In this paper, we further elaborate on a syntactic ambiguity between slurs and epithets first noticed in Orlando, Eleonora & Andrés Saab. 2020b. A stereotype semantics for syntactically ambiguous slurs. Analytic Philosophy 61(2). 101–129. Here, we discuss in detail the large theoretical implications of such an ambiguity both for the proper analysis of binominal constructions in Spanish (e.g., el idiota de Juan) and for the way in which it is advisable to model the expressive content slurs and certain epithets (those deriving from slurs) have. As for the first aspect, we contend that mainstream approaches in terms of predicate inversion for binominal constructions cannot account for why slurs lose their predicative import when occurring as epithets in binominal environments. In consequence, we propose a new analysis for epithets both in simple occurrences and in binominal constructions. This analysis derives the above-mentioned ambiguity as a type of structural ambiguity, according to which certain slurs can occur in predicative and in non-predicative positions. When they occur as predicates, they have a mixed semantics (McCready, Eric. 2010. Varieties of conventional implicatures. Semantics & Pragmatics 3. 1–57) reflected both in the truth-conditional and the expressive dimensions, but when they occur as epithets, the truth-conditional dimension is lost and only the expressive content survives. As for the second aspect, we defend a stereotype semantics, according to which stereotypes are modeled as Kratzerian modal bases (i.e., set of propositions) in virtue of which stigmatizing theories of human groups are reflected in a parallel, expressive dimension of meaning. This way of modeling some kinds of expressive contents explains how different slurs and epithets manage to communicate different theories about particular human groups, which are the target of derogation.