A frame-semantic approach to English communication verbs

Abstract

This paper presents a unified Frame Semantics account of say, tell, speak, and talk. All four verbs instantiate the [Communication] frame but diverge in (i) foregrounded Frame Elements (FEs), (ii) syntactic realizations, and (iii) discourse functions. Corpus evidence with FrameNet-style FE tagging shows: say foregrounds Message (SVO/SVO +that); tell foregrounds Addressee with a caused-possession profile (SVOO, NP+that, NP+to-inf); speak highlights Mode/Language and verbal performance (intransitive + PP: to/with/about); talk foregrounds Interaction/Topic (intransitive + PP). These contrasts form a functional continuum ranging from content-centered through recipient-centered and mode/formality-centered to interaction-centered communication. The analysis links FE foregrounding to argument structure and discourse roles. Pedagogically, it explains recurrent learner errors and informs instruction on register-sensitive verb choice. Computationally, it benefits reporting-verb detection, semantic role labeling, and indirect-speech identification. Limitations include potential subjectivity in FE boundaries and limited genre coverage; future work will model register effects and test cross-linguistic correspondences. Overall, the findings support Frame Semantics as a robust lens connecting lexical meaning, syntax, and discourse in communicative verb classes.