Defining pragmatics (review)

Defining pragmatics (DP) is a survey and evaluation of definitions of pragmatics. Ariel rejects all except a contrast between grammar/code and inference/pragmatics: 'it's a grammatical code if it correlates between some linguistic expression and some meaning or use in a conventional manner. It is pragmatic if the correlation between the form and the meaning or use is mediated by some inference' (19). This implies that pragmatics is nonconventional—a peculiar interpretation if one accepts the definition of convention in Lewis 1969. A writes, 'the goal of a grammar/ pragmatics division of labor is to distinguish two types of explanations for linguistic utterances, not to argue for their absolute psycholinguistic separation under all circumstances' (110). DP does not succeed in distinguishing grammar as code from pragmatics as inference, partly because, as A admits, '[m]ost, if not all language use involves both grammatical and pragmatic aspects' (144; see also p. 19).

DP consists of nine chapters. An introductory chapter, 'What's under the big-tent pragmatics?', answers the title question as follows: 'anything relating to discourse and communication (exclusive of pure language structure)' (11). A distinguishes PROBLEM SOLVERS from BORDER SEEKERS. The former focus on problems that had no solutions according to extant formal grammars; A lists Charles Fillmore, Georgia Green, Jeanette Gundel, Susumu Kuno, George Lakoff, Robin Lakoff, Ellen Prince, Jerrold Sadock, and Sandra Thompson, for whom 'every aspect of the communicated message is somehow coded' (7). Border seekers typically distinguish between coded 'what is said' and inferred/implicated 'what is meant', applying Gricean theory to rework problems of syntax and semantics. A names Laurence Horn, Gerald Gazdar, Stephen C. Levinson, Robyn Carston, and those who contributed to Cole 1981—which, confusedly and confusingly, includes problem solvers Fillmore, Green, Prince, and Sadock!

There are an additional three parts to DP. Part 1 (Chs. 2 and 3) deconstructs pragmatics, Part 2 (Chs. 3 and 4) reconstitutes pragmatics, and Part 3 (Chs. 5-9) maps the 'big tent'. Part 1 finds that the many definitions of pragmatics render it incoherent. Ch. 2, 'Surveying multiple-criterion definitions for pragmatics' (24-34), describes pragmatic meaning as contextually dependent. Semantics is said to be abstract, literal/explicit, and truth-conditional; pragmatics is contextually conditioned and non-truth-conditional, and it accounts for nonliteral and implicit meanings. There is passing reference to the non-truth-conditionality of speech-act types such as questions, imperatives, and expressives, which A inaccurately stipulates thereby to be wholly pragmatic (see Searle 1969, Bach & Harnish 1979, Allan 2006). The much disputed meanings of and versus logical conjunction are assigned to pragmatics via implicature. Politeness effects, hedging, and discourse larger than a sentence are assigned to pragmatics (but the problem of what constitutes a sentence is not broached, so Harry fell asleep and he dreamed of Hermione would be dealt with nonpragmatically, whereas Harry fell asleep. He dreamed of Hermione must be accounted for pragmatically). The Hebrew mi she nominalizer is restricted in application to VIPs, which is grammatical knowledge (234); its proper use, however, is pragmatically determined (41, 234). Surely, this is true of most language expressions.

A toys with the possibility that grammar is a product of the left hemisphere of the brain, and pragmatics a product of the right hemisphere (47-48). Later, she demonstrates that the reality is more complicated (83-86). In her view, codes are quickly and automatically accessed and determinate, while pragmatic inference is slower because the meaning is contextually modulated; but Recanati (2004) and Jaszczolt (2005) have claimed that many inferences are also automatic. Violations to grammar are unacceptable but violations to pragmatic expectations are acceptable if reasonable explication can be given. 'Inferred meanings are implicit, nonconventional and therefore cancelable' (52). Were inferred meanings not conventional, how would language users so readily understand one another? Each speaker would be like Humpty Dumpty. For A, 'conventional' simply means coded/grammatical. The only criterion A provides for distinguishing code from inference is 'the wise-guy test': only encoded meanings can be forced by an uncooperative interlocutor when...