Factors Affecting the Learning of Braille

SLATER E. NEWMAN, PH.D. ANTHONY D. HALL, M.S.

CHARLES J. RAMSEUR, M.ED. DARRYL J. FOSTER, B.A.

DAVID B. GOLDSTON, B.A. BURNEY L. DECAMP, B.A.

SUELLEN P. GRANBERRY-HAGER

JACKIE L. LOCKHART, B.A.

WILSON L. SAWYER, B.S.

JAMES E. WHITE, II, B.A.

Dr. Newman is professor of psychology at North Carolina State University. At the time this research was done, Mr. Hall was a graduate student in psychology at North Carolina State University, and the others were undergraduates, all in psychology, except Mr. Sawyer, who majored in computer science.

Department of Psychology, North Carolina State University, Raleigh, North Carolina 27650.

Abstract: In three experiments, the subjects' task was to learn the names of the first 10 symbols of the braille alphabet. In Experiments 1 and 2, visual examination of the symbols during the study trials and during the test trials enhanced learning. Of particular interest was the finding that subjects who studied the item visually but were tested haptically learned faster than did those who studied the items haptically and were also tested haptically. In Experiment 3, similar results were obtained when standard-size braille symbols were used during the study trials. When large braille symbols were used, visual study of the symbols had no effect. The authors discuss some implications of these results for braille training.

Although the braille system has existed for approximately 150 years, recent reviews of the literature (Foulke, 1979; Harley, Henderson, & Truan, 1979) reveal little information about how braille is learned. This is especially true in the case of people who are exposed to it for the first time. The literature can be readily summarized and appears to allow the following generalizations:

1. Faster learning occurs when braille symbols are presented visually or visually and haptically than when they are presented only haptically (Burns, 1944).

2. Crossmodal facilitation occurs both from haptic to visual (Merry, 1931) and from visual to haptic modalities (Burns, 1944; Hulin & Katz, 1934).

3. It makes no difference whether a symbol and its name are presented simultaneously or whether one immediately precedes the other (Russell, 1970).

4. Performance on a roughness discrimination test does not predict rate of braille learning (Russell, 1970).

5. Rate of learning braille with the left versus the right hand appears to be a complex function of age, gender, handedness, and of the order in which each of the hands is tested (Harris, 1980; Rudel, Denckla, & Spalten, 1974).

Since Merry's study involved only two subjects and Hulin's and Katz's only five, reliable information about the learning of braille is at present quite sparse.

This article describes three experiments recently completed in our laboratory. In each one, as in the studies just summarized, the subjects' task was to learn the names of certain symbols in the braille alphabet. In all three experiments, we investigated the crossmodal effects and obtained information about the effects of study time, item size, and test time on learning the names of the braille symbols. The main dependent variable was the number of correct responses during training.

Experiment 1

In the first study, we compared crossmodal and intramodal effects. This study differed from the crossmodal studies of Merry (1931), Hulin and Katz (1934), and Burns (1944) in one or more of the following ways:

First, we used a study-test procedure, which enabled us to evaluate independently the effects of study modality and test modality.

Second, the subject rather than the experimenter determined both study time and test time. This enabled us to determine the amount of time used on both the study trials and the test trials, although it also, of course, resulted in between-treatment differences in the amount of time used on both study trials and test trials. We chose this procedure, however, because we viewed the experiment as an exploratory one and were uncertain about what would be a reasonable amount of time to allow subjects for examining the items, particularly haptically.

Third, all subjects had the same number of trials during training and were tested again, half crossmodally and the rest intramodally. On this test, the experimenter controlled the time: all subjects had 10 seconds to respond with the name of each item as it was presented.

Method. Sixty-four sighted undergraduates at North Carolina State University served as subjects. All were right-handed, and none had had previous experience with braille. An equal number of males and females participated in each treatment; a separate counterbalanced order was used for assigning males and females to each treatment.

The students' task was to learn the name for the braille symbol for each of the first 10 letters of the alphabet. (Preliminary work suggested that most subjects would not learn the entire list during the first trial.) A Perkins Brailler was used to emboss each symbol on a white 5 x 8 card. The same items were used for both visual and haptic conditions.

The procedure was similar to the one often used in paired-associate learning experiments, except that the students paced themselves during both study and test trials. We told them initially about the modality in which the items would be presented in both study and test trials. We also said they could examine each item as long as they wished during both sets of trials. On study trials, they were instructed to let the experimenter know when they were ready for the next item by saying "next." On test trials, the experimenter presented the next item immediately after the students called out the name of the item displayed.

All students had three study trials, each followed by a test trial. The items were presented in a different order in each study and test trial. During the study trials, the experimenter called out the name of each item as it was presented. [For a description of the display apparatus, write to Anthony D. Hall, Department of Psychology, North Carolina State University, Raleigh, N.C. 27650.] When the items were presented visually on study or on test trials, the student looked at the symbol but did not touch it. During haptic presentation, the student's right hand was covered to preclude visual examination of the symbols; the cover, however did not contact the hand. The student explored each symbol by rubbing the flat portion of the right index finger over the symbol.

Following the third test trial, all students were tested again: half crossmodally, the rest within the same mode. On this test, each item was presented for 10 seconds. The order of the items differed from that used in any of the previous study or test trials. A tape recorder was used to record the students' oral responses during study and test trials. After the fourth test, we determined the students' handedness and then dismissed them.

For the training part of the experiment (the first three trials) the design was a 2 x 2, completely balanced factorial one in which the main independent variables were study modality (visual or haptic) and test modality (visual or haptic). Two sets of orders were used, each with half the students in each treatment,

For the fourth test, half the students in each treatment were tested in thesame mode as they had been tested before. The remainder were tested in the other mode, thus making for a 2 x 2 x 2 design.

Results

Table 1 presents the means for each treatment on each of the first three trials. A separate 2 x 2 analysis of variance was done for each variable: the number of correct responses, study time, test time, and total time. The results of each analysis are discussed below.

In relation to the number of correct responses, all three main effects were significant (study mode: F[1,60] = 6.82, p<.05; test mode: F[1,60] = 5.38, p<.05; trials: F[2,120] = 102.86, p<.001) but none of the interactions were. Inspection of the mean indicated that visual examination of the items during study and test trials enhanced learning, and, of course, performance improved over trials.

With regard to study time,the effects of study mode (F[1,60]=89.66), trials (F[2,120] = 65.14) and their interaction (F[2,120] = 24.58) were all significant (p<.001). Examination of the means shows that the students spent less time studying the items visually than haptically and as training proceeded. The visual haptic differencewas much larger on the first than on the third test.

Significant effects were obtained for the following in relation to test time: (study mode: F[1,60] = 7.74, p<.0l; test mode: F[1,60] = 80.73, p<.001; study mode x test mode: F[1,60] = 4.91, p<.05; trials: F[2,120] = 74.39, p<.001; test mode x trials: F[2,120] = 3.29, p<.05). Subsequent examination of the means for these effects showed that 1) the students took less time when they studied or were tested visually, 2) the difference between visual and haptic treatments on tests was greater for students who studied the items visually, 3) time spent on the test trials declined during training, and 4) this decrease was more marked for students who were tested haptically than for those tested visually.

In the results for total time, only the effects of study mode, (F[1,60] = 61.90), trials (F[2,120] = 123.92), and their interaction (F[2,120] = 16.16) were significant (p<.001), as was the case for study time. Students who studied the items visually took less time during training than did those who studied them haptically. Total time per trial declined during training, particularly for those who studied the items haptically.

Table 2 presents the rank-order correlations for item-difficulty and types of errors between pairs of treatments. For each treatment, the total number of correct responses was counted for each of the 10 symbols. Rank-order correlations for item difficulty were then done for these data for each pair of treatments. As shown in the table, all the correlations are positive and four are significant (p<.05). The order of difficulty of the items (summed across treatments) from most to least difficult was D, J, H, I, E, F, C, B, G, and A.

Table 2. Rank-order Correlations for Item Difficulty and Types of Errors between Pairs of Treatments: Experiment 1.

Treatments Items Correct Types of Errorsc

V-V & V-H .305 .596

V-V & H-V .573a .649

V-V & H-H .330 .558

V-H & H-V .710a .536

V-H & H-H .814b .665

H-V & H-H .864 b .599

ap<.05 cp<.0l in all cases.

bp<.0l

The number of times each item elicited the name for every other item was determined for each treatment and rank-order correlations for types of errors were done for each pair of treatments. As Table 2 indicates, all the correlations are positive and all are significant (p<.0l). The most frequently occurring errors (summed across treatments) from most-to-least frequent were EI and JD; CB and IE; DJ; DH; FD, HJ, and JH; and IC, where the first letter stands for the item presented and the second letter for student's response.

Table 1. Mean Study Time, Test Time, and Total Time (in seconds): Correct Responses on Each Trial; and Efficiency Indexa Experiment 1.

Treatment Mode Test 1 Test 2 Test 3

Time Correct Time CorrectTime Correct Efficiency

Study Test Study Test Total Responses Study Test Total Responses Study Test Total Responses Index

Visual Visual 56.4 53.9 110.3 7.4 51.0 41.8 92.8 8.3 35.9 32.5 68.4 9.3 10.9

Visual Haptic 50.9 98.3 149.2 5.2 50.4 77.1 127.5 6.4 40.4 65.4 105.8 8.9 18.6

Haptic Visual 147.3 72.4 219.7 4.8 108.2 60.2 168.4 6.8 85.1 45.8 130.9 8.4 25.9

Haptic Haptic 167.2 102.1 269.3 4.8 90.4 76.3 166.7 6.2 75.1 68.1 143.2 7.6 31.1

a Total time during training/Total correct responses.

We also examined the error patterns by determining whether the students responded with the name for a symbol that had more than, less than, or the same number of dots as the symbol presented. In the 90 cells of the error matrix for the items used in this study, there were 33 possibilities that an error would involve naming a symbol with more dots, another 33 cells for errors involving the name of a symbol with fewer dots, and 24 cells where the error involved naming a symbol with the same number of dots as the symbol provided. The data show that the mean errors per cell for these three types of error were 2.76, 3.58, and 14.83, respectively. The proportions of these errors (to total errors) were .161, .209, and .630, respectively. Thus students were more likely to err by using the name for a symbol that had the same number of dots as the symbol presented than by using the name of a symbol with either more or fewer dots.

The mean number of correct responses for the eight groups on test 4 are listed in Table 3. An analysis of variance of these data showed no significant effects (p>.05).

Table 3. Mean Number of Correct Responses for Each Treatment Mode on Test 4: Experiment 1

Study Mode Test Mode Mean

Tests 1-3 Test 4

Visual Visual Visual 8.9

Visual Visual Haptic 7.6

Visual Haptic Visual 8.3

Visual Haptic Haptic 7.5

Haptic Visual Visual 8.4

Haptic Visual Haptic 8.0

Haptic Haptic Visual 7.4

Haptic Haptic Haptic 6.4

Discussion

Analyses of the results during training indicated that when the students studied the items visually rather than haptically, they took less time on both study and test trials and made more correct responses. Visual presentation of items during test trials also resulted in quicker and more accurate responses, although study time was not affected. Accuracy improved with training, and this was accompanied by students' using less time during both study and test trials. Although the decline was more marked in the haptic (study and test) conditions, the students were still taking more time on the third test trial than were those who studied or were tested visually. On the fourth test, however, when test time was equated across treatments, the effects of neither the study nor test mode during training affected performance, regardless of whether the fourth test was in the same mode as the tests during training or in a different mode. Concomitantly, performance on the fourth test was independent of the total amount of time spent during training.

The results for performance during training suggest that visual presentation, whether during study or test trials, is more efficient than haptic presentation. The last column in Table 1 presents the means for a derived measure (admittedly imperfect) in which the total time during training for a condition is divided by the total items correct during training in that condition. Examination of these means suggests that the most efficient condition is the visual-visual condition, followed by the visual-haptic, haptic-visual, and haptic-haptic conditions. Of special interest is the fact that the students who were tested haptically but studied visually did better than did those in the haptic-haptic group. Thus, although students in the visual-visual group performed best, not all students who studied and were tested in the same mode did better than did students who studied and were tested in different modes.

The rank-order correlations are also of some interest. They suggest that, at least for the items used here, there is some consistency, both in item difficulty and in the type of errors made, even across modalities.

The use of a self-pacing procedure in Experiment 1 resulted in the confounding of time and accuracy. Thus the results are not as definitive as they might have been if the experimenter had controlled both study time and test time. This was done in the second experiment, which investigated the effects of study time and test time as well as study and test modalities.

Experiment 2

In Experiment 2, the experimenter controlled the rate at which the items were presented on both study trials and on test trials. Thus, the study trials for half the subjects in each treatment lasted 5 seconds; for the other half, the trials lasted 10 seconds. The same rates were also used for test trials. In other respects the procedure was similar to that used in Experiment 1, except that all subjects had five rather than three trials during training and there was no post-training test corresponding to test 4 in the first experiment.

Method. Sixty-four right-handed male undergraduates at North Carolina State University participated in this study. All of them had normal vision, either with or without glasses, and had no previous experience with braille. They were assigned to treatment groups by a counterbalancing procedcure, and each was tested individually.

The design was a 2 x 2 x 2 x 2 in which the independent variables were Study Modality (visual or haptic), Test Modality (visual or haptic), Study Time (5 or 10 seconds), and Test Time (5 or 10 seconds).

All students had five study trials, each trial followed by a test trial. In each study trial, the experimenter presented the item and called out its name, and the student examined it either visually or haptically for either 5 or 10 seconds. On the test trials, the items were presented either visually or haptically for 5 or 10 seconds. During this interval, the student examined the symbol and called out its name. The items were presented in a different order during each study and test trial. Two different sets of orders were used, each with half the students in each treatment. All oral responses were tape-recorded.

Results. Table 4 presents the means for correct responses over all trials for each treatment. An analysis of variance was carried out on these data, and the following significant effects were obtained: study mode: F(1,48) = 75.95, p<.001; test mode: F(1,48) = 28.35, p<.001; study time:F(1,48) = 11.09, p<.0l; and study mode x test time: F(1,48) = 4.55, p<.05. Examination of the means showed that the students performed better when the items were presented visually either on study trials or on test trials and, of course, when they had more time to study. The difference between visual and haptic study modes was greater at the shorter test time, whereas the difference between the visual and haptic test modes was greater at the longer test time.

Table 4. Mean Number of Correct Responses during Training:

Experiment 2 (time in seconds)

Study Study Visual Test Mode Haptic Test Mode

Mode Time Test Time Test Time

5 10 5 10

Visual 5 45.3 46.0 24.8 43.0

10 46.5 49.3 37.8 40.3

Haptic 5 30.3 25.5 19.8 22.8

10 36.0 32.8 29.0 27.5

Duncan's test indicated that the students who studied the items visually and were tested visually did better than students in the other three groups. The only other significant difference was that students who studied the items visually and were tested haptically performed better than did those who studied the items haptically and were also tested haptically. Examination of the means on each trial showed that these between-treatment differences appeared on the first trial and continued throughout training.

The total number of correct responses for the 10 symbols for each of the four combinations of study modality and test modality (collapsed across study and test times) was determined and rank-order correlations were performed for each pair of treatments. The results are presented in Table 5. All six correlations are positive and all are significant (p<. 05). The order of difficulty of the items (summed across all treatments) from most to least difficult was H, D, F, E, J, I, C, G, B, A.

Table 5. Rank-order Correlations for Item Difficulty and Types of Errors between Pairs of Treatments: Experiment 2.

Treatments Items Correct Types of Errorsc

V-V & V-H .873b .430

V-V & H-V .700a .530

V-V & H-H .639a .551

V-H & H-V .776b .624

V-H & H-H .727a .517

H-V & H-H .939b .799

ap<.05 cp<.0l in all cases.

bp<.0l

An error analysis similar to that in the first experiment was done for each of the four combinations of study and test modality (collapsed across study and test times), and rank-order correlations were again calculated for each pair of treatments. These correlations are presented in Table 5. All these correlations are positive and, again, all are significant (p<.0l). The most frequently occurring errors (summed across all treatments), from most to least frequent, were DF; CB and HD; IE; DH; EI; FH, GH, and HF; and FD and HJ.

The mean number of errors per cell for "added-dot," "missed dot," and "same dot" errors were 6.00, 6.52, and 19.79, respectively; the proportions of these types of errors to total errors were .223, .242, and .535, respectively.

Discussion

The results of this experiment indicate that visual examination of the braille symbols during study trials or test trials leads to better performance than does haptic examination of the symbols. The orders of means for the four modal-combination groups in this experiment parallel those from Experiment 1 when number correct, study time, total time, or the efficiency index is used. The better performance of the visual-visual group in this study also replicates Burns's findings (1944).

The results for item difficulty and for errors are also similar to those obtained in the first experiment. The results of both studies suggest that for the items used in these experiments (i.e., the letters A-J), the order of difficulty, and the error patterns are relatively consistent between treatments and between experiments.

Of perhaps greatest interest are the results comparing the performance of the visual-haptic group with that for the haptic-haptic group. This comparison showed that students who studied and were tested in the same mode (the haptic-haptic group) did less well both on the first trial and throughout training than did students who studied and were tested in different modes (the visual-haptic group). Several models of paired-associate learning (Greeno, 1970; McGuire, 1961; Newman, 1964) offer a possible explanation for these findings; visual (as compared with haptic) examination of the items during the study trials leads to the earlier establishment of a stable discriminable encoding for each item. Such an encoding is still appropriate when the items are presented for haptic examination on test trials. In the third experiment, we assessed one implication of this explanation.

Experiment 3

In Experiment 3, all three independent variables were manipulated during the study trials, The major variable of interest was the size of the braille cell. For some subjects, the items were presented in standard-size braille during the study trials; for the remaining subjects, the items were presented in large (or jumbo) braille. In other respects, several conditions replicated those of Experiment 2. Thus during the study trials, half the subjects examined the items visually and the other half examined them haptically. Also half had 5 seconds to examine the items while the remainder had 10 seconds. On the test trials, all items were presented in standard braille for 10 seconds of haptic examination.

Our main prediction was that a significant interaction would occur between item size and study mode. We expected the difference between the visual and haptic study conditions to be substantially greater when standard braille (as compared with large braille) was presented during study trials. This would occur if the large braille items were sufficiently discriminable that visual presentation would provide little additional benefit. The standard braille items, however, being relatively less discriminable, would profit from visual examination.

Method

A 2 x 2 x 2 design was used in which size of braille (standard or large), study time (5 or 10 seconds), and study modality (visual or haptic) were manipulated. In addition, two different sets of orders were used, half with each treatment. Each of six experimenters ran one subject in each of the 16 treatments.

Ninety-six right-handed, sighted, male undergraduates at North Carolina State University- none of whom had had previous experience with braille-were randomly assigned to treatments in a counterbalanced order. Again, each student was dealt with individually.

The procedure was similar to the one used in Experiment 2 with two major exceptions. During the study trials, half the students in each treatment group were exposed to the items in standard braille; the other half were exposed to the items in large braille. During the test trials, the items were presented to all students in standard braille for haptic examination for 10 seconds. The braille items were presented on Thermoform sheets rather than on the 5 x 8 cards used in the first two experiments.

Results

The mean number of correct responses for each treatment are presented in Table 6. An analysis of variance revealed the following significant effects: study modality: F(1,88) = 21.55, p<.001; study time: F(1,88) = 11.37, p<.001; study modality and study size: F(1,88) = 8.10, p<.0l. Examination of the means showed that the mean for the visual mode exceeded the one for the haptic mode when the items were presented in stan-dard braille but not in large braille. Comparisons also showed that large braille enhanced the performance of the haptic students but not the visual students. Again, those who had 10 seconds to examine the symbols performed better than those who had only 5 seconds. For each comparison just described, the differences appeared during the first trial and continued throughout training.

Table 6. Mean Correct Responses during Training: Experiment 3

Study Mode Study Time Item Size

(in seconds) Standard Large

Visual 5 35.9 33.7

10 39.5 37.8

Haptic 5 23.5 30.7

10 29.3 35.4

Analyses of item difficulty and types of errors were done as in Experiments 1 and 2. This time the number correct and the types of errors were collapsed across study times. The rank-order correlations for both are presented in Table 7. Again, all the correlations are positive for item difficulty, and all are significant (p<.0l). The order of the items (summed across treatments) from most to least difficult is D, H, J, F, I, E, G, C, B, and A.

Table 7. Rank-Order Correlations for Item Difficulty and for Types of Errors Between Pairs of Treatment Combinations: Experiment 3

Treatment Combinationsa Item Difficultyb Types of

Errorsb

Visual-large & Visual-standard .988 .793

Visual-large & Haptic-large .952 .761

Visual-large & Haptic-standard .945 .753

Visual-standard & Haptic-large .936 .735

Visual-standard & Haptic- standard .924 .787

Haptic-large & Haptic-standard .948 .818

acollapsed across study-time.

b p<.0l in all cases.

The correlations for types of error are also all positive and significant (p<.0l). The most frequently occurring errors (summed across treatments) from most to least frequent were JG; FD and JH; DF; EI; DH and HD; HJ; CB; DG and HF. The number of errors per cell for "added dots," "missed dots;' and "same dots" were 9.06, 9.85, and 26.33, respectively; the proportions of these errors were .238, .258, and .504.

Rank-order correlations for item difficulty and for type of errors for the data (based on all subjects) of each pair of experiments are presented below (all correlations for both measures are positive and signficant, p<.0l):

Experiments Difficulty Type of Error

1 and 2 .818 .801

1 and 3 .897 .823

2 and 3 .915 .893

Thus, for the braille symbols used in these experiments, there is a considerable degree of consistency between experiments in relation to both order of difficulty and type of errors.

Discussion

The fact that the visual group performed better than the haptic group for standard-size braille but not for large braille supports our main prediction. It also suggests that visual presentation of the symbols during the study trials facilitates subsequent haptic test performance by enhancing the discriminability of the items. Therefore, as Tulving (1979) proposed, retrieval of an item is a function not only of the relationship between the encoding and retrieval contexts for that item but also of the uniqueness with which that item is encoded.

The results for study time also replicated those of Experiment 2 - i.e., the longer study time enhanced test performance. The results for item difficulty and patterns of errors were also similar to those obtained in the first two experiments, indicating some stability in performance over a large number of experimental conditions.

Conclusions

The results of the three experiments permit us to draw the

following conclusions:

  1. Visual examination of items during study trials enhances subsequent test performance, independent of the test modality. The effect is greater with the shorter test time (5 versus 10 seconds) and with standard but not with large braille.
  2. Visual examination of items during test trials enhances test performance independent of the study modality. The effect is greater with a longer test time (10 versus 5 seconds).
  3. Longer study time aids learning independent of the study mode and the size of the braille cell.
  4. Longer test time is accompanied by better performance during learning.
  5. Each of the above effects is evident throughout training.
  6. The order of difficulty of the items A-J is relatively stable

across a fairly wide range of experimental conditions.

7. The patterns of errors are also quite similar across a variety of experimental conditions.

8. When subjects make an error, they are more likely to respond with the name of a symbol that has the same number of dots as the stimulus than with the name of a symbol that has either fewer dots or more dots.

Three findings deserve special attention: (1) the incidence of various types of errors, (2) the better performance of the visual-haptic group than of the haptic-haptic group with standard braille, and (3) the faster learning in the haptic-haptic condition when large braille was used.

Nolan and Kederis (1969, Experiment 1) reported that "missed dot" errors accounted for 86 percent of the incorrect responses in a study in which blind, skilled braille readers were asked to identify 55 different braille symbols. They also reported that the number of "missed dot" errors was directly related to the number of dots in a symbol and greatest for symbols with dots on the bottom row.

In our three experiments, "missed dot" errors accounted for not more than 25 percent of the total number of overt errors. This difference in outcome may derive from between-experiment differences in subjects' experience with braille and, to some extent, at least, from differences in the set of stimuli used in the two studies. Our symbols had fewer dots than Nolan's and Kederis's (means of 2.50 and 3.31 dots per symbol, respectively) and used only the two top rows of the braille cell, whereas theirs used all three rows. With more dots per symbol and with all three rows used, the error patterns of our subjects might have been more similar to theirs.

The faster learning with standard-size braille in the visual-haptic than in the haptic-haptic condition may be of more than theoretical interest. These results suggest that vision may be of some advantage in the study of braille, even when subsequent reading of braille is done haptically. For example, if a person has reason to believe that he or she may become so visually handicapped that a knowledge of braille might be useful, these results suggest that it would be desirable for that person to attempt to learn braille visually while he or she is able to do so. However, the following may influence the effectiveness of this training: (1) the motivation to learn braille may be lower before such learning becomes necessary and (2) learning braille visually may lead to visualizing the braille symbols when braille reading must be done haptically, which could, in turn, reduce efficiency. Our results also favor allowing students with residual vision to use that vision in learning braille. A survey of experienced British teachers of braille produced mixed opinions (48 percent in favor) about this use of residual vision (Tobin, 1971).

The fact that our 'subjects learned faster when large rather than standard braille was used in the haptic-haptic condition may also have implications for training procedures. Tobin's survey showed that approximately half the British teachers (53 percent) advocated using large braille during the early stages of instruction, whereas the remainder advocated standard-size braille. Our results seem to support the use of large braille during early training. An experiment reported by Tobin (1971) also provides some support for the use of large braille.

We undertook our research in the hope of becoming better able to identify the processes involved in learning braille and how they operate. We recognize that sighted adults, even when required to examine the symbols haptically, may behave quite differently than congenitally blind children or adults. We also recognize that they may differ from children or adults who learned to read print before becoming blind. Finally, we are also well aware that behavior during learning under restricted laboratory conditions may differ considerably from behavior in a more natural setting (Neimark, 1976; Newman, 1957). Thus we do not expect our results to be directly applicable to the design of programs for teaching braille, either for those who will be learning braille haptically or for those who will be learning it visually. We hope, however, that this will not always be so and that the results of our research program will contribute to a general understanding of learning and memory in the haptic domain and perhaps to a subsequent understanding of the processes involved in learning to read print. 0

We thank Dora Grimes, Governor Morehead School for the Blind, for advice, Gene Anthony, Division of Exceptional Children, North Carolina Department of Public Instruction, for making available to us a Perkins Brailler,. John Calloway, Governor Morehead School for the Blind, for Thermoform supplies and the use of Thermoform equipment and Donald Mershon for suggestions about these experiments. We appreciate also the comments of Laurel Holloman, Donald Mershon and Rachel Rawls about an earlier version of this manuscript. A paper describing Experiment 1 was presented at the 1979 meeting of the Psychonomic Society in Phoenix, Experiment 2 was described in a paper presented at the 1980 meeting of the Southeastern Psychological Association in Washington and Experiment 3 in a paper presented at the 1980 meeting of the American Psychological Association in Montreal.

References

Burns, B. K. A comparative study of visual and tactual learning of braille. Unpublished master's thesis, Wittenberg College, Springfield, Ohio, 1944.

Foulke, E. Investigative approaches to the study of braille reading. Journal of Visual Impairment and Blindness, 1979, 74, 298-308.

Greeno, J. G. How associations are memorized. In D. A. Norman (Ed.), Models of human memory. New York: Academic Press, 1970.

Harley, R. K., Henderson, F. M., & Truan, M. B. The teaching of braille reading. Springfield, Ill.: Charles C Thomas, 1979.

Harris, L. J. Which hand is the "eye" of the blind?: A new look at an old question. In J. Herron(Ed.),Neuropsychology of left-handedness. New York: Academic Press, 1980.

Hulin, W. S., & Katz, D. Transfer of training in reading braille. American Journal of Psychology, 1934, 46, 627-631.

Lowenfeld, B., Abel, G. L., & Hatlen, P. H. Blind children learn to read. Springfield, Ill.: Charles C Thomas, 1969.

McGuire, W. J. A multiprocess model for paired-associate learning. Journal of Experimental Psychology, 1961, 62, 335-347.

Merry, R. V. An experiment in teaching tactual reading to seeing subjects. Pedagogical Seminary and Journal of Genetic Psychology, 1931, 39, 409-413.

Neimark, E. D. The natural history of spontaneous mnemonic activities under conditions of minimal experimental constraint. In A. D. Pick (Ed.), Minnesota symposia on child psychology. Minneapolis: University of Minnesota Press, 1970.

Newman, S. E. Student vs. instructor design of study method. Journal of Educational Psychology, 1957, 48, 328-333.

Newman, S. E. A mediation model for paired-associate learning. In J. P. DeCeeco (Ed.), Educational technology. New York: Holt, Rinehart & Winston, 1964.

Nolan, C. Y. & Kederis, C. J. Perceptual factors in braille word recognition. New York: American Foundation for the Blind, Research Series No. 20, 1969.

Rudel, R. G., Denckla, M. B., & Spalten, E. The functional asymmetry of braille letter learning in normal, sighted children. Neurology, 1974, 24, 733-738.

Russell, H. K. The effect of order of presentation on the programmed learning of braille. Dissertation Abstracts International, 1969-70, 31, 5038B. (University Microfilm No. 71,4420).

Tobin, M. J. Programmed instruction and braille learning. Birmingham, England: Research Centre for the Education of the Handicapped,1971.

Tulving, E. Relation between encoding specificity and levels of processing. In L. S. Cermak & F. I. M. Craik (Eds.), Levels of processing in human memory. Hillsdale, N.J.: Erlbaum, 1979.