By Lennart Fritzsche Comment Closed

We are happy to announce a talk by Cécile Meier (Frankfurt) in the Semantics Colloquium.

The talk will take place on campus in IG 4.301.
If you wish to participate virtually via Zoom, please contact Lennart Fritzsche for the link. 

Date: December 18, 2025

Time: 4 pm – 6 pm c.t.

Title: Adjectival Horn Scales Conzeptualized 

Abstract: 
Pairs of adjectives forming a Horn Scale like <smart, brilliant> are empirically quite well investigated with respect to whether they trigger scalar implicatures or not. The inference of a scalar implicature to the negation of the extreme adjective is rare and more readily derived if the extreme adjective is mentioned in the context (Doran et al. 2009). That mentioning is the not the only factor that governed the derivation of the implicature is until now under discussion. More recently, Hu et al. (2022) argued by investigating Language Models like GPT2 that the expectedness of the stronger alternative captures scalar implicature rates regardless whether the more informative alternative is mentioned in the context or not. They find that „conceptual expectedness“ (similarity-weighted surprisal) is more apt to capture the empirical psycholinguistic data than „string expectedness“ (raw surprisal). 

In this talk I propose a method of conceptualization for the pattern of pragmatic enrichment from degree semantics that allows for a fresh look on scalar diversity (Tiel et al. 2016) and across-scale and within-scale variation. Unmarked stems like smart in the Positive combine preferably with a silent universal degree quantifier — the positive operator (Seuren 1978, von Stechow 2009). I propose that unmarked adjectives in the positive are, in fact, force variable. High expectedness for a stronger adjective when using a weaker one may induce domain widening for the weaker one and universal weakening is possible (Rullmann et al. 2008). The weak adjective seems to be coerced into an existential reading. As part of a Horn Scale adjectival scale-mates must have the same restriction. Expecting a contrary or contradictory adjective does not have this widening effect: only in the existential reading smart is a Horn scale mate of brilliant. This research contributes also to the theory of alternatives. What counts as an alternative or a Horn Scale is not just a list of words. Rather underlying quantificational patterns (existentials and univerals) could play a key role in deriving scalar implicatures (Buccola et al. 2021).