Using data from interviews with high school students, I first adduce evidence that lends support to Schourup's (1985) claim that the United States English adolescent hedge like is a discourse particle signalling a possible slight mismatch between words and meaning. Such a particle would generally be included in a grammar in a post‐compositional pragmatic component, but, surprisingly, like also affects basic semantic attributes. These include both truth‐conditions and the weak/strong distinction—though only in existential there and sluicing sentences. I argue that the differential behaviour of like in various constructions selecting weak NP's stems from the restricted free variable it introduces, a variable which only there and sluicing require. This variable is available for binding, quantifier interpretation and other syntactic‐semantic processes, yet is pragmatically conditioned. Indeed, I show that, due to its formal properties, like can be interpreted only during the assignment of model‐theoretic denotations to expressions, along the lines of Lasersohn's (1999) pragmatic haloes. These results support the idea that weak/strong is not a unitary distinction and suggest that the various components of grammars must be organized to allow information from pragmatic/discourse elements to affect basic compositional semantics.
Stored only when both papers are in the corpus — 146 citations exist upstream.
Imported from OpenAlex (CC0) — nothing here is generated.