Processes Involved in Braille Reading
LINDA PRING, Ph.D.
Dr. Pring is with the Department of Psychology, University of London, Goldsmith's College, London SE14 6NW, U.K. The research was carried out at the MRC Developmental Psychology Project, London.
Abstract: The purpose of this investigation was to consider whether the input modality and orthographic differences in braille reading produce different word recognition strategies for blind as compared with sighted persons. Blind children, blind adults, and sighted children were compared as to the extent to which they showed a pseudohomophone effect, which is held to reflect phonological coding. While in the sighted individuals there was strong evidence for such an effect, this was not so with the blind persons. The results were interpreted to indicate a differential allocation of attention to levels of word processing between the groups of readers.
Braille is a script used by blind and partially sighted people. Each alphabetic letter is represented by one symbol made up of embossed dots in a 2 x 3 matrix. Grade 2 braille also makes use of letter group contractions, such that orthographic units, for example, TION and TH, are represented by just one (or at the most two) symbols. This is the grade of braille that children learn to read in British schools from age five years. In order to read a single word in braille, one finger is moved across the embossed characters. Thus, in contrast to sighted reading, the written information is perceived through the tactual modality and occurs in a sequential fashion over time. One purpose of this investigation is to consider whether these differences in modality and orthography affect subsequent information processing.
In current studies of the reading process, one focus of attention has been the role of phonological coding. In attempting to investigate this, it has been found useful to distinguish two possible ways that an internal phonological representation of a printed letter-string can be derived.
The first retains close links with semantic knowledge, and may be best conceptualized as "addressed" phonology (Patterson, 1981). This is analogous to a look-say strategy in reading, but here it refers to the process of locating the phonological representation in long-term memory (LTM). In the look-say process, it is the meaning of the word that is located in LTM.
An alternative way of deriving a phonological code that does not need to access the stored phonological representations of the whole word in memory may be best termed "assembled" phonology (Patterson, 1981). The principle here is that phonological units corresponding to a letter or letter clusters can be combined to form a complete phonological representation of either known words, novel words, or nonwords, such as SLINT or BRANE. The latter example of a nonword had been called a "pseudohomophone," since its phonological realization is homophonous to an English word (BRAIN). The assembled phonological code then is constructed in a letter-by-letter manner. Thus these two possibilities, addressed and assembled phonological processing, differ in several crucial ways and may be the cause of a reading difference between blind and sighted persons. Previous results (Pring, 1982) indicated that when the task is that of reading aloud, which necessitates the retrieval of an appropriate phonological code, blind children seem not to differ from sighted children in their means of gaining such a code.
There has been little evidence to suggest that phonological coding plays an integral part in reading single words for meaning. Several studies on this issue have made use of a lexical decision task, where the reader is simply required to decide whether or not a letter array is an English word. In this task, readers must consult their internal word store in order to make a decision, and this process seems likely to be similar to an early stage of reading for meaning. Few studies have found phonological effects when the letter-strings formed a word (Coltheart et al., 1979), but a phonological effect has emerged in dealing with nonwords. In this situation a pseudohomophone, such as BLOO, takes longer to reject than does a nonword, such as PLOO. The phonological code for BLOO presumably accesses its homophonic word equivalent (BLUE) and causes a slow rejection rate. This phonological effect must arise from the use of assembled phonology, since there are no stored representations of nonwords.
In the experiment reported here one airn was to investigate the role of assembled phonology in blind persons, not as it relates to reading aloud but rather in a task similar to reading silently, namely, a lexical decision task. The question is whether, in contrast to reading aloud, blind persons would differ from sighted persons in their use of phonology in such a task. Braille is difficult to learn not just because of the modality of input but also because the orthography presents little redundancy. An error in perceiving one dot in one position of a letter will always lead to misidentification of that letter. Indeed, reading errors made by blind people are nearly always of this type (Nolan & Kederis, 1969). It may then be much less useful for blind people to rely on letter-by-letter phonological processing, and they may instead attend primarily to forming a tactual code of the whole word and use this in a touch-say strategy to achieve recognition. If this were the case, blind persons might tend to rely on a direct method of lexical memory access, based on the pattern of the whole word. Such a strategy of carrying out a lexical decision task would involve rejecting letter-strings simply because they are not recognized as words. In this case pseudohomophones would not differ from nonwords, since neither of these letterstrings have stored representations.
Subjects
Three groups of subjects participated in the experiment. One group was made up of seven boys and three girls from a residential school for the blind. They were either totally blind or had some light but no pattern perception. They were all congenitally blind. Children with brain damage, retardation, or additional handicaps were excluded. The children ranged from normal (IQ 96) to highly intelligent (IQ 130), as assessed by school and medical reports. They were all considered to be good readers for their age, and most started learning braille at the age of 5 years. The mean age was 12 years 1 month (SD 1 year 9 months).
A second group consisted of five congenitally blind students. They were all male, aged between 22 and 30 years, and participating in a three-year physiotherapy course. The students had all been blind since birth and had no pattern vision at all. The physiotherapy course is highly competitive, and all the students were of average or above average intelligence and skilled braille readers.
The third group was 18 sighted children, 11 boys and 7 girls, from a London primary school. The mean age was 10 years 6 months (SD 7 months). The two groups of children, though of differing mean chronological age, were comparable in terms of reading comprehension (Nolan & Kederis, 1969). Teachers were asked to carry out vocabulary and spelling tests after the experimental procedure, and the two groups were found to be equivalent in these respects.
Materials
Words: Thirty-five common English words were selected from the Kucera and Francis (1971) word frequency tables. Their mean frequency was 108 occurrences in the total sample. The words were between four and six letters in length but not matched to the nonword stimuli. Grade 2 braille was used for the blind persons.
Nonwords: Eighteen pseudohomophones, such as SNOE and GURL, and 18 nonwords, such as SNOL and GORL, were included. A nonword was generated by changing just one letter in one position of a pseudohomophone. Thus the nonwords acted as paired controls for the pseudohomophones. Examples of the three types of stimuli follow:
Pseudo-Words homophones Nonwords
PAPER BLOO PLOO
ROSE BATTEL BANTEL
WINDOW CHERCH CHERF
BOAT GURL GURE
CHILD TRANE TRAKE
BALL BODDIE BODDIL
RAIN BRAIK PRAIK
Apparatus
For the sighted children, slides were projected onto a screen by a projector. The stimuli were printed in uppercase letters at the center of each slide. Onset of presentation was synchronized with a digital timer. This timer was stopped at the onset of a vocal response (YES or NO) through a voice microphone attached to the child.
For the blind persons, each stimulus was brailled on a 4 x 6-inch index card with the use of a Perkins Brailler. The cards could be placed, one at a time, in a braille cardholder designed to give the subjects a tactual start position, where they rested their reading finger(s). The letter-string was less than an inch above this position. An electronic signal generator presented an auditory start signal to the subject and simultaneously started a digital timer.
For the blind children, the timer stopped at the onset of a vocal response (either YES or NO) picked up by a voice microphone attached to the subject. The reaction times and errors were noted by the experimenter.
The blind adults made a manual rather than vocal response, by touching one of two reaction keys in order to make a YES or NO response. These touch keys were flat, measuring 5 x 10 cm and placed about 5 cm to the right and 2.5 cm above the letter array (the YES key) and 5 cm to the right and 2.5 cm below the letter array (the NO key). An auditory start signal began a digital timer, and the manual response stopped the timer. One of the two lights (green and white) and one of two buzzers (high pitch, 4 kHz; low pitch 800 kHz) was activated depending on whether the YES or NO response was made.
Design
The aim of the experiment was to investigate how the groups would differ in their speed of response to pseudohomophones relative to matched nonwords. Consequently a Split-plot ANOVA was used to analyze the data, the between-subject variable being thegroup's term, and the within-subject variable being the type of nonword pseudohomophone and legal nonword.
The total number of people in each group differed and indeed was rather small for the blind adult group. Therefore, for this group some caution is necessary in interpreting the results.
Here it was decided not to compare the YES decision to words with the NO decision to nonwords, because first, the words were not strictly matched with the nonwords for number of letters, and second, the interest was focused on the comparison of the pseudohomophones and nonwords that were closely matched. However, as a matter of general interest the mean reaction time for the words as well as those for the pseudohomophones and legal nonwords are presented.
Procedure
The three groups of subjects were allowed between 12 and 20 practice trials, followed by a random sequence of the 35 words and 36 nonwords. A different random sequence was given to each blind person, but the same sequence (permanently placed in the carousel) was shown to the sighted children. Each person was tested individually, and RTs and errors were noted by the experimenter.
Each blind child was fitted with a voice microphone and sat at a desk. On the desk was placed the braille cardholders and the child became familiar with both the tactual start position on the holder and the position of the printed braille stimulus on a card placed in the holder.
The child was instructed to move his or her finger(s) from the start position to the braille stimulus as soon as an auditory start signal was given and then to decide as quickly as possible whether or not the stimulus was an English word, responding YES.
Speed and accuracy were both emphasized in the instructions, and none of the children had any difficulty with the task.
A similar procedure was undertaken with the blind adults, although in their case the response was manual and not vocal. This was because there was an initial interest in any possible hand-of-response effects. Each subject participated in two experimental sessions, on two consecutive days. Their response was made with alternate hands on alternate days. However, no hand response differences emerged, nor did any interactions with stimuli. Therefore the data were collapsed for use in this analysis.
The sighted children were seated about six feet from a screen onto which the stimuli were displayed. The instructions were similar to those for the other two groups, namely, to decide whether or not the printed letter-string was a word and to respond YES if it was or NO if not.
Results
Table 1 presents the mean of the median reaction times (in msec), the standard deviation, and the percentage error made in response to the stimuli involved in the experiment.
Nonword stimuli. The split plot analysis of variance carried out on the pseudohomophone and nonword data revealed that the main effect of the group was significant (F=25.66, p<0.0001). The sighted children were significantly faster in overall reaction times than the other two groups (t = 2.08, P<0.05).
Table 1: Mean of the Median Reaction Times (in msec), Standard Deviation, and
Percentage Error in Blind Children, Sighted Children, and Blind Adults
Subjects Words Pseudo-homophones Nonwords
Sighted Children
Mean 980 1525 1368
SD 291 733 557
% error 3.6 21.9 6.8
Blind Children
Mean 2143 2960 3158
SD 449 597 923
% error 2.0 8.5 4.5
Blind Adults
Mean 2168 2737 2831
SD 517 818 982
% error 1.0 7.8 5.6
The main effect of type of nonword (pseudohomophones and nonwords) was not significant (F = 0. 8, ns), but the interation of the group factor with the type of nonword factor was significant (F= 3.39, P<0.05). The sighted children took significantly longer to reject the pseudohomophones relative to the nonwords in terms of reaction time (t = 2.20, P<0.05) and made significantly more errors (t = 4.77, p<0.001). This was not the case for the blind children, who showed no significant differences either in reaction time (t = 0. 86, ns) or errors (t = 1. 64, ns), or for the blind adults (RT: t = 0.77, ns; error: t = 0.84, ns).
Word stimuli. A one-way ANOVA confirmed that the sighted children were significantly faster than either the blind adults or the blind children (F = 30.69,P< 0.0001).
Discussion
In this experiment the sighted children took significantly longer to reject pseudohomophones, such as BLOO or BRANE, than matched nonwords, such as PLOO or PRANE. No such differences emerged for either blind children or blind adults. The pseudohomophone effect found here with sighted children and well known with sighted adults (Rubenstein, Lewis, & Rubenstein, 1971) may be attributed to the interference of using phonological letter-by letter coding. The conflict presumably occurs when the real word match (e.g., BLUE) is located by the phonological code representing the pseudohomophone (BLOO). The resulting confusion is reflected not only in the long reaction times for pseudohomophones but also in the high error rate.
In contrast, there was no evidence of the influence of this interference process in the time it took blind children and blind adults to reject nonwords. Their error rate was also not significantly different for pseudohomophones relative to nonwords. Susanna Miller's work (1975, 1981)has indicated that blind children can encode both tactual and phonological features of braille letters and may also rely on letter-by-letter processing. However, in contrast to Millers' work, which tested short-term memory coding, this experiment was concerned with single word and nonword recognition. The results provided here suggest that with such a task the blind did not construct a phonological letter-by-letter code. This may be because they needed to allocate relatively more attention to the feature-analysis, perceptual process and thus have less residual processing capacity to proceed with phonological analysis. This differential allocation of attention to the perceptual level and whole word processing in word recognition may arise primarily because of the low redundancy of the braille script, which in this instance leads to an advantage of the blind over the sighted readers. The blind are not misled by the phonological ambiguity of the pseudohomophones and, therefore, readily identify them as nonwords. The braille orthography is structured such that each component of a braille letter is essential for correct identification. This means that braille letters have no redundancy at all, a fact not at all true of print (Huey, 1908).
Regarding the modality of input in braille reading, Walisten and Lambert (1981) have argued that the braille orthograpby iteself makes reading difficult and provides harder feature-detection processing than print. It appears that the initial perception process in the tactual modality is hard and indeed takes significantly longer than in the visual process. Even experienced braille readers can read at only about half the rate that sighted readers achieve (Nolan & Kederis, 1969). Here, even though some reaction times by the blind were very fast, overall they took about twice as long to make lexical decisions compared with the sighted. These relatively long reaction times may thus be explicable in terms of slow tactual perception rather than slow information processing.
I would like to acknowledge the help of the Royal National Institute for the Blind in this report. In particular, I would like to thank Mr. Bolton, the Headmaster of Dorton House School, and also the children who cooperated in the research. Finally, a special thanks to Dr. B. Hermelin for her helpful suggestions and encouragement in this study.
References
Baron, J., & Strawson, C. (1976). Use of orthographic and word specific knowledge in reading words aloud. Journal of Experimental Psychology, 2(3), 386-393.
Barron, R. W. (1978). Access to the meaning of printed words: Some implications for reading and learning to read. In F. B. Murray (Ed.), The recognition of words. IRA Series on the development of the reading process. Newark, DE: International Reading Association.
Coltheart, M., Besner, D., Jonasson, J., & Davelaar, E. (1979). Phonological encoding in a lexical decision task. Quarterly Journal of Experimental Psychology, 31, 489-507.
Frederiksen, J. R., & Kroll, J. F. (1976). Spelling and sound: Approaches to the internal lexicon. Journal of Experimental Psychology, Human Perception and Performance, 2, 361-379.
Huey, E. B. (1908). The psychology and pedagogy of reading. Boston: MIT Press (new edition, 1968).
Kucera, H., & Francis, W. M. (1971). Computational analysis of present-day American English. Providence, RI: Brown University Press.
Liberman, I. Y., Shankweiler, D., Liberman, A. M., Fowler, C., & Fischer, F. W. (1977). Phonetic segmentation and recoding in the beginning reader. In A. S. Reber and D. L. Scarborough (Eds.), Toward a psychology of reading. Hillsdale, NJ: Lawrence Eribaum Assocs., Inc.
McClelland, J. L., & Rumelhart, D. E. (1981). An interactive activation model of context effect in letter perception: Part 1, An account of the basic findings. Psychological Review, 88(5), 375-407.
Millar, S. (1975). Visual experience or translation rules? Drawing the human figure by blind and sighted children. Perception, 4, 363-371.
Millar, S. (1981). Cross-modal and intersensory perception and the blind. In R. D. Walk & H. L. Pick (eds). Intersensory perception and sensory integration. New York: Plenum.
Morton, J. (1980). The logogen model and orthographic structure. In U. Frith (Ed.), Cognitive processes in spelling. London: Academic Press.
Morton, J., & Patterson, K. (1980). A new attempt at an interpretation, or an attempt at a new interpretation. In M. Coltheart, K. Patterson, & C. Marshall (Eds.), Deep dyslexia. London: Routledge & Kegan Paul.
Nolan, C. Y & Kederis, C. J. (1969). Perceptual factors in braille word recognition. New York: American Foundation for the Blind Research Series, No. 201.
O'Connor, N., & Hermetin, B. (1978). Seeing and hearing and space and time. London: Academic Press.
Patterson, K. E. (1981). Neuropsychological approaches to the study of reading. British Journal of Psychology, 72, 151-174,
Pring, L. (1982). Phonological and tactual coding of braille by blind children. British Journal of Psychology, 73, 3 51-3 59.
Rubenstein, H., Lewis, S. S., & Rubenstein, M. A. (1971). Evidence for phonemic recoding in visual word recognition, Journal of Verbal Learning and Verbal Behavior, 10, 645-657.
Thorndike, E. L., & Lorge, I. (1944). The teachers' wordbook of 30, 000 words. New York: Bureau of Publications, Teachers' College.
Wallsten, T. S., & Lambert, R. M. (1981). Visual braille and print reading as a function of display field size. Bulletin of The Psychonomic Society, 17(l), 15-18.