Obligatory speaker-oriented rhetoricality: The case of Cantonese 'me'↓-utterances

Abstract

Rhetorical utterances employ the same sentence types as ordinary questions and assertions, yet differ systematically in how they engage with discourse structure and speaker commitment. Prevailing accounts analyze rhetoricality as Common-Ground–oriented, deriving rhetorical interpretations when the Issue raised by an utterance is already closed in the shared discourse context. This paper proposes a refinement of that view, arguing that rhetoricality is fundamentally speaker-oriented: an utterance is rhetorical when it raises a proposition that conflicts with the speaker’s own public discourse commitments. I develop this proposal through a case study of Cantonese sentence-final particles, focusing on me1 and its high-falling tonal variant me↓. While me1 signals epistemic uncertainty by raising a proposition without committing the speaker to it, me↓ obligatorily yields a rhetorical interpretation in which the speaker rejects the prejacent despite contextual evidence that might support it. I argue that me↓ is compositionally derived from me1 and an additional low boundary tone (L%), where me1 suspends speaker commitment and L% introduces a requirement of incompatibility between the prejacent and the speaker’s discourse commitment in the input context. Their interaction produces a stable configuration in which an Issue is raised while remaining incompatible with the speaker’s commitments, thereby encoding rhetoricality in the form itself.