Concreteness effects on semantic processing of single Chinese characters using mixed effects modeling of EEG data

Sampo LAW, Yen Na Cherry YUM, Wing-Lam G. CHEUNG

Research output: Contribution to conferencePoster


Compared with abstract words, concrete words are responded to more quickly and accurately in various lexical processing tasks. They have also been found to elicit a more negative-going wave in the N400 time window with stronger effects in the frontal region, often followed by a sustained frontal negativity up to 1000ms post-stimulus onset. Holcomb et al. (1999) accounted for the effect by integrating the context availability (Schwanenflugel, 1991) and dual coding theories (Paivio, 1986, 1991). The N400 reflects activities of a stronger and denser semantic network of concrete words, followed by their activation of visual imagery. Previous studies have mainly focused on nouns, and results of the effect on verbs have been equivocal. This situation is also found in two Chinese studies in which written compound words were examined (Tsai et al., 2009; Zhang et al., 2006). The discrepant findings could be due to confounding from sublexical processes (Zhang et al., 2006). Although Tsai et al. (2009) found concreteness effects on both nouns and verbs, the observation was based on two specific subtypes of compounds. The current study examined the ERP correlates of concreteness effects on semantic processing of single Chinese characters of different form classes (POS), including nouns, verbs and adjectives.

A non-factorial paradigm was adopted in which 434 characters varying in ratings of concreteness, number of strokes, and frequency (Liu et al., 2007) were presented. To enhance ecological validity, the selection of stimuli was guided by the types of characters with respect to orthography-to-phonology mapping that a reader would encounter according to the proportion reported in Shu et al. (2003). Twenty-two native Mandarin speakers and readers of the simplified Chinese script participated in a semantic categorization task in which they made a response if the stimulus belonged to a specified semantic category, i.e. animal, body part, or family member. Linear mixed-effects modeling (LMEM) was applied to estimate statistical effects of fixed factors and random factors of subject and item. LMEM is superior to traditional analyses in dealing with unbalanced data, missing values, and non-sphericity, common problems in ERP studies.

Results and Discussion
Significant effects of POS, concreteness, and interactions were only found in the posterior regions (left, midline, and right) during the N400 and the right-posterior region during 500-1000ms (see Table 1). Greater N400 was associated with more concrete than less concrete nouns, whereas less negative N400 was found for concrete verbs relative to abstract verbs, resulting in an interaction between POS and concreteness. Concreteness continued to modulate neural response to verbs in the same pattern as in the N400 after 500ms. The effect and its interaction with POS were restricted to the right posterior region. The opposite directions of the effect on nouns and verbs can be explained by task demand, where abstract verbs generated greater conflicts with concrete semantic categories denoted by nouns. The late concreteness effect in the right hemisphere is taken to indicate activities of imagery processes linked to concrete items (Dhond et al., 2007; Huang et al., 2010). Copyright © 2016 54th Annual Academy of Aphasia Meeting.
Original languageEnglish
Publication statusPublished - Oct 2016
Event54th Annual Academy of Aphasia Meeting - Llandudno, United Kingdom
Duration: 16 Oct 201618 Oct 2016


Conference54th Annual Academy of Aphasia Meeting
Country/TerritoryUnited Kingdom


Law S., Yum Y.N., & Cheung G.W.-L. (2016, October). Concreteness effects on semantic processing of single Chinese characters using mixed effects modeling of EEG data. Poster presented at 54th Annual Academy of Aphasia Mereting, .


Dive into the research topics of 'Concreteness effects on semantic processing of single Chinese characters using mixed effects modeling of EEG data'. Together they form a unique fingerprint.