EVIDENCE LINKING VISUAL PROCESSING WITH NORMAL AND IMPAIRED READING DEVELOPMENT

Although phonological skills have been by far the most popular candidates for dimensions of reading ability, additional possibilities should not be overlooked. Visual perceptual skill is an alternative which has recently been much neglected. It is self-evident that even the possession of a fully operational phonological system would be insufficient to support literacy skills in a subject who could not retain the written representations of sounds due to a visual problem. Furthermore, the antithetic situation would also seem to hold. The existence of the developmental dyslexic, R.E. (Campbell & Butterworth, 1985), suggests that it is possible to acquire literacy skills despite severe phonological difficulties, apparently by reliance upon strong visual memory skills.

Frequently, visual skills have been taken for granted and it has been assumed that a child will possess visual abilities which will be sufficient to support reading acquisition. For example, Goswami (1986) has argued that children acquire knowledge about the spelling sequences in words through the use of analogies based upon their phonological awareness of onset and rime segments. This theory contains an implicit assumption that children will be capable of perceiving the visual similarity between words and be able to isolate and retain the nonredundant letter groups upon which this similarity rests.

The possibility that pre-existing visual skills may influence reading strategies has already been raised (e.g. Snowling, 1987), but is there any evidence that visual impairments may be implicated in developmental dyslexia? In fact, the theory that developmental dyslexies suffered from a visual impairment was actually pervasive in the literature for about thirty years following the publication of Orton's paper in 1937. In 1979, Vellutino reviewed the area and highlighted some serious methodological inadequacies which cast doubt upon this 'perceptual deficit' hypothesis. Nevertheless, reports of developmental dyslexies with what were

considered primarily visual problems have appeared consistently in the literature. Boder (1973) classified 9% of the developmental dyslexies in her study as suffering from an essentially visual difficulty. Although Boder's work suggested that visual impairments may be less common than phonological difficulties in developmental dyslexia, this does not mean that the role of visual difficulties should be disregarded. Seymour (1986) has identified cases of visual-processor and morphemic dyslexia. These subjects demonstrated analytic weaknesses in tasks requiring comparison of letter arrays or problems in more holistic processes such as the instant recognition of words.

Evidence of visual impairments has been less forthcoming from group studies (for a review see Vellutino, 1979). This should not seem surprising if visual deficits are relatively less common than phonological ones. By taking a group average, difficulties which are suffered by only a few members of the group are likely to be concealed. Another methodological point concerns the experimental tasks which have been used. Vellutino (1979) and Wilding (1989, 1990) have argued that it is difficult to interpret performance on reading-based tasks which use words and letters as stimuli. One reason for this is the difficulty in determining whether dyslexic performance in such tasks reflects underlying problems or reading strategies adopted to compensate for their problems. For example, are dyslexies who utilise a laborious sounding-out strategy impaired in their ability to assign phonology, or in their ability to segment a letter string, in order to assign phonology effectively? To reiterate Ellis' (1985) comment upon the nature of phonological dyslexia, the explanation of a nonword naming deficit need not be a problem with phonological processing, it could equally well result from a difficulty with the visual segmentation of a letter string. Consequently, extrinsic tasks may be a valuable means of extending our knowledge about visual skills in developmental dyslexia, just as they have proved illuminating in relation to phonological processing.

Johnston, Anderson, Perrett & Holligan (1990) used extrinsic tasks to examine the visual and phonological segmentation skills of dyslexic children. They

chose the Auditory Organisation Task as a test of phonological segmentation and the Children's Embedded Figures task to investigate visual segmentation skills. This latter test required the subject to locate a simple shape embedded within a more complex figure. The Embedded Figures test was selected because it was an aspect of visual segmentation which had already been implicated in developmental dyslexia. Goetzinger, Dirks & Baer (1960), Lovell, Gray & Oliver (1964) and Stuart (1967) had all reported that dyslexic children aged between 10 and 15 years old had difficulty with similar tasks. However, these studies assessed dyslexic performance relative to chronological age controls and so it was possible that the observed discrepancies between these groups merely reflected the limited reading experience of the dyslexic children. Johnston et al employed both reading age and chronological age comparisons in their study of ten and a half year old developmental dyslexies who were reading at the seven and a half year level. Overall, the dyslexic children were found to be impaired relative to the chronological age controls in both the visual and the phonological segmentation tasks. In fact, the developmental dyslexies performed at a similar level to their reading age controls on both tasks. Closer examination of the data revealed considerable individual differences. Eight of the 20 dyslexies performed more than two standard deviations below the mean of the chronological age controls on the visual segmentation task and seven were as severely impaired on the phonological task (five of whom were also members of the visually impaired group).

With regard to the developmental dyslexies' difficulties, both visual and phonological, there are at least two possible interpretations of such a result. The skills in question could improve conjointly with reading skill and so be a consequence of the developmental dyslexies' reading problems. Alternatively, the skills may be causally related to reading ability but the superior mental age of the dyslexic children may have enabled them to compensate for their difficulties and to bring their performance at least to the level of the reading age controls. This interpretational problem applies equally to the visual and auditory segmentation task.

In the case of phonological segmentation, there is a wide literature testifying to a reciprocal relationship between phonological skills and reading development. Unfortunately, at present, there is no comparable literature examining visual skills. Therefore, it is only possible to speculate that it seems likely that reading instruction would improve visual segmentation skills. On the other hand, the visual segmentation skills of three of the developmental dyslexies studied by Johnston et al, were above average for their chronological age. Thus, it appears that such skills are not purely a consequence of reading experience.

Vellutino (1979) was reluctant to view the early reports of difficulties with visual disembedding tasks as evidence of a visual deficit in developmental dyslexia. His objection was that these studies had not found problems with other visual tasks like the Block Design subtest of the WISC, or in tests of figure rotation and spatial orientation. However, the specificity of a deficit seems rather a spurious criticism and one that has not been raised in relation to studies of phonological problems. Johnston et al used the lack of an impairment on the Mooney Test, which tests visual synthesis, as evidence that the developmental dyslexies were suffering a segmentation problem. This result could be said to enhance rather than undermine their argument.

Treiman & Baron (1981) have proposed that auditory and visual segmentation skills are manifestations of a general perceptual capacity which has a characteristic maturational course. Their conception of the development of this capacity was derived from studies of visual perception (e.g. Ghent, 1956; Vurpillot, 1976). They suggested that young children were initially holistic perceivers and that the ability to analyse stimuli componentially developed only gradually. Treiman & Baron saw this as a visual parallel of the properties of phoneme perception, namely that although the identification of a spoken word involves the perception of phonemes, this does not necessarily entail phenomenological awareness of the phonemes themselves. Vurpillot (1976) commented that young children viewed line-drawings of objects as very rigid configurations. It was not until they were about 6 years of age that they

were able to use a secondary level of perceptual organisation whereby a line or group of lines could also be independently perceived as part of a substructure.

Kolinsky, Morais, Content & Cary (1987) examined the development of these latter postperceptual visual processes. They employed a version of the 'part- probe' task which consisted of a set of six-segment figures paired with various three- segment parts. The subjects were asked to say whether they could find the part within the larger figure. Kolinsky et al reported that preschool children and some first graders found this task extremely difficult. Second graders, on the other hand, were much more successful. The authors also used this task to test unschooled adults and discovered that they performed at a similar level to the preschool children. This led Kolinsky et al to conclude that the ability to make componential analyses of visual stimuli developed as a result of educational instruction. The nature of this instruction was not specified, although it did not appear to be reading instruction since some of the unschooled adults were described as ex-illiterates. However, neither the literacy achievements of this group nor the nature of the reading instruction that they received was described.

Unfortunately, it is debatable whether this rather abstract task is strictly comparable to the Embedded Figures tasks used in the studies referred to previously. The figures in the part-probe task were geometric line-drawings which may not have had the same holistic properties as pictures of recognisable objects. In a later study. Kolinsky (1989) found that the ability of young children to analyse shapes into parts was influenced by the nature of the shape itself. Parts of open shapes like arrows were easier to perceive than parts of closed shapes like triangles.

Elkind, Koegler & Go (1964) investigated children's perception of figures in which the parts also depicted objects e.g. line-drawings of fruits arranged to form the shape of a man. They concluded that young children can sometimes see the parts of a figure although this depended very much upon the nature of the stimuli. However, the simultaneous perception of the wholes and parts of complex visual stimuli did not appear to develop until around 8 or 9 years of age.

Further research will be necessary before it can be established whether visual segmentation skills are causally related to reading development. At present visual segmentation problems would seem to have the potential to disrupt decoding skills in word recognition and have been associated with the reading difficulties of some dyslexic children. It may be that by studying visual segmentation skills in reading it will be possible to provide a more comprehensive account of reading acquisition.

CHAPTER 4

STUDY ONE