Neural substrates of semantic relationships: Common and distinct left-frontal activities for generation of synonyms vs. antonyms Hyeon-Ae Jeon a , Kyoung-Min Lee b, ⁎, Young-Bo Kim a,c , Zang-Hee Cho a, ⁎ a Neuroscience Research Institute, Gachon University of Medicine and Science,1198 Kuwol-dong, Namdong-gu, Incheon, 405-760, Republic of Korea b Department of Neurology, Seoul National University Hospital, 28 Yeongeon-Dong, Jongno-Gu, Seoul,110-744, Republic of Korea c Department of Neurosurgery, Gachon University of Medicine and Science,1198 Kuwol-dong, Namdong-gu, Incheon, 405-760, Republic of Korea abstract article info Article history: Received 14 March 2009 Revised 11 June 2009 Accepted 18 June 2009 Available online 25 June 2009 Synonymous and antonymous relationships among words may reflect the organization and/or processing in the mental lexicon and its implementation in the brain. In this study, functional magnetic resonance imaging (fMRI) is employed to compare brain activities during generation of synonyms (SYN) and antonyms (ANT) prompted by the same words. Both SYN and ANT, when compared with reading nonwords (NW), activated a region in the left middle frontal gyrus (BA 46). Neighboring this region, there was a dissociation observed in that the ANT activation extended more anteriorly and laterally to the SYN activation. The activations in the left middle frontal gyrus may be related to mental processes that are shared in the SYN and ANT generations, such as engaging semantically related parts of mental lexicon for the word search, whereas the distinct activations unique for either SYN or ANT generation may reflect the additional component of antonym retrieval, namely, reversing the polarity of semantic relationship in one crucial dimension. These findings suggest that specific components in the semantic processing, such as the polarity reversal for antonym generation and the similarity assessment for synonyms, are separately and systematically laid out in the left-frontal cortex. © 2009 Elsevier Inc. All rights reserved. Introduction It is commonly presumed that semantic relations between words – antonymy, synonymy, hyponymy and so on – are reflected in the organization of word storage in the brain (Katz, 1972; Kempson, 1977; Pustejovsky, 1995), but the neural mechanism of such organization, or of word storage in general, still remains to be elucidated. The question we address here is what neural substrates are involved in the two different semantic relations, synonymy and antonymy. The reason we focused on the two semantic relations is the significance of their relations in the structure of mental lexicon. We agree with Jackson and Amvela (2000) who stated that “the two most obvious sense relations are those of ‘sameness’ and ‘oppositeness’, called synonymy and antonymy respectively.” Moreover, as quoted in Table 1 , Hutchison (2003) reported that antonyms and synonyms combined comprise about 40% of responses in association norms, which means, the two relations take up much space in the mental lexicon. Miller and Fellbaum (1991) provided a good definition of synonyms: “two expressions are synonymous if the substitution of one for the other never changes the truth value of a sentence in which the substitution is made.” A weakened version of this definition was also widely accepted: if two expressions are synonymous in a context, even though one substitutes the other, it does not change the truth value of the context. As for an antonym, a lexical circumstance bound with oppositeness is generally accepted (Murphy and Andrew, 1993). Both synonyms and antonyms differ in certain features, yet share some common features. What differentiates semantic relations between words, such as synonymy and antonymy, is the composition of these features (Fillenbaum and Rapoport, 1971). According to this, features that play a major semantic role differ between antonyms, but features that differ between synonyms assume only a minor semantic role. Synonyms are words that are similar in the ‘central’ aspects, and different in ‘minor’ or ‘peripheral’ ones (Cruse, 1986). On the other hand, an antonym for a word is similar in many conceptual dimensions, yet different from the word in one semantically critical dimension. This opposing dimension constitutes the core in the antonymic relationship between two words (Kadesh et al., 1976). For example, fast and slow, they are both features of motion but differ only in reference to plus-speed and minus-speed. open/close and social/ antisocial are also good examples; the former is action verb indicating similar behavior with difference only in the direction they describe and the latter is used for describing the same dimension of personality but with opposite value (Murphy and Andrew, 1993). In summary, both synonyms and antonyms can be generated based on not only similar features but also different features. These different features play a major role in antonyms whereas a minor role in synonyms. NeuroImage 48 (2009) 449–457 ⁎ Corresponding authors. K.-M. Lee is to be contacted at the Department of Neurology, Seoul National University Hospital, 28 Yeongeon-Dong, Jongno-Gu, Seoul, 110-744, Republic of Korea. Fax: +82 2 3672 7553. Z.-H. Cho, Neuroscience Research Institute, Gachon University of Medicine and Science,1198 Kuwol-dong, Namdong-gu, Incheon, 405-760, Republic of Korea. Fax: +82 32 460 8230. E-mail addresses: kminlee@snu.ac.kr (K.-M. Lee), zcho@gachon.ac.kr (Z.-H. Cho). 1053-8119/$ – see front matter © 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.neuroimage.2009.06.049 Contents lists available at ScienceDirect NeuroImage journal homepage: www.elsevier.com/locate/ynimg