Non-local interpretations of expressive adjectives
Expressives such as "damn" convey a negative speaker attitude, but the target of that attitude is often not the expression’s syntactic sister. In "The dog ate the damn cake", the speaker may be understood as upset at the dog, the cake, or the event as a whole. Pragmatic accounts attribute this flexibility to inference, whereas syntactic accounts derive it from structural constraints. This talk examines whether the negative emotional valence of potential targets modulates this choice, and whether the effect is categorical or gradient. It also asks whether expressive emojis (The dog 😡 lies on the sofa) show the same degree of locality as expressive adjectives.
Expressives such as damn convey a negative speaker attitude whose target is not fixed by the sentence’s structure. In "The dog ate the damn cake", the attitude can be construed with respect to the whole proposition (the speaker is upset that the dog ate the cake), the subject referent (upset at the dog), or the object referent (upset at the cake). How the target is resolved, and what constrains it, is debated.
Pragmatic accounts (Potts 2005, 2007) treat expressives as an independent speech act that can target different constituents, with the ambiguity resolved by inference. Within this tradition, the culprit hypothesis (Frazier, Dillon & Clifton 2015) holds that the attitude is preferentially construed with respect to an individual who could be held responsible for the described eventuality. A more recent proposal, the valence hypothesis (Glauch 2026), states that expressive adjectives preferentially target the discourse referent that is more negatively valenced than the other potential targets in the sentence.
If valence guides target selection, it is unclear whether the effect is categorical or continuous. Furthermore, psycholinguistic work suggests that emojis substituting for nouns are processed like words (e.g., Paggio & Ping Ping Tse 2022; Weissman et al. 2024), but this has only been tested for descriptive content. Whether expressive emojis are interpreted as non-locally as expressive adjectives is therefore unknown.
The talk presents two planned experiments on German that address these questions. They aim to test the valence hypothesis against experimental evidence and to extend the study of expressive meaning to the emoji modality.