Conditionals are topics

CONDITIONALS ARE TOPICS John Haiman University ofManitoba Conditional clauses and topics are marked identically in a number of unrelated languages. This is a surprising fact, since they are not usually considered to be related categories. Nevertheless, if formal similarity reflects similarity in meaning, they must indeed be related. A review of analyses ofconditionals (in the philosophical literature) and of topics (primarily in linguistics) reveals that, in fact, their definitions are very similar. Moreover, it is possible to motivate revisions to these definitions by which they become virtually identical. This paper thusjustifies the method ofbasing semantic analysis of a construction on a cross-linguistic examination of its superficial form.* Neither linguists nor philosophers have suggested a coherent explication for ordinary-language conditionals; most have not even entertained the notion that such an explication is possible. Logicians, with a few exceptions, admit that material implication, defined truth-functionally, is a very poor approximation to the meaning of conditionals; and linguists have hardly even attempted definitions. Until a satisfactory definition for a category exists, the sole criterion for identification of its supposed members is common superficial form : in the case of conditional clauses, the presence, in English, of a common conjunction //; in other languages, of a corresponding conjunction, word-order, verbal desinence, or whatever. It is the thesis of this paper that f^clauses in English, as represented in the sentences below, share a common meaning. All are the topics of the sentences in which they occur: (1) a. If Max comes, we'll play poker. b.If Max had come, we'd have played poker. c.If ice is left in the sun, it melts. d.Even if it rains, the game will continue. e.If you're so smart, why aren't you rich? f.If you're so smart, fix it yourself. g.There's food in the fridge, if you're hungry, h. If I was a bad carpenter, 1 was a worse tailor, i. She's over forty, if she's a day. Our definitions of conditionals are primarily the work of logicians. Our definitions of topics, however, are the work of linguists. To a surprising degree, these definitions converge: the most satisfactory definitions of ordinary language conditionals (I refer to the ideas of Ramsey 1931, Stalnaker 1975, and Ducrot 1972, 1973) approximate the entirely independent definitions of topics that have emerged in recent linguistic publications (cf. in particular Chafe 1976). Conditionals, like topics, are givens which constitute the frame of reference with respect to which the main clause is either true (if a proposition), or felicitous (if not). * A preliminary version of this paper was read at the 1976 annual meeting of the LSA. I am grateful to my audience at that meeting, as well as to R. M. W. Dixon and H. C. Wolfart, for a number of improvements. Data on Tagalog and Chinese were kindly supplied by Velma Andrada and Jane Simpson respectively. 564 CONDITIONALS ARE TOPICS565 Similarity in the superficial form of grammatical categories usually reflects an underlying similarity of their meanings. Before reviewing and refining the convergent definitions of conditionals and topics, I therefore wish to show that their virtual identity is reflected in (and thus necessitated by) their superficial similarity in a number of unrelated languages. One or more of the following will always be true in the languages discussed below: (a)The characteristic mark of the conditional and that of the topic will be identical. (b)Both will be identical with a third category, the interrogative. (c)The characteristic marks of the conditional and the topic will be distinct, but one will be paraphrasable by the other. For the initial identification of conditional clauses in languages other than English, we shall consider sentences which translate ex. Ia, the least marked and controversial of conditionals. All constructions with the same superficial form will also be identified as conditionals, for the reason already given. In some cases, we will then identify as conditionals structures which in English are not translated by (/^clauses. Conversely, not all the subordinate clauses in la-i are superficially alike in languages other than English. The belief which motivates this principle of identification is that grammars tend to miss...