Miscue Analysis of Braille Readers
V. Sowell; A. Sledge
Abstract: A comparison of blind and sighted readers, using Goodman's Reading Miscue Inventory, was conducted using 22 blind students enrolled in both public and residential schools. Blind braille readers' miscues were compared to those of sighted readers. Results revealed that similarities were greater than differences. Print and braille readers had approximately the same proportion of miscues in substitutions, omissions, insertions, reversals across words, and more miscues in graphic than sound similarity. Differences were found in more unsuccessful corrections for braille readers.
Few investigators have studied the reading process of blind students who use tactile modes in reading (Harley, et al., 1979; Williamson, 1976). The paucity of research on reading by blind people has neither concentrated on the reading process itself, nor on the mode of input; instead, the main emphasis has centered on speed of braille readers as compared to print readers (Crandell & Wallace, 1974; Lowenfeld, 1973; McBride, 1974; Olson, 1977).
In order to extend our understanding of braille reading, it is necessary to investigate the reading process itself, as well as the simple apprehension of the sensory input. In this manner, we can extend our understanding of the complexities of the reading process as it applies to all readers and not just visually impaired readers.
Reading is a language process. However, that which is directly observable - oral reading, rate-encoded responses to the author's message - represents only a segment of the entire process, the behavioral product. The remainder of the process is unavailable to direct observation in that it takes place within the "black box" of the mind. During the past several decades, many means have been employed to draw inferences about the unobservable aspects of the reading process; these have ranged from introspection to eye-movement studies.
Since 1962, a method has been available for effective evaluation of the reading process without the artificiality of previous approaches. This method makes use of the traditional data of oral reading, story recall and questioning, but applies a method of analysis based on linguistic as well as graphic categories. This method is the Reading Miscue Analysis Inventory (RMI), growing out of the work of Kenneth Goodman (1979). Rather than analyzing oral reading in terms of percentages of word recognition errors and categories such as substitutions, mispronunciations and omissions, oral reading errors are considered in terms of their phonological, syntactic and semantic relationships to the text.
The nine categories of RMI are dialect, intonation, graphic and sound similarities, grammatical function, correction, grammatical acceptability and semantic acceptability, and meaning change. In the RMI, the terms are summarized as follows (Burke & Goodman, 1972):
1. Dialect is defined as the range of language usage of the reader, or what sounds right" to the reader rather than the author's word. These can occur in the form of sound variations such as "pitchur" for "picture," vocabulary variations such as "goed" for "went," and grammatical variations such as "don't" for "doesn't."
2. Intonation miscues involve changes in pitch, stress or pause from those which are expected.
3. Graphic similarities are evident when the reader says "walk" for "walked" or "away" for "any."
4. Sound similarities include "sompin" for "something" or "cannot" for "can't." Ratings for graphic and sound similarities are categorized: high degree, some degree, or no degree of similarity to the text.
5. Grammatical function involves the substitution of a single word or nonword in the text. Miscues are rated as being similar to or different from the text in grammatical function.
6. Corrections are the reader's change or insertion of a different word from the one first read. These are rated according to amount of change in meaning and grammatic and semantic acceptability.
7. Grammatic acceptability is concerned with the ability of the reader in coping with the structure of the text sentence. This might include substituting the phrase "was hoping around" for "was hopping around."
8. Semantic acceptability focuses on the ability of the reader to produce understandable structures. Miscues can occur in semantically acceptable sentences which differ from the text meaning. An example would be to substitute "light" camera for "little" camera.
9. Meaning change refers to the extent to which the reader gets the author's meaning, even though other miscues may have occurred.
In this system, oral reading errors are termed "miscues," wherein the text has miscued or elicited inaccurate oral reading behavior. "Miscue" has a less negative connotation than "error," and serves to reinforce Goodman's idea that not all oral reading errors are to be deplored equally. The syntactic and semantic faithfulness of the error reflects the reader's effectiveness an(i efficiency in responding to print in terms of graphophonics and message.
Comprehension is analyzed in terms of the reader's spontaneous recall of the passage, rather than in terms of responses structured by questions. Salience, organization, and reader-generated inference become apparent in an unstructured recall.
To paraphrase Goodman, RM I provides a window on the reading process. Consideration of the quality of the oral reading and recall permits inferences to be drawn about the reading process in general and in particular, i.e., about the contents of the "black box" (Allen, 1976; Goodman, 1979; Goodman and Burke, 1972a).
Only one published study has been conducted with blind or partially sighted students in regard to miscue analysis (Williamson, et al., 1976). Williamson found that the reading performances of blind and sighted readers differed in that some blind readers applied phonic cues more efficiently than braille readers but did not seem as efficient in applying grammatical and semantic cues. Greater loss of comprehension seemed to result from braille miscues.
Other studies of braille readers have included data on comprehension and achievement. Ashcroft (1960) found that comprehension was affected by association and gross substitution errors. He concluded that readability in terms of word forms and sentence length is not enough and must also include attention to the vocabulary and meaning level of the individual child (p. 23). Reading comprehension in relation to speed was studied by Kederis, Nolan, and Morris (1967), who found that reading comprehension was not affected by devices to control speed of reading. Lowenfeld, Abel, and Hatten (1974) concluded that blind children are equal to comparable sighted children in scores on a standardized test in the fourth grade and superior by the eighth grade (p. 112).
Speed of braille readers as compared to print readers is the only area of reading previously reported in the literature. Lowenfeld (1973) found that braille readers fall behind print readers in speed as they progress through school grades. Olson (1977) studied speed of braille reading and.found a negative correlation between age and rate increase made by braille readers even after training in rapid reading skills. Others (Crandell & Wallace, 1974; McBride, 1974) found that teaching of rapid reading techniques increased speed of braille reading. Speed of reading, however, is only one factor involved in the reading process. Therefore, the study of miscues in braille reading will add valuable knowledge of the reading process of blind readers.
Statement of the problem
Braille and print readers differ obviously in terms of the manner in which information is encoded-graphic versus tactile representations-and initially received visual array versus a pattern of raised dots. The sighted reader can perceive not only the word or phrase on which he is focused, but also graphic information surrounding the point of focus. The reader of braille is able to perceive a portion of the horizontal array of braille cells of the line being touched by the fingers of the right hand, and perhaps the beginning of the line following, as the left hand serves as a place holder. The question arises then as to whether there are significant differences in the manner in which that language is decoded and processed when the methods of encoding and decoding that language-letters of the braille cell, vision or touch-differ to such a marked degree.
Possible dissimilarities between braille and print readers have not previously been investigated in terms of potentially significant factors such as comprehension, speed, and linguistic and graphic components of the reading process. Although print reading procedures have been investigated extensively Uuel & Holmes, 1981; Levin & Williams, 1970; Shirmon & Navon, 1982), little research into braille reading has been accomplished to date (Lowenfeld, Abel, & Hatlen, 1974; Williamson, et al., 1976). Comparison of the two modes of reading is of value in that commonalities and/or differences reveal factors that can enhance understanding of both print and braille operations; thus, more complete data can lead to improved instructional methods.
Purpose
Because miscue analysis has been shown to be a viable method of studying the reading process and of pinpointing reading difficulties, it is necessary to gather data on types and frequencies of miscues of braille readers as compared to print readers. The data gathered reveal similarities and differences in the tactual braille reading process as compared to visual print reading.
The topics investigated were as follows: 1. Do braille readers and print readers exhibit the same types of miscues? 2. Do braille and print readers exhibit the same number and percentages of miscues?
Methods
The investigators used the Miscue Analysis Inventory (Goodman & Burke, 1972a) to analyze oral reading data of 22 braille readers in both public and residential elementary school programs. Scores on these inventories were compared to existing data on sighted readers.
Students ranged in age from 6 to 21 years and all were considered normal in range of intelligence. Braille reading level scores were unavailable in many instances; instead, investigators used trial paragraphs to ascertain appropriate difficulty of reading material in order to elicit 25 miscues for the analysis.
Reading material consisted of brailled copies of Reading Miscue Inventory: Readings for Taping (Goodman & Burke, 1972b). Miscues were recorded on print copies of the same passages. A master print copy has been coded to indicate beginnings and endings of braille lines. The average number of brailled pages per print page was 3. The majority of students read middle level passages; however, each passage was read at least once.
mowed showed
there the these
just was
cannot must
to the from the
in no time
I know that
day done
father from
where which
and the
find finished
Figure 1. Braille cells.
Purposes of the study were explained to the students; they were told they would be asked to read a story aloud and to retell what they had read. A tape recording was made of the oral reading and the unaided recall. The oral reading was coded while the student read, and all tapes were reviewed by one examiner to verify the coding and to eliminate potential limitations of interrater reliability
A Reading Miscue Analysis was made of the first 25 miscues using Goodman and Burke's (1972a) categories of dialect, intonation, graphic similarity, sound similarity, grammatical function, correction, grammatical acceptability, semantic acceptability, meaning change, comprehension, and grammatical relationships.
Graphic similarity was determined on the basis of braille cell configuration rather than traditional print configuration. Graphic (tactile) similarity of braille readers is based on number and position of dots in the braille cell rather than on shape and orientation of print. The braille cell consists of 6 dots, 2 horizontal and 3 vertical rows. Within the 6 dots, everything that can be printed can be constructed in braille. This includes foreign languages, mathematical and scientific notation, diacritical markings, etc. (Figure 1).
Braille readers are limited in the "fixation" to the amount that can be read with one or two fingertips; therefore, phrasing is likely to be more difficult for braille readers. This is partly compensated for in the braille system of contractions. For instance, a braille word can be contracted to the length of two or three cells; Sally is contracted into four cells; showed into three cells; condition into five cells rather than nine letters. Adding to the difficulty is the similarity of many contractions. There is just one dot of difference in there and these.
Errors can be made chiefly in terms of spacing and sequencing. For instance, a formation can mean different words according to the spacing in the cell. Sequence of dots can indicate different words; dot5<d> is day while <d> dot 5 < o > is done.
Degrees of tactile similarity were defined as follows:
Yes: a) one to two dot difference (+,-) and/or the similar configuration.
that dots 2,3,4,5
the dots 2,3,4,6
Or b) change in position in braille cell.
was dots 3,5,6
just dots 2,4,5
Or c) changes of inflectional endings.
Partial: a) three or more dot difference (+,-)
f/in/d
f/in/i/sh/ed
Or b) configuration difference.
n/ow dots 2,4,6
n/o dots 1,3,5
Or c) one letter difference.
st/ar/t/ed
st/ar/ed
Table 1. Types of miscues.
(Frequencies: Summary Data)
N=22
x S.D. Range %
Substitutions 15.681 3.896 8-22 62.727
Omissions 3.409 2.986 0-14 13.636
Insertion 1.772 1.306 0-4 7.090
Reversal* .045 .213 0-1 .181
Repetition .136 .467 0-2 .545
Dialect .136 .467 0-2 .545
Partial word .909 1.108 0-4 3.636
Non-word substitution 3.045 2.919 0-11 12.181
*no for one
Table 2. Analysis of miscues.
(Percentages: Summary Data)
N=22
X S.D. Range
Graphic (tactile)
similarity Y 50.818 21.234 21-93
P 27.409 17.349 0-60
N 21.318 14.257 0-48
Sound similarity Y 42.545 20.500 10-86
P 30.045 15.373 7-60
N 27.363 13.892 0-52
Grammatical function Y 55.681 16.447 23-77
P 24.727 12.024 0-45
N 19.590 10.017 0-33
Comprehension no loss 55.409 17.360 20-79
partial 10.272 11.251 0-32
loss 34.318 18.653 0-68
Grammatical relationships strength 46.636 14.650 28-72
partial 12.863 10.048 0-32
weakness 31.818 17.608 8-68
overcorrection 8.227 5.887 0-24
Results
Group data were summarized in terms of both type and frequency of miscues, means, standard deviations, ranges and percentages. Comparisons were made with data reported in the literature as to oral reading miscues of print readers.
Types of miscues ranked according to frequency included real word substitutions (63 %), which constituted more than half of total miscues. Non-word substitutions and omissions each represented fewer than 20 percent of the miscues. Repetitions, dialect pronunciations, and partial word mispronunciations were minimal. No reversals of whole words were observed. (Table 1).
The Reading Miscue Analysis identified strong tactile similarities, based on braille cell comparisons; these characterized 50 percent of the miscues. The kinds of errors made on these selections by braille readers chiefly consist of dot spacing errors or omissions, either vertically or horizontally. Examples are as follows:
A. Mowed for showed (1 dot error)
B. The for there (1 dot error)
C. Which for where (1 dot error)
D. Was for just (vertical spacing error)
E. From for father (1 dot omission error)
F. To for from (vertical and spacing error)
Tactile information appeared to be used more efficiently than syntactic and phonemic cues.
Forty-two percent of the miscues had strong sound similarity to the expected response. In terms of effective use of syntax for decoding, strong grammatical function (55 %) and grammatical relationship (46%) were reflected in the oral miscues of these students. Syntactic cues were used more effectively than phonemic cues (5 5 % and 42%, respectively). Miscues resulted in either no loss of comprehension (5 5 %) or complete loss of comprehension (34%), with very little middle ground (10%) (Table 2).
Some repetitions appeared to function to maintain and reinforce meaning as the reader read or scanned ahead; however, they did appear more frequently at the end of a line or page of braille than elsewhere.
Discussion
Comparison with print readers
The proportion of miscues was similar to that reported for print readers by Yetta Goodman (1976): substitutions, omissions, insertions (descending order of frequency). As with sighted readers, reversals across words were not present (Goodman, 1976).
Braille readers were similar to print readers as follows:
a) Proportions of errors: substitutions, omissions, insertions, in descending order of frequency (Goodman, 1976);
b) Reversals across words were not observed (Goodman, 1976); and
c) Miscues had a higher graphic than sound similarity (Allen, 1976).
Unlike sighted readers, for whom 75-90 percent of corrections are successful, less than 40 percent of braille readers' corrections were successful. Further investigation would be necessary to postulate causes of these differences; spacing and positional factors involved in the cell placement may be contributing factors.
These results contradicted Williamson's, et al., research (1976), in which comparative miscue analyses were made of braille and sighted readers. Results of that study were as follows:
1. Braille readers were found to be similar to sighted readers in their use of graphic information; and
2. Braille readers were similar to sighted readers in preserving all of the meaning of the passage.
Trends similar to those reflected in Williamson's data included the following:
1. Braille readers were more efficient than sighted readers in the effective use of phonics.
2. Braille readers' use of syntactic cues was not as strong as that of sighted readers.
3. Braille readers were less effective in retaining at least part of the meaning.
Braille readers are both effective and efficient; they tend not to overcorrect their miscues and to preserve meaning even when giving an unexpected response. Their use of graphic information is much stronger because the tactile stimulus they must decode is much more complex, requiring significantly finer discriminations than those required of sighted readers. The braille is decoded in smaller segments; rather than an eye-voice span there is a finger-voice span.
Goodman and Burke's Reading Miscue Analysis is a useful tool for analyzing braille oral reading. The only adaptation required is to define graphic similarities in terms of braille topography. The braille reading process appears to differ from the sighted reading process primarily in terms of the nature of the stimulus and the sensory input.
References
Allen, P.D. (1976). Some general implications concerning specific taxonomic categories. In P.D. Allen (Ed.), Findings of research in Miscue Analysis: Classroom Implications, pp. 70-75. Urbana, IL: NCTE.
Ashcroft, S.C. (1961). Errors in oral reading of braille at elementary grade levels. In Report of Proceedings on the Conference on Research in Braille, pp. 16-31. New York: American Foundation for the Blind.
Crandell, J.M., Jr., & Wallace, D.H. (1974). Speed reading in braille: An empirical study. The New Outlook for the Blind, 68, 13-19.
Goodman, K.S. (Ed.) (1979). Miscue Analysis: Applications to Reading Instruction. Urbana, IL: NCTE.
Goodman, Y.M. (1976). Developing reading proficiency. In P.D. Allen (Ed.), Findings Of Research in Miscue Analysis: Classroom Implications, pp. 113-128. Urbana, IL: NCTE.
Goodman, Y.M., & Burke, C.L. (1972a). Reading Miscue Inventory: Manual Procedure for Diagnosis and Evaluation. New York: Macmillan.
Goodman, Y.M., & Burke, C.L. (1972b). Reading Miscue Inventory: Readings for Taping; Procedure for Diagnosis and Evaluation. New York: Macmillan.
Harley, R.K., Henderson, F.M. & Truan, M.B. (1979). The Teaching of braille reading. Springfield, IL: Charles C Thomas.
Juel, C., & Holmes, B. (1981). Oral and silent reading of sentences. Reading Research Quarterly, 16, 545-568.
Kederis, C J., Nolan, C.Y., & Morris, J. E. (1967). The use of controlled exposure devices to increase braille reading rates. International Journal for the Education of the Blind, 16, 97-105.
Levin, H., & Williamson, S. (Eds.) (1970). Basic studies on reading. New York: Basic Books.
Lowenfeld, B. (1973). The visually handicapped child in school. New York: John Day.
Lowenfeld, B., Abel, G.L., & Hatlen, P.H. (1974). Blind children learn to read. Springfield, IL: Charles C Thomas.
McBride, V.G. (1974). Explorations in rapid reading in braille. The New Outlook for the Blind, 68, 8-12.
Olson, M.R. (1977). Teaching faster braille reading in the primary grades. Journal of Visual Impairment & Blindness, 71, 122-124.
Shimron, J., & Navon, D. (1982). The dependence on graphemes and on their translation to phonemes in reading: A developmental perspective. Reading Research Quarterly, 17, 210-228.
Williamson, L.E., Allen, S., & McDonald, B. (1976). Braille reading is less efficient than visual reading. (ERIC Document Reproduction Service No. ED 124 870).
Virginia Sowell, Ph.D., assistant vice president for academic affairs and professor of special education, Texas Tech University, Box 4560, Lubbock, TX 79409; Andrea Sledge, Ph.D., assistant Professor Lehman College, City University of New York.