Orthographic Effects of Braille and Print in the Auditory Modality
L. Pring
Abstract: Three experiments were carried out to explore the influence of orthographic codes in the comprehension by blind subjects and sighted subjects of words presented auditorially. In the first experiment, the task was to detect a target word, which rhymed with a previously given cue word, and to press a reaction-time key when it was heard. Rhyme pairs that shared similar orthographic codes (e.g., lake and take) were detected by both groups significantly faster than were rhyme pairs that did not share similar orthographic codes (e.g., lake and ache). The results suggested that a braille derived orthographic code, as well as a print-derived code, affects auditory word processing. The second and third experiments compared the two groups' speeds in detecting letters presented auditorially in letter-number pairs.
In learning to read braille, blind children have to overcome many difficulties. One problem is that there is little redundancy in the script and therefore the danger of misperceiving individual letters is high (Pring, 1984). Another difficulty is that most children are taught Grade 2 braille, which, for the sake of speed, uses abbreviations or contractions of certain clusters of letters. Thus, for instance, the contraction used to denote one is also used for the word honey. The contractions do not refer to sound units but to orthographic, or spelling units. Ambiguity may also arise because Braille, like print, has some written symbols that can be represented logographically as well as alphabetically, although Henderson (1982) showed that there may be several different bases for distinguishing between the two representations. The one that is emphasized here is the degree of correspondence with the spoken language.
There is some evidence that written symbols that contain no explicit information about sound, such as Chinese script or the logographic representation of numbers (e.g., 2 and 4) may be "read" via different processing strategies than are symbols that contain information about sound, for instance, a syllabic script like Japanese Kana or the alphabetic representation of numbers (e.g., two and four).
The author wishes to acknowledge the special contributions of Dr J. Rusted, who helped devise and test Experiment 1, and Dr B. Hermelin for her helpful criticisms of and encouragement in the research, and to thank the staff and children of Chorley Wood College for Girls for their participation in the study.
The logographic as well as the alphabetic representation of numbers in English has, to a limited extent, allowed a comparison of the reading strategies that can be adopted to deal with logographic and alphabetic symbols. Furthermore, it has been suggested that brain damage can be selective in its effect on alphabetic and logographic processing (Hecaen & Kremin, 1976; Sasanuma, 1980).
The two forms for representing numbers also occur in braille script. A number can be denoted in an alphabetic form (e.g., three: in uncontracted braille) or 3:
It is interesting to note, however, that braille is similar to Hebrew in that the numbers 1 to 9 share the same physical representations as the first nine letters of the alphabet (A to I), although in braille this letter-number confusion is clarified for the reader by the sign indicating that a number follows.
Objectives of the study
The question addressed in this article is whether the shared physical representations of certain letters and numbers in braille influence the performance of a cognitive task. For example, there has been extensive discussion about how far the influence of features derived from the reading process (orthography) affects the auditory recognition of words (Doctor, 1978- Holley-Wilcox, 1977; Swinney et al., 1979; Tanenhaus & Seidenberg; Warren, 1974). Donnenwerth-Nolan (1981) provided strong evidence that the influence of both semantic and orthographic codes can be detected in an auditory rhyme-matching task.
In their experiment, the task was to press a button on recognition of the target rhyme (in this case, tie). Donnenwerth-Nolan, Tanenhaus, and Seidenberg (1981) found that pie= tie pairs of words were recognized as rhymes significantly more quickly than were pie= rye pairs; the critical feature, they argued, is that in the pie= tie pairs, the orthographic forms overlap, whereas in the pie= rye pairs, the orthographic forms differ. The researchers suggested that a word, whether heard or seen, can automatically activate equivalent internal representations, in this case, orthographic representations. In view of this finding, the study described in this article attempted to ascertain whether the letter-number confusion in braille exists in an auditory task.
The first objective was to determine whether congenitally blind readers of braille would be influenced by an orthographic code in the auditory modality. The question was whether the blind subjects, in carrying out a rhyme task in this mode, would utilize the tactually derived braille code. If the blind subjects should fail to differentiate the orthographically similar rhyme pairs (pie= tie) from the orthographically different rhyme pairs (take=ache), then one would have to assume that a different sensory modality and script result in qualitatively different coding processes (see O'Connor & Hermelin, 1980).
Experiment 1
The purpose of this experiment was to ask whether Donnenwerth-Nolan, Tanenhaus, and Seidenberg's (1981) findings of the effect of the reading experience on an auditory processing task would extend to blind readers of braille. Braille uses the same orthography as visible writing, in the sense of mapping graphemes onto phonological sequences, but it uses different symbols to stand for the graphemes. Also, the format of braille, in contrast to print, is relatively invariant; there are no differences in upper or lower case letters and, since braille uses a predeterminant cell size, no differences in handwriting. Therefore, the following question arises: in spite of such different script characteristics, is the same processing found in sighted readers also found in blind readers?
Subjects
Twelve congenitally blind girls from a residential elementary school for blind and partially sighted students participated in the experiment. They were all skilled readers who were in advanced-level courses, which reflects their high standard of academic achievement. The girls, aged 17-19 years, were either totally blind or had some light perception but no pattern recognition. 'I'welve first-year undergraduate students at Goldsmith's College, University of London, also took part in the experiment. They were aged 18-20 years and were paid a nominal fee for participation.
Materials
Two audiotapes of words were prepared, each consisting of instructions to the subject, 8 practice trials, and 80 experimental trials. Each trial consisted of a cue word followed briefly by a monitor list of 3 words, one of which (the target) rhymed with the cue word.
The target could either be similar to the cue word in its visual, tactual, and orthographic features (e.g., pie= tie) or dissimilar (e.g., pie= rye). Tape 1 consisted of 12 experimental trials with similar pairs and 12 experimental trials with dissimilar pairs. The similar and dissimilar pairs did not share the same word cue. Tape 2 consisted of alternate pairs to those in Tape 1; that is, if Tape 1 contained flame= name, Tape 2 would contain the corresponding dissimilar pair, flame = claim.
Thus, each tape contained 24 critical trials (12 similar pairs and 12 dissimilar pairs) and 56 filler trials. Although the filler trials also contained similar and dissimilar pairs, the target appeared either in the first or third position in the monitor list. Only on critical trials did the target appear in the second position-the position least affected by shifts of attention and expectancy.
The 48 critical pairs of words (cue and target) used in the experiments were taken from Donnenwerth-Nolan, Tanenhaus, and Seidenberg (1981) and were matched for length, the frequency of rhyme production, and word frequency. Unusual spellings, uncommon words, homophones, homographs, and homonyms also were avoided. The other words used in the monitor lists were common words, semantically unrelated to the target words.
Apparatus
A Phillips stereo-spatial radio-cassette recorder was modified for use in the experiment. The recorder was connected to a digital timer and to a reaction-time (telegraph) key held by the subject. The cassette tapes were prepared so that the cue word was followed by a two-second pause followed by three words in quick succession.
Electronic circuitry monitoring voltage levels were manually primed simultaneously with the onset of the first word and prior to the target word, so that, on the initial vocalization of the target word, a signal was simultaneously registered on the tape.
This signal, unheard by the subject, activated the digital timer. The timer stopped when the subject depressed the reaction-time key. Thus, the reaction time noted by the experimenter measured the time from the presentation of the target word to the response made by the subject, indicating that the target word had been identified.
Design
There were two lists of rhyme pairs (List A and List B), and alternate subjects listened to either list. Both lists had orthographically similar and dissimilar word pairs. To control for possible variations in individual stimuli, corresponding word pairs were used in the similar category of one list (e.g., tie =pie) and the dissimilar category in the other list (e.g., rye =pie).
An ANOVA was then conducted to compare the blind subjects' and the sighted subjects' recognition of the similar and dissimilar rhyme pairs; an item analysis was carried out as well.
Procedure
Each subject was tested individually. The subject sat at a desk and was given a reaction-time key to hold. The cassette tape recorder was then started. The tape first presented the instructions, gave 8 practice trials, and then proceeded with two sets of 40 trials, a pause occurring in the middle of the experiment. The recorder could be switched off at any time so the experimenter could be sure the subject understood the instructions before beginning the experiment. The instructions explained the rhyme-detection task. They noted that a word would be presented, followed by a pause and then by three words, one of which would rhyme with the cue word; when the subject recognized the word that rhymed with the cue word, the subject should depress the reaction-time key as fast as possible.
Results
Table 1 presents the mean reaction times and standard deviations in milliseconds for the similar and dissimilar rhyme pairs in Experiment 1. Errors were negligible and occurred on rare occasions when the subject failed to detect a rhyme until shortly after the trial had finished. A comparison of performances on Tape 1 and Tape 2 revealed no significant differences.
Table 1.. Mean rhyme-detection times and standard deviation in milliseconds of the similar and dissimilar pairs in Experiment 1.
Group Similar Dissimilar
Pairs Pairs
Sighted Subjects
Mean 419 457
SD 95 87
Blind Subjects
Mean 426 515
SD 149 169
A split-plot ANOVA was carried out with one between-subject variable (group) and one within-subject variable (pair type). The analysis revealed that the speeds of the overall group did not vary but that all the subjects were significantly faster in responding to similar rhyme pairs than to dissimilar ones F(1,21) = 17.57, p<.004. The interaction did not approach significance F(l, 21) = 1.71, p<.21. Related t-tests supported the finding that the group of blind subjects and the group of sighted subjects responded significantly faster to the similar than to the dissimilar word pairs.
The item analysis, which was carried out for each group of subjects, also revealed significant differences between the two types of rhyme pairs: blind subjects, t(22)=3.223, P<.0l; sighted subjects, t(22) = 2.093, p < .05. Eleven of the blind subjects and four of the sighted subjects responded better to the similar word pairs.
Discussion
In this experiment, there were no overall differences in the speed with which the blind subjects and the sighted subjects detected rhymes. Thus, the results of Donnenwerth-Nolan, Tanenhaus, and Seidenberg (1981) were replicated with respect to the identification of similarly and differently spelled pairs of words. Both the sighted subjects and the blind subjects responded significantly faster to the similar pairs than to the dissimilar ones. This finding lends support to Donnenwerth-Nolan, Tanenhaus, and Seidenberg's (1981) suggestion that judgments about rhymes, which would seem, at first sight, to involve only phonological code matching, appear also to involve the activation of other codes connected with experiences with words in other contexts. Thus, the activity of reading is not irrelevant to considerations of auditory word processing.
Although Experiment 1 provided evidence that a braille-derived orthographic code can influence auditory processing, the question remains whether an orthographic code in braille subsumes more script characteristics than it does in print because of the invariant nature of the graphemic symbols. If that is so, then it might be useful to investigate whether the influence of shared physical representation of numbers and letters in braille has a psychological effect on the performance of an auditory task.
Experiment 2
Experiment 2 aimed to investigate whether the shared physical representations in braille of letters and numbers would interfere with the auditory recognition of letters and numbers. The experiment involved a letter-detection task in which the crucial comparison was between letters and numbers that share their braille representations and those that do not.
The subjects were asked to listen to pairs of items-two numbers, two letters, or one number and one letter. Their task was to press a button every time a letter was present in a pair. Since the letters A to I and the numbers 1 to 9 are represented by the same symbol in braille but not in print, the responses of the blind might well differ from those of the sighted. Thus, the pairs B and 2 (shared in Braille) and B and 7 (different in braille) might result in different reaction times for the blind subjects but not for the sighted subjects. However, this would be the case only if the representations of numbers evoked by the blind subjects was in the logographic form (2), not in the alphabetic form (two).
Although it was not known whether the blind subjects would activate a logographic code representing a digit rather than an alphabetic code, it was hoped that hearing a letter first might activate the letter's orthographic representation. If the blind subjects subsequently heard a digit that shared the orthographic representation, then some interference in their response could be predicted.
If the digit was presented first in the pair of items and its representation was in alphabetic form, then the orthographic code of the subsequent letter would not interfere with their response (this assumption was tested directly in Experiment 3).
Thus, if an interference effect of shared braille representations was to emerge in letter-number pairs, the presentation of a letter as the first in the pair would strongly demonstrate such an effect because only the letter needs to be detected for the decision to be made.
Subjects
Eleven of the 12 congenitally blind girls who participated in Experiment 1 also participated in Experiment 2; the twelfth girl was unavailable for testing. A different group of sighted readers who previously were included in another study volunteered for Experiment 2. They consisted of 10 women and 1 man, aged 19-35 years, the majority of whom were students.
Materials
One cassette tape was recorded for the experiment; it contained instructions a practice session, and 72 experimental trials. Pairs of items were recorded with 250ms interstimulus intervals in a pair, which was sufficiently brief to ensure that the subjects would hear both items before attempting to respond. The interval between pairs of items was 3 seconds to ensure time for the subjects to respond.
There were 9 letter-number trials that have shared representations in braille (A and 1, B and 2, C and 3, D and 4, E and 5, F and 6, G and 7, H and 8, and I and 9),, and nine letter-number trials that do not share the same representation in braille (e.g., B and 5, E and 2). There were also 9 letter-letter combinations (e.g., M and V, P and R), 36 number-number combinations (e.g., 8 and 9, 3 and 6), and 9 dummy number-letter combinations (e.g. 3 and R, 6 and T).
Apparatus
The apparatus was the same as that described for Experiment 1. In Experiment 2, the subject pressed the reaction time key when a letter was present in the pair of items auditorily presented. The digital timer began at the onset of the initial vocalization of the first item in the pair, regardless of whether that item was a letter. However, pairs of items that did not contain any letter did not have the signal that began the digital timer. The go-no-go nature of the task was a characteristic of the timer operations as well as of the subject's response.
Design
The pairs of items were recorded onto the cassette tape in random order with the proviso that approximately half the A-1, B-2 type-items came before the A-4, F-1 type-items and half came after. Correct responses to the items would elicit equal numbers of go, no-go responses.
A split-plot ANOVA was conducted on the data with one between-subject factor-group (blind and sighted)-and one within-subject factor-type of pair (shared braille representation and different braille representation). In addition, an item analysis was carried out primarily to assess whether the stimuli obtained consistently similar mean detection times or, for example, whether a rank ordering from A-1 to I-9 could be arranged.
Procedure
The subjects were tested individually and received a similar procedure to that described for Experiment 1. The subjects were asked only to press the button if one (or more) letters were present in a pair of items. The pairs were described as either two letters, two numbers, or one letter and one number. The instructions stressed speed with accuracy in the letter-detection task, and each subject received 10 practice trials.
Results
Table 2 presents the mean detection time and standard deviations in milliseconds across the "shared" and "different" representations in the two conditions. Errors were negligible and inconsistent and did not exceed three per subject.
Table 2. Mean letter-detection times and standard deviations in milliseconds for the blind and sighted subjects in Experiment 2.
Letter-Number Pairs
Shared in Different in
Group Braille Braille
Sighted Subjects
Mean 672 683
SD 160 165
Blind Subjects
Mean 731 685
SD 190 184
The split-plot ANOVA revealed no significant differences with the two main factors (group and type of representation); the blind subjects and the sighted subjects did not differ in overall speed, and the "shared" A-1, E-5-type pairs did not obtain significantly different reaction times from the "different" A-4, E-2 pairs. However, the ANOVA revealed a significant interaction between the two variables, F(1,20)=5.10, P<.036. Planned comparisons showed that even though the sighted subjects did not differ in their detection speeds for the "shared" braille representation pairs and the "different" braille representation pairs, the blind subjects were significantly slower in dealing with "shared" pairs than with the "different" pairs, t = 2.341, P< .05 >.02.
The mean reaction time and standard deviation across items did not reveal any inconsistency across items, nor was there a significant correlation between the reaction times, the item pairs obtained, and their alphabetic sequence (A and 1, B and 2, and so forth). The item analysis supported the subject analysis; that is, the difference between the "shared" braille representations and the "different" braille representations was significant for the blind subjects but not for the sighted subjects, t(8)=2.991, P<.025; t(8)= 1.836, n.s.
Discussion
The results provide evidence that the detection of letters in letter-number pairs is delayed because letters and numbers in braille can share physical representations. The delay might arise from uncertainty about the nature of the stimulus.
The psychological consequence of retrieving shared representations might have led to the facilitation of responses because of the congruence of the evoked letter-number codes; however, it did not. Instead, letter-detection speeds were relatively faster when letter-number pairs did not have common characteristics and no ambiguity existed.
Although the delay in detecting letters may seem surprising since the target letter came first, this order of presentation seems to have led to automatic orthographic coding of both stimulus items and a consequent uncertainty of response. This influence was not apparent for the sighted subjects, who had no experience with braille. However, before concluding that it was due primarily to the specific characteristics of braille and the resulting activation of the orthographic codes, one should consider whether semantic associations between the serial positions of letters and of numbers might not also have played a part.
Experiment 3
If a subject hears a number (e.g., 1 or 3) the number may prime the letter of the alphabet that has the corresponding serial position (e.g., A or C). If such semantic connections between the stimuli are influencing the performance of subjects, then positioning the letter after the number (in contrast to the experiment just described) might lead to a greater semantic priming effect of the number on the letter because the priming of the ordinal position of the letters by the numbers would strengthen the semantic connection between them.
Also, if the representation of digits should happen to be in alphabetic form, then the orthographic code of the subsequent letter would not interfere with the response of braille readers.
Further weight could be added to the interpretation of the previous results by attempting to influence explicitly the particular kind of orthographic representation the subjects evoke. Subjects might think of numbers in terms of their logographic form (e.g., 1 or 4) or their alphabetic form (e.g., one or four). In the latter instance, there should be no difference in the letter-detection speeds of the blind group and the sighted group.
Consequently, a third experiment with two further conditions was run using new stimuli. These stimuli differed in two respects from the previous one.
First, numbers preceded letters. Second, in the logographic condition, only numbers and letter pairs were used, as before. However, in the alphabetic condition, these auditorily presented number-letter pairs were embedded in a list of word-letter pairs, such as table= A and R = ball. The subjects' task remained the same: to press a button when he or she heard a letter in a pair. The purpose of embedding the number-letter pairs in the word-letter list was to bias the subjects into evoking the alphabetic representations of numbers rather than the logographic representations. In Experiment 1, it was shown that the orthographic/alphabetic based code influences auditory processing. Thus, if such a code were evoked, it would have no special connection with letters for the blind subjects (unlike the logographic representation). Therefore, no group differences were expected in this condition.
Materials
The same procedure and apparatus were used to run these two conditions. Two new cassette tapes were constructed. For the logographic condition, the stimuli and tape took a similar form to that described for Experiment 2, except that the letter-number pairs were all recorded as number-letter pairs and the dummy number-letter pairs in Experiment 2 were exchanged for dummy letter-number pairs. For the alphabetic condition, the cassette tape contained mainly word-letter combinations and included 9 number-letter pairs that share their representations in braille, 9 number-letter pairs that have different representations in braille, 14 word-letter and letter-word pairs, 12 letter-letter pairs, and 38 word-word combinations; all the words were common nouns. There were approximately equal numbers of go and no-go trials.
Results
Table 3 shows the mean letter-detection times and standard deviations in milliseconds across the crucial number-letter pairs in the two extra conditions.
Logographic condition
The split-plot ANOVA for the number-before-letter condition revealed that the main effect for group and the main effect for type of representation (shared in braille or different in braille) were not significant.
Table 3. Mean letter-detection speeds and standard deviations in milliseconds for the blind and sighted subjects in the extra, logographic, and alphabetic conditions in Experiment 3.
Alphabetic Condition Logograpbic Condition
Associated Unassociated Associated Unassociated
Number-Letter Number-Letter Number-Letter Number-Letter
Group Pairs Pairs Pairs Pairs
Sighted Subjects
Mean 866 912 832 867
SD 231 222 191 196
Blind Subjects
Mean 789 874 903 891
SD 209 210 263 253
The interaction between these two variables just failed to reach significance, F(1,20)=4.16, P<.06. However, Hays (1963) and Keppel (1972) suggested that when a priori comparisons have been indicated by theoretical considerations, it is permissible to carry out planned comparisons.
'I'he results of such comparisons, which should be considered with caution, revealed that for the blind subjects, there was no difference between types of representation (e.g., 5 and E or 2 and E); however, for the sighted subjects, a significant difference emerged such that the 1 and A pairs were responded to significantly faster than were the 4 and A pairs, t= 2.59, p <.02.
The mean reaction time and standard deviation across items did not reveal any inconsistency across items, nor was there a significant correlation among the reaction times, the item pairs obtained, and their alphabetic sequence. However, t-tests supported the finding just mentioned that the sighted subjects' speed of detecting letters in letter-number pairs with semantic associations (e.g., 1 and A) was significantly faster than for those with no obvious connection (e.g., 4 and A), and for the blind subjects, this comparison was not significant, t(8)=2.672, P<.05>.02, t(8)=1.621, n.s.
Alphabetic condition
The ANOVA carried out with the number-before-letter pairs in the alphabetic condition showed that there was no significant difference between the groups but that there was a significant main effect of type of representation such that one and A pairs were responded to significantly faster than four and A pairs F(l, 20) = 16.86, P < .0005. The interaction of the two main variables was not significant.
The item analysis again showed no significant correlation between the 1 and A, 2 and B, C and 3 pairs and the speed with which letters were detected from those pairs. The absence of a correlation was probably due to the small number of trials; only one trial was provided for each pair.
Discussion
In the alphabetic condition, in which there was a high proportion of verbal items, no differences between the performance of the blind subjects and sighted subjects were found and there was no interaction. This finding is in line with the prediction that in this condition, the auditorily presented numbers might evoke alphabetic representations (two and four), which would be equivalent in both groups and have no special significance for braille readers. In addition, however, both groups showed a significantly greater facility in detecting letters from semantically associated pairs (e.g., three and C and six and F) than from unassociated pairs (e.g., three and E and six and A). A likely explanation for this effect is that some underlying semantic association between the sequential order of the numbers and the letters facilitated the recognition of letters. Thus, the word three primed or predicted the third letter of the alphabet, C.
In the logographic condition, the sighted subjects again showed a similarly significant "semantic facilitation" effect. Turning to the performance of the blind subjects, it was hypothesized that they would again show an interference effect for the detection of letters in associated pairs rather than in unassociated pairs because of the shared braille number-letter representations.
However, the strength of semantic facilitation and the fact that it would speed up the responses to the associated pairs was not considered. As Table 3 shows, it is apparent that both the blind subjects and sighted subjects responded faster to the semantically associated pairs. Therefore, it is possible that the blind subjects' performance may have reflected the operation of opposing processes when they occurred.
For the blind subjects, there appears then to have been a trade-off between a semantic facilitation effect, demonstrated in the alphabetic condition, and the interference of the braille representation, demonstrated earlier in Experiment 2.
General conclusions
The three experiments attempted to demonstrate that blind people as well as sighted people are influenced by internal orthographic codes in auditory word processing tasks. Experiment 1 confirmed that in an auditory rhyme task, in which the spelling of word pairs might be considered irrelevant, there was nevertheless a significant effect of the orthographic representations of the word pairs involved. In addition, the blind subjects were also affected by the ambiguity of the braille representation of numbers and letters, even when no reading was involved. Clear evidence for this effect was found in Experiment 2 and Condition 2 of Experiment 3. In Experiment 2, shared braille representations led to uncertainty and therefore inhibition in detecting letters. It is counterintuitive, perhaps, that experience with reading braille should affect auditory processing at all, particularly in the present context in which conditions were such that the subsequent presentation of a digit name that shares a braille symbol with a letter delayed the detection of that letter. However, it was tentatively hypothesized that this order of stimulus items might bias the subjects toward using an internal logographic representation of a digit in preference to an alphabetic one.
That this assumption may have been justified was given some support in the results of Experiment 3, in which the order of presenting the items was reversed (that is, the digit was heard before the letter). In Condition 1 of Experiment 3, the use of verbal items as well as the reversal of the order of stimulus items in a pair appears to have led the blind subjects to utilize alphabetic codes of digits in preference to logographic ones. Again, this may seem to be a surprisingly effective manipulation of stimulus conditions.
In Condition 1 of Experiment 3-in contrast to the pattern of results in Experiment 1 and Condition 2 of Experiment 3 - no difference between the performance of the blind subjects and the sighted subjects was apparent.
Unfortunately, although the performance of the two groups was different in Condition 2, the predicted interference effect did not emerge. Nevertheless it was argued that, in light of an unexpectedly strong semantic facilitation effect, this should not have been surprising.
Further research needs to be carried out to determine the exact conditions necessary for a blind person to utilize logographic or alphabetic representations of ambiguous braille symbols. If print afforded a similar set of ambiguous symbols, as it does in the case of the letter O and the number 0, a similar interference effect between items, to the one demonstrated for the blind subjects might have been expected for the sighted.
In addition, the results refer to similarities between Grade 1 braille and numeric codes. In the condition in which words were used, it should be remembered that the children were familiar with Grade 2 braille, which uses contractions. Therefore, it is possible that the spelling codes evoked by the children in response to the auditory stimuli ought, in some instances, to have taken a Grade 2 form. However, even such a code would retain strong alphabetic elements.
Certain practical implications can be drawn from these findings. One should be aware that material given to a child auditorily may be translated by the child into an alternative form. It might be useful to keep this in mind when interpreting spelling errors, for example. Children might benefit from having their attention drawn explicitly to the ambiguities between rhyming words that sound as if they should be spelled the same when presented auditorily but that are actually spelled differently. Similarly, even spoken letters and numbers may be cause for confusion because of their shared braille representations. Thus, when teaching algebra, for example, one should take special care to minimize possible interferences.
References
Doctor, E.A. (1978). Studies of reading comprehension in children and adults. Unpublished Ph.D. thesis, University of London, England.
Donnenwerth-Nolan, D.S., Tanenhaus, M.K., and Seidenberg, M.S. (1981). Multiple code activation in word recognition: Evidence from rhyme monitoring. Journal of Experimental Psychology, 7, 170-180.
Hays, W.L. (1963). Statistics for Psychologists. New York: Holt, Rinehart & Winston.
Hecaen, H., and Kremin, H. (1976). Neurolinguistic research on reading disorders resulting from left hemisphere lesions: Aphasic and "pure" alexias. In H. Whitaker and H. A. Whitaker (Eds.), Studies in Neurolinguistics, 2. New York: Academic Press,
Henderson, L. (1982). Orthography and Word Recognition in Reading. London, England: Academic Press.
Holley-Wilcox, P. (1977). The effect of homophony with auditory presentation of stimuli. Paper presented at the meeting of the Midwestern Psychological Association, Chicago, May 1977.
Keppel, G. (1973). Design and analysis: A researcher's handbook, Englewood Cliffs, N.J.: Prentice-Hall.
O'Connor, N., and Hermelin, B. (1978). Seeing and hearing and space and time. New York: Academic Press.
Pring, I,, (1984). A Comparison of the word recognition processes of blind and sighted children. Child Development, 55, 1865-1877.
Sasanuma, S. (1980). Dyslexia in Japanese: Clinical features and underlying mechanisms. In M. Coltheart, K. Patterson, and J.C. Marshal (Eds.), Deep Dyslexia. London, England: Routledge & Kegan Paul.
Seidenberg, M.S., Tanenhaus, M.K., Leiman, J.M., and Bienowski, M. (1982). Automatic access of the meanings of ambiguous words in context: Some limitations of knowledge based processing. Cognitive Psychology, 14(4), 489-537.
Swinney, D.A., Onifer, W., Prather, P., and Hirshkowitz, M. (1979). Semantic facilitation across sensory modalities in the processing of individual words and sentences. Memory & Cognition, 7(3), 159-165.
Warren, R. (1974). Association, directionality and stimulus encoding. Journal of Experimental Psychology, 102(l), 151-158.
Linda Pring, Goldsmitb's College, University of London, New Cross, London SE14 6 NW England.