Processing of denotative and connotative alternative meaning of words after left- and right-hemispheric lesion U. Nocentini a,b , K. Kahlaoui a,* , P. Roberts c , K. Eck a , F. Giroux a , P. Goulet a,d , Y. Joanette a a Centre de recherche, Institut universitaire de ge ´riatrie de Montre ´al, Canada b Fondazione Santa Lucia, Roma, Italia c De ´partement d’orthophonie, Universite ´ d’Ottawa, Canada d Centre de re ´adaptation Lucie-Bruneau, Montre ´al, Canada Accepted 6 July 2006 Introduction Both left and right hemispheres play important—though presumably complementary—roles in everyday verbal communication. Evidence derived from both neurologically intact participants and clinical popula- tions suggest that the right hemisphere not only contributes to the acti- vation of peripheral aspects of word meanings but is also necessary for the processing of figurative language. Thus, (Brownell, Potter, Michelow, & Gardner, 1984) found differences in the sensitivity of the two hemi- spheres to semantic relationships in left-hemisphere damaged (LHD) and right-hemisphere damaged (RHD) participants. LHD participants showed impairment in identifying denotative relationships of words whereas participants with RHD had difficulty to process connotative relationships. Their study suggests a double dissociation between the lex- ico-semantic abilities of LHD and RHD participants. However, the interpretations of these data and their replicability have since been ques- tioned (Nocentini, Goulet, Roberts, & Joanette, 2001). The purpose of this study was to revisit the contribution of the left and the right hemi- spheres to the processing of the denotative and connotative meanings of words. In order to achieve this goal, we contrasted RHD and LHD indi- viduals’ performances while processing the denotative and connotative meanings of isolated words. Since the Brownell et al. (1984) study used only adjective words as stimuli, we implemented two versions of this experiment using adjectives (Experiment 1) and nouns (Experiment 2) as stimuli in order to control for the possible impact of the grammatical class. Experiment 1 Methods Participants Three groups of participants were tested: a group of LHD participants, a group of RHD participants, and a group of neurologically intact partic- ipants. All of them were right-handed native speakers of English. Exclu- sionary criteria for brain-damaged participants included multiple infarcts, history of drug or alcohol abuse and history of psychiatric and neurological illness. Control participants were matched as closely as pos- sible to the brain-damaged groups for age, gender and education. Exclu- sion criteria included a history of neurological and/or psychiatric disorders. Stimuli The stimulus set consisted of 8 adjective words (e.g., cold, deep, loving, shallow, warm, hateful, wise, and foolish) that were combined into 28 pos- sible pairs. Pairs were categorized according to the nature of the semantic similarity shared by two members of the triad. There were six kinds of relations: antonym, domain, metaphor, polarity, domain/polarity, and no relation pairs. Procedure Participants were asked to express judgements of semantic similarity between two members of each triad. Each of the triad was printed in a single card and the order of the three words was random. Participants were instructed to pick two stimuli that were the most similar in meaning or that best go together. Thus, in completing such a task, the participant had to ignore some components of meaning and focus on some others. Results Both analyses of variance (ANOVAs) and hierarchical cluster analyses were realized on similarity percentages. Similarity percentages were based doi:10.1016/j.bandl.2006.06.059 * Corresponding author. Fax: +1 514 340 3548. E-mail address: karima.kahlaoui@umontreal.ca (K. Kahlaoui). www.elsevier.com/locate/b&l Brain and Language 99 (2006) 8–219