#AUTHOR: Flanigan, Patrick J., Ph.D; Joslin, Elizabeth S., M.S.
#TITLE: Patterns of Response in the Perception of Braille Configurations
#CATEGORY: Perception, Braille
#PUBLICATION: Outlook, October 1969
#ABSTRACT: Investigates the relationship between the interval from stimulus presentation to response
Patterns of Response in the Perception of Braille Configurations
PATRICK J. FLANIGAN, Ph.D. and ELIZABETH S. JOSLIN, M.S.
Dr. Flanigan is assistant professor of education., counseling and behavioral studies, Center on Behavioral Disabilities, University of Wisconsin, Madison. Miss Joslin is a research assistant at the Center.
The report of the AAWB-AAIB Joint Uniform Braille Committee to the 1958 American Association of Instructors for the Blind stated: "There are many who have considerable misgivings about Standard English Braille, Grade II, as it stand today."17 The report suggested that educators and researchers attempt to improve the existing system in a number of ways, by among other things, conducting studies in the rate of braille reading and applying techniques of programmed instruction to the teaching of braille.
Both the classroom teacher and educational researcher should be concerned with slowness in practicing braille configurations, for they are extremely important to development of braille skills. A major link in the total reading process is the relationship of sensory and perceptual discrimination variables relevant to the process.
Purpose of the Study
This study was undertaken to examine the relationship between the interval from stimulus presentation to response, as indicated by discrimination of the stimuli on a programmed tachistoscopic instructional device. Another purpose of the study was to determine the effect of a remediation program upon this relationship and upon the consequent speed of reading standard braille materials. The length of the interval between stimulus and response, along with notable differences from previous research in the rank order of difficulty in individual letter perception.
Twenty-seven Subjects were selected randomly from the student population at the Wisconsin School for the Visually Handicapped. They ranged in chronological age from 110 to 212 months, in mental age from 93 to 269 months, in grade placement from 3 through ungraded high school (approximately grade nine level), and in measured IQ from 65 to 144. Table 1 includes data description of the experimental group and of the control Ss- i.e., those subjects who were not exposed to our devices.
All Ss were used to establish normative data in phase 1 of the study; later they were randomly divided into experimental and control groups for phase 2, the remediation phase. During phase 1, the experimental group was treated on the programmed learning device, the tachistoscopic device mentioned earlier; the control group received an equal amount of treatment with traditional braille materials.
The study was divided into two phases. The ultimate objective of phase 1, the presentation phase, was the identification of an optimal range of perception time for use in phase 2. Normative data on (1) the relationship between stimulus presentation time and number of errors and (2) the variation in error count for each letter would lead to a rank order of difficulty for the 26 letter configurations. It was hypothesized that, as stimulus presentation time decreased, number of errors would increase.
How the Tachistoscopic Device Was Used
Before the experiment began, Ss were introduced to the device and allowed to examine each component and its operation. After this, S had contact with the presentation unit only. In program 1, the implementation of phase 1 (96 random order letters, described later), stimulus presentation time was varied on 10 trials from 1 sec to 0.1 sec in decreasing order. Readiness and response intervals were arbitrarily held constant at 4 and 6 sec respectively. During this phase, the Experimenter advanced the program after errors, rather than allowing the missed letter to be re-presented. Each S received one trial a day for 10 days. Four Es worked with the Ss without regard to continuity of experimenter-subject combinations. Two Ss were run simultaneously on two identical machines in the same room. There was, however, no competition between Ss. No verbal encouragement, discouragement, or reinforcement was given by the E. At the end of each trial the S was thanked. An individual record of the errors made on each trial was kept for the 27 Ss. Program 2, the remediation program, was developed from the raw data of program 1.
Phase 2 employed 26 of the 27 Ss, randomly divided into two groups of 13, one experimental and one control. The groups were formed by taking a list of the Ss in alphabetical order by grade level and assigning every other S to the experimental group beginning with S 1. One S was dropped because of illness.
In program 2 (151 braille configurations, described later) each S in the experimental group received nine 15-min training periods on the discrimination machine. Program 2 was used during each period; if completed before the end of the training period, it was repeated. The settings for the readiness and response intervals were again held constant at four and six sec respectively.
Stimulus presentation times during the experimental treatment periods were as follows:
a. remediation periods 1, 2, and 3-0.3 sec
b. remediation periods 4, 5, and 6-0.2 sec
c. remediation periods 7, 8, and 9-0.1 sec
Remediation Was at Hardest Presentation Times
Stimulus presentation times were chosen during the remediation phase on the basis of the data from phase 1, which showed the lowest mean proportion of errors for the total group to occur at 0.6 sec, and the greatest mean proportion of errors to occur at 0.1 sec. On this basis we decided to remediate at the three most difficult times-0.3 sec, 0.2 sec and 0.1 sec; it was anticipated that improved discrimination at these speeds would produce improvement at slower presentation speeds. On the same basis, and because the 0.3-sec time setting was halfway between the setting at which the mean proportion of error was the least and the greatest, the data from the 0.3-sec presentation time of program 1 in phase 1 were selected as the pretask criterion measure and the program was repeated (presentation time, 0.3) as a posttask measure.
Verbal presentation of the remediation program to experimental Ss was similar to that used in phase 1. S was asked at the beginning of the period not to talk and was given no verbal reinforcement, either positive or negative. One major difference in the operation of the experimental device was that, whereas during phase 1 the E advanced the program upon receipt of an erroneous response, during phase 2, when an incorrect response was registered, the same stimulus configuration was re-presented automatically by the device until a correct response was elicited. If S exceeded the arbitrary limit of 10 consecutive errors on a single configuration, E advanced the program to the next configuration.
Control Ss used in phase 2 also received nine timed 15-min training periods. This group used the same materials utilized in constructing program 2, except that they were prepared in traditional braille form on thermoplastic film (Brailon). Procedures utilized in initiating the control group's treatment were the same as those experienced by the experimental group which used the automated device. At the end of the 15 min period E would say: "You may stop now, that is enough practice for today." Control Ss received no reinforcement during their treatment periods; however, if they were not attending for several minutes, E would say: "This is your time to practice letters; keep practicing until I tell you to stop, then you may read a story."
Program 1 of phase 1 contained each letter of the alphabet; all 26 appeared at least once, and the frequency of any single letter in the program ranged from one to nine. No letter appeared twice in succession. Frequency of appearance of the letters in program 1 is shown in Table 2.
A-1 | F-3 | K-5 | P-1 | U-9 |
B-2 | G-3 | L-3 | Q-3 | V-3 |
C-1 | H-2 | M-2 | R-5 | W-7 |
D-3 | I-3 | N-4 | S-3 | X-7 |
E-4 | J-3 | O-2 | T-5 | Y-7 |
Z-5 |
Table 2
Frequency of Letters in Program 1
Program 2, which was used in the treatment phase, was developed from the raw data obtained from program 1 of phase 1. Because the frequency of appearance of the letters in program 1 varied, the percentage of possible errors on each letter for all 27 Ss on the total of the scores for six presentation times was used to determine the relative difficulty of the 26 letter configurations.
The 17 letters used program 2 (remediation program) and their frequency of appearance are shown in Table 3.
T-20 | I-10 | E-9 | H-6 |
R-12 | O-10 | J-8 | S-6 |
Z-12 | F-9 | V-8 | T-6 |
N-10 | Q-9 | Y-6 | W-5 |
M-5 |
Table 3
Frequency or Appearance of Letters in Program 2.
The tachistoscopic tactual discrimination (TTD) machine employed in this study is an experimenter-controlled, machine-paced device designed to facilitate increases in the speed of recognition and accuracy of discrimination of individual braille configurations. The three components of the device are:
1. Console unit. This unit contains the mechanism necessary for the control of readiness, presentation, and response interval timing, as well as number of recording devices.
2. Program unit. This unit employs a 52-panel, six-deck stepping switch, each deck corresponding to a solenoid, or pin, in the presentation unit, and three female cinch Jones connectors with 100 contacts in each. The program is wired on the male cinch Jones connectors which are easily snapped into place on the outside of the program unit. In this manner power is supplied only to those contacts that correspond to the presentation pins the programmer wishes to be presented.
3. Presentation unit. This unit, the only part of the device with which the S comes into contact, presents to him a stimulus-response panel made up of a presentation plate and six response keys. The presentation plate, a braille head from a Perkins Brailler, presents a preprogrammed stimulus in the form of blunt metal pins which project 0.015 in. above the surface of the plate. The length of time the pins remain exposed is controlled by the presentation timer. The response keys are connected to the coincidence circuit in such a manner that only when all of the correct keys are depressed will a correct response be recorded and the program advanced to the next braille configuration. If at any time during the response sequence an incorrect key is depressed, a relay is tripped making a correct response impossible for that presentation. In this case the discontinuous or response timer starts the readiness timer and the sequence is repeated. If a correct response is registered, a solenoid on the stepping switch is energized, stepping it to the next position; a correct response is recorded on the correct response counter and the next preprogrammed configuration is presented.
The following measures were employed to judge the effectiveness of training on the tactual discriminating device. With the exception of IQ measurement and machine data, all measures were administered before phase 1 and again at the conclusion of phase 2. At the conclusion of phase 2, program 1 was readministered to both experimental and control Ss at a 0.3 sec presentation time level. The data obtained were compared to those for the 0.3 sec presentation time data gathered in phase 1 to determine the effect of machine training on the responses made in the machine task. Total responses, responses on remediated letters, and responses on nonremediated letters were analyzed. It was predicted that there would be no difference between the groups on these variables.
Criteria for Measuring Effectiveness
The effectiveness of the training program on traditional braille reading was measured by the following tasks:
a. Traditional braille reading tasks. A 5-min oral reading task, using material adapted and transcribed into traditional braille from Science Research Associates, Inc., was used to determine the traditional braille reading speed of the S. Oral questions were administered after each individual braille-reading evaluation to insure appropriate comprehension of the reading materials. No analysis was undertaken in this regard. Measurements on the basis of this reading were number of braille words read in 5 min, errors, and retracings. The specific braille materials read by each S were appropriate to the reading achievement level of all Ss.
b. Braille configuration recognition task. This included letters and punctuation marks and was used to measure the improvement (in terms of speed of perception) in the perception of traditional braille configurations. The measure in this regard was in terms of errors.
In the analysis of the error counts on the ten time-set intervals, it was found that the Ss committed more errors on the 1.0-sec trials than they did on the middle speed ranges (the three ranges were arbitrarily defined by the Es as follows: slowest, 1.0, 0.9, 0.8, middle, 0.7, 0.6, 0.5, 0.4; and fastest, 0.3, 0.2, 0.1). The pattern of a decreasing mean number of errors from the upper range to the midrange and an increasing number of errors from the midrange to the lower range was found for all Ss for each stimulus presentation time.
It was noted that the optimum presentation speed (that yielding the lowest mean score) was 0.6, with the number of errors greatly increasing from 0.5 through 0.1. Although the presentation speed for the lowest mean errors varied from grade to grade (grade level curves are not presented), the three fastest presentation speeds (0.3, 0.2, 0.1) showed a sharp upward trend for all Ss.
Table 4
Rank Order of Difficulty for 26 letters, A to Z
Table 4 presents the letters evidencing the greatest number of errors in rank order and the probability of error for each. The letters chosen for the remediation program in phase 2 were those in the upper half in the total range of error probability.
Table 5
Rank Order Correlations of Letter Difficulty for Three Speeds of Presentation
Table 5 presents an additional analysis of rank order correlation made between various levels of presentation time: fast and medium, fast and slow, medium and slow, and total range with the fastest, medium, and slow ranges. These correlations indicated more stability for the rank order of letters between the fast and medium ranges and the total range than for the other comparisons supporting the choice of a 0.6-to-0.1 range as a basis for the choice of braille letters to be used in the experimental remediation phase.
The program of 96 random-ordered letters was used in the collection of normative data on all Ss, experimental and control, and was used again as a poststudy criterion measure. A summary of relevant data appears in Table 4.
Table 6
Data in Table 6 confirm the hypothesis that there are no significant differences between the groups in the three pretraining machine variables. In the posttreatment analyses a statistically significant difference (t = 2.23, p < 0.05) in favor of the experimental Ss was obtained on the braille letters which were remediated. Investigation of differences between the groups on other posttreatment variables of total program and nonremediated letters indicated no significant differences.
Further inspection of Table 6 also reveals that there were no significant differences within groups on pretreatment to posttreatment scores for the total program, remediated letters scored separately, and/or nonremediated letters scored separately. Improvement between pretasks and posttasks was shown by the experimental group in the total program and the remediated letters percent scores. A decrease in percent of correct responses from premeasurements to postmeasurements was shown in the nonremediated letters. The control group, on the other hand, showed a slight decrease in the percentage of correct responses between premeasurements and postmeasurements on the total program on all remediated and nonremediated letters treated separately. This decreasing trend more nearly approached significance for the nonremediated letters.
Effectiveness of Discrimination Training on Tachistoscopic Device
Additional analyses were undertaken to study the effectiveness of discrimination training on the TTD machine upon the speed and accuracy of reading traditional braille materials. T-tests measuring the differences between the means of two independent samples were computed. between experimental and control groups on each of the traditional braille variables. These data are summarized in Table 7, which indicates no statistically significant differences between the groups on any of the pretreatment variables. There were also no statistically significant differences between the groups on the posttreatment measurement variables. Thus', no significant treatment effects were noted on traditional braille reading ability of the Ss.
Table 7
In order to measure the possible relationship between machine and traditional measurement scores, rank order correlations were employed to investigate the relationship, if any, between machine presentation timing and the speed of reading traditional braille material. Investigators used correct response scores as a measure of this presentation time variable, and correlations were computed using not only the optimal presentation time indicated by phase 1 (0.6 sec) but also the premeasurement and postmeasurement presentation time of 0.3 sec. Results are summarized in Table 8.
Table 8
There was a significant coefficient of correlation (0.06) for the experimental group on the postmeasurement task. This is significantly different from 0 (t = 2.49, p < 0.05). This relationship does not exist in either of the premeasurement comparisons. It is, therefore, assumed, based upon the derived data, that training on the automated device had the effect of an increased relationship between these two variables. Inspection of the coefficients of correlation in t-values for the control Ss indicates that the relationship differed between the two pretask comparisons. The relationship between the optimal presentation time response scores and the words read for 5 min showed a correlation coefficient of 0.43 and a t-value of 1.57, which approaches a significant difference from 0. While this indicates a positive relationship, the correlation coefficients and t-values for both precorrelations and postcorrelations relating reading speed with the responses at 0.3-sec presentation time showed an increasing nonsignificant negative relationship between the two variables. It may be deduced, from data given in Table 8, that in the control group the fast readers made more errors than the slow readers-a condition which became more pronounced in the posttask correlations-while in the experimental group the positive relation between correct responses on the two tasks became more pronounced.
The primary aim of this study was to examine the relationship between the interval from stimulus presentation to response and the ability to discriminate the stimuli on a tachistoscopic programmed instructional device. The secondary aim was to utilize normative data in the development and implementation of a remedial program to determine the effects of machine training in rapid presentation and discrimination of single-cell configurations on several machine and traditional-task variables. Other variables considered in the analysis of the data included the relationship of fast, medium, and slow presentation times to the rank order difficulty of the 26 braille letter (alphabet) configurations; the relationship between speed of traditional braille reading, responses on the tachistoscopically presented configurations, and two presentation times (0.3 and 0.6 sec.); and the relationship between IQs and errors committed at the optimal presentation time of 0.6 sec.
The results of the study. reported here have implications for several of the recommendations of the aforementioned AAWB-AAIB Joint Uniform Braille Committee. A review of the literature of tactual stimuli presented tachistoscopically (12, 14, 17) or in programmed instruction (10, 14, 16, 17) provided no research information on the optimal speed of presentation of braille configurations. This investigation did provide some data on this subject. The graphing of the data obtained by time trials indicated that, for the total group of Ss, the number of errors committed at the perception time settings of 1.0 sec and 0.9 sec was greater than at settings of 0.7, 0.6, and 0.5 sec. The fewest errors occurred at 0.6 sec. The number of subject errors increased markedly on trials 0.4 through 0.1. The general trend of greatly increasing errors at the faster presentation time seems to indicate some perception speed parameters. There appears to be little doubt that the middle time range-from 0.7 sec to 0.4 sec inclusive-allows adequate time for perceptual difficulties. This does not mean, however, that individual braille readers perceiving configurations within the context of reading traditional braille materials do so at this speed.
Correlation Between IQ and Total Errors
Nolan16 has indicated that, for younger beginning braille readers, there seems to be a correlation between IQ and total errors made at the optimal. presentation speed of 0.6 sec. A Pearson Product Movement Correlation was computed to determine whether such a relationship existed. The resulting coefficient of correlation (-.45) showed a significant inverse relationship between IQ and errors on this measure (t = 2.49, p 0.05). These data suggest that Ss with higher IQs make fewer errors on tachistoscopically presented braille letters than Ss with lower intelligence.
Nolan5 has proposed several factors for the study of the relationship between tactual perception and beginning braille reading. Among the most important of these is the ability to perceive and reproduce tactual patterns. In the normative data phase of the study, the concern was with this variable, though not with beginning braille readers. Fertsch6 has indicated that traditional braille reading habits become established by the third grade level and do not seem to change noticeably with increased age and reading experience.
Future studies with automated learning devices should evaluate experimental programs with young beginning braille readers, since modifying braille reading behavior at these ages would seem most feasible. The tachistoscopically presented tactual stimuli of individual braille letter cells and the called for haptic responses of depressing a corresponding pattern of six response keys did not require a knowledge of stimulus meaning. Utilizing the concepts of programmed instruction-i.e., informational presentation that required Ss to make overt responses, the processes of immediate feedback, and individual pacing in exposing the experimental materials-enabled the Es to use the learning device for measurement purposes, rather than the traditional methods of measurement. All Ss possessed the necessary skills to operate the device efficiently.
The results of the data collected in phase 1 of the study on errors for the presentation times 0.6-0.1 sec inclusive for all 27 Ss provided a rank order of difficulty for the 26 letter-cells. This analysis was concerned only with the total error count for each letter and was not with the individual or grade-level variance of the Ss. We compared these results with rank orders obtained by Kederis12 who used tachistoscopical presentation and oral responses to word presentation, and found both similarities and marked differences. Of the 17 letters which were found most difficult, and which were subsequently used in a remediation study involving the same Ss, 11 were in agreement with both Kederis and Bruklen, although there was little agreement on the exact order of difficulty of the remaining six letters. Of the remaining six letters it was found that I, M, and E were in disagreement with Kederis who placed G X, and L in the 17 most difficult letter group, and F, V, and H were in disagreement with Burklin who placed D, B, and K in the 17-most-difficult range. If consideration was only given to the 10 most difficult letters, we again find considerable differences in rank order among the three studies being compared. Complete agreement on all three studies was found only in the letters R, Z, and N.
Further inspection of the letters on which a proportionately high number of errors were made indicates that the errors considered above were not homogeneous in suspected derivation. Ashcroft2 and Hanley8 have given considerable attention to reversal and alignment problems. These data confirmed the need for concern with this type of error.
The results obtained in this study further suggest considerable similarity in the particular letters which are problematic to the braille reader when knowledge of these letters is arrived at through either oral or tactual responses to the tactually presented stimuli. Although the rank order of difficulty varies considerably, the haptic mode of response (single successive perception of a whole configuration) provides results which seem to confirm most of the difficulty of reversible cells and cells containing four or more dots. Although the response required did not preclude the knowledge of the cell meaning, there was no way to determine, within the scope of this experiment, whether the error indices were in any way influenced by mediating cognitive responses or by tactual errors in making the response. It would seem to these investigators that errors-at least at the faster presentation speeds-would be most likely to result from faulty perception. This would lend support to the opinion of many students of the problem that the present braille code is lacking in varied characteristic forms.
The remediation training program used in phase 2 of the study was designed to remediate those letters in the upper range of probability of errors perception to their rank order probability of error. The resulting data did indicate a significant difference (t = 1.05, p < 0.05) in the speed of perception of braille configurations as a function of training. As the design of the remediation program was, in this case, based on the diagnosis of overall group errors at all presentation speeds, it would seem feasible that remediation programs constructed for individual subject profiles would demonstrate even greater differences. The proportion of remediated letters to non-remediated braille letters used as a measurement task was small. It was, therefore, unlikely that a measure of increased correct responses for either group would have been significant.
Device May Prove Beneficial as a Training Tool
The experimental Ss involved in the remediation phase of the study demonstrated an increase in reading rate which was 3.5 words per min faster than the control group evaluated during the same period. The experimental Ss also demonstrated an increase in traditional braille reading ability of 5.7 words per min over the course of the investigation. These results, while statistically insignificant, do have some practical significance, since during the evaluation period only nine 15 min periods of remedial instruction were given to each S. These data suggest that this unit (the automated learning device) may have potential beneficial effects in training people to read braille in the traditional manner.
1. Ashcroft, S. C. "Cues from Teaching Machines for Programmatic Educational Planning," The International Journal for Education of the Blind, 12: 51-59 (1962).
2. Ashcroft, S. C. "Errors in Oral Reading of Braille at Elementary Grade Levels," Ann Arbor, University Microfilms, 1960.
3. Ashcroft, S. C. "Programmed Instruction and Teaching Machines in the Education of the Blind," Proceedings of the International Conference on Technology and Blindness, Vol.1, Panel III, pp. 201-208 (1963).
4. Burklin, K. "Touch Reading of the Blind" (translated by Frieda K. Merry). New York: American Foundation for the Blind, 1932, 1-51.
5. Coffey, J. L. "Programmed Instruction for the Blind." Paper prepared at Battelle Memorial Institute under the auspices of the National Institute of Neurological Diseases and Blindness and the National Institute of Health, pp.1-29.
6. Fertsch, P. "An Analysis of Braille Readings," The Outlook for the Blind, 40: 128-131 (1946).
7. Graham, M. D. "Psychological Research and Braille: The Need for a Program of Research and Development," American Foundation for the Blind Research Bulletin, No.2, 1962, pp.94-114.
8. Hanley, L. P. "A Brief Review of the Research on Braille Reading," The International Journal for the Education of the Blind, 10: 1-6 (1961).
9. Hildreth, G. H. Teaching Reading: A Guide to Basic Principles and Modern Practices. New York: Henry Holt & Co., 1958.
10. Irazoqui, C. "A New Way to Learn Braille-The Audio-tact" Electro-Mechanical Consultants, Inc., New York.
11. Karp, S. "An Experiment Using Revised Stimuli Presentation," American Foundation for the Blind, Research Bulletin No.2 1962, pp.12-20.
12. Kederis, C. "The Legibility of Braille Characters," Unpublished master's thesis, University of Louisville, 1963.
13. Morris, J. E., and Nolan, C. Y. "Discriminability of Tactual Patterns," The International Journal for the Education of the Blind 9: 50-54 (1961).
14. Nolan, C. Y. "A Program of Research in Braille Reading," The International Journal for the Education of the Blind 8: 18-20 (1958).
15. Nolan C. Y. "Cues in the Tactual Perception of Patterns," Unpublished Progress Report. Louisville: American Publishing House, 1960.
16. Nolan, C. Y., Morris, I. E., Kederis, C. I., and Fieg K. E. "Annual Report for Fiscal 1964 of the Department of Educational Research," Louisville: American Printing House for the Blind, 1964.
17. Proceedings of the American Association of Workers for the Blind, July 27-August 1, 1958, p.57.
18. Stockton, G. H. "Effectiveness of Programmed Learning in Braille Instruction for the Adult Blind." Unpublished doctoral thesis, University of Wisconsin, 1965.
This investigation was supported in part by research and demonstration grant number RD1167-S from the Vocational Rehabilitation Administration, U.S. Department of Health, Education, and Welfare, Washington D. C.