One mainstream theory of dyslexia suggests that the reading difficulties experienced by dyslexic individuals are caused by poor underlying phonological representations, which impair their ability to make the necessary links between graphemes and phonemes.
However, this suggestion is usually based on evidence from tasks which do not assess phonological representations directly, but rather by means of metalinguistic tasks which involve manipulating phonological units such as phonemes. Since such tasks can at best provide only equivocal and inconclusive evidence about implicit representations, there is a need to investigate representations while eliminating as many metalinguistic factors as possible. Additionally, since the ability to perform metalinguistic tasks involving segments can be enhanced by the individual’s competence in alphabetic literacy, it is also important to test phonological representations in a way that does not confound segments with alphabeticism.
This study tested a group of 21 dyslexic university students on a range of tasks which addressed both their representation and manipulation skills. To escape the overlap between segments and alphabetic letters, the representations of conventional segmental contrasts (eg /k/ vs /g/) were compared with the representation of suprasegmental contrasts (as seen in minimal pairs such as 'toy factory and toy 'factory), which have no orthographic counterpart. Phonological manipulation skills were also tested, by means of Pig Latin and Spoonerism tasks which required participants to manipulate both segmental and suprasegmental units (eg extracting the segment /b/ from consonant clusters and the main stress from SWW or WSW stress patterns).
The results showed that although the dyslexics were impaired relative to controls when they were required to manipulate both segmental and suprasegmental components of words, no evidence was found for a suprasegmental deficit in the representation tasks. Since the suprasegmental tasks were as closely analogous to the segmental tasks as possible, this finding suggests that in cases where phonological contrasts can exist in mental representations independently of orthography, dyslexics may be impaired more in the ability to manipulate these contrasts rather than in the representation of them per se.
The implications of this suggestion for typical literacy acquisition are discussed in the context of the argument that rather than considering the literacy acquisition process as one of mapping phonemes to graphemes, phonemes themselves must be acquired through a process of metalinguistic analysis which is driven by the acquisition of literacy.