Abstracts

Rationale: Semantic dementia (SD) is a neurodegenerative disorder, in which a selective semantic deficit follows from atrophy centred on the anterior temporal lobe (ATL). SD patients also present surface dyslexia, characterized by a strong frequency-by-consistency interaction. In the epilepsy field, the language-dominant ventral part of the ATL (vATL) has been known as the basal temporal language area (BTLA). By means of a direct exploration of the BTLA, we aimed to clarify the role of BTLA in reading Kanji, which is a semi-opaque orthography that consequently requires a greater contribution from the semantic system for accurate reading (Fushimi et al., 1999). Methods: Subjects were 4 patients with intractable temporal lobe epilepsy who underwent presurgical evaluation with subdural electrode implantation over language-dominant vATL (IRB C533). We used 120 kanji words consisting of two characters from Fushimi et al. (1999), and classified them into three types (40 for each): consistent (C: associated with only one pronunciation), inconsistent typical (IT: two or more possible ways of pronunciation but the correct pronunciation is the most common), and inconsistent atypical (IA: the same as IT except the correct pronunciation is in the minority). 56 Kana non-words with 2 to 5 mora were also employed from the SALA battery (Japanese version of PALPA). The patients completed the Kanji word (Patient 1-4) and Kana non-word (Patient 2-4) reading tasks. In half of the trials, electrical stimulation (50 Hz, 0.3 ms duration, 5 mA, 3 sec) was delivered to the core BTLA defined by presurgical evaluation, time-locked to the presentation of each word or non-word. The reaction time (RT) was evaluated. The data were analyzed with a 3-way ANOVA [subjects, Kanji type (C, IT, IA) or Kana-mora (2&3 vs. 4&5 mora), stimulation (real vs. sham)] separately for Kanji word and Kana non-word reading. Results: For Kanji word reading, all of the main effects were significant (p < 0.05). Significant interactions were observed between Kanji-type and stimulation (F = 5.085, p = 0.007). Post-hoc analysis showed a significant Kanji-type x stimulation interaction between C and IA (F = 11.333, p = 0.001). For Kana non-words reading, the main effect was not significant for stimulation (F = 2.466, p = 0.119). No significant Kana-mora x stimulation interaction was observed (F = 3.268, p = 0.073). Conclusions: Direct stimulation of BTLA prolonged RT in reading Kanji words, but not in Kana non-words. The significant Kanji-type x stimulation interaction (i.e., longer RT in IA words) implies that the language-dominant vATL is not only a crucial area for semantic processing but also for the contribution of semantics to language processing. Consistent with existing evidence for an important role of this region in language processing, these results suggest that it is clinically important to preserve BTLA for Kanji processing in epilepsy surgery. Funding: KAKENHI 26282218, 15H01664, 15H05874, 16K19510