Effects of Phonological Neighborhood Density on Accuracy of Picture Naming in Aphasia

• ¹ University of Iowa, Communication Sciences and Disorders, United States

Introduction Phonological neighborhood density (PND) has been shown to impact the ease of word production. In speakers without aphasia, some researchers have found increased PND to be facilitative (e.g., Vitevitch, 2002), while others have found it to cause competition (e.g., Newman & German, 2005). As speakers age, the effects of PND may become more competitive due to weakened lexical connections (Gordon & Kurczek, 2014). In speakers with aphasia, increased PND appears to facilitate word production (e.g., Middleton & Schwartz, 2010). The aim of this project is to analyze the combined effects of age, PND, and aphasia on word retrieval. Methods Twenty-one native English-speaking participants with various types and severities of aphasia named 200 single-syllable drawings. Latency and accuracy of each naming response were recorded; speech errors were later transcribed and coded by at least two independent coders. We hypothesized that participants with weight lesions would show less facilitation due to PND, while those with decay lesions would show maintained facilitative effects. Results Using a median split of weight and decay parameters, participants were categorized into 3 groups: low-weight (LW, n=7), high-decay (HD, n=7), and a group that clustered around the median of both weight and decay (Med, n=7). A positive correlation was found between PND and average naming accuracy overall (r=0.18, p=0.01). However, subgroup correlations were significant only for the more impaired LW and HD participants (LW: r=0.18, p=0.01; HD: r=0.16, p=0.03, Med: r=0.11, p=0.13). Unexpectedly, reaction time (RT) was not significantly related to PND for the group as a whole (r=-0.06, p=0.4) or for any subgroup (LW: r=-0.02, p=.78; HD: r=-0.02, p=.82; Med: r=-0.11, p=.11). To test for speed-accuracy trade-offs, we correlated RTs with accuracy within each group. Both HD and Med groups showed strong negative correlations (rs>0.9, ps<0.01), indicating that as accuracy increased, RT slowed. However, the LW group (the slowest but most accurate) did not show this effect (r=-0.54, p=0.21), perhaps suggesting a different strategy in this group. Finally, we investigated the influence of age on RT by using the correlation between RT and PND as an indicator of PND facilitation, and relating the size of this correlation to age. Although there were too few participants to yield significance in any of the groups, different patterns were observed: both HD (r=0.69) and Med (r=0.15) groups showed positive correlations between age and PND facilitation, whereas the LW group (r=-0.09) showed a negligible but negative correlation. This suggests an interaction between age and connection weight on density effects (cf: Gordon & Kurczek, 2014). Discussion Preliminary correlational analyses indicated that individuals who are more impaired in either weight or decay demonstrate more naming benefit from increased PND than individuals with milder impairments, at least in terms of accuracy. Reaction times were not directly influenced by density, but showed different patterns in the different groups, partially influenced by age. Ongoing analyses will examine the influence of PND on patterns of error responses. These results suggest the importance of considering characteristics of both the anomia profile and the demographic profile of participants.