Abstract
This study examines the phenomenon of polarity-reversed ellipsis in Japanese, focusing on cases where the anaphoric expression soo (“so”) refers to a negated proposition implied by a preceding statement. While earlier research, notably Yagi, Sakamoto, and Tatsumi (2022),has analyzed polarity reversal primarily in connection with syntactic Neg-raising (NR) predicates such as omou (“think”), this paper demonstrates that polarity-reversed interpretations extend beyond NR predicates to include non-NR predicates such as kangaeru(“consider”)and handan-suru (“judge”). Drawing on two large-scale acceptability judgment surveys with a total of 466 participants (students and faculty),the analysis shows that polarity-reversed soo is systematically licensed in contexts where the negated predicate functions as a denial,activating a discourse representation of ¬P (not-P) shared among interlocutors. In addition, chi-square tests conducted on acceptability judgment data from the two surveys (N = 466) revealed statistically significant differences in several conditions, particularly with respect to response type in dialogic contexts, while also indicating that such effects are not uniform across all structures. These findings indicate that the licensing of polarity-reversed ellipsis relies not solely on syntactic identity constraints but also on pragmatic inference and discourse-level factors. By integrating empirical evidence and cognitive-pragmatic analysis, this study deepens our understanding of the interaction between negation, anaphora, and discourse structure in Japanese. It also highlights the need for models of ellipsis resolution that incorporate pragmatic and interactional considerations, offering insights for theoretical linguistics and Japanese language education.
https://ror.org/01n09qc98 
