Reading Difficulty

Saturday stream 4 Session 09.00 - 11.10 Length 25 minutes http://www.bdainternationalconference.org/2001/presentations/sat_s4_a_3. htm

Reading difficulty characteristics in dyslexic and hearing-impaired students (in Hebrew) Ruth Engel and J. Rosenhouse (1) Dept. of Technology and Science Education, Technion -I.I.T., Haifa 32000 Israel (2) , Dept. of General Studies, Technion - I.I.T., Haifa 32000 Israel [email protected] Abstract This study examines reading difficulty characteristics in Hebrew in three reading-impaired populations. Two are groups of dyslexics: 100 readers with impaired auditory perception and 100 readers with impaired visual perception. The third group comprises 61 readers with deep/severe hearing impairment. All were elementary schools students in the second to sixth grades. The subjects were tested with a conventional Hebrew reading test. It examined types of reading errors, (e.g., changes of phonetic structure or word content), self-correction in reading, reading speed, sequential/holistic reading, the effect of reading texts with and without the Hebrew diacritical vowel signs ("punctuation"), and the effect of meaningful or meaningless text material on the amount of reading errors. The literature describes distinctions between various kinds of reading disability related to auditory impairment and visual perception, and the definition of dyslexia as being one category or including sub-groups. Our research hypothesis was that similar characteristics of reading difficulties would be found amongst auditory perception-impaired students and hearing-impaired students, and that they would differ from those of students with impaired visual perception. Our findings support this hypothesis. Many of the sub-tests revealed similarity in the reading difficulties between the hearing impaired students and those with impaired auditory perception vs. the visually impaired. An unexpected finding revealed that 4th grade students in all the groups were a special sub-group in each group. These findings suggest, in accordance with a major research approach, that dyslexia should be defined in terms of dyslexia sub-groups rather than as a single category.

Introduction Reading difficulties are examined among other features of reading research in general (Just and Carpenter, 1987, Chall, 1982, Rayner, 1989), and the study of processes in reading comprehension in particular (see, e.g., Cornoldi

and Oakhill, 1996, Morrow, 1989, Olson, 1994, Perfetti, 1994, Whitney et al., 1995, Rosenhouse et al. 1997, Shanahan and Neuman, 1997). Reading disability as a primary developmental feature of reading difficulty is examined in its various aspects, including holistic or sequential reading, types of errors in reading, reading which involves (or not) the semantic (content) factor, reading rate, etc. Basically, reading disability is defined as a mismatch between the phonetic and graphic structure of the word. Reading disability is characterized in (at least) three different ways: 1) Content, when the reading error is caused by substituting the word with a synonym in e.g., 'throw/fling'; 2)Phonology, when the reader changes the phonological structure of the word as in 'guest/just'; or 3) Form, when the reader wrongly replaces the word with a wrong one which has a similar holistic graphical structure (e.g., of consonantal roots) 'fighter/fighting'. The research of reading disability in Hebrew raises language-specific issues. In Hebrew, most of the vowels are not indicated by letters as in the Latinbased alphabet, but by diacritical points and marks . Usually, these marks are omitted, except in texts for beginning readers and in some special texts (mainly the Bible and poetry) (cf. Glinert, 1989). Thus, reading Hebrew resembles going in a loop: In most cases, readers can decode the meaning of a single word by the general context (of the sentence, the topic etc.). Due to the above, the acquisition of Hebrew reading in the preliminary stages requires the support of vowels or "pointing" (i.e., with diacritical vowel marks - hereafter "pointing"), and only later students move on to reading "unpointed" texts. In our examination of the mismatch between the graphic and phonological structure, we therefore refer also to vowel pronunciation (which actually changes the word), whether or not "pointing" appears in the text . Reading disability occurs due to a few reasons among which are the motivational-emotive factor (Bentin, 1992), the environmental factor (Vernon, 1979), the mental-cognitive factor (Rutter and Yule, 1975), the psychological-sensory factor (Rudel and Denkla, 1976) and the neurodevelopmental factor (Geschwind, 1985). Dyslexia, as a reading impairment, is often attributed to primary neuro-developmental damage. One of the dilemmas in research is defining the dominant neuro-developmental cause that explains dyslexia (Rahmany, 1990). Significant discussions between researchers revolve around the dominance of auditory or visual perception as causes of reading disability. Auditory perception impairment is considered to be a cause for reading disability (Vernon, 1966), and especially among hearing impaired populations (Just and Carpenter, 1987). Auditory perception, which is involved in the acquisition of reading skills, is usually related to phonological and phonetic competence, to the ability to distinguish between similar speech sounds, to identify the phonetic structure of the word, and to the competence in grapheme-phoneme matching (Goswami and Bryant, 1990, Morais et al. 1986, Marschark and Harris, 1996).

Frith (1985), testing a dyslexic population, found support to the correlation between the impairment in auditory perception and auditory discrimination ability, and between reading disability. Frith (ibid.) examined their auditory perception ability by tests which required identifying phonetic word structures, identifying incomplete words (with deleted letters) and forming words by joining letters, as well as by pronouncing meaningless (nonsense) words. Bradley and Bryant (1983) found that it was possible to expect adequate reading ability from children with good phonological awareness on the basis of a test for identifying similar sounds. Children with poor phonological ability, however, were slow and revealed difficulty in reading acquisition. Birch and Belmont (1965), Jorm (1977), Holmes and Peper (1978) and others support the correlation between reading disabilities and auditory perceptual impairment. In most cases, researchers include auditory discrimination ability and auditory analysis in testing auditory perception of dyslexics (Jorm, 1977). In such cases, a dyslexic student reveals difficulty in splitting a word into its components while maintaining, i.e., the sequence of speech sounds in auditory word completion. Sharan and Sharan (1978) refer to a particular visual-auditory generalization causing readers to exchange a written word (whose meaning they know), with another word, e.g., "grass" for "herb", and to split a word into its components (syllables), e.g., "cet" / "faucet". Correlation between impaired auditory short-term memory and poor reading is also mentioned in Sharony, 1990. The idea that auditory perception is the dominant reason explaining reading difficulties has met with opposition by other researchers who suggest visual perception as the dominant reason for dyslexia. According to these researchers, due to the impaired visual perception, readers have difficulties with the holistic perception of the word which they read sequentially. Their reading errors include changing the order of letters (graphemes) in a word, or letter (phoneme) deletion and insertion (Kinsbourne and Warrington, 1962, Warrington and Rabin, 1971, Warrington and Shallice, 1979, Bradshaw, 1975). These findings led to new approaches distinguishing between categories of reading disability according to the involved perceptual impairment. Instead of "dyslexia" as one category, sub-groups of dyslexia have been suggested. Boder (1973) for example, distinguishes auditory-phonological dyslexia from visual dyslexia. A similar approach is found in Marshall and Newcomb (1973) and in Shallice and Warrington (1975) who define "deep" (phonemic) dyslexia and "superficial" dyslexia. The "deep" dyslexia is related to the auditory perceptual impairment, whereas the "superficial" dyslexia relates to the visual perceptual impairment. The following error types characterize deep (phonemic) dyslexia (Shallice and Warrington, 1975):

1. The subject's errors are semantic rather than phonological. S/he substitutes the written word by another word related to the content of the sentence. 2. Word type affects the errors: The subject can read a word better when it represents an object and not function words (prepositions). 3. The dyslexic subject shows difficulty in reading meaningless words. Frith (1985) who stresses the perceptual-auditory factor as a cause for reading difficulties in dyslexics, deals with the problem also from the point of view of hearing-impaired readers. In his opinion, beyond the argument about reading comprehension, most of the children gradually move on to holistic reading. This process is due to the change from a visually distinguished logographic strategy to a phonological or alphabetic strategy, in which letters are translated into phonemes since they (the graphemes) represent the phonological system. This phenomenon explains according to Frith (ibid.) the difficulties of hearing-impaired readers who are usually known to have poor phonological abilities. The hearing-impaired person, whether growing in a hearing or in a deaf home where one or both parents are deaf, was found to have lower phonological skills than those of the hearing population. Waters and Doehring (1990) found in seven to twenty years old hearing-impaired subjects difficulty in phonological coding ability, which is one of the abilities examined in auditory perception functioning in dyslexics. Harris and Beech (1995) found in hearing-impaired subjects poorer phonemic awareness than in hearing subjects. The same study claims that up to the age of seven years, the hearing-impaired child hardly relies on phonological coding. Gradually, with age, the child acquires this skill based on developing "inner language". Evidence for difficulties which characterize hearing-impaired readers also exists in other studies (cf. Truax, 1978, Soderbergh, 1985,). Bellugi (quoted in Geschwind, 1985) presented evidence that deaf children, born in deaf families, acquired sign language at a rate parallel to that of oral language acquisition in normal hearers. Hence, relying on Geschwind's definition of reading as the ability to "derive meaning from any kind of visual representation of language" (Geschwind, 1985:198) the hearing-impaired child is expected to read, as it were, without any difficulty. Still, it has been found that reading is the most difficult academic challenge for the hearingimpaired child (Marschark and Harris, 1996). DiFrancesca (1972) found that reading skills of hearing-impaired children increase only by 0.2 score per year. Vernon (1972) found that between the age of 10-16 years, the hearing-impaired student's reading achievement increases by little more than 0.1 score. Also Alan (1983) and Marschark (1997) explain this population's reading difficulty as caused by their impaired phonological channel.

Such findings support the theory about the role of impaired auditory perception in reading disability and determine the background for the comparison between two populations with reading disability: dyslexics with a good peripheral hearing system, but whose auditory perception is impaired, and a non-dyslexic hearing-impaired population whose peripheral hearing system is impaired. These dilemmas and research findings are the basis of our present research. The Assumptions and Aims of the Present Research Our research examines the assumption that there exist three different types of reading disabilities related to different types of impaired perception: an impaired auditory perception, an impaired visual perception, and an impaired peripheral hearing system. Our research assumptions were that if, indeed, different types of dyslexia (by perception channel) existed, we would find: 1) similar characteristic reading difficulties in subjects with reading disability due to auditory perception impairment and hearing-impaired subjects, who suffer from reading difficulties. 2) We expect these impairment types to be different from those of the dyslexic subjects with impaired visual perception. This leads to the major research question in this study: Is it possible to find in the framework of a comparative study similar characteristics of reading difficulties in these populations, based on a physiological source, on the one hand, and on a neuro-developmental (auditory or visual) perception impairment, on the other. Method of Work THE SUBJECTS Three research groups participated in this study, all studying in the 2nd to the 6th elementary school grades. They were selected from the 2nd grade up, since in the first grade children usually struggle with basic reading acquisition. The first group included 20 children per grade, that is, 100 children, who were diagnosed as dyslexic by the first author and suffered only from impaired auditory perception. The second research group included 53 hearing-impaired students in the same grades who were diagnosed as having normal visual perception. The breakdown of this group by class is as follows: 2nd grade - 8 students; 3rd grade - 20 students; 4th grade - 11 students; 5th grade - 10 students; 6th grade - 4 students (see Table 1.). Their hearing impairment was severe or deep, with a hearing level decrease ranging between 70-115 dB, and most of the values ranging between 90-100 dB. The control group consisted of 100 (20 per grade) dyslexic students (also dignosed by the first author) who suffered from impaired visual perception.

The dyslexic groups had received no corrective intervention prior to the tests. The dyslexic children came from Haifa and Northern Israel, as did about half of the hearing impaired children. The latter were students in an integrative school in Haifa. The rest of the hearing impaired students came from various places in the center of Israel, and were tested in the "Shema'" Center for the Hearing-Impaired in Tel-Aviv. Some of the hearing-impaired children talked at a near-normal quality, but others had considerable speaking problems. Therefore, a teacher was always present during the tests, and stopped the reading whenever an error occurred and specified the error to the observing researcher. This procedure was not necessary with the dylsexic groups. The three groups will be henceforth referred to as follows: Auditory - the dyslexics with auditory perception impairment; HI - the hearing-impaired group, and Visual - the dyslexics with visual perception impairment. PROCEDURE The subjects who were diagnosed as reading impaired were examined by a Hebrew diagnosing kit for auditory perception functioning. Following Jorm (1977) and Sharan (1978) we were assisted by the auditory perception tests included in Kidron (1987) which are accepted for most of the diagnosing tasks in Israel: the auditory discrimination test (based on Wopman adapted by Hoga), auditory distinction between syllables (following Dar'in) and the auditory classification test (identification of opening, medial and final speech sounds in a word). The dyslexia diagnosing tests (in the Kidron set, 1987) involve also visual perception functioning tests based on the MVPT test, and include the examination of shape fixation, distinction of form and background, holistic perception, visual sequence and visual memory. The examination of the students' reading skills included the following categories of reading characteristics and difficulty types: 1. Reading meaningful and meaningless text. By this test, a comparison between reading aloud a list of (15) meaningful vs. (15) meaningless (nonce, artificial) words was done. In this part we compared reading of words with and without meaning by the number of errors made by the subjects during reading. Each error was marked numerically for scoring and statistical analysis as follows (cf. Figure 1): -1 - In reading meaningful texts there are more errors than in reading meaningless texts; 0 - In reading meaningful and meaningless texts the number of errors is the same; 1. 1 - In reading meaningful texts there are less errors than in reading meaningless texts.

2. Sequential or holistic reading. The length of the passage (ranging between 100-189 words) was adapted to the student's grade. The term "sequential" means a reading technique which takes single graphemes, unit by unit, uniting them only at the end of the word into a sense unit (i.e., a word), e.g. reading the word 'general' as [ge-ne-ra-l]. The term "holistic" refers to reading the word as a whole, relying on its visual pattern. Figure 2(a,b,c) summarizes the findings. 3. Self-correction in reading. This variable measures the number of spontaneous stops each child made in order to correct any kind of errors in his/her reading (aloud). This measure involves the child's control over the text s/he reads, which requires the use of semantic contextual analysis by an auditory/semantic feedback system. 4. Reading rate. This variable was measured by the number of words read aloud per minute(WPM) as in Engel (1997). The student's reading was stopped after one minute. The raw numbers of each group were averaged by groups and by classes (with standard deviations) is shown in Table 2 and in Figure 5. 5. "Pointed" vs. "unpointed" text: Comparison between the effect of reading texts with or without "pointing". The texts were taken from Engel (1987), and Kidron (1977). The analysis was also done by group and grade, as in the rest of the variables, and raw results were marked by the following coding (cf. Figure 4(a,b,c)): -1 - the "pointed" text involves more reading errors; 0 - the "pointed" and "unpointed" texts yield the same number of errors; 1 - the "pointed" text involves less reading errors. 6. Reading errors: Tests and analysis were made for the following categories of reading errors. (See Figure 6 a, b, c): a - wrong pronunciation of vowels. This is a more or less language specific feature, since as noted, in Hebrew vowels ("pointing") are not written as consonant letters, and errors in vowels may change the word (even if consonants are correctly read). b - reading errors in suffixes and "b/k/l/m" prefixes (for this term see endnote iv). c - errors due to the child's changing the visual form of the word; cf. e.g. Hebrew /hitbonen/ '(he) looked at' vs. reading it wrongly as /hitlonen/ '(he) complained'. d - change in the auditory form (the sound pattern) of the word; cf. e.g. Hebrew [he'etik] '(he) copied' vs. reading it as [ekita] (a non-word). e - change in the content or meaning of the word as a result of the error, changing it into a totally different word.

Findings In this section we summarize mainly verbally the findings of each reading category. Altogether we had six test variables, three groups of subjects, five grades per group, and the interactions amongst them. Due to the large number of statistical tests which analyze the many different variables by different methods, we do not present numerical tables. Let us note here that Kruskall-Wallis test was used to analyze ordinal distributions (X square) such as the reading of meaningful/meaningless words, the effect of pointing, and the holistic/sequential reading style. Two-tailed ANOVA, Wilcoxon, and Duncan tests were used for multi-population comparisons when the variables were sequential as in, e.g., the reading rate. When the interaction between the variables was significant, one-tailed ANOVA was used. READING MEANINGFUL/MEANINGLESS TEXT Inter-grade comparison The HI group There is no clear evidence that the distribution of the difference in error quantity in reading meaningful or meaningless text changes with time. That is to say, the distribution is fixed in time (p=0.154). In the 3rd, 4th and 6th grades (but not in the 2nd and 5th grades) HI significantly tend to a situation where they make more errors in reading meaningful than meaningless words. In the 2nd and 5th grades there is no statistical evidence for this trend. The Auditory group The distribution of the difference in reading meaningful vs. meaningless texts is not fixed throughout the grades (p=0.02). The contribution of the 4th grade is a major source contributing to this effect, for in this grade there are more reading errors in reading meaningful texts than meaningless texts than would be expected by the equal distribution hypothesis. In all the grades, except in the 4th grade, there is no significant difference in the errors made in reading meaningful and meaningless texts. In the 4th grade reading meaningful texts raises significantly the rate of reading errors (p=0.008). The Visual group There is no clear evidence that the content has a different effect along the grades (p=0.10) (in the place test comparison p = 0.48). In all the grades except in the 4th grade it is possible to say that in reading meaningful texts there are significantly less errors than in reading meaningless texts. In the 4th grade this is not significantly evident (p=0.10). Inter-Group Comparison Grades 2,4,5: At the 5% level of significance the visual group tends to improve its reading when the text is meaningful compared to the HI and Auditory groups. There is no significant difference between the HI and the Auditory groups.

Grade 3: There is no significant difference between the HI and the Auditory groups and between the Auditory and Visual groups. The difference between the Visual and HI is significant: The Visual group tends to make less errors in reading meaningful text than the HI. Grade 6: At the 5% level of significance the three populations differ from one another: The Visual group tends to read meaningful texts with less errors than the HI and Auditory groups; the Auditory group reads meaningful texts with less errors compared to the HI. That is to say, the probability that HI would read meaningful texts with many errors is higher than the matching probability of the Auditory group. Considering all grades together, at the 5% level of significance, the Visual group tends to read a meaningful text with less errors than the Auditory group, and the Auditory group - with less errors than the HI group. Figure 1 shows that there is hardly any interaction between the dyslexics (both Visual and Auditory groups) and grade. (The above finding, based on all the grades together, is therefore valid). SEQUENTIAL AND HOLISTIC READING Inter- grade comparison The HI group A significant difference was found between the grades (p=0.045). The source of this difference is mainly in the 2nd grade, where the prevalent reading style is sequential, whereas in the other grades, the holistic type is dominant. Between grades 3 to 6 there is no significant difference (p=0.384) (cf. Figure 2). The Auditory group A significant difference along the grades was found (p=0.001), but without the findings of grade 2 the difference is not significant (p=0.465). The Visual group There is a significant difference in reading style along the grades (p=0.001), but without grade 2 the difference is not significant (p=0.062). In grade 2 the prevalent reading style is sequential, and it remains so in the higher grades Inter-Group Comparison For this comparison the three reading styles were scored as follows: 1- sequential reading; 2 - mixed (sequential and holistic); 3 - holistic. At the 5% level of significance in the 2nd grade, no difference was found between the HI and the Auditory group. The HI group, more than the Visual

group, tends towards the holistic reading style. In addition, the Auditory group tends to be significantly more holistic than the Visual group. The HI and the Auditory groups differ in this respect from the Visual group (p=0.0009). Third-sixth grade: There is a difference between the groups in reading style distribution (p=0.001). At the 5% level of significance, there is no difference between the HI and the Auditory groups. Each of the HI and the Auditory groups tend to read in the holistic style more significantly than the Visual group. Relating to all the grades and inter-group differences, the Auditory and HI groups read significantly more holistically than the Visual group. There is no significant difference between the HI and the Auditory group (p=0.22). SELF CORRECTION IN READING Inter-grade comparison Self-correction in reading was also examined statistically (cf. Figure 3). No significant difference in the probability to self-correction was found between groups in the 2nd through 6th grades. The hypothesis that each population had a fixed probability to self- correction was not refuted (HI: p=0.206; auditory: p=0.756; visual: p=0.876). In the HI population the probability for self correction was especially low in the 3rd grade. Inter-group comparison As to inter-group differences in this test, at the 5% level of significance this probability is not the same in all the groups and at least two are different. The probability estimation for self correction is as follows: HI: 0.38, Auditory: 0.58, Visual: 0.55 At the 5% level of significance only the difference between the HI and the Auditory groups is significant (p=0.016). Likewise, there is a significant difference between both dyslexic groups (together) and the HI (p=0.013). In both cases, the HI have a lower probability for self-correction. In the 5th and 6th grades there was no difference between the three groups at the 5% level of significance (p=0.792). READING RATE In a two-tailed ANOVA examination which was made in order to study the effect of grades, the studied group and the interaction between group and grade, we found an interaction at the 5% level of significance (p=0.0001). In other words, inter-group differences in reading rate depend on the grade. Inter-grade comparison

In the 2nd grade there is no difference between the populations. In the higher grades, the number of WPM of the HI increases with the grade. In the Visual group the increase in reading rate is the most moderate. In the Auditory group the increase is moderate and then, after the 4th grade, their reading rate becomes equal to that of the HI. In the 2nd and 4th grades there is no significant difference between the groups. In the 3rd grade, the HI group has a significantly higher number of WPM than the Visual or Auditory groups. In the 5th and 6th grades there is no significant difference between the HI and the Auditory groups, and the Visual group differs significantly from the other two groups. Inter-group comparison (in WPM) The HI group In the 2nd grade the reading rate is especially low. It is also lower than that of the other groups. In the 3rd and 4th grades the number of words is equal, and higher than in the 2nd grade. In the 5th and 6th grade, the average number of WPM is significantly higher than in the lower grades. The Auditory group The number of WPM in the 4th grade is significantly different from both the lower and higher grades. In the 5th and 6th grades the number of WPM is different from and significantly higher than in the 4th grade. The Visual group The number of WPM in the higher grades (4th through 6th grades) is significantly higher than that of the lower grades (2nd and 3rd grades). EFFECT OF READING "POINTED" AND "UNPOINTED" TEXT Inter-grade comparison The HI group At the 5% level of significance the effect of pointing on the number of reading errors is not equal for all the grades. In the 2nd and 6th grades there were less errors with pointed texts than could be anticipated according to the hypothesis of equal distribution per grade. When the five grades are examined together, there is no statistical evidence to support the hypothesis that pointing affects the number of errors. The Auditory group At the 5% level of significance the pointing effect on the number of reading errors is not equal for all grades. Two facts contributed to this finding: In the 2nd grade there were too many cases in which "pointing" helped reading (i.e., less reading errors) and in the 6th grade there were too many errors in reading "pointed" texts . Throughout the grades there is a growing tendency to make more errors in "pointed" texts. Apparently "pointing" interferes with the reading process (cf. Figure 4), although this has not been statistically tested.

In the 2nd grade reading a pointed text decreased significantly the number of reading errors; in the 3rd and 5th grades there was no statistical evidence for a difference between reading "pointed" and "unpointed" texts. In the 4th and 6th grades there was statistical evidence that reading "pointed" texts increased the number of errors. In general, throughout the grades, in this group "pointing" had an interfering factor. The Visual group There is no statistically significant difference between the 2nd through 6th grades in the distribution of the effect of "pointing" on the errors (p=0.243). In all the grades there are less errors when reading "pointed" texts than "unpointed" texts. In the 2nd and 3rd grades there are significantly less errors in reading "pointed" texts (p=0.0001, and p=0.0003 respectively). The same findings are true for the 5th grade. For the 4th and the 6th grades there is no statistical evidence of this trend. Inter-Group Comparison At the 5% level of significance, in 2nd, 3rd and 5th grades the HI andAauditory groups are similar. The Visual group tends to make significantly less errors while reading "pointed" texts compared to reading "unpointed" texts than the HI and Auditory groups. In the 4th grade, at the 5% level of significance, only the difference between the Visual and Auditory groups is significant in that the Visual group tends to make significantly less errors in reading "pointed" texts than the Auditory group. There is no significant difference between the HI and the Visual or Auditory groups in this grade. In the 6th grade, at the 5% level of significance, there is a significant difference between the HI and the Auditory group, as well as between the Auditory and the Visual groups. The Auditory group tends to make significantly more errors in reading pointed texts. There is no significant difference between the HI and the Visual group in this grade. Considering all grades together at the 5% level of significance, the Visual group tends significantly to read "pointed" texts with less errors than the HI and the Auditory groups. There is no significant difference between the HI and the Auditory groups. THE TYPES OF READING ERRORS Inter-grade comparison In both the 2nd and 3rd grades there is a significant inter-group difference in the error type distribution. At the 5% level of significance all the groups differ: In the HI the most prevalent error types are categories a, b, c; In the Auditory group the most prevalent error type is category c; In the Visual group the most prevalent error types are categories a, d. In the 2nd grade category no. e does not occur at all in the HI and Auditory groups. In this grade it is found in only one Visual subject. In the 3rd grade of the HI no errors of category d and e occur.

In the 4th grade at the 5% level of significance the HI differ significantly from the Auditory and Visual groups, between whom there is no significant difference. In the 5th grade at the 5% level of significance there is significant difference between the groups: In HI, errors from categories b, c are frequent, whereas errors of categories d, e do not occur at all.; in the Auditory group there are no category b errors; in the Visual group there are more category d errors than in the HI and the Auditory groups. The Visual and Auditory groups also differ regarding errors of category c. In the 6th grade at the 5% level of significance there is no difference between the HI and the Auditory groups. Errors of category b were not found at all. The Visual group differs from the Auditory one: More category c errors were found in the Auditory group whereas errors of category d are more frequent in the Visual group. Errors of category b were not found in the dyslexic groups, but in the HI it is found in all except the 6th grade. In the 5th and 6th grades of the HI no errors of categories d and e were found. Comparison of visual and auditory pattern errors (categories c, d) This comparison examines the difference in the distribution of errors of category c and d throughout the grades in altogether 207 students. At the 5% level of significance the distribution of these errors was not the same in all the three groups. Dependency was found between group and category type as follows: No significant difference was found between HI and auditory groups; Each of HI and auditory groups differed significantly from the visual group; These findings result from the fact that in the HI and Auditory groups errors of category c are frequent, whereas in the Visual group many category d errors occur. It should also be noted that in the Auditory and Visual groups the difference in error distribution along grades is not significant, i.e., the distribution is stable throughout the grades. In contrast, in the HI group a significant difference was found which is due to the fact that in the 4th grade category c errors were not found at all and in the other grades almost only category c errors were found. Testing each grade separately at the 5% level of significance yielded the following findings: In the 2nd grade - the difference between the HI and Auditory groups is not significant. The HI and the Auditory groups differ from the Visual group (mainly due to category 3 vs. category d errors). In the 3rd grade, the difference between the HI and Auditory groups is not significant (in both mainly category c errors occur). The difference between

the Auditory and Visual groups is significant (p=0.011). The Visual group differs significantly from the HI (the Visual group has mainly errors of category 4, whereas in the HI group mainly errors of category c occur). In the 4th grade - HI and Auditory groups differ significantly (HI have mainly category 4 errors, the Auditory group has mainly category c errors). The HI and Visual groups are not significantly different (mainly category d errors occur); neither are the Auditory and Visual groups. In the 5th and 6th grades HI and Auditory groups are not significantly different. But both HI and Auditory groups differ significantly from the Visual group (mainly due to errors in category c in the HI and Auditory groups vs. errors of category d in the Visual group). Discussion and summary General reference to the research hypotheses

1. The hypothesis about the similarity in reading difficulty characteristics of populations with auditory perceptual channel impairment (due to physiological or neuro-developmental origins, i.e., the HI and Auditory groups) has not been refuted. 2. Our findings do not disclaim the hypothesis that the HI and Auditory populations, impaired in their auditory perception, share characteristics of reading difficulties which differ from those of the population with visual perception channel impairment. 3. The findings of this research support hypotheses of previous studies which distinguish between reading difficulties due to impairment of the auditory vs. the visual perceptual channels. Thus, this research also supports the description of dyslexia as consisting of sub-categories instead of viewing it as a single category. Discussion of the findings Reading meaningful texts The Visual group reads meaningful better than meaningless texts. The HI group reads meaningless better than meaningful texts. The Auditory group is not affected by the presence or lack of meaning in the text, although in the 4th grade the presence of meaning detracts from their reading quality. This picture may be explained by different causes. The HI: Since their reading is mainly holistic-structural, they probably search their internal vocabulary for a similar pattern (word); when they fail to find it, they linger on the meaningless word and then decode it more accurately. The Visual group members read sequentially and take rather a long time in reading; understanding the meaning of the text (when there is a meaning) supports the sequential reading process.

The Auditory group, on the other hand, reads holistically and faster and does not dwell on the meaning. It falls, in fact, between the HI and the Visual groups, since its auditory channel is impaired, but its hearing system is better than that of the HI. Therefore, compared to the HI, it has better linguistic skills which help it in reading meaningful texts, but not to the same extent as it helps the Visual group. Sequential and holistic reading In the 2nd grade all the students use the sequential reading strategy. Later on, the Visual group continues to use this strategy, while the HI and the Auditory groups move more and more towards holistic reading. We offer two explanations for this phenomenon, related to the different kinds of impairment: Normal readers pass on to global reading after acquiring (seqential) technical reading. Since the Visual group members have an impaired visual perception channel, which limits their ability to perceive whole elements, these students continue reading in the sequential style. The HI and the Auditory groups, on the other hand, due to their impaired auditory perception channel, use a sight-based (visual) reading technique which is not mediated by speech sounds or generalization of sounds and graphic symbolization. Self-correction in reading The findings of this variable refute the hypothesis that there would be similarity between the auditory and HI groups and difference between them and the Visual group. The HI tend less than the two other groups to self correction in reading, but this difference is not statistically significant. In addition, there is no significant difference in this respect in all the grades (i.e., development along years). Up to 5th and 6th grades, the lowest self-correction rate is that of the HI. This finding is in conformance with the fact that their reading is not semantic, and they do not rely on the content to correct errors. In the 5th-6th grades there is no significant difference between the groups in terms of self-correction. Still, the HI have the lowest probability for self correction. In this case also we may explain this finding by the lack of semantically-supported reading. On the one hand, this is a feature which characterizes the communication of HI: To a large extent, they tend to apply technical-automatic reading "oral Hebrew" which they know less well as it may be considered their second language compared to Hebrew Sign Language. The fact that in the higher grades there is no significant difference between the groups could be explained by the fact that the HI, too, begin at this stage to be aided by the semantic component of the text. They are also probably able to control themselves better than in earlier grades as a result of their maturation and general progress in studies, including language and literacy. This hypothesis, and whether this trend continues later on, should be tested with older subjects. Reading rate In the 2nd grade, reading rate is similar in all the three groups. This finding matches the sequential reading style of all three groups. In the 3rd grade the HI are the fastest readers (in reading aloud). In the 5th and 6th grades the reading rate of the HI and the Auditory groups is similar. It differs, however, from the Visual group, which has the slowest reading rate.

It should be noted that all the subjects read at a slower rate than the average normal readers (Engel, 1997), even after the 2nd grade, which is the stage when the HI (and the Auditory group) move on to holistic reading. In the two dyslexic groups the slow rate can be explained by the combined effect of various reading difficulties. The Auditory group cannot use the contribution of semantic content efficiently, whereas the Visual read sequentially, which slows down their processing of the material; thus both groups lack the reading flow and speed of the normal reader who already identifies and reads the following word while still processing previous words. The effect of reading "pointed" text The visual group reads "pointed" texts better than both the HI and Auditory groups. Between the latter two groups there is no significant difference. "Pointing" does not help the HI's reading, and it actually detracts from the reading quality in the auditory group. Phrased differently: the higher the grade, the more interfering factor "pointing" becomes for the HI. A likely explanation for the difference between the populations apparently involves (again) the different reading styles of the groups: The Visual group reads sequentially, and therefore pays attention to every mark in the written line; it takes longer to read and process the additional vowel-indicating marks, but they do add information to enable better reading. The other two groups read globally, holistically. In the process of word identification, comparing a word with the existing vocabulary in their long-term memory, the "pointing" adds more data than needed for processing. This increases the information load, and becomes a detracting factor. Reading errors types Similarity was found between HI and Auditory groups in that most of their errors relate to the visual form of the word. These groups differ from the Visual group whose main errors involve the word's auditory pattern (sound pattern). It may be possible to explain the type of errors in the HI and the Auditory groups by the fact that these groups do not have sound mediation and rely mainly on the visual channel in their reading. Accordingly, they cannot assess fast enough the link between the read letter and its phonological-linguistic meaning, nor do they trace errors in words through the auditory channel. In contrast, the Visual group is mainly aided by auditory information, since its visual channel is impaired. Unlike the Visual and the Auditory groups, the HI have also errors in vowels, suffixes, and "b,k,l,m" prefixes. Regarding this kind of error we suggest an explanation linking the structure of the Hebrew language and speech with their specific impairment. Due to the use of the Hebrew Sign Language, which differs in many respects from oral Hebrew, HI children are not sufficiently proficient in the rich morphological structure of Hebrew in the first grades of school and make morphological errors in both reading and spontaneous speech. They do not always use the "b,k,l,m letters" since these are not part of the lexemic word indicated by the root consonants. (They are also morphological, syntactic and semantic elements, i.e., morphemes and

syntagms with certain meanings, which they do not perceive well enough.) As soon as they identify the written word by its consonantal root letters, which in Hebrew supply it its basic meaning, they think they understand it (by linguistic and non-linguistic contextual cues) and do not pause to study its (partly redundant) morphological pattern. Only for the 6th grade we did not find such errors in the HI group. Maybe at this stage their linguistic knowledge (expressed in general oral language and literacy) is already more deeply rooted, so that it may actively contribute to decoding the written text more than in earlier grades. Comparison with Shallice and Warrington (1975) Shallice and Warrington (1975) focused on reading disability and impaired auditory perception and suggested a sub-group of dyslexia, i.e., phonemic dyslexia. Since our research touches upon this question from a different aspect, we are comparing here our findings with those of Shallice and Warrington (1975):

1. These scholars found about 33% semantic errors in their study vs. 66% of the "visual" kind in their subject. Our research seems to support this finding. The HI (mainly) do not rely on the content and therefore their reading is not "semantic", and they make many semantic errors; the Auditory group does not show preference to reading meaningful or nonsense texts, but nevertheless make also semantic errors. However, the Visual group is significantly different from both those groups in use of the semantic content of the text (and closer to the results of Shallice and Warrington (1975)). 2. Shallice and Warrington (1975) refuted the hypothesis of "auditory" (sound) errors in their subject. Our findings corroborate this view regarding the HI and Auditory groups (and are in contrast with the Visual group's results). Findings concerning the HI This study sustains the view that reading is the "most difficult academic challenge for the hearing impaired" (Marschark and Harris, 1997).We found that in comparison with the other two populations, which can be defined as reading impaired, the HI group was even weaker. Although its reading rate was the highest of the three groups, this finding was significant only in the 3rd grade. The HI population's reading was the least supported by the meaning of the text, so that their reading was not semantic, and they were least prone to self-correction of reading errors, and the types of errors they made were the most heterogeneous of the three studied groups, and included prefixes such as "b,k,l,m" and word suffixes. These difficulties seem to result from the correlation between their reading and language knowledge expressed in their speaking skills. Is the 4th grade special?

The 4th grade revealed specific features when the parameters and groups findings were crossed. These findings include the following:

1. Reading meaningful and nonsense texts: The Auditory group had an increase of errors when reading meaningful texts. In the other grades of this group there is no difference between reading meaningful or nonsense texts. As to the Visual group, no significant evidence was found in the 4th grade that in reading a meaningful text there were less errors compared to the other grades. 2. In the Auditory group: Reading rate in this grade is statistically related neither to the lower nor to the higher grades. 3. In the 4th grade of the Auditory group "pointing" contributes to the errors, contrary to the other grades of this group. Also from the point of view of inter-group differences, in this grade there is significant difference only between the Visual and the Auditory groups (unlike other reading characteristics). 4. In the 4th grade of the HI group, there are hardly any visual pattern (category c) errors. In the other grades almost only visual form errors occur. 5. In comparing reading errors of category c and category d in the 4th grade with these types of errors in the other grades, the HI and Auditory group differ. The HI and the Visual group, however, do not differ significantly between themselves, nor do the Auditory and Visual groups differ from each other. These findings about the 4th grade raise questions as to the origin and explanation of these differences. Can we consider the dyslexic and HI population at this grade as being in a special developmental stage, at least concerning reading skills? Is it in any way related to the fact that the 4th grade is an important stage of maturation for the child, as implied by the structure of syllabi for this grade in elementary schools? A possible research question for further study would, thus, be whether the 4th grade begins a new stage in the development of reading skills among reading-impaired children as those studied here. Conclusion This research studied three groups of children with impairments that lead to reading difficulties. We found differences between the groups with impaired auditory or visual perception. These differences support the approach which considers dyslexia as an "umbrella" term which includes sub-groups of various kinds of reading impairment. The research revealed typical features for the reading difficulties of members of each of these groups. These differences depended both on impairment type and on grade, i.e., the child's developmental stage. In these respects, the HI and Auditory groups are more similar to each other than any of them to the Visual group. Our research hypotheses were thus corroborated. The HI children were found to have the lowest reading scores among the three groups. This finding was, in fact, anticipated, since reading acquisition

problems in HI children are still considered their greatest academic challenge, which teachers, educators and researchers all over the world have not solved so far. The study is relevant to other research topics such as the study of language problems among HI (with various levels of impairment), the study of reading processes in general, development stages of impaired beginning readers, the passage from the first stage, i.e., technical reading, to higher reading skills, differences between sequential and holistic readers, differences between reading patterns of male and female readers, etc. The findings of this research may have implications on intervention manners, e.g., in the strategies of teaching reading to members of the studied populations. When dealing with Hebrew, for example, one should consider using "pointing" with impaired visual-perception students even in the higher grades and avoiding the use of "pointed" texts in HI and students with impaired auditory perception, due to the different reactions to "pointing" in these populations. It might be also advisable to train members of the HI and the Auditory groups to read meaningful texts and use "reading comprehension" techniques, especially in the 4th grade, and meticulously train sequential reading to members of the HI and Auditory groups in order to decrease their reading errors. The findings about 4th grade students in the three groups of our research are very interesting, and deserve a separate future study, as do our findings about errors specific to the Hebrew language. Acknowledgments This study was supported partly by a Technion VPR grant no. 2001576 for the Enhancement of Research, and partly by the Ministry of Education. The authors extend their thanks to the hundreds of dyslexic and hearing impaired children in Haifa and Tel-Aviv, the school principals and Shema' TelAviv director, for their invaluable help in conducting the experiments. The authors cordially thank Prof. A. Cohen, Head of the Statistical Consulting Laboratory, at the Technion School of Industrial Engineering and Management, and especially Dr. O. Barnett, for their help with the statistical analysis. References Allen, T.E. (1974, 1983) Patterns of academic achievement among hearing impaired students, in: A.N. Shidroth and M.A. Karchmer (eds.) Deaf Children in America, San Diego: Hill Press, 161-206. Bentin, S. (1992) The effect of learning to read in the first grade on the development of children's phonological awareness, Megammot, 38, 442-455 (in Hebrew).

Birch. H. and L. Belmont (1965) Auditory-visual integration in brain damaged and normal children, Developmental Medicine and Child Neurology, 7: 3-7 Bradley, L. and P.E. Bryant (1983) Categorizing sounds and learning to read a causal explanation, Nature, 301, 419-421. Bradshaw, J. L. (1975) Three interrelated problems in reading: a review, Memory and Cognition, 3, 123-134 Boder, E. (1973) Developmental dyslexia: a diagnostic approach based on three typical reading spelling patterns, Developmental Medicine and Child Neurology 15, 663 ff. Chall, J.S. (1982) Stages of Reading Development, New York: McGraw-Hill. Colarusso, R. P. and D. Hammill (eds.) 1972 . M.V.P.T. Motor Free Visual Perception Test, San Rafael, CA: Academic Therapy Publications. Cornoldi C. and J. Oakhill (eds.) (1996) Reading Comprehension Difficulties Processes and Intervention, Mahwah, NJ: Lawrence Erlbaum Associates. DiFrancesca, S. (1972) Academic Achievement Test Results of a National Testing Program for Hearing Impaired Students, Washington D.C.: Gallaudet College, Office of Demographic Studies. Engel, R. (1987) Dysgraphics - Theoretical and Applied Aspects of Writing Impairment, Tel-Aviv: Ramot Publishing House, the University of Tel-Aviv. Engel, R. (1997) Instrument for locating students with learning disabilities, International Journal of Rehabilitation Research, Vol. 20 , 2 (June), 169-181. Frith, U. (1985) "Beneath the surface of developmental dyslexia", in: Patterson K.E. Marshall, and M. Coltheart (eds.) Surface Dyslexia, Hillsdale, NJ: Lawrence Erlbaum Associates, 301-330 Geschwind, N. (1985) Dyslexia in neurological perspective, in: Geschwind, N. (ed.) Dyslexia, A Neuroscientific Approach to Clinical Evaluation, U.S.A.: Little Brown and Comp. Glinert, L. (1989) The Grammar of Modern Hebrew, Cambridge, New York: Cambridge University Press. Goswami, U. and P. Bryant (1990) Phonological Skills and Learning to Read, Hove: Lawrence Erlbaum Associates Holmes, D.L. and R.J. Peper (1978) An evaluation of the use of spelling - a reply to Forster and Grierson, British Journal of Psychology, 69, 499-503.

Harris, M., and J. Beech, (1995) Reading development in prelingually deaf children, in K. Belson and Z. Reger (eds.) Children's language, vol. 8, Hillsdale, NJ: Lawrence Erlbaum Associates, 181-202. Jorm, A. F., (1977) Parietal lobe function in developmental dyslexia Neuropsychologia, 15, 841844. Just, M.A. and Carpenter, P.A.(1987) The Psychology of Reading and Language Comprehension, Boston: Allyn and Bacon. Kidron, R. (1987) Diagnostic Didactic Test, Jerusalem: Ministry of Education. Kinsbourne, M. & Warrington, E. K. (1962) A disorder of simultaneous form perception, Brain, 85, 461-486 Marshall , J, C., & Newcombe, F. (1973,) Patterns of paralexia: a psycholinguistic approach, Journal of Psycholinguistic Research, 2, 175-206 Marschark, M. (1997) Growing Up Deaf, New York: Oxford University Press Marschark, M. and M. Harris (1996) Success and failure in learning to read: The special case(?) of deaf children, in: C. Cornoldi and J. Oakhill (eds.) Reading comprehension Difficulties - Processes and Intervention, Mahwah, NJ: Lawrence Erlbaum Associates, 279-300 Morais, J., Beetelson, P., Cary, L, & Alegria, J., (1986) Literacy training and speech segmentation, Cognition, 24, 45-64. Morrow, L.M. (1989) Literacy Development in the Early Years: Helping Children Read and Write, Englewood Cliffs, NJ: Prentice Hall. Olson, R.K. (1994) Language deficits in "specific" reading disability, in: M.A. Gernsbacher (ed.) Handbook of Psycholinguistics, San Diego: Academic Press, 895-916. Perfetti, C.A. (1994) Psycholinguistic and reading ability, in: M.A. Gernsbacher (ed.) Handbook of Psycholinguistics, San Diego: Academic Press, 849-894. Rahmany, L. (1990) Mind and Learning, Tel-Aviv: Papyrus (in Hebrew). Rayner, K. (1994) The Psychology of Reading, Englewood Cliffs: Prentice Hall Rosenhouse, J., D. Feitelson, B. Kita and Z. Goldstein (1997) Interactive reading aloud to Israeli first graders: Its contribution to literacy development, Reading Research Quarterly, 32:2, 168-183.

Rutter, M. and Yule, W. (1975) The concept of specific reading retardation, Journal of Child Psychology and Psychiatry, 16, 181-198 Rudel, R.G. and Denkla, M.B. (1976) Relationship of IQ and reading score to visual, spatial, and temporal matching tasks, Journal of Learning Disabilities, 9, 169-178 Shanahan, T.and S.B. Neuman(1997) Literacy research that makes a difference, Reading Research Quarterly, 32:2, 202-210. Shallice , T., and Warrington, E. (1975) Word recognition in a phonemic dyslexic patient, Journal of Experimental Psychology, 27, 187-199 Sharan, S. and Y. Sharan (1978) Learning Impairment and Its Correction, Merhavia: Po'alim Library (in Hebrew). Sharony, V. (1990) "What is learning impairment", in: Problems in Special Education, Tel-Aviv: The Open University (in Hebrew). Soderberg, R. (1985) Early reading with deaf children, Prospects, 15, 1, 7785. Truax, R.R. (1978) Reading and language, in: R.R. Kretschmer and L.W. Kretschmer (eds.) Language Development and Intervention with the Hearing Impaired, Baltimore: University Park Press, 279-310. Vernon, M. D. (1979) Variability in reading retardation, British Journal of Psychology, 70, 7-9. Vernon, M. D. (1972) Mind over mouth: A rationale for total communication, Volta Review, 74, 529-54 Vernon, M. D. (1966) Perception in relation to cognition, in A.H. Kidd and J.L. Rivoire (eds.) Perceptual Development in Children, New York: International Universities Press. Warrington, E.K. and P. Rabin (1971) Visual span of apprehension in patients with unilateral cerebral lesions Quarterly Journal of Experimental Psychology, 23, 423-431. Warrington, E.K and T. Shallice (1979) Semantic access dyslexia, Brain, 102, 43-63 Waters, G. S. and D.G. Doehring (1990) Reading acquisition in congenitally deaf children who communicate orally: insights from an analysis of component reading, language, and memory skills, in T. H. Carr and B.A. Levy (eds.) Reading and its Development, San Diego: Academic press, 232-373

Whitney, P., D. Budd, R.S. Bramucci and R.S. Crane (1995) On babies, bath water and schemata: A reconsideration of top-down processes in comprehension, Discourse Processes, 20, 135-166. Table 1. Breakdown of Hearing Impaired subjects per grade and region Grade Haifa Region Tel-Aviv Region Total 2nd

7

1

8

3rd

7

13

20

4th

6

5

11

5th

5

5

10

6th

4

---

4

Total

29

24

53

group

2nd grade 3rd grade 4th grade 5th grade 6th grade

HI average STD

14 5.4

49.6 5.4

50.8 11.2

61.3 15.8

66.41 17.9

auditory average STD

17.40 7

20.95 10.8

43.2 13.4

59.9 15.7

61.3 15.6

visual average STD

17.85 6.2

20.75 12.34

42.5 15.7

43.3 19.4

48.4 16.6

Figure Captions Figure 1. Effect of reading meaningful or meaningless text in the three groups by grades (High scores indicate that meaningful text helps reading. The asterisk marks significance) Figure 2. Sequential, holistic and combined reading in the three groups by grades a: Hearing Impaired (HI), b: Auditory Perception (AP), c: Visual Perception (VP) Figure 3. Proportion of self-correction in reading in the three groups Figure 4. Effect of reading "pointed" and "un-pointed" text in the three groups by grades

a: Hearing Impaired (HI), b: Auditory Perception (AP), c: Visual Perception (VP) Figure 5. Average number of words per minute (WPM) in the three groups by grades Figure 6. Technical reading errors by the "Hearing Impaired" in the three groups by grades a: Hearing Impaired (HI), b: Auditory Perception (AP), c: Visual Perception (VP) Notes a - In Classical Hebrew all the vowels were indicated by diacritical signs, in addition to 4 letters used to indicate long vowels ("mater lectionis"). In Modern Hebrew, new rules have been added by the Academy of the Hebrew Language, so that the use of these "mater lectionis" letters has expanded to more cases than in the past. b - It was later found that hearing impaired and deaf children often acquire sign language even earlier than hearing children acquire oral langauge. c - The ages of children of this group varied more than in the dyslexic groups, because as is well known, hearing impaired children are generally slower in developing their language and literacy skills than hearing children. We therefore ignore individual ages in this study and refer to grades only. d - "Too many" compared to their occurrence probability according to the equal distribution hypothesis. e - Probably as a result of the mediation of the structure of Hebrew Sign Language they do not cosnider it important to distinguish between, e.g., /lamad/ ??? (he studied) and /lamda/ ???? (she studied) where the last letter is not pronounced at all, or between /lamad/??? (he learnt) and /lilmod/ ????? (to learn) where the first letter heads the infinitive form of the root.