Role of inflectional regularity and semantic transparency in reading morphologically complex words: Evidence from acquired dyslexia

Abstract

We report two patients with acquired phonological dyslexia who have great difficulty reading affixed words. Experiment 1 demonstrates that both patients' reading performance is influenced by the apparent morphological status of words by comparing the patients' reading of suffixed and pseudo-suffixed words. Experiment 2 was designed to examine reading performance of both regularly and irregularly inflected words. Experiment 3 examines the patients' reading of derivational forms with particular emphasis of the role of ‘semantic transparency’. Experiment 4 tested both patients' reading of prefixed words. Finally, Experiment 5 examined performance on a lexical decision task using affixed words. These data support models in which regularly formed inflections and semantically transparent derived forms are subjected to decomposition during processing, whereas irregularly inflected forms and semantically opaque forms may be represented independently. Data are discussed with regard to current ‘dual mechanism’ models of morphological processing as well as connectionist perspectives, with particular emphasis of the types of data that will ultimately be necessary to arbitrate between the rival theories.

Keywords:

• Morphology
• Cognitive neuropsychology
• Prefixes
• Semantic transparency
• Acquired dyslexia

Notes

¹All stimuli lists were formulated to control for as many psycholinguistic variables as possible, including word length (defined here as the number of letters in each word). However, regular inflections are often longer (given the addition of a suffix) than irregularly inflected words and the lists necessarily had small differences in letter length. The role of letter length in each patients' reading is fully addressed in the Discussion.

² Pilot testing in both patients demonstrated no difference as a function of exposure time within the range of 250–1000 ms. The percentage of stimuli that are affixed varied from one experiment to another; however, testing revealed that neither patients' performance differed when suffixed words were presented in blocks (that is, all words were suffixed) compared to when only 20% of the words were suffixed.

³Words from large orthographic neighborhoods are named more quickly than words from small orthographic neighborhoods (Sears, Hino, & Lupker, 1995). However, significant neighborhood effects are typically reported only for low frequency words and when large and small neighborhoods differ greatly in size. Given that the present stimuli are not low frequency and differ in neighborhood size by very little, we conclude that it is very unlikely that orthographic neighborhood size can account for the present data. However, future experiments should consider both neighborhood size and neighborhood frequency effects.