Institut Jean Nicod

Accueil du site > Séminaires & Colloques > Séminaires > Archives > 2012-2013 > CPR > May 14 : Ernesto Perini (Universidade Federal de Minas Gerais, (...)

May 14 : Ernesto Perini (Universidade Federal de Minas Gerais, Brésil)

Mardi 14 mai 2013 de 16h30 à 18h30
Institut Jean-Nicod, ENS, 29, rue d’Ulm 75005 paris, Salle de réunion, RDC.

Ernesto Perini (Universidade Federal de Minas Gerais, Belo Horizonte, Brésil).

Title : "Green Leaves Again : an assessment of Kennedy’s and McNally’s solution to Travis’ puzzle"

Pia paints the russet leaves of her maple green, and says :
1. The leaves are green
A botanist asks if she has green leaves for a study. She answers :
2. The leaves are green,
1 seems to be true, and 2 false.
Kennedy and McNally 2010 explain the different evaluations of 1 and 2 by postulating that they have terms with different senses : Color adjectives have anongradable sense, denoting “the property of manifesting some property which is correlated with the color in question,” and a gradable one denoting that the object has the appropriate quantity or quality of the color. In 1, ‘green’ has a gradable meaning, and in 2, the nongradable one.
They have two arguments to show that color terms ambiguous : (i) the gradable sense, but not the nongradable sense, admits degree modifiers, and (ii) crossed-sense anaphora are not allowed. Both arguments are inconclusive, in each case there is an alternative explanation. I will focus on (i). We can find already in Zwicky and Sadock 1975 reasons to think that distributional arguments such as (i) are not always enough to establish the ambiguity of expressions. In this case, I propose the following reasoning, building on Zwicky and Sadock, to refuse the ambiguity of color adjectives : (a) if color adjectives were ambiguous, this ambiguity should be realizable in some language ; (b) supposing that this is not the case,
they are either underspecified or there is an unsystematic ambiguity ; (c) the different readings of color adjectives are systematic ; (d) if the supposition in (b) is right, they are underspecified.
We can also extend the argument of Kripke 1977 concerning defining definite descriptions to establish (d). Supposing a language in which color adjectives have a gradable sense, the nongradable use will be available. Unlike Kripke’s result, however, we will reach a literal use due to a conceptual extension, and not a non literal use due to our fallibility.
Kennedy and McNally propose perspicuous representations of the intended readings of 1 and 2, but they still have to show that such representations belong to the object language, and not to the metalanguage of the theorist – and a distinction in the metalanguage doesn’t show by itself an ambiguity in the object language. Their argument is inconclusive. Moreover, it seems more reasonable to explain Travis’ puzzle as a case of conceptual extension, rather than two different concepts associated with the same word.