521 ensure consistency of instruction and procedure throughout all 20 classrooms. It is important to note that while the program itself was highly structured, some measure must be given to the proper assignment of students to conditions. From the results of this research, it has been found that the student acting as tutor must have a thorough knowledge of the subject matter presented to the tutee. ABOUT THE AUTHOR Lee Epstein is presently functioning as a child-school psychologist in private practice in I^ouisville, Ky., andholdsan adjunct faculty position at Indiana University Southeast in the Division of Education. His major professional interests include studying children with school phobias and investigating the teaching of reading to learning disabled children. Dr. Epstein is presently completing a study of the effects of exercise and

jogging on the elimination of school phobias. Requests for reprints should be sent to Dr. Epstein at Room 0028, Hillside Hall, Indiana University Southeast, 4201 Grant line Rd., New Albany, Ind. 47150. REFERENCES Christoplos, F.: Keeping exceptional children in regular classes. Exceptional Children, 1973, 39, 569-571. Croth, R.I., Whelan, R.J., Stables, J.M.: Teacher application of behavior principles in home and classroom environments. Focus on Exceptional Children, 1970, 1, 7. Findley, W.J., Bryan, MM.: Ability grouping: 1970. Athens, Ga.: Center for Educational Improvement, University of Georgia, 1971. Gbss, R.M., Griffin, J.B.: Affective education for the inservice elementary and special education teacher. Bloomington, Ind.. Center for Innovation in Teaching the Handicapped, 1973. Niedermeyer, F.C., Ellis, P.: Remedial reading instruction by trained pupil tutors. Elementary School Journal, 1971,71, 400-496.

Using the Test of Language Development with LanguageImpaired Children Phyllis Newcomer, EdD, and Donald D. Hammill, EdD

T

he Test of Language Development (TOLD) (Newcomer & Hammill 1977) is an individually administered diagnostic instrument that measures seven semi-independent aspects of children's spoken language. It adheres loosely to a linguistic model of language structure and includes two phonological subtests, word articulation and word discrimination; two semantic subtests, oral vocabulary and picture vocabulary; and three syntactic subtests, sentence imitation, grammatic understanding, and grammatic completion. Three subtests —

66

word discrimination, picture vocabulary, and grammatic understanding — involve receptive language abilities, while three subtests — word articulation, oral vocabulary, and sentence imitation — tap expressive skills. One subtest — grammatic completion — measures both receptive and expressive ability. Word discrimination measures the child's ability to differentiate between speech sounds presented in a same-different word pair format. Word articulation requires the child to emit spontaneously speech sounds in response to Journal of Learning Disabilities

522 pictures. For the picture vocabulary subtest the child must select one of four pictures that matches a stimulus word spoken by the examiner. Oral vocabulary involves the child's ability to define words presented orally. Grammatic understanding requires that the subject select one of three pictures that best matches an oral stimulus sentence. For sentence imitation the child must imitate spoken sentences. Grammatic completion is a cloze technique requiring the use of correct morphemes to complete spoken sentence stems. Thus far several investigations have been conducted to establish the reliability and validity of the TOLD students when used with normalspeaking children. Data accrued from the administration of the TOLD to a sample of 115 standard-English speakers between the ages of 4 and 9 indicate that the subtests are sufficiently reliable (internal consistency coefficients at or exceeding .80) to be used for specific diagnosis with individual children (Brown, Newcomer, & Hammill undated). There is also evidence (Hammill & Newcomer 1977) that the subtests have criterion validity, since they correlate highly with various criterion tests of specific linguistic abilities, such as the Peabody Picture Vocabulary Test (Dunn 1965), the TemplinDarley Articulation Tests (1960), the Auditory Discrimination Test (Wepman 1958), the Test of Auditory Comprehension (Carrow 1973), the oral vocabulary subtest of the Wechsler Intelligence Scale for Children (1949), and the Northwestern Syntax Screening Test (Lee 1969). An equally important but heretofore unconducted study of the TOLD's usefulness as a measure of language skills pertains to the validity and the reliability of the subtests when they are used with language-impaired children. While several procedures could be employed to demonstrate such validity, in this study we chose to test the TOLD's ability to differentiate between groups of children who have relatively normal language abilities and groups who are language-impaired in some manner. Such a study would contribute to a knowledge of the TOLD's Volume 11, Number 8, October 1978

construct validity. Presumably the test profiles of children who have either speech and/or language problems will show consistent patterns of weaknesses and will vary significantly from the profiles achieved by normal children. Failure to obtain evidence of significantly depressed performances by language-impaired children would indicate that the TOLD subtests lack the diagnostic sensitivity that is necessary to identify language problems. In addition, failure to establish test reliability with a language-impaired sample would suggest that the subtest items are inappropriate when used with such children. Consequently, this study was undertaken to (1) compare the TOLD subtest scores achieved by a group of deviant-language children with those obtained by a group of standard-English speakers who were matched for age and sex, (2) make a similar comparison between the performances of an articulation-problem group and a matched control group, (3) compare the subtest performances of the articulationproblem and the deviant-language groups, and (4) compute the internal-consistency reliability of the TOLD subtests and total when it is used with both language-deviant and articulationproblem groups of children.

SAMPLE SELECTION Thirty-two English-speaking, intellectually normal children between the ages of 4 and 9 (mean = 5.4) served as subjects. All of them had been diagnosed as having speech and/or language problems and were receiving therapy at the Speech and Hearing Clinic at The University of Texas, Austin. Sixteen of the children were classified as having poor articulation, 11 as having language difficulty, and five as having both kinds of problems. The five subjects who had both types of problems were excluded from that portion of the study dealing with group differences. However, their test scores were added to both the articulationproblem and language-impaired groups for the 67

523 purposes of determining the TOLD's internal consistency. T ratios were used to compare the test performance of the language-impaired children with that of the normal groups having the same age and background characteristics. Internal consistency reliability was computed by the Kuder-Richardson formula 20.

RESULTS AND DISCUSSION The t ratios associated with the mean scores for the deviant-language group and its control group were all significant at the .01 level of confidence. This indicates that all of the TOLD subtests differentiate between children with language problems and those who speak normally. Consequently, one might expect that the administration of the TOLD to a child with a heretofore unidentified language handicap would result in an accurate diagnosis of the language disability in any or all of the seven linguistic skills that are measured. The mean score comparison between the articulation-problem group and its control group are particularly interesting. The t ratios associated with five of the seven subtests are highly significant (.01 level), demonstrating that children with articulation problems are less proficient than their normal-speaking peers in all areas but word discrimination and picture vocabulary. Examination of the mean scores and the size of the t ratio indicates that word articulation, as expected, represents the most significant deviation from the norm. Notably, the two tests that presented no problems to the experimental group were both receptive language tasks and required little or no verbalization. Possibly the children's articulation problems depressed their performance in all the expressive language areas tested. However, it is equally feasible that the members of the articulation-problem group had alternate types of language handicaps that were less pronounced than their articulation problems. Comparisons of the mean scores of the deviant-language and articulation-problem 68

groups reveal clarifying information about this point. The deviant-language children performed significantly more poorly than the articulationproblem children on all subtests except word articulation and sentence imitation. While it is true that children who are diagnosed as having articulation problems do not match normal children in their expressive language performance, they do significantly better than children who have been diagnosed as having deviant language except in the area that involves their pronounced disability, word articulation, and in sentence imitation, a task that is greatly impeded by faulty speech mobility. Actually, on the sentence imitation subtest, the articulationproblem group was able to repeat an average of 6.0 sentences as compared with 1.1 for the deviant-language group, but the large variance in their performance accounted for the insignificance of the t test. In any event, it appears that for the most part the TOLD results support the previously determined clinical classifications of these language-impaired children. Those children classified as having deviant language had far more severe, pervasive language problems than did any other group of children, including those who had been classified as having articulation problems. The final analysis in this study deals with the internal consistency reliability of the TOLD when it is used with two groups of languageimpaired children and the five children who evidenced both types of problems. According to Guilford (1956), a reliability coefficient of .80 is necessary when a test is to be used for the purposes of individual diagnosis. Examination of the coefficients reported for the deviantlanguage children shows that the TOLD subtests and total score have excellent reliability. Only the coefficients for the word articulation (.79) and the grammatic completion (.78) subtests fell slightly below the required level. In the case of the articulation-problem group, the picture vocabulary (.75), oral vocabulary (.76), and grammatic understanding (.73) are slightly below the desired .80 coefficient. For the most Journal of Learning Disabilities

524 part, these internal consistency coefficients suggest that the test is as reliable when used with language-impaired children as when used with normal speakers of standard English.

CONCLUSION The results of this study support the validity and reliability of the TOLD subtests when they are used with language-impaired children. A group of children who had been classified as having deviant language performed significantly more poorly than a matched group of normal children on all seven subtests skills. Children with articulation problems were inferior to normal speakers on all but two receptive tasks. Even more interesting is the fact that the TOLD results appear to confirm the clinical classifications of the language-impaired child, since those with deviant language performed significantly worse than those with articulation and sentence imitation. ABOUT THE AUTHORS Phyllis L. Newcomer is an associate professor of education at Beaver College in Glenside, Pa. She earned her doctorate in special education at Temple University in 1973. Dr. Neucomer has worked as a teacher of emotionally disturbed children and as a school psychologist. Before moving to Beaver College, she was an assistant professor and program associate with the

leadership Training Institute in learning Disabilities at the I diversity of Arizona, and then an assistant professor of special education at The University of Texas at Austin. She has coauthored a book on psycholinguistics. Her primary interests are in children's language development and the relationship of language to learning disabilities. Donald D. Hammill has worked as a speech pathologist in the public schools and was an adjunct professor of special education at The University of Texas at Austin. After obtaining his doctorate at The University of Texas in 1963, he took postdoctoral work in speech and language problems of brain-injured children at Wichita State University with Dr. Orvis Irwin. He has co-authored several hooks in the areas of learning disabilities and psycholinguistics Requests for reprints should be addressed to the authors at 5555 N. Lamar Blvd ., Suite C-101. Austin, Tex. 78751.

REFERENCES Brown, L., Newcomer, P., Hammill, D.: The Reliability of Language Tests. Unpublished manuscript, Department of Special Education, University of Texas, Austin, undated. Carrow, E.: Test for Auditory Comprehension of language. Austin, Texas: Learning Concepts, 1973. Dunn, L: Peahody Picture Vocabulary Test. Minneapolis: American Guidance Service, 1965. Guilford, /.. Fundamental statistics in psychology and education. New York: McGraw-Hill, 1956. Hammill, D., Newcomer, P.: Statistics and Construction Manual for the Test of Ijinguage Development. Austin, Texas: Empiric Press, 1977 Lee, L.: Northwestern Syntax Screening Test. Evanston, III.: Northwestern University Press, 1969. Newcomer, P., Hammill. D.. The Test of language Development. Austin, Texas: Empiric Press, 1977. Templin, M., Darly, F.: The Templin-Darley Tests of Articulation. Iowa City. Bureau of Educational Research, State University of Iowa, 1960. Wechsler, D.: The Wechsler Intelligence Test for Children New York: The Psychological Corporation, 1949. Wepman, J.: Auditory Discrimination Test. Chicago: language Research Assoc. 1958.

Antecedent and Contingent Modeling to Teach Basic Sight Vocabulary to Learning Disabled Children Jo Hendrickson, PhD, Michael Roberts, PhD, and Richard E. Shores, EdD

I

odeling has been defined as occurring when an individual exhibits a behavior similar to that just previously performed by another person or "model" (Frayer & Klausmeier Volume 11, Number 8, October 1978

1972, O'Leary & O'Leary 1972). Researchers and educators have long recognized that modeling may be a critical factor in numerous learning situations (Anderson & Faust 1974, Bandura

525 1965, Haring & Gentry 1976, Skinner 1968). Few studies, however, have systematically investigated the use of modeling to teach reading, arithmetic, spelling, and language (Roberts 1977), and those appeared only recently. Haskett and Lenfestey (1974), for example, wanted to interest preschool children in looking at books. Introducing more books into the classroom did not cause the desired increase. Having an adult model "reading," that is, pick up a book, sit down, and read aloud, did result in an increase in the children's use of books. In a series of arithmetic experiments, Smith and Lovitt (1975) taught 8- to 11-year-old learning disabled youngsters to perform various computational processes. They used a demonstration where the teacher verbally worked the first problem on the child's arithmetic worksheet as a permanent model technique: the completed problem remained as a referrant for the child as he worked problems on his own. All children generalized the processes to problems of a similar type, and some generalized across response classes. Jobes (1975) trained learning disabled children to spell using a combination of verbal and written models. She also assessed the effect on peers who observed the target children learning by this technique. The observers also gained spelling skills, but not to the level of the directly taught youngsters. Strorner (1975) ameliorated written reversals in special and regular education children with a modeling strategy coupled with feedback and verbal reinforcement. Other researchers (Clark & Sherman 1975, Hendrickson 1977, Hester & Hendrickson 1977, Lahey 1971, Twardosz & Baer 1973, Zimmerman & Pike 1972) have employed modeling tactics to teach normal and handicapped populations language responses. Acquisition and generalized responding were noted by these investigators. A review of these studies indicated that the most common form of modeling was what we term "contingent modeling" — that is, the modeled response occurred after a child erred. How70

ever, another form of modeling is to present the model prior to the child's response (antecedent modeling), thereby virtually ensuring a correct response. To date, the antecedent modeling strategy has been used most often when teaching motor responses and simple language responses to developmentally young children.

PROCEDURE A single-subject, multiple-baseline (Baer, Wolf, & Risley 1968), and multiple-treatment design was used to assess the differences between the two teaching procedures as applied to two primary-school-aged boys diagnosed as severely reading disabled. One-to-one training and probe sessions occurred daily. An antecedent and contingent modeling tactic was used to train each of two different sets of words to each student. Training was followed immediately by probes on all response sets. Probe data were used as criteria for phase change.

Antecedent Modeling The trainer presented the word card and said, "This word i s . . . , " and then asked, "What is this word?" If the child repeated the modeled word correctly, he was praised and the question was repeated without the model until three correct responses were attained. If the child misread the word or paused three seconds, the modeling sequence was repeated.

Contingent Modeling This method replicates antecedent modeling training in every detail except for when and how the modeling occurs. The subject was first presented the word card without a verbal model and asked, "What is this word?" If he answered correctly, he was praised and the question was repeated for three correct responses. For each error the child made by miscalling the word, the child was told, "No, this word is...," and the question was repeated. Errors made by not responding or by stating, "I don't know," were Journal of learning

Disabilities

Using the test of language development with language-impaired children.

521 ensure consistency of instruction and procedure throughout all 20 classrooms. It is important to note that while the program itself was highly str...
664KB Sizes 0 Downloads 0 Views