Scalar Implicatures: Pragmatic Inferences or Grammar?

Scalar Implicatures: Pragmatic Inferences or Grammar? * Chungmin Lee Dept of Linguistics Seoul National University Seoul 151-742, Korea clee@snu.ac.kr Abstract. This talk discusses the nature of different kinds of scales and controversies over issues on the generation of scalar implicatures, particularly those in complex sentences involving disjunction and another operator in its scope, and so on. The pragmatic position based on Gricean reasoning in opposition to the grammatical position based on alternative semantics and LF syntax employing the exhaustivity (Exh) operator will be examined. The context-driven view and the default view largely still within the pragmatic position will also be discussed. In doing so, the talk will offer my position that scalar implicatures are motivated by Gricean pragmatic reasoning but that they are deeply and crucially rooted in the grammatical devices of Contrastive Topic (CT), overt or covert. CT requires PA (pero/aber) conjunction, i.e. ‘concessive But’ and that’s why scalar implicatures begin with but and its equivalents cross-linguistically. The CT operator rather than the exhaustivity (Exh) operator must be represented to be related to the previous discourse and the forward concessive conjunction. Keywords: scalar implicatures, Gricean reasoning, grammatical system, disjunction, context-driven vs. default, exhaustivity, Contrastive Topic, PA (concessive) conjunction. 1. Introduction *I express my gratitude to the audiences of LELNS 2006 in Tokyo and of ICKL 2006 in Guadalajara. I also deeply thank Rachel Roxas for her gracious patience in editing the Proceed’s with her team. Copyright 2008 by Chungmin Lee 30 22nd Pacific Asia Conference on Language, Information and Computation, pages 30–45 1. 1 Scales There are many gradable adjectives, to which different degrees of the state of an entity can be assigned such as clean and dirty (Kennedy 1999). The adjective clean shows the degree of no dirt on its subject entity (maximum standard, total-universal), as an <e, d> type function, taking an entity and returning a degree. Its antonym dirty, on the other hand, starts from the minimum degree of having dirt on an entity (minimum standard, partial-existential) (Yoon , Lee ). These adjectives, but not un-gradable adjectives such as alive and dead, can easily be modified by degree adverbs such as a little and very, and intensifiers such as terribly and awfully. But the so-called un-gradable adjectives can also generate scalar implicatures, as will be discussed later. Gradable adjectives typically constitute scales with degree modification on the same single adjectives (intra-lexically). In contrast, scales in connection with scalar implicatures may be different in the sense that they are typically formed by a set of scalar alternatives of different lexical items, mostly predicates. Horn’s (1972) scalar predicates are related by entailment asymmetrically – a stronger one entails the weaker one in the scale but not vice versa. A sentence containing a scalar value or item can generate a scalar implicature when the scalar value is replaced with a stronger item from the scale, resulting in an alternative sentence. All such stronger alternative sentences are implicated to be negated (or false): (1) ScalAlt(φ) = {φ’: φ contains scalar …

Content truncated to 500 words