech perception training program. Similarities and differences between these L1 groups in their primary confusions were determined for onsets, nuclei and codas utilized in spoken English. An analysis in terms of syllable constituents is more meaningful than analyses in terms of phonemes as individual phonemes have differing articulatory and acoustic structures depending on their roles in the syllable and their phonetic environments. An important observation is that only a few of all the possible confusions that might occur do occur. Another interesting characteristic of confusions among syllable constituents is that many more confusions are observed than those popularly cited, e.g., the /r/ v /l/ for Japanese speakers. As noted by many, the perceptual problems encountered by learners of English are conditioned on the relations between the sound-structures of English with each talker’s L1. These data suggest that the intrinsic similarities within of the sounds of English also play an important role.

An approach to the analysis of relations between syllable and sentence perception in quiet and noise in the Speech Perception Assessment and Training System: Preliminary results for ten hearing-aid users (2014)
James D. Miller
The Journal of the Acoustical Society of America, 136 (4), 2311

Abstract

Logistic functions relating abilities to indentify syllable onsets, nuclei, and codas in quiet and noise as a function of snr are measured. Syllable perception is the product of these individual abilities. It is found that syllable perception in noise is highly correlated with syllable perception quiet. The relation of sentence perception in the SPATS sentence task with SPATS syllable constituent perception is examined. As shown years ago at the Bell Labs, only modest levels of syllable identification are needed to support nearly perfect levels of sentence perception. Here, it is found that sentence perception in quiet and noise is correlated with syllable perception in quiet, the use of inherent context provided by syllable perception (Boothroyd and Nittrauer (1988)), and with the use of situational context, independent of syllable perception. Finally, the effects of speech perception training on these relations are examined for each of the ten hearing-aid users studied. [Work supported by NIH/NICD Grant R21/R33DC011174 “Multi-site Study of the Efficacy of Speech Perception Training for Hearing-Aid Users,” C. S. Watson, PI. Data supplied by cooperating sites: Medical University of South Carolina, J. Dubno, Site PI; University of Memphis, D. Wark, Site PI; and University of Maryland, S. Gordon-Salant, Site PI.]

Computer-based perceptual training as a major component of adult instruction in a foreign language (2013)
Charles Watson, James D. Miller
IGI Global.

Abstract

Millions of adult learners have acquired good-to-excellent literacy in English, but most of them continue to have difficulty with oral communication in that language. The more obvious their problem is with pronunciation, which varies from just noticeably “foreign” to very difficult to understand, the less apparent, but possibly fundamental to their overall skill level in English, is their difficulty in recognizing spoken sounds, words, and phrases of that language. Contemporary research has shown that adults are capable of learning to perceive a new language quite accurately, through systematic training. Perceptual skills acquired through such training are likely to contribute to continuing improvement in pronunciation and to an increasing vocabulary. Such a training program is described in this chapter, the Speech Perception Assessment and Training Program for ESL (SPATS-ESL). After 20-30 hours of training with this program, most ESL students are shown to achieve near-native recognition of the sounds of English and of words in sentences spoken at normal conversational rates.

Telephone Screening Tests for Functionally Impaired Hearing: Current Use in Seven Countries and Development of a US Version (2012)
Charles Watson, Gary R. Kidd, James D. Miller, Cas Smits, Larry E. Humes
Journal of the American Academy of Audiology, 23 (10), 757-767

Abstract

An estimated 36 million US citizens have impaired hearing, but nearly half of them have never had a hearing test. As noted by a recent National Institutes of Health/National Institute on Deafness and Other Communication Disorders (NIH/NIDCD) Working Group, "In the United States (in contrast to many other nations) there are no readily accessible low cost hearing screening programs…" (Donahue et al, 2010, p. 2). Since 2004, telephone administered screening tests utilizing three-digit sequences presented in noise have been developed, validated, and implemented in seven countries. Each of these tests has been based on a test protocol conceived by Smits and colleagues in The Netherlands. Purpose: Investigators from Communication Disorders Technology, Inc., Indiana University, and VU University Medical Center of Amsterdam agreed to collaborate in the development and validation of a screening test for hearing impairment suitable for delivery over the telephone, for use in the United States. This test, utilizing spoken three-digit sequences (triplets), was to be based on the design of Smits and his colleagues. Research Design: A version of the digits-in-noise test was developed utilizing digit triplets spoken in Middle American dialect. The stimuli were individually adjusted to speech-to-noise ratio (SNR) values yielding 50% correct identification, on the basis of data collected from a group of 10 young adult listeners with normal hearing. A final set of 64 homogeneous stimuli were selected from an original 160 recorded triplets. Each test consisted of a series of 40 triplets drawn at random, presented in a noise background. The SNR threshold for 50% correct identification of the triplets was determined by a one-down, one-up adaptive procedure. The test was implemented by telephone, and administered to listeners with varying levels of hearing impairment. The listeners were then evaluated with pure-tone tests and other audiometric measures as clinically appropriate. Study Sample: Ninety participants included 72 who were volunteers from the regular client population at the Indiana University Hearing Clinic, and 18 who were recruited with a newspaper ad offering a free hearing test. Of the 90 participants, 49 were later determined to have mean pure-tone thresholds greater than 20 dB hearing level (HL). Data Collection and Analysis: The primary data analyses were correlations between telephone test thresholds and other measures, including pure-tone thresholds and speech recognition tests, collected for the same participants. Results: The correlation between the telephone test and pure-tone thresholds (r = 0.74) was within the range of correlations observed with successful telephone screening tests in use in other countries. Thresholds based on the average of only 21 trials (trials five through 25 of the 40-trial tracking history) yielded sensitivity and specificity values of 0.80 and 0.83, respectively, using pure-tone average((0.5, 1.0, 2.0 kHz)) >20 dB HL as the criterion measure. Conclusions: This US version of the digits-in-noise telephone screening test is sufficiently valid to be implemented for use by the general public. Its properties are quite similar to those telephone screening tests currently in use in most European countries. Telephone tests provide efficient, easy to use, and valid screening for functional hearing impairment. The results of this test are a reasonable basis for advising those who fail to seek a comprehensive hearing evaluation by an audiologist.