Projects per year
It is increasingly clear that when trying to say a word, activation of words related in meaning can interfere with production of that word. Evidence supporting this position is available from priming and word picture interference tasks with both unimpaired speakers and people with aphasia (e.g., Howard, Nickels, Coltheart, & Cole-Virtue, 2006; Schnur, Schwartz, Brecher, & Hodgson, 2006; Schriefers, Meyer, & Levelt, 1990). However, it is less clear the extent to which semantically related words influence word production without previous presentation of this ‘neighbour’. Different effects of semantic neighbourhood are reported in the literature, and disentangling these effects is complicated by the use of different measures of semantic neighbourhood (e.g., Bormann, 2011; Kittredge, Dell, & Schwartz, 2007; Mirman, 2011). This study had three parts:Experiment 1 investigated the interrelationships between the six measures of semantic neighbourhood that have been used in previous studies: rated semantic neighbours (e.g. Bormann, 2011); raw contextual neighbours, categorical contextual neighbours (e.g. Kittredge et al., 2007); near feature neighbours, distant feature neighbours (e.g., Mirman, 2011); association neighbours. Using principal component analysis, four independent factors were derived that related to four types of neighbour: Featural, associative, contextual and distant semantic neighbours.Experiment 2 used these factors to determine the effects of semantic neighbourhood on the picture naming of 86 target pictures for whom all variables were available by 50 unimpaired speakers (data from Székely et al., 2003). The semantic neighbourhood factor scores determined in Experiment 1 were used, together with ‘lexical’, ‘sublexical’ and visual complexity factor scores in regressions predicting either RT or accuracy. While the overall model was significant in predicting RT (R2=.276, F= 4.257, p<.01), and both lexical and sublexical factors were significant, no semantic neighbourhood factor showed a significant effect on naming RT. The overall model was not significant for accuracy (R2=.086, F=1.044, p>.05).Finally, Experiment 3 analysed the effects of semantic neighbourhood on picture naming of the same items from Experiment 2 for a large group of individuals with aphasia (n=193) from the Moss Aphasia Psycholinguistic Project Database (MAPPD; Mirman et al., 2010). Using Generalised Mixed Effects models for binomial outcomes, we examined the effects of the same 7 factors on accuracy and error type and, for a reduced group for whom degree of semantic and phonological production impairment could be established (n=140), the interactions of these factors with impairment. For both groups, the feature-based neighbourhood factor predicted accuracy but did not interact with degree of semantic or phonological impairment. Nevertheless, the final model with per participant random slopes for feature-based factor was significantly better than the same model without random slopes (Χ2(3)=8.15, p=0.04), showing that the effect differed significantly across participants. In terms of errors, words with many feature-based neighbours were more likely to be correctly named than to result in omissions and phonological errors, and were more likely to result in a semantic error than to be omitted. The full theoretical implications of these results will be discussed but appear to be best accommodated within interactive theories of word production.
Bibliographical noteCopyright the Author(s) 2019. Version archived for private and non-commercial use with the permission of the author/s and according to publisher conditions. For further rights please contact the publisher.
- Semantic Neighbours
- spoken word production
- lexical access
- Semantic neighborhood density