CHlLDREN'S PERCEPTION 0F TEMPORALLY DlSTORTED SENTENTIAL APPRGXEMATIONS AND NORMAL SBJTERCES Thesis far the Degree of M. A. memsm STATE UNIVERSITY ANNE K. FLAHERTY . 1974 THFSIS 1]. "y fibmbn “ - ‘4”an E E": 5- " . 11' 1,; {i _‘d ' 2 av Q! h H v 13' ‘ . I .J k ;\L 98. Thesis MMA. Flaherty, Anne K. Children's Percebtion of Temporally . Distorted Sentential Approximations Normal Sentences andE ‘5' 1 w BINDING av 3' HUM} & 8096' 300K BINDERY INB. UBRARY BINDE RS gramaropy. mommy! w“ A ”g.’ RETURNING MATERIALS: )V‘ESI_J PIace in book drop to LJBRARJES remove this checkout from your record. FINES wiII ‘— be charged if book is returned after the date stamped beIow. ABSTRACT CHILDREN'S PERCEPTION OF TEMPORALLY DISTORTED SENTENTIAL APPROXIMATIONS. AND NORMAL SENTENCES BY Anne K. Flaherty A review of the literature suggested that auditory perceptual processing and short-term memory interact with each other and are temporally biased. One way to investi- gate this concept is to control stimulus duration and vary the silent interstimulus interval in an auditory recall task. The stimuli used in auditory recall tasks have been primarily digits and word lists, which do not adequately assess higher central nervous system processing and the functions of short-term memory. It is necessary to use more syntactically and semantically meaningful material. The purpose of this study was to explore the under- lying system involved in perceptually processing speech stimuli by imposing syntactic and semantic constraints upon stimuli of a sentential nature. This study investi- gated the effects of modified silent interstimulus interval ratio of sentential approximations to normal English upon Anne K. Flaherty the auditory perceptual processing of normal second and fourth grade children. In order to increase syntactic and semantic com- plexity, ten three-word and ten five-word first order sentential approximations, ten three-word and ten five- word second order sentential approximations, and ten three- word and ten five-word normal sentences were constructed. The sentential approximations and normal sentences were read by a male speaker from a Second Level standardized reading list of monosyllabic words. These forty sentential approximations and twenty normal sentences were made into three experimental conditions, each condition with a specific silent interstimulus interval size. Word duration remained constant (normal speaking rate) and three inter- stimulus interval sizes (200msec, 400msec, and unaltered) were used. Sixty second and fourth grade children from an elementary school in southern Michigan served as subjects. All subjects had normal hearing as assessed by school screening tests. Each of the three experimental condi- tions--sixty sentences--was presented via a tape recorder under earphones to ten second graders and ten fourth graders. The experimenter using standardized instructions asked the subjects to repeat exactly what they heard on the tape. Anne K. Flaherty An item error analysis was performed on the total number of words recalled for the sixty "sentences." Any words substituted, deleted, or in wrong order were con- sidered in error. The results of this study demonstrated that grade, silent interstimulus interval, order of approximation of stimuli to full grammaticality, and sentence length were important factors in auditory perception of speech and language. Recall performance, on sentential stimuli, of the fourth grade group was substantially higher than the second grade group for all three silent interstimulus interval conditions. For both grades there was a small difference as a function of interstimulus interval size, with the unaltered condition showing the highest recall accuracy and the 400msec condition showing the poorest recall accuracy. For all three interstimulus interval conditions the children in both grades recalled three-word "sentences" more accurately than five-word "sentences." The effect of sentential order revealed that highest recall accuracy was obtained on the normal sen- tences, whereas the first order sentential approximations showed the lowest recall accuracy. These data supported the contention that increasing the order of sentential approximations provided more cues to perception, thereby aiding recall. The interstimulus interval by sentence length by order interaction reveals that for all Anne K. Flaherty interstimulus interval conditions, there were only slight differences for the three-word sentences as a function of order of approximation. However, for five-word sentences, there were substantial differences in percent correct performance as a function of order of approximation. The interaction between sentence length and order demonstrated a breakdown in performance for the first and second order conditions for the five-word sentences, whereas essen- tially no breakdown in performance as a function of order was found for three-word sentences. The interaction between grade, interstimulus interval, and sentence length showed that both second and fourth graders maintained high scores for all three inter- stimulus interval conditons at the three-word sentence length condition. However, the children in both grades showed a reduction in accuracy of performance on five-word sentences for all three interstimulus interval conditions. Finally, for all measures, as grade level increased, percentage correct scores improved. This investigation sought to study the simultaneous interactions of these variables upon the perceptual analysis of language, especially as related to the contributions of word duration, interstimulus interval, and sentence length embedded in sentential approximations and normal sentences. The findings of this investigation support implications which may be important in discussing theories of normal Anne K. Flaherty development of speech and language and the applications of these theories to clinical situations. CHILDREN'S PERCEPTION OF TEMPORALLY DISTORTED SENTENTIAL APPROXIMATIONS AND NORMAL SENTENCES BY Anne K. Flaherty A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Audiology and Speech Sciences 1974 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 Master of Arts Degree. Guidance Committee: 1' J4 ii ACKNOWLEDGMENTS I wish to express my sincerest gratitude to Dr. Daniel S. Beasley, my thesis director, and to Dr. Leo V. Deal and to Dr. William F. Rintelmann, the members of my guidance committee for their personal and intellectual contributions of time and assistance in the preparation of this thesis. I would also like to express appreciation and special thanks to Mr. Paul Woodley, Principal of Laingsburg Elementary School, and to the students who so willingly participated as subjects in this study. I further thank the teachers and staff who gave their valuable time for the benefit of this study. I graciously thank my family and particularly my dear friend and confidant, Bill, who so willingly and unselfishly gave their time, encouragement, and support throughout this endeavor. iii TABLE OF CONTENTS Page LIST OF TABLES . . . . . . . . . . . . . Vi LIST OF FIGURES . . . . . . . . . . . . Vii Chapter I. INTRODUCTION . . . . . . . . . . . 1 Auditory Perception and Short-Term Memory . l Sentential Stimuli and Short-Term Memory . 4 Previous Investigations Involving Temporally Distorted Stimuli. . . . . 6 Statement of the Problem. . . . . . . 9 II. EXPERIMENTAL PROCEDURES. . . . . . . . 13 Subjects . . . . . . . . . . . . 13 Design and Stimuli. . . . . . . . . l3 Recording and Stimulus Generation Procedures. . . . . . . . . . . 17 Presentation Procedures . . . . . . . 21 Analysis . . . . . . . . . . . . 23 III. RESULTS . . . . . . . . . . . . . 24 Effect of Interstimulus Interval . . . . 26 Effect of Sentential Order . . . . . . 26 Effect of Sentence Length . . . . . . 31 Effect of Grade Level. . . . . . . . 32 ~ IV. DISCUSSION . . . . . . . . . . . . 33 Trends of Prior Investigation . . . . . 33 Interstimulus Interval and Order of Sentential Approximation . . . . . . 35 Sentence Length. . . . . . . . . . 38 iv Chapter Page' Grade Level . . . . . . . . . . . 39 Implications for Therapy and Future Research . . . . . . . . . . . 39 Language Acquisition and Development . . 39 AudiOIOgy O O O O O O O O O O O 42 Implications for Future Analysis of Present Data . . . . . . . . . . 44 V. SUMMARY AND CONCLUSION . . . . . . . . 45 REFERENCES 0 O O O O O O O O O O O O O 49 APPENDICES Appendix A. Revised Word List Basal Vocabulary ON fig g9 (Houghton-Mifflin) (1966) . . . 52 B. First and Second Order Sentential Approximations and Normal Sentences Used in This Study. . . . . . . . . 54 C. Standardized Instructions Given to Subject Prior to Each Experimental Session . . . . . . . . . . . . 58 D. Answer Form Used to Transcribe Subjects' Responses. . . . . . . . . . . . 59 E. Tables of Mean Percentage Correct Scores . . 63 F. Subjects' Raw Scores for Each Experi- mental Condition . . . . . . . . . 68 LIST OF TABLES Table Page 1. Examples of the First and Second Order Sentential Approximations and Normal Sentences used in this study . . . . . . 15 2. Means and standard deviations in Msec of graphic level recordings for the three experimental conditions. . . . . . . . l6 3. Mean number of correct reSponses at each interstimulus interval size for second and fourth grades and First Order and Second Order Sentential Approximations and Normal Sentences. . . . . . . . . 25 4. Overall mean percent correct scores for Grade, Interstimulus Interval, Order of Sentential Approximation and Sentence Length across each of the other conditions . . . . . . . . . . . . 63 5. Mean percentage correct scores for Interstimulus Interval by Grade and Sentence Length . . . . . . . . . . 65 6. Mean percentage correct scores for Interstimulus Interval by Order and ' Sentence Length . . . . . . . . . . 66 7. Mean percentage correct scores for Order by Grade and Sentence Length . . . . . ‘. 67 vi LIST OF FIGURES Figure 1. Mean number of sentences correctly recalled as a function of Grade and Interstimulus Interval for Two Orders of Sentential Approximations and Normal Sentences . . . . . . . 2. Mean number of sentences correctly recalled as a function of Sentence Length and Order . . . . . . . 3. Schematic of recording situation and equipment . . . . . . . . . 4. Experimental Design. . . . . . . 5. Mean percentage correct scores: Grade Level by Interstimulus Interval . . 6. Mean percentage correct scores: Grade Level by Interstimulus Interval and Sentence Length . . . . . . . 7. Mean percentage correct scores: Grade Level by Order of Sentential Approximations. . . . . . . . 8. Mean percentage correct scores: Sentence Length by Interstimulus Interval and Order of Sentential Approximation . vii Page 10 19 22 27 28 29 30 CHAPTER I INTRODUCTION A review of the literature revealed a major interest in the area of speech perception by several inves- tigators. Researchers, in looking for data concerning human information processing, have been concerned with the physical characteristics of the stimuli including such dimensions as duration and time and also with the physio- logical variables of the individual's perceptual system, including such factors as attention, prior experience, and short-term memory. The purpose of this investigation is to study children's auditory perception abilities, as reflected in the influence of temporal distortion of sentential stimuli. Auditory Perception and Short-Term Memory Several descriptions and models of short-term memory have been prOposed (Aaronson, 1967; Broadbent, 1957; Pollack, 1959; Sperling, 1963). Most of these models view memory as a two-stage process. The first stage is an unstable, large capacity storage system having a fast decay time. It is a parallel processing system whereby more than one stimulus item can be dealt with simultaneously. Stage II is characterized by a small storage capacity system with a slower decay system. Stimulus items are received into Stage II in an ordered series from Stage I. Thus, if items arrive in Stage II at too rapid a rate, interference of the perceptual analysis of the message may occur. Rapid decay will occur if stimulus items are left too long in Stage I, again interfering with perceptual analysis. It has been suggested (Aaronson, 1967) that the three parameters of (1) Presentation rate, (2) Interstimulus interval, and (3) Stimulus duration may be manipulated in order to investigate the short-term memory characteristics of auditory perceptual processing. The process of time compressing or expanding the duration of the stimulus and/or the silent interstimulus interval alters the presentation rate. The importance of varying presentation rates in the functioning of auditory short-term memory has been presented by Aaronson (1967). An increase in presentation rate is thought to allow for information processing prior to loss of information from short-term memory shortage. It is thought that decreased presentation rates provide more response time to process the stimuli thereby allowing the utilization of individual memory strategies. Shriner and Daniloff (1970) showed the importance of the interstimulus interval for resynthesis of words by varying the silent interphonemic interval of words and presenting these words to children. In the task employing meaningful stimuli with an increase in the silent inter— phonemic interval, resynthesis abilities of segmented CVC syllables decreased to about a silent interphonemic interval of 200msec, with no further decrease to 400msec intervals. However, the task employing non-meaningful stimuli showed a reverse pattern of no decrement in resynthesis abilities to 200msec and then a decline out to a 400msec interval. These results may be an indication of the importance of semantic constraints in a task requiring auditory perceptual pro- cessing. Aaronson and Sternberg (1964), in holding word duration constant and varying interstimulus interval, found that an increase in presentation rate will yield a decrease in recall accuracy, thereby indicating that interstimulus interval is an important parameter in auditory perceptual processing. Beasley and Shriner (1973) investigated the rela- tionship of word duration and interstimulus interval by covarying the temporal length of these two parameters. These temporal manipulations were performed upon first and second order sentential approximations, and the subjects (normal hearing yound adults) were required to recall the sentential sequences. Results demonstrated that stimulus duration, the silent interstimulus interval size and the order of approximations of stimulus materials to full grammaticality were important factors in auditory perception. Word duration was shown to be an important factor in the recall of sentential stimuli (Beasley and Shriner, 1973). It was found that the number of items correctly recalled increased as word duration increased. The major difference in recall accuracy between first order and second order sentential approximations was at the 300msec word duration level, indicating that 200msec and 400msec are nearing anchor points beyond which no decrease or increase, respectively, in recall accuracy of items would be expected. These same findings may apply to the variable of silent interstimulus interval in a recall task. The importance of the silent interstimulus interval in a task involving recall of sentential stimuli has not been adequately defined. It is necessary then to control word duration and vary the silent interstimulus interval in order to determine its importance in an auditory recall task. Sentential Stimuli and Short-Term Memory The current theoretical bases of short-term memory as related to time must be considered incomplete, partly because empirical investigations have failed to consider sentential stimuli adequately. The traditional auditory perceptual testing paradigm (e.g., as used in audiology) uses stimuli such as clicks, pure tones, digits and word lists. It may be contended that non-sentential acoustic stimuli primarily assess the peripheral mechanism (Bocca and Calearo, 1963; Calearo and Lazzaroni, 1957; Bocca, 1967) and that central nervous system perceptual functioning is not adequately assessed by word lists (Pollack, 1967). Jerger (1960) discussed the principle underlying the con- cept of reducing stimuli redundancy for the purposes of audiological evaluation and referred to this as the "Subtlety Principle." He proposed ". . . that the subtlety of the auditory manifestation increases as the site of lesion progresses from peripheral to central" (p. 4). Jerger included in his theory the Bottleneck Principle where he noted that ". . . the process of a complex auditory stimulus encounters a very real bottleneck in the eighth nerve and lower brain stem" (p. 4). Thus, it may be hypo- thesized that increasingly more subtle stimuli are needed to assess the lower brain stem and cortical functioning of auditory processing. That is, determining normal func- tioning at the cortical level must be accomplished with measures requiring subtle decisions on the part of the listener (Miller, 1965). The term "multiple-cueing," introduced by Harris (1960), described the number of cues available to the listener including time and order characteristics. Adequate assessment of higher order neurological functioning as related to language processing is restricted unless multiple-cueing characteristics of normal language are somehow utilized. Speech perception, in this view charac- terized by redundancy reduction and multiple-cueing, then, becomes related to the broader problem of determining the bases for language perception and comprehension. As long as the perceptual processing of speech and language is studied using words spoken in isolation, the possibility of the existence of larger perceptual units which may modify current theories of auditory perception (Liberman, 1969; Jacobson, Fant, and Halle, 1965; Chomsky and Halle, 1968) and short-term memory (Aaronson, 1967; Broadbent, 1957) would not be revealed. It also may be argued that the study of pure tones and single-word stimuli fail to test the subjects' competence in handling sentential stimuli, since in fact sentences and phrases are the very "core" of oral language processing. Investigators have shown that intelligibility increases as the number of alternatives is restricted in message sets (Pollack, Rubenstein, and Decker, 1960; Pollack and Pickett, 1964). The fact that the subject must draw from an open or undetermined message set is another problem with testing methods currently employed (Speaks and Jerger, 1965). By imposing syntactic and semantic constraints through the use of approximated sentential Stimuli, the subject is able to choose from a narrower range of alternatives. Previous Investigations Involving Temporally Distorted Stimuli In previous investigations involving responses to temporally distorted speech, children (Maki, Beasley, and Orchik, 1973) tended to have more difficulty than adults (Beasley, Schwimmer, and Rintelmann, 1972) with a discrimi- nation measure that required responses to an open message set. Differences in the performance of adults as compared to children in auditory processing may be due to the fact that there is a difference in familiarity with language, as adults have experienced a longer period of language use. It has also been said that possibly an open set message is more affected by reduced temporal redundancy than a closed message set (Maki, Beasley, and Orchik, 1973). A preliminary study with children (Flaherty and Cribbs, 1973) was conducted to investigate the effects of a modified silent interstimulus inteval (200msec) and order of sentential approximations (first and second order) as compared to an unaltered interval. The number of items correctly recalled in the first and second order sentential approximations and normal sentences was analyzed. There were six listeners, three second graders, and three fourth graders. Each listener received all experimental condi- tions. The effect of interstimulus interval in this study suggested that the ability of children to recall normal sentences was consistently more accurate than the experi- mentally treated interval for both orders of sentential approximations and for both grade levels, as shown in Figure l. The results of the Flaherty and Cribbs study also indicated that recall accuracy was aided when the order of _. N (I 0 Mean Number of "Sentences“ Correctly Recalled 5 0 Second Grade D Fourth Grade ‘ \\\\\\\\\\\\\\\\\\\\\\\\\\Vi r? I I 3 g I \\\\\\\\\\\\\\\j\v First Second Normal Sentential Approximations and Normal Sentences Figure I. Mean number of sentences correctly recalled as a function of grade and interstimulus interval and two orders of sentential approximations and normal sen- ences. sentential approximations was increased, approaching that of normal sentences. It appears that recall accuracy improved with an increase in grade level. The effect of sentence length in relation to mean number of words recalled for first and second order senten- tial approximations demonstrated that three-word sentential approximations were more readily recalled than the five- word sentential approximations. For the normal sentence condition, there was no difference in recall accuracy as a function of sentence length, each being recalled with one hundred percent accuracy (see Figure 2). Statement of the Problem There seems to be little doubt that auditory per— ceptual processing is temporally-biased. The exact nature of the temporal factors involved in this processing is as yet undetermined. The current theory of short-term memory, based predominantly upon non-sentential stimuli, cannot be considered complete unless it is extended to include sen- tential stimuli. It is not considered complete because the non-sentential stimuli, such as pure tones, clicks, digits, and word lists do not adequately assess higher central nervous system perceptual functioning. Using sentential stimuli, in order to obtain information concerning higher central nervous system perceptual functioning, Beasley and Shriner (1973) gathered normative data for adults. Mean Number of "Sentences” Correctly Recalled 10 First Order E] Second Order I Normal Sentences A \\\\\\\\‘S Three Word Five Word SENTENCE LENGTH Figure 2. Mean number of sentencescorrectly recalled as a function of sentence length and order. 11 There now exists the need to begin investigation in the same direction with children in order to answer questions concerning their auditory functioning during childhood. It is necessary to determine how normal hearing children respond to the parameters of stimulus duration, interstimulus interval, order of sentential approximations and sentence length, and to determine whether this ability is a function of age and can be considered developmental. Thus, the purpose of this study was to explore the underlying system involved in processing speech stimuli by imposing syntactic and semantic constraints upon the sen- tential input and by simultaneously controlling the amount of time given to select among the sentential stimuli and normal sentences presented. The effects of modified silent interstimulus interval ratios of sentential approxi- mations to normal English upon the auditory perceptual processing of normal second and fourth grade children was investigated. Specifically, the following questions were studied: 1. What would be the recall accuracy of second and fourth grade children for first order and second order sentential approximations and normal sentences? 2. What would be the recall accuracy of second and fourth grade children for three-word and fiVe-word sentential approximations and three-word and five— word normal sentences? 12 What would be the recall accuracy of second and fourth grade children for silent interstimulus intervals of 200msec, 400msec, and ulaltered--an interstimulus interval of normal speaking rate--, respectively. What would be the effects of interactions of the above factors and their associated levels upon recall accuracy? CHAPTER II EXPERIMENTAL PROCEDURES Subjects Thirty second grade and thirty fourth grade children from the Laingsburg Elementary School served as .subjects. All children had been recently screened for hearing by the Intermediate School District, and were judged to have hearing within normal limits (re: ISO, 1964). Each of the three experimental conditions (200msec, 400msec, and unaltered--an interstimulus interval of normal speaking rate) was presented to ten second and ten fourth grade children. (See Figure 4, page 22, for the pictorial representation of the design.) Design and Stimuli The stimulus materials were similar to those used in a previous investigation of recall accuracy by Flaherty and Cribbs (1973). The stimuli consisted of three experi- mental tapes: I (1) Ten three-word and ten five—word first order sen— tential approximations, each with a silent interstimulus interval of 200msec between each word; ten three-word and 13 14 ten five-word second order sentential approximations, each with a silent interstimulus interval of 200msec between each word; ten three-word and ten five-word normal sen- tences, each with a silent interstimulus interval of 200msec between each word. (2) Ten three-word and ten five-word first order sen- tential approximations, each with a silent interstimulus interval of 400msec between each word; ten three-word and ten five-word second order sentential approximations, each with a silent interstimulus interval of 400msec between each word; ten three-word and ten five-word normal sen- tences, each with a silent interstimulus interval of 400msec between each word. (3) Ten three-word and ten five-word first order sen- tential approximations, each with an unaltered interstimulus interval; ten three-word and ten five-word second order sentential approximations, each with an unaltered inter- stimulus interval; ten three-word and ten five-word normal sentences, each with an unaltered interstimulus interval. The sentential approximations to full grammatical sentences were ordered in a manner similar to that described by Speaks and Jerger (1965). For the purpose of this investigation the sentential approximations to full gram- matical sentences were constructed using words taken from the Basal Vocabulary of the primary reader workbook, 9N W§_§Q (Teacher Edition) (Second Level) (Houghton Mifflin, 1966). One hundred monosyllabic words chosen randomly by 15 Table l.--Examples of the First and Second Order Sentential Approximations and Normal Sentences Used in This Study. FIRST ORDER SENTENTIAL APPROXIMATIONS 1 go girl zoo 2 men zoo pop nest is 3 take hot girl 4 red play late I . call 5 food jar wish SECOND ORDER SENTENTIAL APPROXIMATIONS 1 men will do 2 I put up to like 3 that man of 4 here all of stop tree 5 snow time will NORMAL SENTENCES l I will play. 2 We went to the zoo. 3 The cat played. 4 That girl is not good. 5 I knOw you. 16 m.H NN.NH HMEHOZ N.N HM.NH HOUHO GGN 00v mmo. ov.NH HOUHO HMH ems. ma.e Hmsuoz mam. mN.m Hmflho USN OON am 2 aowumfiwxoummd oomz CH oflumm HmH Hm>nmucH msasfiwumnoucH HMflHCOUGOm MO HOUHO .mc0wuaocoo Housmeuomxm mean» on» How mmcflpuoowu Ho>mH ownmmum mo 0mm: cfl mcoflum«>m0 pumocmum ppm mcmozll.m magma 17 the experimenter were taken from the ON WE GO list and randomized into a new list (see Appendix A). For the first order sentences, the words were randomly chosen from the revised list. Ten three-word and ten five-word sentential approximations were constructed to constitute the first order sentential stimuli. Construction of the second order sentences was accomplished by choosing the first word randomly from the revised list. The experimenter then asked another indi- vidual to choose a word from the list that may follow word one in a sentence. Word three was supplied by a second individual who, without knowing word one, chose a word from the list to follow word two. The process was con- tinued until ten three-word and ten five-word "sentences" (sentential approximaions) were constructed by the experi- menter using words from the revised list of words from 9N WE g9. (See Appendix B for a complete listing of the sentential approximations and normal sentences used in this study). An additional undistorted three-word and five-word first order sentential approximation was prepared as prac- tice stimuli. Recording and Stimulus Generation Procedures The experimental stimuli were recorded onto a master magnetic tape recording by a male speaker who spoke General American speech and was trained in phonetics. The speaker 18 used a microphone (Electro-voice 635A) and a tape deck (Ampex AG 440-4) to record the words in a sound treated room. Conversational pitch and effort level with minimum inflection were used in making the recordings. The reason for minimum inflection was to minimize the interaction of prosodic cueing effects that might influence the data. A pause was left between words in order to facilitate the experimenter in the location of the initiation and termina- tion of each word for splicing purposes. The speaker monitored his vocal intensity on a VU meter (which peaked between 0 and -3 dB VU) located in the recording suite. A schematic of the recording situation and equipment is shown in Figure 3. One tape of ten three-word and ten five-word first order sentential approximations, ten three-word and ten five-word second order sentential approximations and ten three-word and ten five-word normal sentences was recorded with an interstimulus interval associated with a normal speaking rate. This tape was unaltered. Two copies were made of this tape at 7 l/Zips using an Ampex 601 and an Ampex 600 tape recorder. Two copies of the master recording of ten three-word and ten five-word first order sentential approximations, ten three-word and ten five-word second order sentential approximations, and ten three-word and ten five-word normal sentences were made at 7 l/Zips using an Ampex 601 and an Ampex 600 tape recorder. Each of these copies of the master recording was prepared as experimental 19 2.6236 25 55.25 05388. E 28628 .n 230E mmomoomm ma 4... ¢ro¢¢ 64 xma2< uzozaomoi mo.o>.. omhom4u <30 20 stimuli by varying the size of the silent interstimulus interval (200msec and 400msec, respectively). The 200msec silent interstimulus interval tape was prepared first. The acoustic initiation and termination point of each word was aurally determined using an apparatus comprised of a play- back recorder head coupled to a pre-amplifier and amplifier/ speaker. These acoustic boundaries were then marked on the tape, and the tape was cut at these boundaries. The existing interval of tape between the word was discarded. A silent interstimulus interval equal to 200msec (determined .to be 1.5 inches of tape) was then spliced between each word of the 200msec c0py. The same procedure was followed for the remaining interstimulus interval condition of 400msec (determined to be 3.0 inches of tape). A response time of five seconds of blank tape was spliced between each sentential stimulus and normal stimulus on the experimental tapes. Each experimental tape was played through a high speed graphic level recorder (Bruel & Kjar 2305) (paper speed-30mm/sec; writing speed-250mm/sec; 50dB; 20Hz). The silent interstimulus intervals between words were hand measured in millimeters and converted into milliseconds. This procedure has been described elsewhere by Beasley and Shriner (1973). Any silent interstimulus interval in error by approximately : 30 milliseconds was reprocessed. The tape which consisted of the twenty first order sentential approximations, the twenty second order 21 sentential approximations, and the twenty normal sentences with an unaltered interstimulus interval between each word provided a baseline for the "sentences" whose interstimulus interval had been modified. The carrier phrase "Number " (which was in fact the number of the sentence) preceded all sentences. Presentation Procedures Each experimental condition presented to a sub- group of twenty children, ten per grade, consisted of ten three-word and ten five-word first order sentential approxi- mations, ten three-word and ten five-word second order sentential approximations, and ten three—word and ten five- word normal sentences with the silent interstimulus interval 0f 200msec and 400msec and unaltered, respectively. The total breakdown can be seen in Figure 4. The listeners were seated in chairs in a room adjacent to the principal's office at the Laingsburg Elementary School. Each child received the experimental tape binaurally at 7 l/2ips via a tape recorder (Ampex 600) and via earphones (TDH 39-10Z) housed in biscuit type cushions (MX-4l/AR). The intensity level was set at 70 to 75 dB sound pressure level (SPL) (re: .0002 dynes/cmz). There was occasional peaking at 80 dB SPL. The ambient noise level in the test room was measured at 60 to 65 dB SPL on the C scale of a sound level meter (Bruel and Kjar Type 2203) using a sound field con- densor microphone (Bruel and Kjar Type 4131). This ambient 22 6239 3225.236 .e 050E whom—63m u: N .2 mun—mo “9.2.25 o: N «a. mun—mo . .038 00¢ 2 new 3. muomo .03.: OON 329 5.50“. 326 2.83 23 noise level was_sufficiently low as not to interfere with the subjects' listening task. After a brief greeting period, standardized verbal instructions were given to each subject (see Appendix C). If there were questions, the instructions were repeated. No subject required more than one repetition of the instruc- tions. Each subject was tested individually. The responses of the subjects were written by the examiner on an answer sheet (see Appendix D). The sub- jeCt's responses were also recorded via a tape recorder (Sony TC 106A) and played back at a later time in order to allow the examiner to check her original written record against the playback tape. Analysis The number of items correctly recalled was the score for each subject. There were ten subjects (second and fourth grade children, respectively) per condition for a total of sixty subjects (see Figure 4). There were twenty subjects at each silent interstimulus interval size (200msec and 400msec) and twenty subjects at the unaltered inter? stimulus interval condition (N). CHAPTER III RESULTS The results of this study support the thesis that speech perception can be investigated by imposing semantic, syntactic, and temporal constraints upon sequential stimuli. The overall results demonstrated that grade, silent inter- stimulus interval, order of approximation of stimuli to full grammaticality, and sentence length were important factors in auditory perception of speech and language. It was found that as order of sentential approximations to full grammatical sentences increased, recall accuracy increased. Recall accuracy scores decreased as sentence length increased from three words to five words. As the silent interstimulus interval ratios increased the number of items correctly recalled decreased. For all measures, as grade level increased, percentage correct scores improved. The mean correct score at each silent inter- stimulus interval size for the second and fourth grades for both orders of sentential approximations and normal sen- tences is presented in Table 4. These results, discussed 24 25 .om ou Honda muoom manfimmom Essflxmfi m spas .mmocmucom auxwm ou muomnndm so» «0 noncommou on» mucomoumou momum>m Somme n.ma m.mH m.mH Hmuoa o.om m.ma m.ma HaEhoz v.ma N.va m.vH Homuo cam suusom n.na m.mH N.¢H nacho and o.na m.ma m.oa Hmuoe m.ma o.o~ «.ma Hmfiuoz m.na n.ma h.ma nacho can pcooom m.ma m.~a m.va nacho umH cmumuamcs oov com Ho>mq momma AommEV oNflm Hm>umucH mSHDEHumeucH «.moocmucmm HmEHoc new mcoflumexoummm aneucmucom nacho pcooow new nacho umuwm 0cm mopmum QUHDOM mam mcoomm you ouwm Hm>umucfi msHsEflumuoucfl some um mochQmou uomuuoo mo Hones: cmozll.m canoe 26 below, can be found in Appendix E and graphically illus- trated in Figures 5-8. Effect of Interstimulus Interval Figure 5 shows that the recall performance of the fourth grade group was substantially higher than the second grade group for all three interstimulus interval conditions (200msec, 400msec, ulaltered). For both grades there was a small difference as a function of interstimulus interval, with the unaltered conditon showing the highest recall accuracy and the 400msec condition showing the poorest recall accuracy. Figure 6 illustrates that for all three interstimulus interval conditons, both the second and fourth graders recalled three-word sentences more accur- ately than the five-word sentences. Figure 6 also illus- trates that in the 400msec condition and the unaltered condition the fourth graders recalled five-word sentences with more accuracy than the second graders. Effect of Sentential Order Figure 7 shows that for both the second and fourth grades there was a small difference in performance as a function of both orders of sentential approximations and normal sentences. The highest recall accuracy was obtained on the normal sentences, whereas the first order sentential approximations showed the lowest recall accuracy. These data support the contention that increasing the order of 27 IOO J ‘ 200 msec. D ] 7O 60 (8 O PERCENT CORRECT 01 0 IO a. O 400 msec. . Unaltered z '9 Second Grade Fourth Grade 'Figure 5. Mean percentage correct scores: Grade level (second grade, fourth grade) by interstimulus interval (200 msec., 400 msec., unaltered). 28 I00 200 msec. D 90 400 msec. I V Unaltered ‘ \\\\\\\‘ PERCENT CORRECT «b 0| 0 O 01 O 20 IO Second Grade Fourth Grade Second Grade Fourth Grade THREE WORD FIVE WORD Figure 6. Mean percentage correct scores: Grade level (second grade, fourth grade) by interstimulus interval (200 msec., 400 msec., unaltered) and sentence length (three word, five word). PERCENT CORRECT I00 90 80 7O 60 50 4o 30 20 IO 0 29 First order U Second order I 7 Normal ‘ Second Grade Fourth Grade Figure 7. Mean percentage correct scores: Grade level (second grade, fourth grade) by order of sentential approximation (first order, second order, normal sen- tences). 30 .3352: 66.5: .395 2.03» .396 ES V .2502 SEEREaao .220» E 396 ecu 822.0 s is coo»... 00¢ roe»... 003 6225 8.26:» L.095 Ucooom .520 «new... 1.32.: an Abba! 0): 4:03 00.5.: «0:0. 00:0. - D .50 6200» 89:3 32523 So: .0 8:3“. 0.53 n 903 m use: m P6: n P5: m 9.03 n 0 O O IN 0 to Q' 1038800 .LNBOHSCI O O k\\\\\\\\\\\ \\\\\\\\ \. \ 09.2.25 .oonE 00¢ .oomE CON 31 sentential approximation provided more cues to perception (multiple-cueing effect), thereby aiding recall. Figures 8 a, b, and c display the interaction between interstimulus interval, sentence length, and order of approximations. For all interstimulus interval condi- tions (200msec, 400msec, unaltered), there were only slight differences for the three-word sentence as a function of order of approximation. However, for five-word sen— tences, there were substantial differences in percent correct performance as a function of order of approximation. It can be seen in Figures 8 a, b, and c that as a function of order of approximation, the mean scores for performance on five-word sentences rank highest to lowest as follows: normal sentences; second order sentential approximations; first order sentential approximations. These results were found for all three interstimulus interval conditions but were more dramatic for the 200msec and 400msec condition than for the unaltered condition. In comparing three-word and five-word sentence length, this figure demonstrates a breakdown in performance for the first and second order conditions for the five-word sentences, whereas essentially no breakdown in performance as a function of order was found for three-word sentences. Effect of Sentence Length Figure 6 shows the interaction between grade, interstimulus interval, and sentence length. This figure 32 shows that both the second and fourth grade children main— tained high percentage scores for all three interstimulus interval conditions (200msec, 400msec, unaltered), at the three-word sentence length condition. However, the children in both grades showed a reduction in accuracy of performance at the five-word sentence length condition for all three interstimulus interval conditions, with the lowest mean scores associated with the 400msec condition and the highest mean scores associated with the unaltered condition. Finally, this figure shows a small difference in percent correct scores as a function of grade level for all three interstimulus interval conditions for the five-word sentence length condition. The fourth graders showed a slightly higher performance score than the second graders for the 400msec and unaltered interstimulus interval conditions. No differences were found between grades for the 200msec interstimulus interval condition. Effect of Grade Level The effect of grade level as it interacts with all variables has been discussed in the above text. The inter- actions discussed above followed the general trend for grade level comparison: as grade level increased, errors decreased. CHAPTER IV DISCUSSION The results of this study indicated that grade level, silent interstimulus interval size, order of approximations to full grammaticality, and sentence length interacted at various levels. It can be seen that specific variables in certain instances took precedence over others in affecting the recall accuracy of children in this study. Earlier investigators had not studied the simul- taneous interactions of these variables upon the perceptual analysis of language, especially as related to the contri- butions of word duration, interstimulus interval, and sentence length embedded in sentential approximations and normal sentences. These findings support implications which may be important in discussing theories of normal development of language and the applications of these theories to clinical situations. Trends of Prior Investigation In reviewing the variables independently of each other, the results of this study support earlier 33 34 investigations. No previous studies had combined the fac- tors used in this study using school age children. However, the same trends have been found for preschool children and school age children for an imitative or recall task. The trend for the number of items correctly recalled to decrease as the size of the silent interstimulus interval increased supported the findings of earlier studies employing the use of silent interstimulus interval ratios (Beasley and Shriner, 1972; Beasley and Beasley, 1973). The fact that recall accuracy increased as the order of sentential approximations to full grammatical sentences increased agrees with earlier studies (Speaks and Jerger, 1965; Beasley and Shriner, 1972). The higher order sen- tential approximations approaching full grammaticality involve syntactic-semantic concepts (words) which provide multiple-cueing effects to aid in auditory perceptual processing. The trend for recall accuracy to decrease as sen- tence length increased supports the findings of earlier studies of sentence length, item length, and short-term memory (Miller, 1965; Aaronson, 1967; Schuckers, Shriner, and Daniloff, 1971; Pantalos, Schuckers, and Hipskind, 1972). When a string of items (words or digits) has exceeded the immediate or short-term memory span of an individual, that individual tends to delete items from the string. Miller (1956) found that the short-term memory span was usually seven "items" long, plus or minus two. 35 The present study showed that more words were recalled incorrectly from five-word sequences than three-word sequences. Finally, recall accuracy increased as grade level increased (Beasley and Acker, 1971; Smith, 1972; Beasley and Beasley, 1973). Interstimulus Interval and Order of Sentential Approximation Interstimulus interval size is a major parameter in the perceptual analysis of auditory input according to Aaronson (1967). She states that as interstimulus interval increases, more time is allowed for processing. Interstimulus interval, an artifically introduced variable, was shown to play a significant part in this study. Figure 5 shows that the number of items correctly recalled did not increase as the interstimulus interval size increased. It may be seen in Figure 5 that the unaltered interstimulus interval condition (interstimulus interval of normal speaking rate) showed the highest recall accuracy and the 400msec interstimulus interval condition showed the poorest recall accuracy. This may indicate that there is no optimal interstimulus interval ration size for' syntactically and semantically constrained stimuli.r It can be seen that the effect of interstimulus interval is not crucial as the stimuli approach full grammaticality. The higher order sentential approximations and normal sentences involving syntactic-semantic concepts, provide multiple- cueing effects which aid auditory perceptual processing. 36 Subjects have less time to analyze words separately as multiple-cueing increases. They tend to chunk the words at this level (Miller, 1962). As the order is increased, the material is analyzed in longer chunked units; and it even appears that with more familiar material (normal sentences) the listener may identify several words at a time. The verb phrase, around which the rest of the sen- tence revolves (Foder, Bever, and Garrett, 1969), has been discussed as the basic unit of perception. The increase in the linguistic aspect of the stimuli increases the complexity of the sentential material.. However, when linguistic constraints are increased, there are increased multiple cues used for perceptual analysis. The processing of lower order sentential material would result in a larger number of recall errors than higher orders, as the lower orders of sentential approximations are not as linguistically bound as higher orders. This concept can be seen in the results of this study. The highest recall accuracy was obtained on the normal sentences, whereas the first order sentential approximations showed the lowest recall accuracy. The interstimulus interval does not affect the ability to discriminate order if the events carry high information content (such as words in sentences). This study refutes the contention that decreasing the inter- stimulus interval results in a decrease in the ability of children to discriminate order as the events occur in close 37 succession. The results showed that poorest recall accuracy occurred at the longest interstimulus interval condition, i.e., 40msec and the highest recall accuracy occurred at the unaltered interstimulus interval, determined to be the shortest interstimulus interval condition. It is contended that temporal factors, such as interstimulus interval, may play a dominant role in the analysis of auditory input at the lower, more peripheral neural centers. As input becomes more sentential in nature, it is analyzed at a higher more central neural level. At this point, time between elements of the stimulus material plays a less significant role. It is here that order of approximation to full grammaticality becomes a significant factor in the correct analysis of the input. The relationship between grade and interstimulus interval size, which showed that the recall accuracy per- formance of the fourth graders was substantially higher than the recall accuracy performance of the second graders for all three interstimulus interval conditions, suggests per- ceptual processing is developmental relative to the para- meters investigated. Perhaps, this performance is due to the difference in the experience of the two age groups. The older children (fourth grade) have experienced a longer period of language use and are more familiar with the language than are the younger children (second grade). These older children are therefore able to recall language (words, sentences) with more accuracy. 38 It is indicated from the data that the fourth graders appear to process higher order stimuli which carry syntactic and semantic cues more proficiently than the second graders. This suggests that older children are developing a more sophisticated perceptual processing system. At this time these older children are beginning to analyze the more complex stimuli in "chunked" groups as opposed to each word independently. Sentence Length The results indicated that recall accuracy decreased as sentence length increased. It was evident that there was a greater number of recall errors on the five-word sequences (sentential approximations and normal sentences) than on the three-word sequences. When sentence length increased to five items, the increase in number of recall errors may have indicated that short-term memory capacity and/or processing capacity had been exceeded. When immediate memory limits have been exceeded, recall becomes selective and must reflect strategies used by the listener in processing the information. Further support that five-word sequences may have exceeded short-term memory span can be seen when counting the total number of sentential approximation and normal sentence deletions. Thirty-five five-word "sentences" were completely deleted, whereas only eight three-word "sentences" were completely deleted. The strategies used by individuals in selecting 39 words to recall may very well be reflected by the words subjects recalled. The relationship between grade and sentence length reflects upon a developmental process for short-term memory. The capacity of short-term memory imposed greater limits upon second graders than upon fourth graders. Thus, it appears that more efficient use of short-term memory increased with age. It appears from the data that short- term memory processing is limited at the second grade level. However, further research on older children would be necessary before it could be determined if a peaking and/or leveling for the development of short-term memory had been achieved by the fourth grade. Grade Level When interactions involving grade level were exam- ined, there were no exceptions to the trend. For all measures, as grade level increased, percentage correct scores improved. Implications for Therapy and Future Research Language Acquisition and Development It may be hypothesized that language acquisition is based upon the establishment of the temporal patterns of perceptual processing (Berry, 1969). The ability to analyze sequential auditory events must be firmly 40 established on a temporal basis in order for the language perceptual processor to function optimally. Language acquisition begins when this is made possible. It may be stated that before language acquisition could occur, the cortex would have to be at a certain level of maturation. Studies involving children with "delayed language" have suggested that delay may be related to malfunctioning "auditory memory." These children are said to have impeded memory spans, and their ability to sequence information is hampered resulting in distorted syntactic relations. Katz (1971) has set forth the hypothesis that the problem with children displaying depressed speech and language skills may lie in the disruption of clear auditory perception. One important skill in auditory perception is phonemic synthesis. Katz refers to phonemic synthesis as "the task of remembering individual phonemes; keeping them in correct sequence; blending the individual items together and repeating the word that is formed by these elements" (p. 19). The author indicated that the reason a person might fall down in a phonemic synthesis task is that such a task taps several auditory perceptual functions of the brain. In his research, Katz had no definite control of the parameters used in this phonemic synthesis task. It is necessary to consider the interstimulus interval and set up definite parameters to test this concept. With no such parameters (i.e., interstimulus interval) set up in this 41 type of task, it is difficult to determine whether the child is simply memorizing the words or perceptually synthesizing the words. The hypothesis of interrelated perceptual-productive disorders (e.g., children with articulatory deficits have associated language difficulties) has been set forth by Shriner and Daniloff (1970). These authors noted that normal children do poorer resynthesizing "meaningless" silent interval spanned consonant-vowel-consonant words than "meaningful" consonant-vowel-consonant words. They suggest that the child applies semantic and syntactic rules to the "meaningful" stimuli but does not do so for the "meaningless" stimuli. The authors indicated that the more systems brought into play (such as syntax, semantics, etc.), the easier the task becomes for the child, facili- tating short-term memory recall. Implications for therapy and for teachers arise from the second grade limitation on short-term memory. It appears that the child is only processing shorter units effectively in any learning or recall task. This may give evidence to the difficulty younger children have in follow- ing directions. If a child is limited in his processing of longer sentences, it may follow that, with repetition, the child can eventually acquire meaning out of a longer sentence. Language therapy should not fail to consider the experience of the child and the familiarity of the child 42 with regard to language acquisition. These factors may play an important role in the process of a child's developing language. They also may help indicate the sophistication of the child's perceptual processing system. Existing in the area of mental retardation is a great need for guidance in speech and language programs. In regards to this study, a comparison should be made between the trends found in the normal population and the trends characterized by the mentally retarded population. If similar patterns are found, programs for the retarded could parallel regular language programs. However, language therapy techniques which optimally aid the retarded indi- vidual's perceptual processing and recall behavior should be employed where abnormal or no trends were found. Audiology It has been suggested that the use of distorted speech tests may contribute to what is known of the inte- grative aspects of message processing (Bocca and Calearo, 1963; Bocca, 1967). Simple psychoacoustic signals, such as pure tones, fail to elicit differential responses of the central nervous system; that is, central nervous system pathways provide enough intrinsic redundancy to allow satisfactory cortical integrative and interpretive centers to act upon these simple signals. In order to assess these higher centers, the use of distorted speech which increases the complexity of the message and reduces the extrinsic 43 redundancy of this complex message is advocated. When such a distorted message is passed into a damaged pathway, per- ceptual processing will suffer. It is also suggested that if verbal integration is to be studied, meaningful senten— tial stimuli must be used. This is because highly redundant messages will be least affected by peripheral lesions (middle ear, cochlear, and eighth nerve pathologies). Sentential approximations, variously modified, may serve the purpose of assessing processing of higher cortical centers. Simply distorting word lists may not be suffi- cient. Audiological diagnostic and prognostic testing seeks to appraise a patient's auditory perceptual processing of everyday speech. It is reported that word lists are not adequate for this purpose (Webster et al., 1965). In order to overcome these problems of assessment, it may be neces- sary to use verbal sequences with semantic and syntactic constraints in audiological testing. The degree of semantic and syntactic constraints placed upon the stimuli must be limited in order that the sentential sequences are not full grammatical sentences. It is known that as the sentential sequences approach that of normal language, the inherent redundancy makes the material too easy for testing purposes. Also known is that distorted stimuli require the listener to respond to an open message set, which makes the material too difficult. A compromise may be met by the use of first and second order sentential approximations. Varying the interstimulus interval size 44 of these sentential approximations may provide a method of assisting the differentiation of listeners who have lesions of the central nervous system from listeners with normal hearing. Implications for Future Analysis of Present Data In order to obtain knowledge of the interactions of the variables under investigation, it was necessary to score "sentences" as correct or incorrect. However, infor— mation about the syntactic and semantic mispercpetions may possibly be apparent from further analysis of various recall errors. For example, an analysis of item errors, namely those errors on different parts of speech may give some idea which parts of speech are developed in language first and which are recalled with the greatest accuracy. The same type of analysis could be conducted for word sub- stitutions in relationship to semantics and order errors in relationship to syntax. CHAPTER V SUMMARY AND CONCLUSION Numerous investigators have reported that the per- ceptual processing of auditory stimuli is influenced by temporal factors. The exact nature of this process is as yet undetermined. The current theoretical bases of short- term memory as related to time must be considered incom- plete, partly because empirical investigations have failed to consider sentential stimuli adequately. It was contended that pure tones, clicks, digits, and word lists did not adequately assess higher central nervous system functioning. Rather, adequate assessment of higher order neurological functioning as related to language processing is restricted unless multiple language cues are utilized. This requires imposing semantic and syntactic constraints upon the stimuli which is accomplished through the use of first and second order approximations to full grammatical English sentences. The purpose of this study was to explore the system involved in processing speech stimuli by imposing 45 46 semantic and syntactic constraints upon the sentential input and at the same time controlling the amount of time given to select among the sentential stimuli and normal stimuli presented. The investigation studied the effects of modified interstimulus interval ratios of sentential approximations to normal English upon the auditory per- ceptual processing of normal children. Hopefully, the results may be applied to the current theories and models of auditory perception, particularly in terms of the acquisition and development of language perception and comprehension. Specifically, the following questions were inves- tigated: 1. What would be the recall accuracy of second and fourth grade children for first order and second order sentential approximations and normal sene tences? 2. What would be the recall accuracy of second and fourth grade children for three-word and five-word sentential approximations and three-word and five- word normal sentences? 3. What would be the recall accuracy of second and fourth grade children for silent interstimulus intervals of 200msec, 400msec, and unaltered, respectively? 47 4. What would be the effects of interactions of the above factors and their associated levels upon recall accuracy? The overall results demonstrated that interstimulus interval, order of approximation of stimuli to full gram— maticality, sentence length, and grade were important factors in auditory perception of speech and language. As the silent interstimulus interval ratios increased, the number of items correctly recalled decreased. As the order of sentential approximations to full grammatical sentences increased, recall accuracy increased. Recall accuracy scores decreased as sentence length increased from three words to five words. For all measures, as grade increased, percentage correct scores improved. Recall performance of the fourth grade group was substantially higher than the second grade group for all three silent interstimulus interval conditions (200msec, 400msec, and unaltered). The effect of sentential order revealed highest recall accuracy was obtained on the normal sentences and lowest recall accuracy was obtained on the first order sentential approximations. Children in both grades (second and fourth) main- tained a high level of recall performance at the three-word sentence length condition, but the children in both grades showed a reduction in accuracy of performance at the five- word sentence length condition. 48 Results of grade level performance indicate a need for further research. Although the total error rate dif- ferences of second and fourth graders were minimal, it is necessary to test the recall ability of older children on this present task in order to determine the grade level at which this decreasing error rate stabilizes. REFERENCES REFERENCES Aaronson, D. 1967. Temporal factors in perception and short-term memory. Psychol. Bull., 67, 130-44. Aaronson, D., and Sternberg, S. 1964. Effects of pre— sentation rate and signal-to-noise ratio on immediate recall. Paper presented at Eastern Psychol. Assoc., Philadelphia. Beasley, D., and Acker, J. 1971. The effect of sentence length, sentence type, word type, stress on four and five year olds. Unpublished study, Michigan State University. Beasley, D.; Schwimmer, S.; and Rintelmann, W. Intelligi-A bility of time-compressed CNC monosyllables. 1. Speech and Hearing Res., 15, 340-50. Beasley, D., and Shriner, T. 1973. Auditory analysis of temporally distorted sentential approximations. Audiology, 12, 262—71. Berry, M. 1969. Language Disorders pf Children: The Bases and Diagnosis (Chapter 3). Appleton-Century-Crofts: New York. Bocca, E. 1967. Distorted speech tests. In Sensorineural Hearing Processes and Disorders. ed. by A. B. Graham. Little, Brown and Co.: Boston. Bocca, E., and Calearo, C. 1963. Central hearing pro- cesses. In Modern Develppments ip Audiology. ed. by J. Jerger. Academic Press: New York. Broadbent, D. 1957. A mechanical model for human attention and immediate memory. Psychol. Rev., 64, 205-15. 49 50 Chomsky, N., and Halle, M. 1968. The Sound Pattern pf English. Harper and Row: New York. Flaherty, A., and Cribbs, J. 1973. Children's perceptions of temporally distorted sentential approximations. (Pilot Study) Unpublished study, Michigan State University. Foder, J.; Bever, T.; and Garrett, M. 1969. The develop- ment of psychological models for speech recogni- tion. USAF Contract No. 19 (628)—5705. L. G. Hanscom Field: Bedford. Harris, J. 1960. Combinations of distorted speech: Twenty- five percent safety factor by multiple cueing. Arch. pf Otolaryngol., 72, 227-32. Jacobson, R.; Fant, G.; and Halle, M. 1965. Preliminaries 59 Speech Analysis. MIT Press: Boston. Jerger, J. 1960. Audiological manifestations of lesions in the auditory nervous system. Lapyngoscope, 70, 417-25. Katz, J. 1971. Auditory perception training for children with learning disabilities. Menorah Medical Jour- nal, 2, 18-29. Maki, J.; Beasley, D.; and Orchik, D. 1973. Children's perception of time-compressed speech using two measures of speech discrimination. Unpublished study, Michigan State University. McKee, P. et a1. 1966. Basal Vocabulary from QN_W§ g9, Teacher's Guide, Houghton-Mifflin Co.: Boston. Miller, G. 1956. The magical number seven, plus or minus two: Some limits on our capacity for processing information. Psychol. Rev., 63, 81-97. Pantalos, J.; Schukers, G.; and Hipslind, N. 1973. Sen- tence length-duration relationships in an auditory assembly task. Submitted to Journal pf Communica- tion Disorders. Pollack, 1.; Johnson, L.; and Knapf, P. 1959. Running memory span. J. Exp. Psychol., 137-46. Pollack, 1.; Rubenstein, H.; and Decker, L. 1960. Analysis of incorrect responses to an unknown message set. J, Acoust. Soc. America, 32, 454-58. 51 Pollack, I., and Pickett, J. 1964. Intelligibility of excerpts from fluent speech: Auditory vs. Struc- tural context. JJ. Verbal Learn. Verbal Behav., 3, 79-84. Pollack, I. 1967. Language as Behavior. In Brain Mechanisms Underlying Speech and Language. ed. by C. Millikan and F. Darley. Grune and Stratton: New York. Schuckers, G.; Shriner, T.; and Daniloff, R. 1973. Audi- tory assembly of segmented sentences by children. J. Speech and Hearing Res., 16, 116-27. Shriner, T., and Daniloff, R. Reassembly of segmented CVC syllables by children. J, Speech and Hearing Res., 13. Smith, A. 1972. Short-term memory and imitation: Effects of sentence length, sentence type, word type, stress and grade. Unpublished study, Michigan State University. Speaks, C., and Jerger, L. 1965. Method for measurement for speech identification. J. Speech and Hearing Res., 8, 185-94. Sperling, G. 1963. A model for visual memory tasks. Human Factors, 5, 19-31. Webster, J.; Davis, H.; and Ward, W. 1965. Everyday speech intelligibility. J. Acoust. Soc. America, 38, 668. APPENDICES APPENDIX A REVISED WORD LIST BASAL VOCABULARY pg pp pg HOUGHTON-MIFFLIN (1966) APPENDIX A REVISED WORD LIST BASAL VOCABULARY pg pp pg HOUGTON-MIFFLIN ONU'I \I 10 11 12 13 14 15 16 17 (1966) a red pan sand girl zoo high stop cry with eye was all ten dog late boy set feet jar but is like come had in my six her to not men I we milk ring dot pop that said man take toy they out will wish lock on me sat see put hot good snow tell the it food see eat ride dish sun do play 90 52 18 19 20 21 22 23 24 25 two of here box big car five day 53 call cat him up us you nest store wood yes no name end door run one 699 know work word tree this time would APPENDIX B FIRST AND SECOND ORDER SENTENTIAL APPROXIMATIONS AND NORMAL SENTENCES USED IN THIS STUDY 11. 12. 13. 14. 15. 16. go girl men take red food snow jar name red time wish snow egg Pop up tree up yes zoo hot play jar tak sand mil toy all Pop Pop call 200 jar APPENDIX B FIRST AND SECOND ORDER SENTENTIAL APPROXIMATIONS AND NORMAL SENTENCES USED IN THIS STUDY First Order Sentential Approximations zoo pop nest is girl late I call wish e play big man man k good girl sun us sun pop tree milk dish is name ten cry food milk said me word 54 17. 18. 19. 20. 55 nest ring sun me play wish milk nest name will jar dog cat high man hot 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. that 56 Second Order Sentential Approximations men will do I put up to like man of all of stop tree here snow stop dog feet wish nest man good late like is uP time will egg run with good nest girl is big ten good with her said milk one sun sun sat with milk girl work time to wish you food one said five feet pop end to I milk pop come zoo zoo food man eat egg take up take do hot wish 13. 14. 15. 16. 17. 18. 19. 20. 57 Normal Sentences I will play. We went to the zoo. The cat played. That girl is not good. I know you. She ran to the store. Stop the car. We play in the snow. You work late. You do like the snow. I am big. That boy is with me. I eat here. I was a good boy. I like milk. That boy will play here. I like her. Five men came to work. Lock the door. You take that cat out. APPENDIX C STANDARDIZED INSTRUCTIONS GIVEN TO SUBJECT PRIOR TO EACH EXPERIMENTAL SESSION APPENDIX C STANDARDIZED INSTRUCTIONS GIVEN TO SUBJECT PRIOR TO EACH EXPERIMENTAL SESSION I want you to listen to what the man is saying on the tape recorder. He will say the number of the sentence "Number " and then the words. Some sentences will be longer than others. When the man stops speaking, I want you to tell me exactly what you heard him say. If you can't remember everything, repeat as much as you can remember. Do you have any questions? (Answer questions) let's try a couple together. (Play two undistorted prac- tice sentences--one three word and one five word) Any questions? (Answer questions) Fine, let's begin. 58 APPENDIX D ANSWER FORM USED TO TRANSCRIBE SUBJECTS' RESPONSES 10. 11. 12. l3. 14. APPENDIX D ANSWER FORM USED TO TRANSCRIBE SUBJ ECTS' RESPONSES RESPONSE FORM GRADE CONDITION First Order go girl zoo men zoo pop nest is take hot girl red play late I call food jar wish snow take play big man jar sand man name milk good girl sun red toy us time all sun pOp tree wish pop milk snow pop dish is name egg call ten pop zoo cry food milk 59 15. 16. 17. 18. 19. 20. 60 up jar said. tree up yes me word nest ring sun me play wish milk nest name will jar dog cat high man hot COMMENTS 21. 22. 23. 24. 25. 26. 26a. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 61 RESPONSE FORM GRADE CONDITION Second Order men will do that here I put up to like man of all of stop tree snow stOp dog feet wish nest man good late like is time will egg run with good nest girl is big ten good with her milk one sun said sun sat with milk girl work time to wish you food one said five feet p0p up end to I milk pop come man eat ZOO zoo food egg take up ten do hot wish COMMENTS 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 62 RESPONSE FORM GRADE CONDITION Normal Sentences I will play. We went to the zoo. The cat played. That girl is not good. I know you. She ran to the store. Stop the car. We play in the snow. You work late. You do like the snow. I am big. That boy is with me. I eat here. I was a good boy. I like milk. That boy will play here. I like her. Five men came to work. Lock the door. You take that cat out. COMMENTS APPENDIX E TABLES OF MEAN PERCENTAGE CORRECT SCORES m.mm v.mm «.mm Hmuoe m.as m.ea R.wm woos o>am m.mm c.0m m.mm puo3 onus» numcmq mocmucmm h.He w.mm m.h~ Hmuoa m.e¢ v.ow m.m~ omumuamcs H.mm v.~m m.m~ come cow a.dv h.vm m.hm came com Hm>uoucH msaaaaumuaucH m.av «.mq m.av mumuw Hmuoa acmuo zuusom mcmuw pcooam .mcoHuwpaoo Honuo on» no name mmouoa Acuos o>wm .puoz counuv cannon accoucam can Amooceucom arena: .cowueEonummo Heap Icoucom nacho ncooom .GOADawaoummm Hewucoucmm nacho umuHmV cowuaEonummd Hewuseucmm no nacho .Apououamcs .oamfiooe .uamEoomv Hm>uoucH unasEHDmuuucH .Aopmum nunsow .apeum Economy ownnw you mauoom uoauuoo useouam some Hacua>0rr.¢ canes 63 64 m.~w H.50 m.mm Hmuoe pcmuu m.mw h.vm w.mw Hmuoe H.mm m.mm m.mm Hmfiuoz m.mh h.mh ~.mh Hopuo new m.mn ~.mh m.ah Hmono and coaumaflxoumm< aneucoucmm mo nacho Hmuoa opmuo nunsom opmuo pcooom emscuunooul.e manna 65 0.00 «.mm 0.~m Hmuoa vacuw 5.00 m.mm 5.50 HmuOB m.mm m.mm m.m5 UHO3 m>wm o.mm S.ma m.em euoz some» commuaecs m.m5 m.Hm m.55 Hmuoa m.H0 m.m0 n.5m ©HO3 m>wm 0.50 0.50 0.50 UHO3 wmuau OOmE 00¢ m.am m.Hm m.~m Hmuoa m.05 0.05 0.05 UHO3 m>ww m.mm 0.Nm 0.mm UHO3 Omuau ommE 00m Heuoa opmuo nunsom ammuw pcooom camcoq mocmucom Hm>umucH mDHaEwumuoucH .«ouo3 o>wm .Uuo3 mounuv nausea aocaucam one Romano cannon .ooeum ozooamv accuw an “pauouaaca .oamsooe .oansoo~0 Ha>uaucH wadsswumuoucH How mouoom uoouuoo oomucaouom ceazll.m manna 65 0.00 «.mm 0.Nm Hence ocmuw 5.0m m.mm 5.5m Hmuoa m.mm m.mm m.m5 puoz a>fim 0.00 0.00 m.5m cuoz moms» wauauamca m.m5 m.Hm m.55 Hmuoe m.H0 m.m0 n.5m cuo3 a>wm 0.50 0.50 0.50 vuo3 wanna come 00¢ m.am m.am m.~m Hmuoe m.05 0.05 0.05 puo3 o>am m.m0 0.~m 0.mm puos wanna come 00m deuce acmuw nuuaom opnuw neocom cumcaq monouamm Hm>uaucH msHSEHumuaucH .Aouoz o>wm .nuoz mausuv npmcaq aocaucam can Rowena cannon .oomum pcooamv enema an Acououaecs .oomE0oq .oan50om0 Hm>uaucH msHSEHumuoucH now nauoom uoouuoo ammusoouum ccmzll.m manna 66 v.05 0.H5 0.00 HeuOB ficmuw 0.H0 0.00 0.00 Hence 0.00 0.00 0.00H HeEHoz 0.00 0.00 0.00H Heouo 0cm 0.00 0.N5 0.00 Hevho umH UeHeuHmcfl 0.55 0.00 0.00 Hence 0.H0 0.00 0.00 Heeuoz 0.50 0.00 0.00 HeUHO 0:0 0.00 0.00 0.50 Hecuo uma DemE 00¢ 0.00 0.00 0.00 Heuoa 0.00 0.00 0.00 Heeuoz 0.05 0.H0 0.H0 HeUHO UGN 0.H5 0.0V 0.00 HeGHO “ma oemE 00m Hence cuoz e>wm 0H03 eeuna cofluewaoumm< He>ueucH msHsEHDmHeucH Hefluceunem 00 “echo .Anuos e>wm .0uos eeusuv numceq ecceusem one aeoceucem Heeuoc .cofiuewaoumme Hewuceucem Henuo cacoee .coaueewxoumme Hewuceucee ueouo umuwu0 Hecuo.»n .floeueuaecs .oemfioov .oemE0omv He>ueucH esaafiwumueucH now menace uoeuuoo eunuceonem amezll.0 eance 67 0.00 0.00 0.00 H0009 nacho H.00 0.00 0.00 Hence 0.00 0.00 0.50 0H03 e>Hm 5.00 0.00 0.00H wuoz eeunu Hmfiuoz. 0.05 0.05 0.05 Heuoe «.00 0.00 m.H0 whoa e>fim 0.00 0.00 0.00 0u03 eeunu uecno 0:0 0.H5 0.H5 m.H5 Heuoe 0.00 0.00 0.0v unoz e>wm 0.00 0.00 m.v0 0Ho3 eeunu ueouo pea Heuoa eweuw neuron eoenu ocooem numcen eoceucew cowuefiwxoumm< Hewuceucem mo newno .A0Ho3 e>wm .Uuo3 eeunuv nuwaeq eoceucem can Aeveum neuron .eceum ocooemv eUeHO an Ameoceucem daemon .cowueawxoumme Hewuceuaem Henna ccooem .GOHueE lwxoumme Hewuceuaem uevuo umuwwv Hecno How meuoom uoeuuoo emeuaeouem seezl|.5 eanea APPENDIX F SUBJECTS' RAW SCORES FOR EACH EXPERIMENTAL CONDITION APPENDIX F SUBJECTS' RAW SCORES FOR EACH EXPERIMENTAL CONDITION SECOND GRADE 200 msec Interstimulus Interval Size Subjects lst Order 2nd Order Normal 1 16 18 20 2 16 20' 20 3 16 17 19 4 16 16 20 5 13 16 19 6 14 16 20 7 l3 l6 l9 8 12 9 18 9 l7 l6 19 10 12 13 20 400 msec Interstimulus Interval Size lst Order 2nd Order Normal 11 14 15 20 12 l3 16 20 13 12 11 20 14 15 16 20 15 10 14 20 16 12 13 20 17 13 14 20 18 12 13 20 19 12 ll 20 20 15 14 20 68 69 SECOND GRADE Unaltered Interstimulus Interval Size Subjects lst Order 2nd Order Normal 21 12 15 20 22 12 14 20 23 20 19 19 24 ll 14 20 25 17 18 20 26 l6 19 20 27 16 16 20 28 20 20 20 29 ll 18 20 30 20 20 20 70 FOURTH GRADE 200 msec Interstimulus Interval Size Subjects lst Order 2nd Order Normal 1 16 19 20 2 13 11 20 3 11 15 20 4 12 15 20 5 17 19 20 6 13 14 20 7 14 13 18 8 15 13 20 9 13 13 20 10 18 16 20 400 msec Interstimulus Interval Size lst Order 2nd Order Normal ll l4 13 20 12 13 16 ' 20 13 12 9 20 14 14 13 19 15 15 16 19 16 13 13 20 17 15 17 20 18 17 17 20 19 15 16 20 20 10 12 20 Unaltered Interstimulus Interval Size lst Order 2nd Order Normal 21 16 17 20 22 18 18 20 23 19 20 20 24 17 18 20 25 19 19 20 26 ‘ 15 16 20 27 20 20 20 28 19 20 20 29 18 17 20 30 16 19 20 MICHIGAN STATE UNIV R ITY LI 8 BA lIWIIl'ES 5 6 5 15 8 IIIHIIIIIII 3 0 3 1293 0