Submission 240
When Negation Fails: Diagnosing Semantic vs. World Knowledge Violations
SymposiumTalk-02
Presented by: Frauke Buscher-Schiewe
A longstanding challenge across linguistics, philosophy, and cognitive psychology is to delineate the boundary between semantic anomaly (e.g., Vaccinations are square) and conflicts with general world knowledge (e.g., CDs are square). Both yield deviant statements, but only the former reflects a failure of linguistic composition rather than a clash with experience. Our focus lies on selectional restrictions as instances of semantic knowledge that define the range of conceptually compatible arguments a predicate can combine with. We present a novel approach that examines how such restrictions interact with negation, from which we derive two reliable diagnostics for distinguishing between semantic and world knowledge violations: the FALSUM Test, extending Repp’s (2013) analysis of denial (illocutionary negation), and the Wait-a-Minute! (WAM) Test, building on von Fintel (2004). Our central claim is that only world knowledge violations, not semantic ones, can be coherently rejected by denial or standard WAM responses. We present two judgment studies supporting this view. In Study I (Sentential Negation), only world knowledge violations were rendered acceptable under negation, suggesting that negation presupposes semantic well-formedness. In Study II (WAM), world knowledge violations were rejected by standard WAM, whereas semantic violations required a stronger WAM version. Together, these findings provide converging empirical evidence for two diagnostics that remain robust even for difficult or disputed cases. They advance a linguistically grounded account of how semantic well-formedness constrains the processing of negation and the rejection of claims in discourse, and they represent a promising step toward delineating semantic anomaly.