Kenneth W. Berger

Extra-facial gestures in relation to speechreading

Abstract In a pilot study ( N = 32) it was found that extra-facial gestures, by themselves, elicited responses ranging from identical to similar but non-identical. It was hypothesized that the use of extra-facial gestures would influence the proficiency of speechreading. Twenty sentences based upon corresponding concepts of gestures used in the pilot study were presented to two groups of 16 subjects each. Each of the subject groups received alternate test items without gestures and with gestures. To minimize possible facial gestures a face-mask was worn by the speaker which allowed only his lips to be seen. Results indicated that for the 20 gestures employed in this study, there was a statistically significant increase in the proficiency of speechreading.

The most common words used in conversations

Abstract Unguarded informal conversational vocabulary from a general adult population was sampled in the northeastern Ohio area. The sample produced 25000 words of which there were 2507 different words. A limited vocabulary usage and simple words as reflected in words of small-syllable length were found for conversation as compared with more formal speech and with printed English. The words found in the present study are presented in an appendix. The Appendix gives all the words found, in alphabetical order, and includes variants of the base word where syllable length does not change. The usefulness and application of oral vocabulary as opposed to written vocabulary, and of conversational vocabulary as opposed to more formal speech vocabulary are briefly discussed. Further samplings of conversational speech, in spite of the difficulty as contrasted to printed materials, are recommended, particularly to determine consistency and variability based on geographical areas.

Two-dimension vs. three-dimension viewing in speechreading

Abstract The relatively low correlations between speechreading proficiency as measured by filmed tests and teachers judgements may be due to differences in the depth of image. Two vs. three-dimensional factors were examined by having twenty subjects alternatively speechread the profile of a speaker as seen by silhouette and as seen normally. The normal-profile view produced significantly higher speechreading scores than the silhouette-profile view. Although the silhouette view is an unnatural on it is a pure two-dimension and it is speculated that the visual clues available to the speechreader are greater in a live situation than by film or television, and greater in a filmed or televised view than in a silhouette view.

Relationships Between Various Stimuli for MCL

Five test stimuli (pulsed 1000Hz pure tone, continuous 1000Hz pure-tone, white noise, speech, and music) were employed to determine the reliability of the MCL (Most Comfortable Level) measurement and interrelations among the various test stimuli. The reliability of the MCL, as reflected in correlations, was found to be high on test-retest for each of the five stimuli employed. The MCL variability for each test stimuli among subjects was substantially greater than that of threshold determination for the same stimulus. MCL differences between test stimuli were large. In as much as MCL was found to be more variable than threshold testing on test-retest, between several stimuli, and for the same stimulus over a period of several minutes, it is recommended that Most Comfortable Range (MCR) be used in place of MCL when referring to this determination.

ELECTROMYOGRAPHIC RECORDING DURING WIND INSTRUMENT PERFORMANCE

The description and measurement of the role of the respiratory mechanism in phonation has produced a long and often imaginative history of research. It may seem surprising that respiration factors in wind instrument tone production, which appear to have much in common with vocalization, have on the contrary produced less interest and research, other than considerable descriptive empirical material. The latter, unfortunately, often becloud the issues rather than clarify them. R. H. Stetson (1951) is the first known researcher to systematically investigate the coordination of specific muscle areas during the production of vocalizations. Stetson’s work began in the late 19203, and although uncritically quoted on occasion, remained relatively dormant until a flurry of electromyographic research began appearing in the 1950’s and 1960’s. The awakened interest in the approaches suggested by Stetson produced a number of non-speech EMG studies and were used with speech applications in studies such as those by Campbell (19581, Ladefoged and associates (1958), Draper and associates (19591, Hoshiko (1957), Hoshiko and Berger (1965), Eblen (1961 ), and others. The present study was an outgrowth of a larger electromyographic investigation of speech breathing. The purpose of the present study was to investigate the pattern and sequence of activity of the internal intercostals, the external intercostals, the rectus abdominis, and the external abdominal oblique muscles as inferred from action-potential patterns recorded by surface electromyography. These patterns were obtained during the production of single and multiple articulations and of sustained tones on the trumpet under several tonal frequency conditions.

Gain Requirements of Conductive Hearing Losses

Individuals with significant air-bone gap were presumed to need the gain used by sensori-neural hearing losses, plus one-fifth of the air-bone gap. The amount of operating gain used by 76 conductive and mixed hearing losses was found to closely approximate the predicted amount, but there were large individual differences. As was expected, a gradual increase in gain was used as the size of the air-bone gap increased. There appeared to be minimal relationship between operating gain differences and the air conduction pure-tone threshold pattern, but some relationship to the underlying conductive pathology.

The Effect of Noise on Lip-Reading Performance

Normal hearing young adults received a lip-reading test in quiet (55dB SPL), and in continuous and in intermittent white noise backgrounds of 90dB SPL. Scores decreased from the quiet to the continuous noise to the intermittent noise background, but differences were not statistically significant. It is hypothesised that the predictable nature of the noises allowed the subjects to adapt to them.

In search of a language intervention lexicon

An analysis of the syntactic, semantic and phonological features of a corpus of therapy words drawn form recent publications dealing with remedial language instruction was performed to determine whether these words would in theory support full grammatical acquisition. Percentage of occurrence data were compared with distributional statiscs for adult (Berger, 1967a) and child (Weir, 1962) usage. The results suggested a bias in favor of concrete words referring to everyday objects around the home, nouns, verbs and adjectives, and morphemes of simple syllable and sound structure. In general the words analyzed appeared more suitable for children under 3 years of age than for older children. Of factors guiding the clinicians selection of therapy words, native speaker intuition for distributional characteristics of the language was regarded as the major factor of judgment.

Three experiments in speechreading

Abstract Three speechreading factors were examined: whether the speechreader can identify the /h/ before the vowel in monosyllabic words, whether the speechreader can identify the initial or final consonant voicing in mono-syllabic words not otherwise different, and whether syllable stress can be identified by speechreading. Subjects were undergraduate students in a speechreading course. The identification of the /h/ before a vowel and the identification of voicing were not made on better than a chance basis. Subjects did not identify dissyllabic words spoken with correct stress better than with incorrect stress, but the identification of syllable stress with spoken numbers was better than chance.