This study introduces new data on the Japanese truth-based answering system. I use such new data to compare the empirical coverage of two existing approaches in the literature. A scopal one from Krifka (2013) and a featural one from Servidio et al. (2018). Although their discussion focuses on polarity-based languages, they both suggest that their analysis can be extended to truth-based languages. I show that the former approach, which relies on the scope relation between NegP and TP to capture crosslinguistic variation, fails to account for the fact that Japanese maintains the truth-based system for questions with different negation scope. I further show that a featural approach, which does not refer to structure below the CP-level, can be straightforwardly extended to the novel Japanese data. I further show that a featural analysis fares better with answers to high-negation questions in Japanese and English.