Poster E1, Saturday, August 18, 3:00 – 4:45 pm, Room 2000AB
Semantic diversity affects word processing similarly for PWA with and without semantic deficits: Evidence against the semantic control hypothesis
Curtiss Chapman1, Randi Martin1;1Rice University
Introduction. Although a benefit for high frequency (HF) words over low frequency (LF) words is ubiquitous in language processing, persons with aphasia (PWA) with multimodal semantic deficits sometimes show reverse word frequency effects (WFEs; e.g., in synonymy judgments; Almaghyuli et al., 2012). Hoffman and colleagues suggest that this reversal results from the greater semantic diversity (SemD; Hoffman, et al., 2011) of HF words – i.e., a greater range of associated semantic representations which become activated during word processing and compete with selection of the target word meaning. Difficulty selecting among competing semantic representations is argued to be part of these individuals’ underlying semantic deficit—one of controlled access to semantics (Hoffman, et al., 2011). Our study tested two predictions following from these claims: 1) smaller WFEs should be observed for high than low SemD words, 2) this interaction of frequency and SemD should be greater—with even smaller WFEs for high SemD words—for PWA with semantic deficits than those without, due to their greater sensitivity to competition. Method and Results. Data from 117 PWA from the Moss Aphasia Psycholinguistics Project Database were analyzed, including 81 who had semantic deficits and 36 who did not have multimodal semantic impairments (> 2 sd below control means on verbal and nonverbal semantic tasks). Errors on the Philadelphia Naming Test (PNT) and the Philadelphia Repetition Test (PRT) were regressed on group, SemD, frequency, and their interactions. In PNT and PRT, we observed an effect of group (semantic deficit < non-semantic deficit; p’s<.001), a beneficial effect of HF (p’s<.001), and no SemD effect (p’s>.23). For PNT, we observed the expected interaction of SemD and frequency (p=.001), with larger WFEs for low than high SemD words, but no interactions with group were significant (p’s>.59). Interestingly, the SemD x frequency interaction showed a negative effect of SemD for HF words but a positive effect of SemD for LF words. For PRT, we observed a group x frequency interaction (p=.01), but surprisingly PWA with semantic deficits showed a stronger WFE than did those without. There was a marginal interaction between group, frequency, and SemD (p=.06), such that the semantic deficit group showed an interaction between frequency and SemD like that in picture naming (p<.001), whereas the non-semantic deficit group did not (p=.16). However, further analysis revealed no group difference in WFEs for high SemD words, but larger WFEs for low SemD words in the semantic deficit group. Conclusions. As predicted, WFEs were diminished for high SemD words. However, as PWA with and without semantic deficits showed this interaction in picture naming and only WFEs for low SemD words were stronger for those with semantic deficits in repetition, it is unlikely that this interaction arose from semantic control difficulties in those with semantic deficits. Interestingly, the interaction showed that while high SemD hurt performance for HF words, it improved performance for LF words – perhaps because weak activation of associates for LF words aids in their comprehension whereas strong activation of associates for HF words creates competition.
Topic Area: Control, Selection, and Executive Processes