JOURNAL OF APPLIED BEHAVIOR ANALYSIS 1991)24,153-160 NUMBER I (SPRING 199 1) TEACHING MEMORIZED SPELLING WITH A MICROCOMPUTER: TIME DELAY AND COMPUTER-ASSISTED INSTRUCTION KAY B. STEVENS, A. EDWARD BLACKHURST, AND DEBORAH Borr SLATON UNIVERSTY OF KENTUCKY A computer-assisted instruction program was evaluated that used a constant time-delay procedure to teach 5 students 18 spelling words. In addition to delivering the instructional procedure, the program managed the presentation of training content based on individual student responding and collected instructional data on individual student performance. The procedure was effective at teaching 4 of the 5 students the words, and generalization occurred from the computer-delivered keyboard response format to a teacher-delivered hand-written response format. Maintenance data varied among the students. The study demonstrated the feasibility of using microcomputers to deliver time-delay instruction in special education classrooms and suggested several research questions related to specific features of microcomputer-delivered time-delay instruction. DESCRIPTORS: time delay, computer-assisted instruction, mildly handicapped, spelling, special education Skinner (1986) provided a brief history of teaching machines and discussed the capabilities of computers to deliver effective programmed instruction, especially during the acquisition phase of learning. He emphasized that effective programmed instruction should prime the student for learning by providing correct models that may be easily imitated by the learner and result in accurate responding. Skinner conduded that the "small computer is the ideal hardware for programmed instruction" (p. 110) and has the potential for being an effective teaching machine. Cipani (1985) described the compatibility of time-delay prompt fading with computer-assisted instruction (CAI). Time delay is a systematic instructional procedure containing many of the procedural components outlined by LeBlanc, Hoko, Aangeenbrug, and Etzel (1985) as optimal teaching procedures for students with learning difficulties. Two types of time delay exist, constant and progressive, and both have been successful for learners with mild to severe handicaps. transferring stimulus control from a controlling prompt signaling the correct response to the terminal discriminative stimulus (Snell & Gast, 1981; Stevens & Schuster, 1988). This transfer of control is accomplished by pairing the prompt with the task request until the student can respond to the task request without assistance of the prompt. With constant time delay, a constant time interval is inserted between the task request and the controlling prompt, giving the student time to respond before the prompt if the answer is known or to wait and receive the prompt if the answer is not known. When the student is consistent at making correct responses before the prompt, the behavior is considered learned. Research indicates that constant time delay has been effective in teaching acquisition of a variety of tasks to students who are mildly handicapped. These tasks indude math facts (Cybriwsky & Schuster, 1990; Mattingly & Bott, 1990), sight word reading (Alig-Cybriwsky, Wolery, & Gast, 1990; Ault, Wolery, Gast, Doyle, & Martin, 1990; Both types of time delay have the potential for Wolery, Ault, Gast, Doyle, & Mills, 1990), sciproviding nearly errorless learning conditions by ence and social studies facts (Wolery, Alig, Gast, & Boyle-Gast, in press), and spelling words (KinThe authors thank Judith Mattingly for her cooperation ney, Stevens, & Schuster, 1988; Stevens & Schuster, and efforts during the research project. Address all correspondence to the first author, 229 Taylor 1987). All studies reported that the procedure was Education Building, Department of Special Education, Uni- effective in teaching the targeted skills and that versity of Kentucky, Lexington, Kentucky 40506. learning errors were minimal. For example, Kinney 153 154 KAY B. STEVENS et al. et al. used a CAI time-delay program to teach 1 male student with spelling deficits 15 spelling words in a tutoring setting. The purpose of the present study was to extend the previous time-delay research with mildly handicapped students by (a) examining the efficacy of providing CAI time-delay instruction in spelling in a public school dassroom rather than in a teacherdirected session or one-to-one private tutoring session and (b) adding a review bank feature that permitted the computer to present training trials for only those words in the training set on which the student had not met a predetermined mastery criterion. This change was made to examine the effects of focusing instructional opportunities on words that the student was not spelling correctly. Typically, time-delay studies have conducted training on all items of a training set until criterion is met on the entire set. However, the present study used the unique features of the microcomputer to manage the content items based on each student's response history to "Waiting to Spell" (Stevens & Blackhurst, 1988), a constant time-delay software program. respond to instructional and feedback phrases used in the program, (d) use a simple scanning procedure for finding keys on the keyboard that resulted in typing a minimum of one correct letter per 3 s, and (e) enter responses by typing on the keyboard for each activity induded in the CAI program. All experimental sessions occurred in the resource room while the students sat at the microcomputer with their backs to the dassroom and used the program independently. During sessions, one teacher, four to 10 other students, and the experimenter were present. An Apple Ile® microcomputer system with 128K of random access memory, color monitor, two 5.25-in. disk drives, Zip Chip accelerator, ImageWriter printer, Slotbuster voice synthesizer, and SONY headset were used during all demonstration, probe, and instructional sessions. "Waiting to Spell" (Stevens & Blackhurst, 1988), a software program authored using Apple SuperPILOT and stored on two 5.25-in. disks, provided the constant time-delay program. The dassroom teacher recommended the training content. During screening, the experimenter dictated 30 to 50 words to each student individually; students responded in writing. Words spelled incorrectly on two consecutive sessions were selected METHOD for training. Each student's 18 target words were Subjects, Setting, and Materials divided into three word sets (Sets A, B, and C) of Five students (3 females and 2 males), enrolled six words each. in a suburban elementary school and placed in a special education resource room, served as subjects. Experimental Design A multiple probe design across word sets repThey ranged in age from 1 1 years 3 months to 12 across students (Gast, Skouge, & Tawney, licated years 3 months and were in fifth or sixth grade. Three students were diagnosed as learning disabled 1984) was used to assess the effectiveness of the and 2 were diagnosed as educable mentally hand- training procedures. Following completion of trainicapped according to the criteria of the school dis- ing on the first word set, each subsequent word set trict. Each student's individual education program was then trained with the same procedures. listed spelling as a weakness, and each scored more than one standard deviation below the mean on a Procedure Probe and training sessions. Table 1 presents recent administration of the "Test of Written Spellprocedures for probe, 0-s, and 5-s delay trials. Each ing" (Larsen & Hammill, 1986). The results of direct observation indicated the trial is presented in terms of the instructional anfollowing prerequisite skills: (a) sit and attend to tecedents and feedback presented via the CAI proan instructional task for 20 min, (b) turn on the gram, as well as student response options. Prompts microcomputer and load software, (c) read and consisted of a model of the correct spelling of the 155 TIME DELAY AND CAI Table 1 Procedures for Probe, O-s, and 5-s Delay Trials Possible student Type of trial Instructional antecedent Probe A:a (Task request) "Spell O-s delay A: (Task request) "Spell responses ." (Immediately after the task request, the prompt appears.) A & V: Correct spelling as prompt 5-s delay A: (Task request) "Spell ." (Five s after the task request, if the student made no response, the prompt appears.) Instructional feedback Correct Incorrect or no response Correct after prompt V:b "Good." No feedback A & V: Praise statement V: Scoreboard Incorrect after prompt or no response Correct before prompt A & V: Correction procedure Incorrect before prompt Correct after prompt A & V: Praise statement V: Scoreboard A & V: Correction procedure A & V: Praise statement V: Scoreboard Incorrect after prompt A & V: Correction procedure aA, presented aurally. bV, presented visually. word. Praise statements were one of 10 randomly selected confirmation statements (e.g., "Good for you!"). Two small scoreboards at the bottom of the monitor cumulatively tallied unprompted and prompted correct responses. Correction procedure involved a voiced and printed message that said, "No. Check your answer and press RETURN," which signaled the student to compare the response to the model provided on the screen. This was followed by a repetition of the same trial until a correct response occurred on that word either before or after the prompt. A probe condition, consisting of a minimum of three probe sessions, occurred prior to and following training on each set of words. Probe sessions consisted of 18 trials; the 18 target words were randomly presented via the microcomputer as indicated in Table 1. Throughout all trials, students had 30 s to respond by typing the word and pressing the return key. Students could self-correct prior to pressing the Return key by using the left arrow key to erase. The microcomputer scored probe responses and provided a printout indicating (a) percentage of correct responses per word set per session, (b) total session time, (c) the stimulus word presented during each trial, and (d) the actual student response per trial. During the first training session of each word set, a O-s delay was used; each word was presented randomly five times each for a total of 30 training trials. During subsequent training sessions with that word set, a 5-s delay was used, and the number of trials became dependent on the students response accuracy. Following each training session, a printout was available indicating (a) number and percentage of unprompted correct responses, (b) number of prompted correct responses, (c) number of unprompted errors, (d) number of prompted errors, (e) stimulus word presented during each trial, (f) actual student response per trial, and (g) total session time. Each student was required to meet two training criteria per word set. The first criterion level required that individual words be spelled correctly before the prompt on three consecutive trials. Because the words were intermixed and presented randomly, three consecutive presentations of a single word could occur intermittently across any number of trials. As students met this criterion on individual words, the words were transferred by the computer into a "review bank." When a word moved into the review bank, no further training trials were presented on that word during the session. Instead, remaining training trials were ran- 156 156KAY B. STEVENS et al. domized among the words that had not met the first level criterion. At the beginning of each 5-s training session, the microcomputer program scanned the review bank. If the computer program found that a word or words from the set being trained were in the review bank (indicating the first level criterion had been met during previous sessions), a review trial was conducted on each word. Review trials were identical to probe trials. As a result of review trials, any word(s) spelled correctly remained in the review bank; any word(s) misspelled moved back into the training set. The student then had to meet the first level criterion again in order to move the word back into the review bank. Therefore, the number of training trials per 5-s delay session depended on the number of trails required to get all words into the review bank or a maximum of 30 trials (whichever came first). A maximum of 30 training trials was chosen because it approximated the number of trials per session used in previous time-delay spelling research studies with mildly handicapped students (Kinney et al., 1988; Stevens & Schuster, 1987), and the amount of time needed for 30 training trials matched the preexisting classroom schedule for time allotted to individual spelling lessons. The second criterion level required the student to maintain all six words in the review bank for 2 consecutive days. To meet this criterion, the student had to move all words into the review bank by meeting the first level criterion. When this occurred the computer program presented only review trials on the words. If they were all spelled correctly, all words remained in the review bank and no training occurred that day. The following day, the words were again tested with review trials. If they again were all spelled correctly, the words were considered learned and training on that word set ceased. sessions. Posttest sessions occurred following the first and third sessions ofeach microcomputer probe condition on trained word sets. Fifteen days after training on Set C words, a follow-up test occurred to assess performance in the generalized format. Feedback during generalization probes was withheld until the end of the session and consisted of the teacher telling the students the number ofwords they spelled correctly accompanied by one general positive statement regarding their correct responding. A handwritten format exduding continuous feedback was selected for testing generalization and maintenance because it represented conditions under which students typically were required to demonstrate spelling skills. Reliability Response reliability was evaluated by examining the microcomputer-generated data sheets that recorded each student's actual response and the microcomputer's scoring of each response. The experimenter (a) scored the student's response as printed on the data sheet, (b) compared her score to that of the computer, and (c) checked the computer calculations for total types of responses (correct and incorrect before and after the prompt). Reliability checks were conducted once per phase during probe, 0-s, and 5-s delay phases for a total of nine checks per student. Reliability was calculated by dividing the number of agreements by the number of agreements plus disagreements and multiplying by 100. Response reliability was 100%. Procedural reliability was conducted via extensive previewing of CAI lessons prior to classroom application, as well as 20 direct observation sessions during program implementation. During each observation, the five essential components of the timedelay procedure were checked for accuracy. Reliability was calculated by dividing the number of instructional components delivered accurately by Generalization and Maintenance Probe Sessions the total number of components under observation. Generalization probe sessions consisted of the Procedural reliability was 100%. teacher or experimenter dictating previously trained RESULTS AND DISCUSSION words to individual students and the students writFigures 1 and 2 depict performance data for 2 ing the words on a piece of paper. Pretests for generalization occurred prior to the first training of the 5 students during probe and review trials. 157 TIME DELAY AND CAI 6- Probe Condition I Probe Probe Condition Review Bank Condition Review Bank Set A Set B II m Probe Review Bank Condition Followup IV Set C 5.- W~\ 4. 32tn 1. < 0.- ..*- 0 * * Trainnyno K/A Computer probe orreview i HAandwritten Probe iew 65 CO a) 40) 32- 1. 0 z 6- 5. 0 4 ( 2. 3. Two data points durig 1. 0. 1.- . \ , I . 1 ,1 i 2 3 4 5 6 7 8 9 1011212313414516171819 20 21222324252627 Sessions Figure 1. Number of words spelled correctly during sessions for Jake. computer 1' ,9 8 2930313233343536 7 3839 40 probe, review bank, and handwritten generalization (Data for all 5 subjects are available from the first Therefore, asterisks have been plotted on the graphs author.) Due to the software design, it was im- to mark those sessions. Breaks in the abscissa inpossible to have words in the review bank during dicate Christmas vacation and the 15 days prior to the first two training sessions of each word set. follow-up probes. Probe Review Bank Set A Probe Condition III Review Bank Set B Probe Condition II A ru 6543CO 21 '1 (Handwritten Probe) A zA - 0CO x U) 65T CO 4W 3X 202 0 1 iA - :s O ..-.--. SI * * * * - :z 654- 0,32 a We 2- (Two 1 data points durin o,?nesession -1 2 3 4 5 6 7 8 9 101l1213 l5ig148449202152i3i4252627282930533233435i6ki839io~142T 445 Sessions Figure 2. Number of words spelled correctly during sessions for Todd. computer probe, review bank, and handwritten generalization 158 158 KAY B. STEVENS et al. Four of the 5 students performed similarly following the introduction of training. Figure 1 is representative of the 4 successful students. These 4 students met the accuracy criterion of two consecutive review sessions at 100% correct on six words; this outcome was replicated across three word sets per student. Figure 2 shows performance for Todd. Todd was the exception and was removed from the study after training on two word sets because of the excessive amount oftime he required to meet the training criteria and the lack of generalization and maintenance that occurred following training. In addition, he exhibited frequent inappropriate and noncompliant behaviors during training that disrupted instruction. The application of CAI with noncompliant students has been reported to be problematic in previous interventions. For example, Landeen and Adams (1988) compared the effectiveness of penciland-paper drill and practice with computer-assisted drill and practice on math facts with students with behavior problems. They reported that the students sometimes stopped responding when computer feedback indicated an error and otherwise engaged in a variety of inappropriate behaviors during CAI. Based on these findings, it is likely that CAI may not be best used with some students unless teacher supervision and feedback are also provided. Therefore, the generalizability of findings across subgroups of students with mild handicaps is questionable and requires further investigation. Percentage of errors during training was calculated for the 4 students who completed the study by dividing the number of errors by the number of training trials for each word set. The average error rate across all students was 12.8%; the range of errors was 8.6% to 19.4%. Possibly the most attractive element of the time-delay procedure is the potential for errorless learning due to the option of waiting for the prompt before responding. The 12.8% average error rate reported here is higher than those reported in previous studies involving time-delay with students who were mildly handicapped (Ault et al., 1990; Kinney et al., 1988; Mattingly & Bott, 1990; Stevens & Schuster, 1987). Error rates reported in these studies ranged from 0.4% to 4.8% across all subjects and training trials. Aspects unique to the computer-assisted format used in the present study that may have contributed to the higher error rate indude (a) the microcomputer's delivery ofpositive feedback statements and points on a scoreboard did not serve as reinforcers for correct responding; (b) the correction procedure delivered the same information as the controlling prompt, so guessing and not guessing resulted in similar consequences; and (c) there was no authority figure present during the CAM lessons (e.g., a teacher) whose presence may have inhibited nonwait errors (errors before the prompt). Future systematic investigations of CAI time-delay lessons might manipulate those variables to evaluate error rates. An analysis of probe condition data for the 4 students who completed the study showed that scores on posttraining computerized probes varied and were often below the 100% correct responding exhibited at the end of training. These declines in accuracy during computerized probes are cause for concern because they are more substantial than reported for posttraining probe conditions in other time-delay investigations involving learners with mild handicaps (Kinney et al., 1988; Mattingly & Bott, 1990; Stevens & Schuster, 1987). One procedural difference that may account for these findings is that previous studies continued training on all items in the training set until each student reached criterion on the entire set. This is in contrast to the item-by-item criterion for moving words into the review bank used in this study. It is possible that requiring criterion on an entire set facilitated maintenance because, as individual words met criterion, training trials continued on those words until all words in the set met criterion. In addition, the review bank feature of the present study did not allow the interspersal ofknown and unknown training content, because learned content was omitted from training. Interspersing known items with unknown items during spelling instruction has been found to increase the rate of acquisition and facilitate maintenance when teaching students with mental retardation (Neef, Iwata, & Page, 1980). These findings suggest several areas for future investigation, induding (a) comparing mastery cri- 159 TIME DELAY AND CAl terion by sets and by items to darify the desirability of the review bank feature; (b) comparing various levels of mastery criterion, induding accuracy, fluency, and consistency across sessions; (c) interspersing known with unknown items by either leaving in words after they are learned or induding already known words at the outset; and (d) investigating the effects of supplementing the time-delay procedure with follow-up activities such as handwriting activities and word puzzles. Performance on tests of generalization across presentation and response formats permitted several further analyses. First, prior to training, none of the words were spelled correctly when dictated by the teacher and handwritten by the student. The first generalization data point following training on each word set represented immediate generalization. During that probe, all students responded with 100% accuracy across all sets of words with the exception of 1 student, who misspelled one word in Set C. These findings indicate that immediate generalization from computer training to a teacher-dictated handwritten format occurred without the need for continuous consequences for correct responding. However, the 100% correct responding was not maintained on subsequent generalization probes. In the 1 5-day follow-up probes, 80.6% was the mean correct. After 2 of the students (Mona and Willie) received additional training consisting of two 5-s delay training sessions for Willie and three 5-s delay sessions for Mona, they received higher follow-up probe scores than the other 2 students, who received no additional training. Although follow-up training was not a planned part of this research and occurred only because the 2 students requested it, it may serve as an effective component of a CAl instructional package and warrants further evaluation. When comparing generalization probe scores with microcomputer probe scores occurring within 30 min of one another (computer probes conducted first), generalization scores exceeded microcomputer probe scores on 15 occasions (29.4%), were equal to computer scores on 31 occasions (60.7%), and were lower than computer probe scores on five occasions (9.8%). These findings suggest that errors were more likely to occur during responses made on the computer than during handwritten responding. One explanation might be that some errors were due to incorrect typing rather than to not knowing how to spell the word, suggesting the need for fiuher examination of typed versus written responses. In addition, prompting students to monitor their responses on the screen before advancing the program might be a necessary component of CAI. The CAI constant time-delay spelling program used in this study provided preliminary evidence as to the efficacy of delivering time-delay instruction in a microcomputer format. Results indicated that such an approach to instruction, although in need of refinement and additional research, represents a promising development in technology-assisted instruction. REFERENCES Alig-Cybriwsky, C., Wolery, M., & Gast, D. L. (1990). Use of a constant time delay procedure in teaching preschoolers in a group format. Journal of Early Intervention, 14, 99-116. Ault, M. J., Wolery, M., Gast, D. L., Doyle, P. M., & Martin, C. P. (1990). Comparison of predictable and unpredictable trial sequences during small-group instruction. Learning Disability Quarterly, 13, 12-29. Cipani, E. (1985). Using time delay fading in the design of CAI. The Computing Teacher, 12, 12-13. Cybriwsky, C. A., & Schuster, J. W. (1990). Using constant time delay procedures to teach multiplication facts. Remedial and Special Education, 11, 54-59. Gast, D. L., Skouge, J. R., & Tawney, J. W. (1984). Variations of the multiple baseline design: Multiple probe and changing criterion designs. In J. W. Tawney & D. L. Gast (Eds.), Single subject research in special education (pp. 239-252). Columbus, OH: Merrill. Kinney, P. G., Stevens, K. B., & Schuster, J. W. (1988). The effects of CAI and time delay: A systematic program for teaching spelling.journal ofSpecial Education Technology, 9, 61-72. Landeen,J.J., & Adams, D. A. (1988). Computer-assisted drill and practice for behaviorally handicapped learners: Proceed with caution. Education and Treatment of Children, 11, 218-229. Larsen, S. C., & Hammill, D. D. (1986). Test of written spelling. Austin, TX: Pro-Ed. LeBlanc, J. M., Hoko, J. A., Aangeenbrug, M. H., & Etzel, B. C. (1985). Microcomputers and stimulus control: From the laboratory to the classroom. Journal ofSpecial Education Technology, 7, 23-30. 160 KAY B. STEVENS et al. Mattingly, J. C., & Bott, D. A. (1990). Use of a constant time delay procedure in teaching facts to mildly handicapped students. Exceptional Children, 56, 438-449. Neef, N. A., Iwata, B. A., & Page, T. J. (1980). The effects of interspersal training versus high-density reinforcement on spelling acquisition and retention. Journal of Applied Bebavior Analysis, 13, 153-158. Skinner, B. F. (1986, October). Program instruction revisited. Phi Delta Kappan, pp. 103-110. Snell, M. E., & Gast, D. L. (1981). Applying time delay procedures to the instruction of the severely handicapped. Journal of the Association for Persons with Severe Handicaps, 6, 3-14. Stevens, K. B., & Blackhurst, A. E. (1988). Waiting to spell [Computer program]. Lexington, KY: Department of Special Education, University of Kentucky. Stevens, K. B., & Schuster, J. W. (1987). Effects of a constant time delay procedure on the written spelling performance of a learning disabled student. Learning Disability Quarterly, 10, 9-16. Stevens, K. B., & Schuster, J. W. (1988). Time delay: Systematic instruction for academic tasks. Remedial and Special Education, 9, 16-21. Wolery, M., Alig, C., Gast, D. L., & Boyle-Gast, K. (in press). General and specific attentional responses: Acquisition and maintenance of target, observational, and incidental behaviors. Exceptional Children. Wolery, M., Ault, M. J., Gast, D. L., Doyle, P. M., & Mills, B. M. (1990). The use of choral and individual spelling attentional responses in teaching sight words during small group instruction. Remedial and Special Education, 11(5), 47-58. Received January 26, 1990 Initial editorial decision April 3, 1990 Revisions received June 25, 1990; July 30, 1990; August 20, 1990 Final acceptance September 11, 1990 Action Editor, David P. Wacker
© Copyright 2026 Paperzz