Brain and Language 91 (2004) 7–8 www.elsevier.com/locate/b&l

Errors of lexical selection during high and low semantic competition Tatiana T. Schnur,a,b Adelyn Brecher,a Nick Rossi,a and Myrna F. Schwartza,* a

Moss Rehabilitation Research Institute, Philadelphia, PA 19141, USA b University of Pennsylvania, USA Available online 3 August 2004

In the ‘‘paced-cyclic’’ naming paradigm participants take longer to name a repeated series of pictures when the pictures are from the same semantic category vs. mixed categories, presumably due to semantically mediated competition for lexical selection (Damian, Vigliocco, & Levelt, 2001). Thompson-Schill’s selection hypothesis (ThompsonSchill et al., 1998) maintains that anterior brain damage compromises selection when semantic competition is high. In keeping with this, two anterior aphasics were shown to produce more errors in the homogeneous compared to mixed condition of paced-cyclic naming (McCarthy & Kartsounis, 2000; Wilshire & McCarthy, 2002). The present study replicated and extended this finding using a group study design. It also used error-type analysis to test the hypothesis that competition is the basis for higher error rates in the homogeneous contexts [see McCarthy and Kartsounis (2000) for alternative].

Results

Methods

Error type analysis

Participants

Semantic errors were characterized as any substitution semantically related to the target. Omissions included no-responses and descriptions. Other errors (formal, unrelated, neologisms, abstruse neologisms, and picture part names) were grouped together. We performed three separate ANOVAs on Semantic, Omission, and Other errors following the F1 analysis above. Fig. 1 shows the following pattern of results: More Semantic errors occurred in the homogeneous vs. mixed condition (F(1, 16) = 58.58, p < .0001) and this effect changed (increasing) over cycles (F(3, 48) = 5.20, p < .01). The non-fluent group

Participants were 18 individuals with lesions restricted to the left hemisphere. Mean (and range) for age was 57 years (35–80); education 14 years (10–20); and months post-onset 55 (10–175). Participants were divided into groups according to anterior vs. posterior aphasia presentation. Seven had non-fluent, agrammatic aphasia and 11 had variants of fluent aphasia. Groups were matched for: (1) accuracy on standard naming tests; (2) proportion of semantic errors in standard naming; and (3) lexical–semantic comprehension.

Error rate analysis We performed two ANOVAs on error rates treating subject (F1) and item (F2) means as separate random factors. We included condition (homogeneous/mixed), and cycle (1–4), as within-subject and within-item factors. Group (Non-Fluent/Fluent) was treated as a between-subject and within-item factor. More picture-naming errors were made when pictures were semantically homogeneous (30%) vs. mixed (27%; F1(1; 16) = 14.70, p < .01; F2(1; 71) = 22.13, p < .0001). The semantic interference effect was more pronounced for non-fluent aphasics (F1(1; 16) = 4.56, p < .05; F 2ð1; 71Þ = 6.87, p < .05) and changed (increasing) from cycles 1 to 4 (F1(3, 48) = 3.27, p < .05; F2(3, 213) = 4.93, p < .01).

Procedures Stimuli were 72 line drawings depicting 12 semantic categories, each with 6 exemplars. From these, 12 homogeneous (same-category) sets and 12 mixed-category sets were formed. In the experiment, each set was sequentially repeated four times, with a response deadline of 5 s. Response-stimulus interval was manipulated (1 and 5 s) but reported collapsed here.

*

Corresponding author. E-mail address: [email protected] (M. Schwartz). 0093-934X/$ - see front matter Ó 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.bandl.2004.06.007

Fig. 1. Difference in mean number of errors between homogeneous and mixed conditions for cycles 1–4 for 3 types of errors: Semantic, Omission, and Other.

8

Abstract / Brain and Language 91 (2004) 7–8

made more Semantic errors in the homogeneous vs. mixed condition compared to the fluent group (not shown in Fig. 1; F(1, 16) = 11.52, p < .01). Omissions, like Semantic errors, occurred more often in the homogeneous condition (F(1, 16) = 18.12, p < .001). Other errors by contrast, were more frequent in the mixed condition (F(1, 16) = 28.44, p < .0001); they also changed (declining) across cycles (not shown; F(3, 48) = 2.85, p < .05). No other effects were significant.

anatomical damage to the left inferior frontal gyrus (Thompson-Schill et al., 1998).

Acknowledgments This research was supported by NIH Grant #2 RO1 DC00191-22 (M. Schwartz) and NIH HD007425 (T. Schnur).

Discussion We showed here that aphasics as a group produced more errors in the homogeneous than mixed condition of paced-cyclic naming and that this effect was especially pronounced in those with an anterior presentation. The error-type analysis addressed the nature of the effect. Specifically, in support of the competition account, the homogeneous manipulation caused semantically related words to become activated and misselected, as demonstrated by the pattern of Semantic errors. Omissions also were more numerous in the homogeneous condition. That Omission and Semantic errors pattern together in this way suggests that both result from competitive activation, omissions representing occasions on which competition was not resolvable prior to the response deadline. This is an alternative to the ‘‘threshold model’’ of omissions (Dell, Lawler, Harris, & Gordon, 2004; Laine, Tikkala, & Juhola, 1998) which features sub-threshold activation as the cause of omissions. Having demonstrated that the homogeneous manipulation induces semantically-mediated competition for lexical selection, we conclude that the heightened effect for the non-fluent group is consistent with the theory that left anterior regions subserve selection during high competition (Thompson-Schill et al., 1998). To test the theory directly, we will examine whether this pattern of results correlates with

References Damian, M. F., Vigliocco, G., & Levelt, W. (2001). Effects of semantic context in the naming of pictures and words. Cognition, 81(3), B77– B86. Dell, G., Lawler, E. N., Harris, H. D., & Gordon, J. (2004). Models of errors of omission in aphasic naming. Cognitive Neuropsychology, 21(2/3/4), 125–145. Laine, M., Tikkala, A., & Juhola, M. (1998). Modelling anomia by the discrete two-stage word production architecture. Journal of Neurolinguistics, 11, 275–294. McCarthy, R. A., & Kartsounis, L. D. (2000). Wobbly words: Refractory anemia with preserved semantics. Neurocase, 6(6), 487–497. Thompson-Schill, S. L., Swick, D., Farah, M. J., D’Esposito, M., Kan, I. P., & Knight, R. T. (1998). Verb generation in patients with focal frontal lesions: A neuropsychological test of neuroimaging findings. Proceedings of the National Academy of Sciences of the United States of America, 95(26), 15855–15860. Wilshire, C. E., & McCarthy, R. A. (2002). Evidence for a contextsensitive word retrieval disorder in a case of nonfluent aphasia. Cognitive Neuropsychology, 19(2), 165–186.