Disponible uniquement sur Etudier
  • Pages : 37 (9005 mots )
  • Téléchargement(s) : 0
  • Publié le : 22 novembre 2010
Lire le document complet
Aperçu du document
Available online at www.sciencedirect.com

Brain and Language 105 (2008) 185–198 www.elsevier.com/locate/b&l

Spoken-word processing in aphasia: Effects of item overlap and item repetition
Esther Janse
Utrecht Institute of Linguistics OTS, Utrecht University, Janskerkhof 13, 3512 BL Utrecht, The Netherlands Accepted 16 October 2007 Available online 19 November 2007

Abstract Two studieswere carried out to investigate the effects of presentation of primes showing partial (word-initial) or full overlap on processing of spoken target words. The first study investigated whether time compression would interfere with lexical processing so as to elicit aphasic-like performance in non-brain-damaged subjects. The second study was designed to compare effects of item overlap and itemrepetition in aphasic patients of different diagnostic types. Time compression did not interfere with lexical deactivation for the non-brain-damaged subjects. Furthermore, all aphasic patients showed immediate inhibition of co-activated candidates. These combined results show that deactivation is a fast process. Repetition effects, however, seem to arise only at the longer term in aphasic patients.Importantly, poor performance on diagnostic verbal STM tasks was shown to be related to lexical decision performance in both overlap and repetition conditions, which suggests a common underlying deficit. Ó 2007 Elsevier Inc. All rights reserved.
Keywords: Spoken-word processing; Aphasia; Inhibition; Lexical activation; Priming; Lexical deactivation; Repetition priming

1. Introduction During auditoryword recognition, lexical word candidates compete for recognition and influence each other’s activation levels. Evidence for lateral inhibition between competing word candidates comes from interference effects found with high-similarity word-initial form-overlap between a prime (e.g., difficult) and a following target (e.g., diffident; cf. Monsell & Hirsh, 1998; Slowiaczek & Hamburger, 1992). When theprime is being processed, several word candidates compete. In the TRACE (McClelland & Elman, 1986) model of auditory word recognition, there are direct inhibitory connections between words. An increase in the level of activation of one candidate then automatically leads to a decrease in the activation level of others. McClelland and Elman (1986) claim that this ‘winner-takes-all’ principle makes therecognition process more
Now also at Max Planck Institute for Psycholinguistics, Nijmegen, The Netherlands. Fax: +31 30 253 6406. E-mail address: esther.janse@let.uunl 0093-934X/$ - see front matter Ó 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.bandl.2007.10.002

efficient. Once one of the candidates has been isolated and recognised, the other candidates are decreased in activation.When one of these once-activated candidates is subsequently presented as the next item, recognition of this item is inhibited, relative to unrelated (control) targets. Monsell and Hirsh (1998) describe the competitor interference effect (i.e., slower lexical decision responses to diffident if preceded by difficult than if preceded by an unrelated control word) and the facilitatory effect of repetition(faster lexical decision responses to diffident if preceded by diffident than if preceded by an unrelated control word) as two sides of the same coin: recognition of a word makes it easier to recognise on the next encounter, but at the cost of making similar words (i.e, onset-overlapping words) more difficult to recognise. Whereas theories as TRACE describe the competition process as ‘‘lateral’’inhibition between several lexical candidates, others have argued for bottom-up inhibition: if there is a mismatch between the input form and the lexical entry’s form, activation of that candidate is lowered (Marslen-Wilson, Moss, & van Halen, 1996; Marslen-Wilson &


E. Janse / Brain and Language 105 (2008) 185–198

Warren, 1994). The extent to which activation is lowered may be gradient:...
tracking img