Morphological-compound dysgraphia in an aphasic patient: “A wild write through the lexicon”

Abstract

We describe the case of a dysgraphic aphasic individual—S.G.W.—who, in writing to dictation, produced high rates of formally related errors consisting of both lexical substitutions and what we call morphological-compound errors involving legal or illegal combinations of morphemes. These errors were produced in the context of a minimal number of semantic errors. We could exclude problems with phonological discrimination and phonological short-term memory. We also excluded rapid decay of lexical information and/or weak activation of word forms and letter representations since S.G.W.'s spelling showed no effect of delay and no consistent length effects, but, instead, paradoxical complexity effects with segmental, lexical, and morphological errors that were more complex than the target. The case of S.G.W. strongly resembles that of another dysgraphic individual reported in the literature—D.W.—suggesting that this pattern of errors can be replicated across patients. In particular, both patients show unusual errors resulting in the production of neologistic compounds (e.g., “bed button” in response to “bed”). These patterns can be explained if we accept two claims: (a) Brain damage can produce both a reduction and an increase in lexical activation; and (b) there are direct connections between phonological and orthographic lexical representations (a third spelling route). We suggest that both patients are suffering from a difficulty of lexical selection resulting from excessive activation of formally related lexical representations. This hypothesis is strongly supported by S.G.W.'s worse performance in spelling to dictation than in written naming, which shows that a phonological input, activating a cohort of formally related lexical representations, increases selection difficulties.

• Acquired dysgraphia
• Deep dysgraphia
• Formal lexical errors
• Third route
• Morphology
• Compounds.

We are indebted to S.G.W. and her family for their time and their support. We would also like to thank the speech therapists in Elzach for their support as well as Gerhard Blanken and Brenda Rapp for helpful comments at various stages of this research.

Notes

1 This model includes a single phonological and orthographic lexicon, a view compatible with recent evidence (e.g., Angelelli, Marinellli, & Zoccolotti, 2010). The conclusions of the study do not rest on the model incorporating one or two lexicons.

2 Monitoring accounts of formal errors have been developed in the context of models of spoken word production. It is, at present, unclear whether comparable internal monitoring mechanisms are active during written language production.

3 Nil responses were not excluded in the other assessments of length effects.

4 This experiment was suggested during the preparation of the manuscript and was carried out in December 2011. By that time, S.G.W.’s condition had improved but still, about 10% of her responses were formal errors (36% of all error responses).

5 A length effect in terms of word correct is predicted for all patients with deficits at the letter level. The more the letters in a word, the higher the probability that one will be produced incorrectly. Instead, only a problem in keeping letter active over time—or, in other words, a buffer impairment—predicts what we have called a “super-length effect”. This refers to a disproportionate increase in error rates per letter with increasing word length (see Romani et al., 2002).

6 This may appear as a strength of this proposal since all aphasic patients who have been shown to make high rates of formal errors in speaking also committed segmental errors (e.g., Best, 1995; Blanken, 1990, 1998; Goldrick et al., 2010; McCloskey et al., 2006). However, impairments at the lexical level also produce letter errors since weak activation of lexical representations and competition among lexical representations disrupt selection of the right set of letters (Graham, Patterson, & Hodges, 1997).