This study explores how negation affects long-distance (i.e., embedded) wh interrogative arguments in French. We begin with the observation by Baunaz & Puskás (2008) that wh in-situ constructions are grammatical with clausemate negation but significantly degraded with matrix negation. To investigate this phenomenon, we report on two experimental studies that compare long-distance wh-questions across various classes of interrogative-selecting matrix verbs, including communication, (semi-)factive, perception, and reasoning.

Our findings indicate that wh in-situ constructions are at least as acceptable as their ex-situ counterparts across all configurations tested. This challenges the common claim that negation only interferes with covert wh movement. We delineate a clear hierarchy of acceptability based on the position of negation: positive clauses (no negation) are rated higher than those with embedded negation (Neg2), which in turn are rated higher than those with matrix negation (Neg1). While this pattern consistently applies to communication and (semi-)factive verbs, an exception arises with perception verbs, where embedded negation leads to a significant acceptability decline similar to that caused by matrix negation. This finding aligns with Baunaz and Puskás’ (2008) assertion that the semantics of the matrix predicate influences the effects of negation. Finally, we propose a new perspective for understanding how negation selectively impacts wh extraction.