Frontiers in Psychology (Apr 2014)
What factors predict individual subjects' re-learning of words during anomia treatment?
Abstract
A growing number of studies are addressing methodological approaches to treating anomia in persons with aphasia. What is missing from these studies are validated procedures for determining which words have the greatest potential for recovery. The current study evaluates the usefulness of several word-specific variables and one subject-specific measure in predicting success in re-learning problematic words. Methods: Two participants, YPR and ODH, presented with fluent aphasia and marked anomia. YPR’s Aphasia Quotient on the Western Aphasia Battery was 58.8; ODH’s AQ was 79.5. Stimuli were 96 pictures chosen individually for each participant from among those that they named incorrectly on multiple baselines. Subsequently, participants were presented with each picture and asked to indicate whether they could name it covertly, or “in their head.” Each subject completed a biweekly anomia treatment for these pictures. We performed separate statistical analyses for each subject. Dependent variables included whether each word was learned during treatment (Acquisition) and the number of sessions required to learn each word (#Sessions). We used logistic regression models to evaluate the association of (self-reported) covert naming success with Acquisition, and linear regression models to assess the relationship between (self-reported) covert naming success and #Sessions. Starting with the predictors of covert naming accuracy, number of syllables (#syllables), number of phonemes (#phonemes), and frequency, we used backwards elimination methods to select the final regression models. Results: By the end of 25 treatment sessions, YPR had learned 90.2% (37/41) of the covertly correct words but only 70.4% (38/54) of the covertly incorrect words. In the unadjusted analysis, covert naming was significantly associated with Acquisition, OR=3.89, 95% CI: (1.19, 12.74), p=0.025. The result remained significant after adjustment for #phonemes (the only other predictor included in the final model for Acquisition), OR=3.54, 95% CI: (1.02, 12.35), p=0.047. During 19 treatment sessions, ODH learned 97.8% (44/45) of the covertly correct words and 93.6% (44/47) of the covertly incorrect words. Since most of the words were learned, logistic regression analyses for Acquisition were not performed. In the unadjusted analysis covert naming was significantly related to #Sessions (p<0.001). The result remained significant (p<0.001) after adjustment for #syllables (the only other predictor included in the final model for #Sessions). Specifically, the words that ODH could name covertly were learned on average 2.28 sessions earlier than the other words, 95% CI: (0.98, 3.58). For YPR covert naming and #phonemes were associated with Acquisition, while for ODH covert naming and #syllables were related to #Sessions. Interestingly, word frequency was not identified as an important predictor for any outcome for either subject. It is important to note that for the 35 words common to the two subjects’ word lists, there was very little agreement (kappa=0.11) between the subjects’ self-reports of covert naming success. These result suggests that the subject-dependent property of a word, i.e. the subjective experience of correct covert naming, is a better indicator of the strength of that word’s representation in the person’s lexicon than word-specific properties such as number of phonemes, number of syllables, and word frequency.
Keywords
