LIPREADING PERFORMANCE AS AFFECTED BY. CONTENUOUS AUDITORY DISTRACTIONS Dissertatéon fer the Degree of Ph. D. MICHIQAN STATE UNWERSiTY DONALD E. 0113M 1973 This is to certify that the thesis entitled Lipreading Performance as Affected by Continuous Auditory Distractions presented by Donald R. Ciliax has been accepted towards fulfillment of the requirements for Ph. D. degreein Audiology WZOW t/ / / ajor profefi Date 3" 267~7§ 0-7639 1 2;; 7 \ ' ‘ t ‘ :- amotuG av ~=‘ ‘ ll mm; a sour _ 880K amnm me I LIBRARY amass; ll 1 ABSTRACT LIPREADING PERFORMANCE AS AFFECTED BY CONTINUOUS AUDITORY DISTRACTIONS BY Donald R. Ciliax The major objectives of this investigation were (1) to ascer- tain the effects of different environments (quiet, babble, industrial noise, traffic noise, and music) on individuals being trained in the lipreading process and (2) to determine whether or not certain acous- tical conditions (quiet, babble, industrial noise, traffic noise, and music) function as auditory distractors and, thereby, decrease the lip- reading efficiency of trained lipreaders in a test situation. The experimental procedure consisted of a training program and a test session. Forty normal hearing, normal seeing, college-age adults were randomly assigned to one of five training conditions, each condition representing a different auditory environment in which to learn to lipread. All groups contained eight subjects and were divided evenly between men and women. The five auditory environmental condi- tions (with sound pressure presentation levels in dB) included (1) quiet (ambient noise, 50 dB SPL), (2) babble (90 dB SPL), (3) in- dustrial noise (90 dB SPL), (4) traffic noise (90 dB SPL), and (5) music (90 dB SPL). Each subject was shown, by way of television Donald R. Ciliax monitor, one of six videotapes in which a male Speaker was presenting a set of spondaic words. The task of the lipreader in the training session was to continue to watch the same videotape (given no more than eight times over a two day period) until correctly identifying 90 per cent or more of the vocabulary stimuli. Upon satisfying this training criterion, the subject was considered qualified to partici- pate in the final phase of the experiment. In the test procedure (given within forty-eight hours after training was completed) the same set of vocabulary words was presented five times, on each occasion in a different environmental situation, including a replication of the environment under which the subject had been trained. Data from both the training and test procedures were analyzed statistically. The results of the data suggested the following conclusions: 1. Female subjects as a group lipread significantly better than male subjects in the test session. 2. Female participants needed significantly fewer training trials to qualify for the test session than did male participants. 3. The scores of male and female subjects trained in various sound environments varied as a function of the noise background employed in the test session. No definite pattern of subject reaponse emerged relative to the noise backgrounds. 4. The results from the test session demonstrate that sub- jects trained in various sound backgrounds will achieve comparable scores, irreSpective of the environment in which they were trained. LIPREADING PERFORMANCE AS AFFECTED BY CONTINUOUS AUDITORY DISTRACTIONS BY ’ l b 1 Donald RI Ciliax A DISSERTATION Submitted to . Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Audiology and Speech Sciences 1973 Accepted by the faculty of the Department of Audiology and Speech Sciences, College of Communication Arts, Michigan State University, in partial fulfillment of the requirements for the degree of Doctor of Philosophy. Thesis Committee: « Leo V. Deal, Ph.D. » u Oscar I. Tosi, Ph.D. [2- 277. gar/NA Abram M. Barch, Ph.D. ii ACKNOWLEDGMENTS The writer wishes to express his deepest appreciation to Dean Herbert J. Oyer, whose guidance, encouragement and thoughtfulness made the completion of this investigation possible. The investigator is also indebted to Dr. Lee Deal, Dr. Oscar Tosi, and Dr. Abram Barch for their constructive criticism and helpful suggestions in completing this study. Special acknowledgment is also extended to Mr. Donald E. Riggs for his technical assistance and timely suggestions regarding the experimental procedure. Finally, appreciation is extended to the United States Army whose support made this research possible. iii TABLE OF CONTENTS L/I LISTOFTABLES. . . . . . . . } LIST OF FIGURES 4 Chapter or 1. STATEMENT OF THE PROBLEM (1 Introduction . . . . . . Purpose of the Study . . . . Null Hypotheses . . . . . Importance of the Study Terminology . . . . . . . Organization of the Dissertation m. II. REVIEW OF THE LITERATURE . . . . J 5/ Importance of Visual Cues to Communication Some Effects of Environment on Lipreading The Effects of Noise on Man . . Lipreading Studies in Noise . . Summary . . . . . . . . . III. EXPERIMENTAL DESIGN . . . . . . Subjects . . . . . . . . Materials’ . . . . . . . . Training and Test Environment and Procedures . . . . . . . . IV. RESULTS AND DISCUSSION . . . . . Introduction . . . . . . . Results . . . . . . . . Discussion . . . . . . '. . iv Equipment Page vi vii OMAMMH H \l 11 14 20 24 26 26 27 29 31 35 35 35 51 Chapter 1 Page V. SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS . . . . . . 64 Summary . . . . . . . . . . . . . . . . 64 Conclusions . . . . . . . . . . . . . . . 66 Recommendations . . . . . . . . . . . . . 66 BIBLIOSRAPHY . . . . . . . . . . . . . . . . . . 68 APPENDICES Appendix A. Videptape Word Orders Numbers One Through Six . . . . 71 f B. Individual Stimulus Item Time Sequences for , Videotapes Numbers One Through Six . . . . . . . 77 R C. Forms: Alphabetized Experiment Vocabulary, Experiment Instructions, Training Answer Sheet and Test Answer Sheet . . . . . . . . . . . . . . 83 :1; D. Presentation Order of the Background Noise Conditions and Videotape Vocabulary Sequences Used for Each Group of Subjects Employed in the Training and Test Sessions . . . . . . . . . . . . . 87 LIST OF TABLES I Table A. Page 1. Summary of the Two-Way, Fixed-Effects (Training by Sex) Analysis of Variance for the Number of Training Trials Needed to Qualify for the Test Session . . . . . . . . . . . . . . . 36 J 2. Mean Number of Trials, Tabulated by Sex and Training Condition, Needed to Qualify for the Test Session . . 37 3. Number of Training Trials Needed by Each Subject to Qualify for the Test Session . . . . . . . . 39 4. Summary of the Three-Way, Fixed-Effects Analysis of Variance with Repeated Observations on the Test Measures . . . . . . . . . . . . . . 41 5. Mean Scores in Per Cent as a Function of Training Condition, Sex and Test Measure . . . . . . . . 42 6. Subject Lipreading Scores in Per Cent Across All Test Measures, Grouped by Training Condition and Sex . . . . . . . . . . . . . . . . 43 7. Analysis of Variance for Simple Effects of the Training by Sex by Measure Interaction . . . . . . 46 8. Confidence Intervals Around Differences Between Test Measure Means Using Tukey . . . . . . . . . 48 9. Confidence Intervals Around Differences in the Test Measure Means of Quiet Versus the Combined Noise Stimuli Using Scheffé . . . . . . . . . . . . 52 vi LIST OF FIGURES Figure Page 1. Physical Arrangement of the Equipment and Chair-desks in the Test (Training) Environment . . . . . . . 30 2. Mean Number of Trials Across Training Conditions as a Function of Sex . . . . . . . . . . . . . 38 vii CHAPTER I STATEMENT OF THE PROBLEM Introduction The act of communicating vocally, in which a talker and a listener are alternately involved, is a process which normally utilizes both the auditory and visual sense modalities of each participant. Under advantageous circumstances, i.e., in a quiet, relatively noise- free environment, expectations for a complete and adequate exchange of information are high. However, the intrusion of noise into the listen- ing environment increases the likelihood that the conversation between individuals will be hampered. The focus of most research on noise has been the potential threat of noise producing permanent hearing loss. The extent and type of auditory impairment induced by acoustic insult have been well docu- mented (Kryter, 1970) and will not be discussed further. Instead, what should be understood is the influence of noise as a masker of Speech intelligibility. The majority of published reports evaluating noise as a masker of speech has found noise to be a demonstrably disruptive agent which produces confusion and misunderstanding between the speaker and receiver and induces errors. The masking of auditory functions by noise might cause the listener to use speechreading in order to main- tain satisfactory communication. O'Neill (1954) and others have sought to determine the rela- tionship which vision and audition share in verbal discourse, partic— ularly in environments with unfavorable signal-to-noise (S/N) ratios. When messages are auditorily hampered, either partially or completely, the necessity for implementing lipreading skills is often required if the desired information is to be exchanged. Extensive investigations have previously demonstrated the effects of noise on man with other than auditory tasks, e.g., those of vigilance which rely heavily on the worker's vision (Broadbent, 1958). In addition to assessing the influence of audition on job performance, effects of noise on human attitudes, feelings (Cohen, 1969) and body physiology (Jansen, 1969) have also been ascertained. When lipreading is resorted to under disadvantageous listening conditions, the effect of the environment on this process must be well understood. Berger (1972) has described vision as being subject to "more environmental restrictions than audition from a communications standpoint." The factors of lighting, distance and physical distrac- tions (visual and auditory) are among those which must be considered. Of the parameters often identified as variables in the lipreading process (speaker-sender, lipreader-receiver, code or stimulus, and environment), the factor of environment is the least understood and researched (O'Neill and Oyer, 1961). In particular, the effects of auditory distractions on the lipreading process need to be clarified. Purpose of the Study» The primary objective of this investigation is to establish whether or not various noise stimuli (babble, industrial noise, traffic noise, and music) function as auditory distractors and decrease lip- reading efficiency. If lipreading performance is fbund to be affected, is one type of noise stimulus more disruptive than another? Of equal interest to this study is to determine the effects that certain background noise stimuli (babble, industrial noise, traffic noise, and music) have on the training or teaching of the lip- reading process. Do subjects trained to lipread in noise become more efficient lipreaders or learn to lipread faster than subjects trained in quiet? Null Hypotheses In order to answer statistically the aforementioned queries, the following null hypotheses were formulated: 1. For each subject trained to read lips in the five training environments (quiet, babble, industrial noise, traffic noise, and music), there are no differences when comparing their lipreading scores for the test measures of: (a) quiet and babble (b) quiet and industrial noise (c) quiet and traffic noise (d) quiet and music (e) babble and industrial noise (f) babble and traffic noise (g) babble and music (h) industrial noise and traffic noise (i) industrial noise and music (j) traffic noise and music 2. There are no differences in the lipreading performance of subjects with regard to the various environmental conditions under which they were trained to read lips. 3. There are no differences between the lipreading performance of male and female subjects trained under various environmental condi- tions to read lips with regard to the five auditory test conditions employed in this study. Importance of the Study The contribution of the environment to lipreading success or failure is still largely undetermined, especially when evaluating the effects that relevant or irrelevant auditory stimuli may have on the lipreading process. A study by Leonard (1962) found that continuous ‘white noise, running speech and continuous background music impeded lipreading performance of subjects who were trained under conditions of relative quiet. Other investigations, employing different kinds of auditory distractors have disclosed conflicting findings. There- fore, additional research effort is necessary in order to establish more definitively the effects of auditory distractors on lipreading success. By ascertaining the influence of noise on lipreading perform- ance, the need to make the environment as quiet as possible where people work and live may be illustrated. This study also evaluates the impact that various controlled auditory conditions have on the training process of learning to read lips. To date, only the investigations of Ojima and Nakano (1961, 1962, 1963) have reported using noise while teaching the process of lipreading to experimental subjects. However, their findings regarding the effects of one auditory distractor (white noise) on lipreading performance are unclear. Consequently, additional research is indi- cated if the effects of auditory stimulation during lipreading acquisition are to be understood. By such efforts we may then deter- mine how best to employ auditory stimulation so that it enhances the rate of lipreading acquisition, the skill of the lipreader, or both. Terminology For the purposes of this study, the terms used are defined operationally as follows: Lipreading (speechreading).--A strictly visual, non-auditory process by which a subject correctly identifies Spondaic word items spoken by a talker presented on videotape. Spondaic words.--Words of two syllables spoken with approxi- mately equal stress on both syllables, e.g., baseball, hotdog, etc. Auditory distractors.—-Various sound stimuli continuously presented without interruption. These auditory distractors were as follows: (1) babble--segments from a men's conversational group, a women's conversational group and a cocktail party with both men and women participating; (2) industrial noise--a recording of power plant generators in use; (3) traffic noise--a simulated traffic jam with car horns honking, ambulance sirens blaring and motorcycle engines running; and (4) music--Gershwin's Rhapsody in Blue. Normal hearing:--The ability to pass a pure-tone screening test presented individually to both ears at 25 dB (ANSI, 1969) at octave intervals from 250 through 4000 Hz. Normal vision.--Visual acuity of 20/20 in both eyes (with or without correction) as measured by a Bausch and Lamb, Modified orthorater. Organization of the Dissertation Chapter I contains a statement of introduction about the dele— terious contribution of noise to the communication process and its possible effects on lipreading skills. The exact purpose of the study and its potential importance to the research literature are included. The major terms used in this study are defined and the null hypotheses are stated. Chapter 11 reviews the scientific literature pertinent to this investigation. Previous studies which looked at, among others, the effects of environmental stimuli on lipreading ability, the physio- legical effects of noise on vision and the effects of noise as a dis- tracting agent on various non-auditory tasks are summarized. Chapter III describes, in detail, the subjects, the equipment, the materials and the experimental procedures employed in this study. Chapter IV presents the data collected from this investigation and a discussion of the results. Chapter V incorporates a summary of the problem investigated, the conclusions drawn as a result of this study and the recommenda- tions for further research activity. CHAPTER II REVIEW OF THE LITERATURE Importance of Visual Cues to Communication Traditionally, vision and audition have been considered the distance or lead senses, particularly from the standpoint of message reception (Berger, 1972). Vision has served as the primary sense for the deaf or severely hard of hearing individual, whereas normal hearing persons have relied mainly on audition for effective communication. Lipreading represents the process by which the visual modality functions to increase information flow. It is a useful tool for the hearing impaired person who is occasionally or consistently unable to differentiate clearly the sounds of language. Lipreading assists these individuals because it enables them to identify visually the shape and movement of a speaker's articulators, thereby filling in some of the information which otherwise might have remained unheard or undetected (O'Neill and Oyer, 1961). This visual mode also serves a functional purpose for the normal-hearing. It augments the auditory channel when a noise environ- ment precludes a satisfactory exchange of verbal information. Reliance on vision also helps the normal hearing listener who is communicating with a soft-Spoken individual. The contribution of vision to Speech perception and intelligibility has been investigated extensively, although the issue still remains somewhat in doubt. O'Neill (1954), seeking to determine what role lipreading played in the communication process, fbund that the visual modality made its most significant mark as listening became less favorable. He enlisted thirty-two normal hearing subjects who were exposed to three talkers producing vowels, consonants, words and phrases. In half of the experimental conditions, the participants both heard and viewed the talker, each time listening in a different Speech-to-noise ratio which became increasingly more difficult to understand. In the other situations, the S/N values remained the same, but the subjects only heard the Speaker. In two instances (for consonants and words) the scores for vision alone exceeded the results for audition alone. For vowels and phrases, the non-visual totals surpassed those of the visual, i.e., lipreading. However, in every case the combined approach of audition and vision was most efficient. In addition, O'Neill assessed the relative contribution of vision under S/N comparisons of -20 dB and +10 dB. The results, as stated above, Showed that the visual mode added a more significant increment as the listening condition decreased. For example, the relative contribution of the visual process for words at a S/N ratio of -20 dB was 43 per cent but only 5 per cent at a S/N ratio of +10 dB. Sumby and Pollack (1954) also investigated the contribution of vision to Speech intelligibility in the presence of competing noise. A talker using live voice presented Spondaic words, monosyllables and trisyllabic phrases to 109 listeners who wore ear- phones. These stimuli were presented unimodally (separately to the eyes and to the ears) and bimodally (where vision and audition are combined) under various S/N conditions, from 0 dB (the most favorable) to -30 dB (the poorest). In general, the authors found, as did O'Neill in 1954, that as the Speech-to-noise ratio decreases, the advantage and benefit from vision in comprehending Speech increases. Another Study attempting to quantify the contribution of vision to Speech comprehension was carried out by Neely (1956). Thirty-five normal hearing listeners viewed a trained talker from various profiles and distances. The investigator determined that the visual modality, when combined with the auditory mode, increased the reception of speech by almost 20 per cent over hearing alone. There- fore, the importance of lipreading was upheld especially in the en- vironment with high noise content. Karlovich (1968) investigated the sensory interaction of the auditory and visual modalities. Eight normal hearing adults parti- cipated in an alternate binaural loudness balance (ABLB) task in which a 1000 Hz tone was balanced at sensation levels at 20 dB and 50 dB. In the first of three conditions, no visual Stimulus was presented during the ABLB task. In the last two Situations visual stimulation (a Short duration StroboscOpic light flash) was introduced. In the one case the visual Stimulus was synchronized with the standard audi- tory Stimulus during the ABLB task, whereas in the second instance, the visual stimulus was synchronized with the comparison auditory stimulus. The results Showed that the visual stimulus facilitated 10 the perception of loudness, especially at the higher sensation level. Consequently, this experiment demonstrated that there is interaction between the sensory modes of vision and audition. Sanders and Goodrich (1971) also discovered the value of the speechreading process in procuring information where auditory reception is hampered because of failing listening conditions. Of interest to these authors was the role of vision and audition (both separately and collectively) in the reception of Speech when auditory stimuli were selectively distorted. This research deviated from those studies cited previously in that the amount of available information to the auditory channel was controlled by filtering rather than the manipulation of the Signal-to-noise ratio. The results demonstrated, in part, that when the listening Situation is severely limited, as exemplified by a low-pass filter condition, the effect of combining the visual and auditory senses immeasurably augments comprehension, whereas each modality separately contributes substantially less. / Each of the above investigations demonstrates that vision and audition, when used in combination, provide more information to the receiver than when used separately. Also the value of vision appears to increase aS the listening Situation becomes more difficult. What apparently occurs is the reduction of information provided by the auditory signal. No longer is the listener able to predict accurately the content of the communication and thus increasingly relies on visual cues to supplement information. In disagreement with the above findings are the results obtained by Gaeth (1967) who also evaluated the relative efficiency of 11 the auditory and visual senses as individual and bimodal contributors to the process of communication. He found that either one of the Single modalities was superior or equal to the combined audio-visual presentation. Although some controversy may still exist as to the approximate contribution of these two primary communication channels, each is very important in its own way, depending on the particular environmental condition existing at that moment. Some Effects of Environment on Lipreading_ Vocal communication between individuals often occurs under physical circumstances over which neither the sender nor receiver has any appreciable control. Therefore, the success of the interchange of ideas is necessarily dependent upon the environmental factors which exist at that moment. In the preceding paragraphs the importance of lipreading to message reception was discussed. When, for whatever reasons, the desired acoustic Signals are not received by the listener, research has demonstrated the value of using lipreading. It is, there- fore, incumbent that the environmental factors which could facilitate or undermine the effectiveness of the process of lipreading be under- stood. These factors include lighting, the distance between Speaker and lipreader, the transmission medium of the message, and visual and auditory distractions. The latter (auditory distractions) will be considered separately in the section Lipreading Studies in Noise (p. 20). 12 Lighting Few Studies have investigated the effects of lighting on lip- reading ability. Thomas (1962) trained twenty adults to lipread thirty test sentences under illumination which remained constant throughout the practice lessons. Afterwards, these same stimuli were presented again under Six different intensities of light as measured in foot candles (lumina) in order to evaluate the effect of decreasing light levels on lipreading performance. The results showed that diminishing light intensities did not Significantly change lipreading scores. Even under the most adverse conditions of luminescence (0.5 lumen) satisfactory lipreading was possible. Jackson in 1967 (Berger, 1972) found that relatively inexper- ienced lipreaders scored better when the light source was Strate- gically placed below and in front of the talker rather than above the Speaker in the conventional manner of classroom lighting. With the more experienCed lipreaders, light placement made little difference. Distance Between Speaker and Lipreader Numerous studies have attempted to establish the optimum dis- tance from which lipreading can be taught and practiced. Cavender in 1949 (O'Neill and Oyer, 1961) had subjects speechread from distances up to eighteen feet. No significant test differences were found. Similarly, Mulligan in 1954 (O'Neill and Oyer, 1961) uncovered no statistically important distinctions between scores by Speechreaders viewing a filmed talker at distances of S, 10, 15 and 20 feet, although the best results occurred at ten feet. 13 Neely (1956) placed subjects at intervals of 3, 6 and 9 feet from the Speaker. No Significant differences could be noted at those three distances. Berger (1972) concluded that Speechreading performance should not be affected by distances up to twenty-four feet provided the observer's vision is within normal limits. However, for practical purposes, speechreading is best carried out from five to ten feet away, a distance representative of most conversational speech situations. Transmission Medium An important variable in lipreading is the medium through which the information passes from the Speaker to the receiver. The transmission link can be person to person (live), films, or television (closed circuit or videotape). Each of these channels has been used experimentally for purposes of training or testing speechreading ability. Berger (1972) concludes that the major fault of live Speech- reading presentations is the inability of the Speaker to duplicate exactly the original Speech stimuli time and time again (a factor of reliability). Filmed tests, not without limitations, are criticized for being a two-dimensional process, whereas speechreading in a live situation is three-dimensional (Berger, 1972). Oyer (1961) found that lipreading could be taught success- fully through the medium of closed-circuit television. A group of thirty-two normal hearing college students was given ten weeks of training. In a post-training filmed test situation they made fewer mistakes than in the initial test session. 14 Cancel (1970) also employed television in his speechreading investigation. This researcher, however, used the process of video- tape to present the test stimuli (bisyllables and sentences). This study was deve10ped to test the lipreading Skills of Spanish-Speaking individuals who were either hearing impaired or deaf. Visual Distractions Miller (1965) examined the effects of an assortment of selected visual distractions on lipreading performance. Fourteen subjects lipread a female speaker under four conditions: (1) without distraction (the control condition), (2) a flashing light, (3) a Spinning disc, and (4) nonpurposeful hand movements (made by the Speaker as the test materials were presented). The only Significant and influential visual distractor was the nonpurposeful movement of the hands. Keil (1968) investigated the consequences of background stimuli on lipreading performance. The peripheral visual stimuli used in this study included (1) a neutral, gray background, (2) this same backdrOp with a female projected on each side of the talker, (3) a Still picture of a building with trees and an automobile, and (4) a moving scene of people at a busy Street corner. No Significant differences in lipreading performance were noted for either normal hearing or hearing impaired lipreaders. The Effects of Noise on Man For several years a number of investigations have attempted to determine whether or not noise affects man's (l) ability to lS perform tasks efficiently, (2) physiological reactions, and (3) personal feelings and attitudes. Noise Effects on Task Performance The published reports which have characterized the influence of audition on task performance have arrived at a number of widely diSparate conclusions, partly because the parameters of interest were as equally divergent. The discrepant results may be attributable, among others, to one or both of the following reasons: (1) the nature or demands of the work performed and (2) the type of auditory Stimulus employed, its intensity, duration and/or mode of presentation. The Nature of Task Performance Cohen (1969) concluded that the majority of those Studies evaluating non-auditory sensory and perceptual experience (and citing noise as a disruptive influence) involve performance tasks of vigi- lance in which the subject is required to monitor visually a number of clocks or dials for the purpose of detecting slight changes in movement or intensity of light. Broadbent (1954) established that the task efficiency of a group of ten subjects became impaired in the presence of a 100 dB SPL noise but not in a noise level of 70 dB SPL. The vigilance Operation was to monitor a display panel of several Steam-pressure gauges which were difficult to read. These results contrast with those of a Similar experiment reported in the same study where a different group of twenty subjects, 16 completing an easier type of vigilance assignment, manifested no discernible effects from the noise. Jerison (1959) conducted an investigation with a group of male subjects whose monitoring task was to observe an arrangement of three clocks for a duration of two hours in conditions of "quiet" (83 dB SPL of environmental noise) and noise (114 dB SPL). The results demonstrated that the noise had no appreciable effect on work performance for the first hour and a half. However, during the remaining thirty minutes performance efficiency deteriorated under the noise condition, although no comparable change occurred during the last half-hour interval in the quieter condition. The investigator concluded that noise, if intense enough, will lessen task capability provided duration time of exposure is sufficient and the task requires substantial Shifts in task attention, e.g., monitoring several clocks. This same investigator in an earlier Study (Jerison, 1957) had found no differences in efficiency for a simple vigilance task (watching one Stepping clock) carried out by twenty subjects under similar conditions of exposure (noise, 112.5 dB SPL; quiet, 79 dB SPL of ambient noise) and duration (one and three-quarter hours). Cohen (1969) also concluded that tasks conducted in noise which "demand less than the total skillfulness of the individual" can be successfully completed by drawing upon reserve capacity. This stands in contrast to any task (most likely vigilance) which requires constant surveillance and involvement. Here performance has been found on occasion to falter when noise is introduced. 17 Auditory Stimuli: Types, Intensity, Duration and Mode of Presentation A portion of the conflicting data has been attributed to the influence of the various acoustical Stimuli employed in the different investigations or to any number of parameters or combinations of vari- ables such as sound pressure level, duration of exposure and mode of presentation of the auditory Stimuli. Cohen (1969) referred to Studies by Wbodhead in 1959 and by Forwalt in 1965 in which they found, independently, that the probable sources behind poorer performances were random, intermittent noises rather than the steady-state, continuous type of noise. Broadbent (1957) found that a 100 dB SPL high frequency machinery noise (filtered to contain frequency bands above 2000 Hz) contributed to twice as many errors in performance than did a low frequency sound (filtered to contain frequency bands below 2000 Hz) of Similar sound pressure. However, at levels of 80 dB and 90 dB SPL no such effects appeared. Nevertheless, Broadbent recommended eliminating or reducing the high frequency components in noise for improved performance. Jerison's (1959) study showed that high noise levels (114 dB SPL) over a period of time produced increased error frequency in com- parison to low intensity noise (83 dB SPL) for work tasks which can be described as demanding and repetitive. Kryter (1970) reiterated, from results he reported in 1950, that noise as such is not reSponsible for performance decrements in non-auditory work situations. Any changes, he felt, are ephemeral, 18 relatively insignificant, or tied to worker attitudes or feelings of motivation which are subject to fluctuation. Physiological Assessments Lipreading is a procedure which exclusively relies upon the visual modality, and it represents the sole means by which the test stimuli in this current research effort can be identified. Therefore, it is incumbent to understand the potential role of noise on the integrity of the visual process. In 1941 Stevens (Cohen, 1969) found that intense acoustic levels had the effect of slowing down the rate of eye movement and, concomitantly, producing a focusing problem for the identification of near and distant objects. Stevens surmised that the noise pre- cipitated a restraining effect on the ciliary muscles which regulate the lens covering the pupil. Cohen cautioned, however, that such changes in visual capacity as a result of Stimulation from noise are minimal except at the very highest exposure levels. Two studies reviewed by Kryter (1970) reached different con- clusions regarding the effects of noise on Sight. Loeb, in 1954, found vision unaffected by 115 dB SPL of broadband noise, whereas the subjects in a 1954 Rubenstein investigation reacted unpropitiously to intensity levels as low as 75 dB SPL. In Kryter's Opinion, the major portion of research has shown noise to have fairly negligible effects on vision. According to Jansen (1969) any acoustic Stimulus, irrespec- tive of its meaning to the listener, might potentially influence the individual's physiological integrity. The critical determinants are 19 the intensity and frequency of the stimulus. His experiments have demonstrated that alterations take place in pupil function and in the eyes' peripheral circulatory system. When young adult subjects were exposed daily for thirty to Sixty minute sessions in 90 dB SPL of white noise over three years, they evinced measurable pupil dili- tation. Jansen demonstrated that there is a direct relationship between the intensity and bandwidth of a Stimulus and pupil size. He found that as either intensity or bandwidth was increased, so did the magnitude of the visual organ increase when subjected to white noise at exposure levels of 70 dB SPL or greater. However, noise delivered at 95 dB SPL or less had no apparent path010gica1 side effects on vision. A previous investigation by Jansen in 1959 had confirmed the occurrences of such Side effects when intensities exceeded the 95 dB SPL level. Personal Feelings and Attitudes The presence of noise or other acoustic agents, when in close proximity to human subjects, can Significantly influence feelings and attitudes, especially toward job satisfaction and work productivity. These feelings have been found to evoke both negative and positive responses. Cohen (1969), in particular, has evaluated the effects of annoyance on human emotion and response when caused by divergent auditory stimuli. Annoyance, he contends, can be due to a qualitative or quantitative characteristic of the stimulus or the result of the type of message which the Stimulus conveys. He cited the Goodfriend and Cardinell study of 1963 in which hospital patients reported being 20 upset having overheard the conversations of hospital employees who were discussing other patients' conditions and Operations. This led to unpleasant feelings which were precipitated by sounds whose content was disruptive and irritating though not physically intense. Stein (1963) discovered that music positively influenced the therapy of disturbed patients. For example, those individuals who were emotionally upset reSponded favorably to melodic, rhythmic music, whereas those who exhibited depression were stimulated by bouncy, fast-paced music. Kryter (1970) described instances where music has been used in industry to elicit changes in worker morale in an effort to increase production, especially among employees performing monotonous tasks. Wyatt and Langdon in 1935 (Kryter, 1970) found no consistent relation- ship between the use of music and increased work output. Kryter's general conclusions were that if beneficial effects do occur, they will be small and transitory at best in regard to worker motivation and production. Another type Of acoustic stimulus which has been found to be disturbing is roadway traffic. Wilson in 1963 (Cohen, 1969) dis- covered that such noise, when heard at home, was felt to be an intrusion to rest and privacy. LipreadingStudies in Noise Of Special interest to this investigation are those studies which have evaluated the lipreading performance Of subjects who were being eXposed to various combinations of acoustic Stimuli and 21 intensity levels or who were being trained to lipread under different kinds and conditions of noise. Leonard (1962) employed twelve normal adult subjects who were initially trained under conditions of quiet (55 dB SPL of ambient noise) to lipread a talker repeating fifty monosyllabic and bisyllabic words. The talker was filmed in advance with the test vocabulary inaudible to the listeners. Each subject was required to achieve an 86 per cent correct reSponse criterion of the test words before he could participate in the experimental test session. The experimental session was Similar to the training program in that the vocabulary and test format were the same. However, this time two conditions were different: (1) a total Of four different films was used, each one representing a randomly scrambled ordering Of the test items and (2) lipreading ability was measured under the following background conditions: (a) in quiet (55 dB SPL Of environ- mental noise) and in 80 dB SPL of (b) white noise, (c) running speech, and (d) music. The results in general showed that certain continuously present acoustic Stimuli can reduce lipreading performance substan- tially. Each Stimulus (white noise, running Speech, and music) lowered lipreading performance in a statistically significant manner, although the background music did so to a lesser degree than the others. Pettit (1963) evaluated the lipreading proficiency of thirty- three subjects who watched silent motion-picture films in the pres- ence of different auditory maskers. The participants were evenly 22 divided into three groups depending upon their lipreading skills. One-third were considered to be trained lipreaders, another third were familiar with the process of lipreading, whereas the remaining eleven subjects were classified as untrained, i.e., having little Skill or knowledge of lipreading methods. A total of four lipreading films was made. Each included the same 104 test items (monosyllables). These words had been previously rank-ordered as to ease of lipreadability and divided into four equally difficult subtests of twenty-six words. Each Silent film also used four different talkers vocalizing twenty-six items apiece. Therefore, in all four films, all four Speakers delivered a different subtest of words, whose order of presentation also changed from film to film. Each film was viewed by all participants under the following listening conditions: (1) quiet, (2) babble, (3) traffic noise, and (4) white noise. Quiet represented 55 dB SPL of ambient room noise. The other three stimuli were presented under earphones at 90 dB SPL. All motion pictures contained forty practice items presented in a setting of quiet which preceded the test words. The results demonstrated that Speechreaders performed differ- entially under divergent acoustical environments. Scores were better in quiet than in white noise when these two conditions were con- trasted, whereas the totals for the stimulus of traffic noise were superior when compared individually to both white noise and babble. In each case these differences were statistically significant. Interestingly enough, the comparisons of quiet versus traffic noise 23 and quiet versus babble manifested no statistical differences. The scores of trained Speechreaders were found to be significantly differ- ent from the results of their untrained counterparts, whereas no such distinctions statistically could be made for the comparisons trained versus familiar and familiar versus untrained lipreaders. The most recent study to investigate the effects of noise on lipreading skills (Berger and Lewis, 1972) employed twelve normal hearing subjects who had received Speechreading training. Each participant lipread various monosyllabic word pairs Spoken by a talker isolated in a room and viewed only through a large, double window. The test was conducted under three contrasting listening environments: (1) quiet (55 dB SPL of ambient noise), (2) continuous white noise (90 dB SPL), and (3) intermittent (one-half second on, one-half second off) white noise (90 dB SPL). The authors hypothesized that an intermittent noise stimulus would precipitate more distraction than a continuous Stimulus. The best scores were made in quiet, the worst in the inter- mittent white noise. However, the separations in scores among the three conditions were not Significant. The results are conflicting as to the effect of noise on lipreading performance, in general. The first study (Leonard, 1962) found all Stimuli Significantly and negatively affecting lipreading scores. The Pettit (1963) investigation found some significant score differences but not others, whereas the most recent research (Berger and Lewis, 1972) found small differences between quiet and noise, but these differences were not statistically significant. The types of 24 stimuli have varied, the mode of presentation has varied (sound field versus earphones), but the intensity level of stimulation has remained comparable among these studies. Ojima and Nakano have published three Studies in which noise was introduced while their subjects were learning to lipread. In the first of these reports (1961) ten normal hearing students learned and practiced to lipread words, vowels and syllables under conditions of a white noise masker. Although the authors concluded that practice positively aided lipreading scores, the effect of the masking on performance was not stipulated. The format of the second investigation (1962) was similar to the first. The results showed that the fre- quency of errors made throughout the investigation remained constant. Masking did not significantly contribute to an increased error rate. Also the identification of low vowels was immeasurably better than that for high vowels. The latest of their Studies (1963), which also involved the lipreading training of ten normal hearing subjects under white noise, found that lipreading performance under noise improved with practice although the increases varied from slight to considerable. These variations were attributed to the type of stimulus lipread and individual subject differences. Summary The scientific literature has clearly indicated the role which vision plays in the communication process, particularly in the pres- ence of adverse auditory environmental conditions where the verbal message is received unintelligibly. The successful application of lipreading Skills, however, is subject to a number of variables, many 25 of which could be directly influenced by the physical surroundings which exist at that moment. Of these, the parameter which potentially appears to have the most Significant effect on lipreading performance is noise. The disrupting consequence of noise on some non-auditory tasks has been demonstrated. In addition, noise has been found to alter human feelings, attitudes and physiology. Although a handful of investigations has looked directly at the effects of noise on lipreading performance, disparate conclusions have been drawn. In addition, those few studies which have looked at the influence of auditory stimuli in the acquisition or training of lip- reading proficiency have not completely exhausted this aspect of aural rehabilitative research. Therefore, the necessity for further exploration in this area is justified and indicated. CHAPTER III EXPERIMENTAL DESIGN Subjects IE? Forty normal hearing, normal seeing adults, all of whom were college educated and affiliated with the Audiology and Speech Sciences Department of Michigan State University as either Students, staff or spouses of a Student, served as subjects in the present investigation. The age range of these forty persons (twenty men and twenty women) was twenty to forty-two years with a mean age of 25.75 years, with a standard deviation of 3.73 years. To satisfy the criterion for normal hearing, each participant passed a pure-tone audiometric screening test in each ear given at a level of 25 dB (re ANSI, 1969) at the octave frequencies from 250 through 4000 Hz via Telephonics TDH-49/102 earphones mounted in MX-4l/AR cushions. The examination was completed in an Industrial Acoustics Company (IAC), model 1600 series (ACT), sound insulated suite using a calibrated Grason-Stadler diagnostic audiometer, model 1701. In addition, each individual had to satisfy the visual screening criterion of 20/20 in both eyes (with or without correction) as measured with a Bausch and Lomb, Modified orthorater. 26 27 In both cases, these screening measures were performed and evaluated by the investigator and carried out in the Rehabilitative Audiology Research Laboratory of the Audiology and Speech Sciences Department prior to the subject's involvement in any of the formal experimental procedures. Materials Below is a list of the materials employed in this investigation along with a descriptive note about each of them: The test vocabulary consisted of fifty spondaic stimulus items, i.e., two syllable words such as "baseball" and "hotdog," which were taken from the Central Institute for the Deaf's Auditory Test W-l and Harvard's Psychoacoustic Laboratory Auditory Test Number Nine (Lists I and II). The test vocabulary was placed into Six random orders (see Appendix A) according to accepted Statistical procedures for use with all subjects in the training and test procedures of this study. Videotapes using these Six random orderings of the vocabulary were made. A Cohu, model 20/20, black and white television camera videotaped the full face contour of a native male talker (the investi- gator) as he uttered each one of the fifty vocabulary items. The camera captured the talker's face so that it would appear life Size when played back on a twenty-one inch television screen. Each tape was electronically edited so that every Spondaic item was exactly the same in all versions. For each group of eight subjects one of these six tapes (by random selection) served as the training tape, whereas the five other tapes were utilized in the test session. 28 The construction of each tape was as follows. Each stimulus item was preceded by ten to eleven seconds of blank videotape and approximately 1-1/2 to 2-1/2 seconds of tape Showing the talker ready to present the item from the closed lips position. After the vocab- ulary word was uttered, the talker remained on camera for a period of from three to five seconds before more blank videotape appeared. The time separation between vocabulary items varied between sixteen and eighteen seconds. (See Appendix B for exact time sequences of all Six videotapes.) The total running time of each tape was approximately fifteen minutes. Four background sound tapes were recorded for use with the training and test procedures. These auditory sounds included (1) babble, (2) industrial noise, (3) traffic noise, and (4) music. Each was recorded on Scotch 201 magnetic tape at a Speed of seven and one-half inches per second for approximately Sixteen minutes and played back on an Ampex tape recorder, model AG600. Various incidental materials were used and samples of them can be found in Appendix C. They included: (1) two hundred scoring sheets for use by the forty subjects to record their answers to the Stimuli presented under the five test conditions, (2) approximately 200 scoring Sheets which were used during the training sessions by the forty subjects for their written reSponses, (3) forty vocabulary lists of the fifty spondaic words (listed in alphabetical order) for purposes of familiarization by the subjects, (4) forty instruction sheets describing the training procedures, (5) eight number two lead pencils for writing responses. 29 Training and Test Environment and Equipment Test Environment An eighteen by nineteen foot, windowless sound-treated class- room.1ocated in the Audiology and Speech Sciences Building served as the training and test environment. A schematic diagram of this room (Figure l) pinpoints the placement of the equipment. Equipment A Sony Videocorder Production Deck (model 3600), which accom- modates one-half inch black and white videotape, played the final dubbed versions of the vocabulary in both the training and test pro- cedures. It was located at the rear of the classroom illustrated below and was linked to a television monitor situated at the front of the classroom. In addition to the Videocorder Production Deck, a Sony Video- corder Edit Deck (model 3650) and two Sony black and white television monitors (models CVM-920U and PVJ-SlO) were used together to dub magnetically the vocabulary into the six desired tapes. A Motorola twenty-one inch black and white television set served as the monitor of the videotaped vocabulary. It was physically situated about twelve inches above the eye level of the subjects in the experimental room. Eight chair-desks were arranged in two rows of four seats each. The seats nearest the television monitor were six feet away, whereas the chair-desks in the second row were placed three feet further back. All seats were situated so that all subjects had an unobstructed view of the television set. 30 r [ Loud , speaker TV monitor AAAA Chair-desks AQAA Videotape deck and Tape recorder I ,l/ ‘ Slide projector [ Loud speaker Figure 1. Physical Arrangement of the Equipment and Chair-desks in the Test (Training) Environment. 31 An Ampex, model AG600, tape recorder fed the pre-recorded sound tapes into two Ampex, model AA620, loud Speakers. This recorder was juxtaposed with the Videocorder Production Deck at the rear of the classroom. The auditory background sounds were reproduced by the two Ampex loud speakers which were placed at a zero degree azimuth to the front and rear of the experimental subjects. Each loud Speaker was equi- distant from the nearest row to ensure that all subjects were exposed to equivalent levels of sound pressure. To monitor the desired output of sound pressure from both loud Speakers, a Bruel and Kjaer sound pressure level meter (model 2203) was employed. A Sawyers (model 570AF) two by two slide projector, Situated above the rear loud Speaker and operated by the investigator, projected the apprOpriately numbered sequence of the Spondaic stimulus above the television monitor in order to preclude any subject from writing an answer out of sequence during either the training or test trials. Each appropriate item number was flashed from the moment one stimulus item had been presented until the next stimulus word was to be offered but not during the presentation of any vocabulary item. Procedures Each subject participated in two different experimental pro- cedures. The first was a training procedure, the second, a test procedure. Each training and test format is described below. Prior to their inclusion in the study, all subjects had met the auditory and visual requirements. 32 Training Procedure The forty experimental participants were assigned on a random 'basis to one of five groups. The male and female subjects were assigned separately so that there would be four members of each sex per group. Each aggregate of eight subjects was differentiated only by the type of background environment under which the vocabulary of fifty Spondaic words was to be lipread. Group I lipread in quiet. The ambient noise level of the training environment was approximately 50 dB SPL as determined by the C scale of a sound pressure level meter. The remaining four groups were trained in a sound field of 90 dB SPL (monitored by sound pressure level meter). The types of sound employed in the remaining four groups were as follows: (a) Group II, babble; (b) Group III, industrial noise; (c) Group IV, traffic noise; and (d) Group V, music. The objective of the training procedure was to familiarize the subjects with the experimental vocabulary SO that each participant could learn to lipread this vocabulary and attain at least a correct response criterion of 90 per cent (i.e., at least forty—five correct out of fifty). Two days prior to the training session each subject was given a cOpy of the instructions to be followed during the training period and an alphabetized list of the vocabulary. The instructions, which were read by the investigator prior to the training procedure, were as follows: This is a test of your ability to lipread fifty inaudible words such as "outside" and "cowboy" as they are presented by me on the television monitor in front of you under a background condition of either quiet or auditory Stimulation. On the 33 answer Sheet in front of you, decide what word you saw on the screen and write it in the appropriately numbered Space provided. Prior to the presentation of each stimulus, its sequence number will appear above the monitor. Each vocabulary word is preceded by several seconds of blank videotape and approximately two seconds of videotape Showing the talker ready to present a stimulus item from the closed lips position. After the vocabulary item is uttered, the talker remains on camera for a brief moment before more blank tape appears. After each Spondaic item is presented, write down your answer quickly and then look up and prepare to reSpond to the next stimulus item. About three seconds before each word is presented a pure tone will sound alerting you to the presentation of the next Stimulus item. At any time you may refer to your alphabetized vocabulary list. Guessing is encouraged, but if you cannot guess, be sure to draw a line through that particular response blank. Some items may be very easy to identify, others more difficult. In any event, watch the monitor closely when vocabulary items are being presented. All words are just given once. 00 you have any questions about what I have asked you to do? If the training Situation called for one of the four auditory background tapes to be played, it was Started about one minute before the videotape was run to minimize any possible mistakes because of a surprise reaction to the introduction of the sound. The presentation level of 90 dB SPL for each auditory sound was pre-set on the equipment. All papers from the training session were collected after the last of the fifty word Stimuli was given (considered one trial) and each was graded while the subjects took a rest break. Those who scored 90 per cent or higher were considered to be trained lipreaders for the purpose of this study and dismissed, since they had met the pre- established criterion for participating in the test procedure. For those not meeting this criterion, the same conditions under which they were initially exposed resumed. Each subject was told only his per cent correct score. The above procedure was carried out until the 90 per cent correct reSponse criterion was met. If a subject needed 34 more than four such trials, he was brought back the next day to avoid the possibility of excessive fatigue. Each subject, however, was not given more than eight trial runs to satisfy the training criterion. If unsuccessful, the subject was replaced. Once each subject had met the training requirements, he or she was ready for the test procedure. Test Procedure The test procedure was similar to that of training except for a few changes. The lipreading ability of each subject was now eval- uated under all five conditions used in the training procedure, including the condition under which that particular individual had been trained (quiet, babble, industrial noise, traffic noise, and music). The experimental vocabulary remained the same, but for each of the five test situations a different videotaped ordering of the Spondaic stimuli was presented in an effort to preclude vocabulary order effects. All other conditions remained the same. In the test session a rest period of approximately ten minutes was given after the second and fourth trials. The papers were not corrected during this test phase. For each group of eight subjects, the sequence in which the background sounds was presented was also different to preclude order effects, as it had been for vocabulary lists. See Appendix D for the order of test and vocabulary sequence presentations. The training and test procedures were both completed within two to three days so that the subjects would not lose their trained efficiency for lipreading the vocabulary. CHAPTER IV RESULTS AND DISCUSSION Introduction This chapter is divided into two main sections: results and discussion. The results include the statistical analyses to which the data were subjected along with a descriptive and, where appropriate, a graphic explanation of what these analyses revealed. Finally, a discussion of the results as they relate to other relevant research is provided. Of the original forty subjects (twenty men, twenty women) who attempted to satisfy the training criterion (90 per cent correct identification of the spondaic word stimuli), thirty-four (fifteen men, nineteen women) qualified in eight trials or less. Six subjects were replaced. Three of the five non-qualifying males had been trained to lipread in music, one in quiet and the other in industrial noise. The one female who dropped out came from the group trained in traffic noise. Their Six replacements subsequently fulfilled the established training criterion and participated in the study. Results Two analyses were employed in this investigation. The first was a two-way, fixed-effects analysis of variance (5 x 2) which was 35 36 calculated to establish the relationship of training condition (quiet, ‘babble, industrial noise, traffic noise, and music) and sex (male, female) to the number of trials needed by a subject to qualify for the test phase of the experiment. The second procedure was a three-way, fixed-effects analysis of variance (5 x 2 x S) with repeated observations on one of the factors (test measures). The lipreading scores obtained from the five test measures (quiet, babble, industrial noise, traffic noise, and music) were analyzed as a function of the treatments of training con- dition (quiet, babble, industrial noise, traffic noise, and music) and sex (male, female). Trials as a Function of Training Condition and Sex Table 1 represents a two-way, fixed-effects (training by sex) analysis of variance on the number of trials needed by subjects to TABLE 1 SUMMARY OF THE TWO-WAY, FIXED—EFFECTS (TRAINING BY SEX) ANALYSIS OF VARIANCE FOR THE NUMBER OF TRAINING TRIALS NEEDED TO QUALIFY FOR THE TEST SESSION . . Degrees of Mean F Probability Source Of Variation Freedom Square Value of Statistic A (Training) 4 4.41 1.04 0.4030 B (Sex) 1 13.22 3.11 0.0880 AB (Training x Sex) 4 1.66 0.39 0.8141 Error (Within) 30 4.24 37 «mualify for the test phase of the study. The results indicate that there is no Significant main effect for training (factor A) and no Significant training by sex interaction (factor AB). However, the sex main effect (factor B) was statistically Significant at the .088 level. The mean data for these trials were tabulated by training condition and sex and appear in Table 2. The row mean for women (across all training conditions) was 3.35 trials versus a row mean for men of 4.50 trials, indicating that female subjects required fewer trials to qualify than did their male counterparts. The analysis of variance found this difference between these two means to be signi- ficant. Figure 2 graphically traces these mean scores for trials as a function of sex across all training conditions. TABLE 2 MEAN NUMBER OF TRIALS, TABULATED BY SEX AND TRAINING CONDITION, NEEDED TO QUALIFY FOR THE TEST SESSION Training Condition Sex . Industrial Traffic . Row QUIet Babble Noise Noise MUSIC Mean Male 4.00 5.50 4.50 3.25 5.25 4.50 Female . 2.75 4.75 3.25 5. 25 2.75 5.35 Column Mean 3.38 5.13 3.88 3.25 4.00 3.93 Table 3 contains the basic information on the number of trials needed by each subject to qualify. These data are grouped by sex and training condition. Note that one member of each sex (subject number twenty-one, female, trained in industrial noise; subject number Mean Number of Trials 1:. 38 Sex -——- Men — — Women 3d 2 . 1.. I I T l ’1 u 0 H4) 00 U o H mm with "-4 0H '0 'H‘H M'H m e; .o u o ‘H o :g m sz «:2 In m a a E— 'e a H Figure 2. Mean Number of Trials Across Training Conditions as a Function of Sex. 39 TABLE 3 NUMBER OF TRAINING TRIALS NEEDED BY EACH SUBJECT TO QUALIFY FOR THE TEST SESSION 5.5 «5 9 C094 44 and 0H "-4“ ”3'0 OH «4+3 (0'0 00 coo-4 H0 00 Goo-t HO no go on n we up as 5 .5 a: a? r :25 a. cn:z a: F'LJ 9-2: «0:2 co F‘LJ 9-2: 1 M Quiet 2 S F Quiet 2 2 M Quiet 3 6 F Quiet 4 3 M Quiet 6 7 F Quiet 2 4 M Quiet S 8 F Quiet 3 9 M Babble 2 13 F Babble 3 10 M Babble 5 14 F Babble 8 11 M Babble 8 15 F Babble 4 12 M Babble 7 16 F Babble 4 17 M Industrial Noise 5 21 F Industrial Noise 1 18 M Industrial Noise 3 22 F Industrial Noise 2 19 M Industrial Noise 2 23 F Industrial Noise 3 20 M Industrial Noise 8 24 F Industrial Noise 7 25 M Traffic Noise 2 29 F Traffic Noise 2 26 M Traffic Noise 1 30 F Traffic Noise 3 27 M Traffic Noise 7 31 F Traffic Noise 6 28 M Traffic Noise 3 32 F Traffic Noise 2 33 M Music 5 37 P Music 2 34 M Music 4 38 F Music 2 35 M Music 6 39 F Music 4 36 M Music 6 40 F Music 3 40 twenty-Six, male trained in traffic noise) qualified in just one trial, whereas all other individuals needed two or more (up to the limit of eight) trials to satisfy the entry requirements for the test phase. Lipreading Scores as a Function of Training and Sex: Sources of Variation Between Subjects Table 4 represents a three-way, fixed-effects analysis of variance (training by sex by measure) with repeated observations on the the last factor. The data contained in this table are discussed separately as a function of the treatment main effects for between subjects (factor A: training; factor B: sex) and their interaction (factor AB: training by sex) and as a function of the treatment main effect for within subjects (factor C: measure) and its appropriate ‘interactions. Factor A, the treatment effect for training, was not found Significant. In addition, its interaction with sex (factor AB) was nonsignificant. However, the treatment main effect for sex (factor B) was significant at the .0747 level. Table 5 summarizes the mean lipreading scores (in per cent) as a function of training condition, sex and test measure. Also pro- vided are the combined cell, column, row, and grand means. Note that the total means for sex (men, 88.60 per cent; women, 92.58 per cent) resulted in a difference which favored the female subjects. This separation between means, as indicated above, was statistically signi- ficant at the .0747 level. Table 6 contains the raw score data (in per cent) and is grouped by training condition and sex as a function of the five test SUMMARY OF THE THREE-WAY, FIXED-EFFECTS ANALYSIS 41 TABLE 4 OF VARIANCE WITH REPEATED OBSERVATIONS ON THE TEST MEASURES .. 2222:. ”22:222.“ .2222. 2222222222. Between Subjects 10156.78 39 A (Training) 641.68 4 160.42 0.69 0.6033 B (Sex) 792.02 1 792.02 3.41 0.0747 AB (Training x Sex) 1766.88 4 441.72 1.90 0.1363 Error (Between) 6956.20 30 231.87 Within Subjects 3041.60 160 C (Measure) 252.68 4 63.17 5.31 0.0006 AC (Training x Measure) 807.52 16 50.47 4.25 0.0000 BC (Sex x Measure) 119.48 4 29.87 2.51 0.0454 ABC (Train. x Sex x Meas.) 435.12 16 27.20 2.28 0.0061 Error (Within) 1426.80 120 11.89 42 coo: vcmho enammoz umOH mm.om mm.~m me.om om.om mo.mm oo.om memo: ceaaou wm.~m oo.em oe.~m om.~m o~.om oo.em mamaom oo.mw oH.~m om.wm om.mm om.nm om.nw can: mn.mm oo.nw oo.mw m~.Hm m~.mm mm.ww ewe: uocwneou o~.om om.om om.vm oo.mm oo.nm oo.mm caused om.Hm om.nn om.mw om.em om.mn om.Hm ode: ammo: ow.mm mn.nm oo.om om.mm om.nm mm.em ado: voaaoaou o~.mm om.wm om.Hm oo.mm om.em om.om oaeaoa emfloz oe.~m oo.nm om.wm oo.~m om.~m oo.~m can: cummmne o~.Hm om.mm m~.oa mn.mm om.~m oo.om new: pocwneou oH.om oo.Hm om.mw om.nm oo.~m om.om caesom omwoz om.~m oo.om oo.~m oo.~a oo.mm om.mw can: Hefiuomoch on.mm mn.~m m~.om om.ow mn.om m~.Nm coo: vocaoaou oo.~m om.vm om.nm oo.~m oo.om oo.mm oHaaoa ov.nw oo.~m oo.nm oo.~w om.nw om.mw ode: cannon om.mm mn.am mu.~m om.om m~.vm m~.wm one: eocwoaou ov.mm om.mw oo.mm oo.om om.~w oo.mm nausea oo.mm oo.vm cm.~m oo.~m oo.uw om.mw can: paged cowuapcou . Omaoz omaoz acacaenh Ohms: . . cannon guano . . . dummeuh Heuuumsch . coaufivcoo mcdoz 30m xom ucwcwdhh change: amok . 328 58%: Sm... oz< mm 20:. m2 oszH ommuoovw> one meowoupcoo omfloc pcsouuxoomv scammom amok cowmmom cucumae mzommmmm meh oz< oszHogm2m mhumhmam mo gnome mu mm oz< wonPHozou mmHoz ozaomuxu