The Weekly Newsletter of MIT Linguistics

LF Reading Group 4/20 — Adele Mortier (MIT)

This week, our very own Adele Mortier is giving a talk at LFRG as practice for her upcoming conference talk at GLOW!


Speaker:  Adele Mortier (MIT)

Date and time: Wednesday 4/20, 1-2pm

Place: 32-D461

Title: It’s Tough to be Pretty: semantic relatedness between tough- and pretty-predicates


Abstract: Tough (1a) and pretty (1b) predicates are two classes of predicates that can take an infinitival clause as complement.
(1) a. Suzi is tough to please. (TC)
(1) b. Roses are pretty to look at. (PC)
The main contrast between the two classes is that tough-constructions allow an “it-variant” (2b)[Rosenbaum, 1967]; while pretty-constructions do not.
(2) a. It is tough to please Suzi. (it-TC)
(2) b. * It is pretty to look at roses. (it-TC)
Another difference is that, unlike pretty predicates, tough-predicates do not seem to take their subject as a semantic argument (3):
(3) Joseph is tough to please. =/=> Joseph is tough.

In this talk, we argue that the matrix subject in TCs is in fact, a proper semantic argument of the tough-predicate, as already suggested by [Bayer, 1990] and [Fleisher, 2015]. We call this argument the reference argument, and understand it as the causer of the toughness of the event denoted by the embedded clause. We argue that this analysis of “tough” can be extended to it-TCs, by showing that English and French it-TCs are not pure expletive constructions but rather extraposed constructions, whereby the reference argument (“it”) refers to the embedded clause (“it”-extraposition, [Rosenbaum, 1967]). We then establish that tough and pretty have the same basic argument structure, modulo some reversal of the argument order: tough takes its subject as reference and states the toughness of the event denoted by the embedded clause, while pretty takes the embedded clause as reference and states the prettiness of its subject. Assuming that pretty imposes an individual-restriction on its subject, this account allows to explain the ungrammaticality of it-PCs.

In brief, our approach develops a more fine-grained and unified semantics for tough and pretty, avoiding lexical ambiguity as posited by [Keine and Poole, 2017] in the case of TCs. It also allows to explain the (un)availability of an it-variant in those constructions. The existence of a reference argument in TCs ends up providing additional support for a base-generation account of tough-constructions.