Even in predictable orthographies: Surface dyslexia in Turkish

ABSTRACT

Purpose

We report here, for the first time, on developmental surface dyslexia in Turkish, a very transparent orthography. Surface dyslexia is a deficit in the lexical route, which forces the reader to read words via the sublexical route, leading to regularization errors.

Methods

To detect surface dyslexia, we used reading aloud of loanwords with irregular conversion of vowel length or consonant allophone, and analysed regularization errors. We also tested the properties of this dyslexia using lexical decision on pseudohomophones, repetition and reading of nonwords, and analyses of reading of different types of words. The children with surface dyslexia were identified out of 175 9–10-year-olds who were tested using a reading aloud screening task and tasks that were designed to detect sublexical (rather than lexical) reading of existing words.

Results

We identified 45 fourth-graders with surface dyslexia. Reading speed was less sensitive to surface dyslexia than regularization errors, as only one-third of the children we diagnosed with surface dyslexia according to their reading errors also showed slower reading than controls. A task of lexical decision of pseudohomophones indicated that 31 of the participants had impairment in the orthographic input lexicon, whereas for 14 others the orthographic input lexicon was intact and the deficit is probably at a later stage in the lexical route – the phonological output lexicon or the connection between the lexica. Nonword reading was intact for the majority of surface dyslexia participants (35 of the 45). None of the surface dyslexia participants showed phonological deficits.

Conclusions

Surface dyslexia can be identified even in transparent orthographies once the relevant stimuli and error analyses are used.

Acknowledgments

We wholeheartedly thank the families of the participating children for their participation and the teachers in the schools for their help and guidance. We thank Şebnem Keleş, Özge Üçpınar, Nupelda Yalçınkaya, and Balca Ağaçsapan for their help in data collection, and Hadar Green for insightful comments. We are deeply grateful to the Language and Brain Lab members for their feedback and helpful advice during the development of FRİGÜ test battery.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. In effect, several researchers made this claim that dyslexia in general (not only surface dyslexia) can only be identified in transparent orthographies according to reading speed (German: Wimmer & Mayringer, 2001; Italian: Brizzolara et al., 2006; and Spanish: Serrano & Defior, 2008; Turkish: Raman, 2011). This may have resulted from the salience of surface dyslexia and from not distinguishing between types of dyslexia. However, there is no reason transparent orthographies would differ from other orthographies in the possibility of identifying types of reading errors other than regularizations. Transparency is only relevant to surface dyslexia, not to all other types of dyslexia. Whereas it may indeed be difficult or impossible to identify irregular words in transparent orthographies unless irregular words are presented, it is not less probable than in other orthographies to detect all other kinds of errors, typical of all other dyslexias in transparent orthographies. Namely, dyslexias such as letter position dyslexia and attentional dyslexia should not be more difficult to detect in transparent orthographies than in less predictable ones. Indeed, other types of dyslexia can be identified in transparent orthographies using the analysis of reading errors.

2. Turkish has a regular final stress position, and a few loanwords include irregular stress position (i.e., banka, posta). However, given that the change in stress is very subtle in Turkish, stress irregularity is not a good tool for detecting regularizations in reading (unlike in Italian, for example).

3. These six children also rule out an alternative explanation for the results. It may have been the case that surface dyslexia would manifest itself either in surface errors with normal speed or, in case the participants monitor themselves and check their response against the phonological output lexicon, which would slow their reading down, slow reading without errors. However, this is not the case for our participants. The six slower readers who did not have surface dyslexia had other types of dyslexia.

4. Surface dyslexia is defined by an impairment in the lexical route. In pure cases of surface dyslexia, the non-lexical route is preserved. Thus, for pure cases of surface dyslexia, in which only the lexical route is impaired, a difference is expected between performance in irregular words and in nonwords, which is predicted to be greater in individuals with surface dyslexia than in the control group. To examine this prediction, we applied Bayesian Standardized Difference Tests (Crawford & Garthwaite, 2007) to compute the probability that differences between irregular word and nonword reading would be larger for the surface dyslexics than for the controls. We reported Bayesian p value together with a Point Estimate (PE) of the abnormality of a given score, associated with the 95% credible interval for the estimation. The Bayesian PE directly provides an estimation of the percentage of the control population susceptible to obtain a lower score than the cases score. The Bayesian Standardized Difference Tests found that 30 of the 45 surface dyslexics showed significantly larger irregular word-nonword differences than the control group (PEs<1.5, ps<.02). Of the remaining 15 children, 10 were found to show a deficit in nonwords. The results show that the vast majority of the participants with surface dyslexia, those who did not have a nonword deficit in addition to their surface dyslexia, indeed showed the expected pattern of a larger difference between irregular words and nonwords.

5. Even the success of typical readers in reading (non-potentiophonic) irregular words may serve as a counter-argument for the no-lexical route alternative explanation. If there were only a sublexical route with checking in the phonological output lexicon, then non-potentiophonic irregular words read via the sublexical route should not be able to find any exact match in the phonological output lexicon. In this case, some correction mechanism is expected to change the word’s pronunciation to an existing word. However, why is it that we only see corrections to the correct irregular pronunciation of the target word and not for other phonologically neighboring words, which would cause other substitution/omission/addition/migration errors of consonants and vowels, contrary to fact. (e.g., a relevant example from English is the word “door”, when the regularized conversion of the word arrives in the phonological output buffer as /dur/, correcting it after failing to find it in the phonological output lexicon may lead to reading it as dear, poor, do, or tour, and not only as the correct door.

Additional information

Funding

This work was supported by the Anadolu Üniversitesi 1503E142; by the Liselotte Adler Lab for Child Development; by the Branco-Weiss Chair for Child Development and Education; and by the Cukier-Goldstein-Goren Center for Mind and Language.