Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Seeking Visual Word Form Area in EEG: A study in three languages and four orthographies
Poster Session D, Saturday, October 26, 10:30 am - 12:00 pm, Great Hall 3 and 4
Hareem Khokhar1,2, Jill McLendon1, Donald Dunagan1, Zahin Hoque1, Dustin Chacón1,2; 1University of Georgia, 2University of California Santa Cruz
[INTRODUCTION] Across writing systems, the left fusiform gyrus ('visual word form area', VWFA) is a key brain area in reading (Dehaene & Cohen 2011). VWFA-localizer tasks are used in studies on word recognition and morphological processing (Tarkiainen et al. 1999; Gwilliams et al. 2016). Can we develop a VWFA-localizer using a typical EEG lab? We examine four language-orthography pairs: English (Roman), Mandarin Chinese (MC) (Hanzi), Urdu (Naskh Arabic script), and Urdu (Nasta`liq Arabic script). Two variants of Urdu were included because they differ in degree of letter overlap and letter segmentation cues (Naz et al. 2013), which affect word-form recognition. In English, Urdu (Naskh), Urdu (Nasta`liq), we identified a cluster on the left ventral surface for the 'Type II Word Effect', although we do not identify a VWFA functional ROI in MC. [METHODS] Participants (N = 34 English; N = 34 MC; N = 28 Urdu) passively viewed words and symbols for 60ms followed by 240ms of blank screen. Brain activity was continuously recorded with a 64-channel EEG. Stimuli were 50 short frequent words (4 letters in English and Urdu; 2 in MC) and length-matched symbols; stimuli were embedded in two levels of noise (Low, High). Within each comparison, we controlled psychophysical factors luminance, height, visual angle subtended, and perimetric complexity (Pelli et al. 2006) (ps > 0.1) [RESULTS] [EEG Sensor Data]. Raw EEG data was band-pass filtered from 0.1–40Hz, then epoched from –100 to 300ms, with baseline correction to –100-0ms. Artifacts were excluded using ICA, and epochs exceeding 100μV peak-to-peak threshold were excluded. Spatio-temporal cluster-based permutation tests (Maris & Oostenveld 2007) were conducted using Noise (Low, High) × Stimulus (Word, Symbol) ANOVAs. Urdu data were analyzed separately for Naskh and Nasta`liq. In all four comparisons, Low vs. High Noise ERPs differed ~100–150ms, and one cluster was identified between ~200–300ms that distinguished Words vs. Symbols (ps < 0.05). [Source Estimates]. EEG data were coregistered with fsaverage template brain. We used sLORETA to estimate source activity (Pascual-Marquis 2002). We used sensor-space clusters as temporal localizers for one-way ANOVAs comparing Word vs. Symbols and High vs. Low Noise. Clustering was constrained to bilateral occipito-temporal regions. In all four comparisons, bilateral occipital activity was greater for Low vs. High Noise stimuli ~100–150ms. In English, a posterior left fusiform cluster showed greater activity for Symbols vs. Words; in Urdu (Nasta`liq), an anterior left fusiform cluster showed greater activation for Words vs. Symbols, both ~200–300ms (ps < 0.05). In Urdu (Naskh), a similar, insignificant cluster was identified as in Nasta`liq (p=0.16). No clusters were identified in occipito-temporal regions in MC.[CONCLUSION] We presented a VWFA localizer task using a typical EEG set-up that can identify portions of left fusiform gyrus involved in early stages of reading in English and Urdu. Unlike previous MEG findings, English VWFA activity was greater for symbols than words; whereas the typical increased activity for words was demonstrated for Urdu. No functional VWFA was identified for MC, which may echo previous findings (e.g. Bolger et al. 2005).
Topic Areas: Reading, Methods