A patient with phonologic alexia can learn to read "much" from "mud pies"

Abstract

People with phonologic alexia often have difficulty reading functors and verbs, in addition to pseudowords. Friedman et al. [Friedman, R. B., Sample, D. M, & Lott, S. N. (2002). The role of level of representation in the use of paired associate learning for rehabilitation of alexia. Neuropsychologia, 40, 223-234] reported a successful treatment for phonologic alexia that paired problematic functors and verbs with easily read relays that were homophonous nouns (e.g. "be" paired with "bee"). The current study evaluates the efficacy of pairing problematic grammatical words with relays that share initial phonemes, but vary in the relationship of their final phonemes. Results showed that reading of target grammatical words improved to criterion level (90% accuracy over two consecutive probes) in all experimental conditions with shared phonology, but remained far below criterion level in control conditions. There was a significant correlation between degree of phonologic relatedness and error rate. Maintenance of the treatment effect was poor as assessed by traditional measurement, however a dramatic savings during relearning was demonstrated during a subsequent treatment phase. The finding that reading can be re-organized by pairing target words not only with homophones, but with other phonologically related relays, suggests that this approach could be applied to a wide corpus of words and, therefore, potentially be of great use clinically. We suggest, within a connectionist account, that the treatment effect results from relays priming the initial phonologic units of the targets.

Figures

Figure 1

Example of a training card. The written target word is presented in isolation on the front of the card (left); on the back of the card the target is underlined and paired with a relay in both its written and pictorial forms (right).

Figure 2

HN’s target word reading accuracy before, during and after training in each condition. For Experiment 1 and Experiment 2, A represents periods of no training; B represents periods of training. Note: After session #14 in the near multiple relay condition, HN became ill and, therefore, was able to attend only two sessions in a one month period. HN’s performance reached plateau after session #16, but training in this condition was continued because of the disruption in her treatment. B2 represents the period of training after her return.

Figure 3

HN’s target word reading accuracy immediately after each condition was trained. Criterion level performance was achieved in all four experimental conditions (black), but not in the control conditions (gray).

Figure 4

a–c. A general connectionist model of phonologic priming in the paired association treatment in the case of a) a random relay (no priming) b) a homophone relay (complete priming) and c) a single stretch relay (partial priming). Components are symbolized as follows: orthography (△); semantics (□); word phonology (); phonemes (○; shared); connections ( intact; weakened); oral reading response (); and insufficient activation is outlined in gray.