Wierzbicka, Anna (1997). Conditionals and counterfactuals: Conceptual primitives and linguistic universals. In Angeliki Athanasiadou, & René Dirven (Eds.), On conditionals again (pp. 15-59). Amsterdam: John Benjamins. DOI: 10.1075/cilt.143.04wie
I claim that the concept of IF (as used in natural language) is a conceptual primitive, which cannot be defined in terms of any other concepts. I also claim that this concept is a lexicogrammatical universal, and that claims that, for example, some languages do not have lexicogrammatical resources to distinguish IF from WHEN, are incorrect.
Furthermore, I suggest that the category of “counterfactuals” is also a semantic primitive, which probably cannot be defined in terms of IF, KNOW, negation, and any other hypothetical semantic ingredients. I argue that the “hard core” of this category includes doubly negative sentences with a past reference and I try to explain (with reference to other known linguistic facts) why this should be the case. I question the view that there is some “continuum of hypotheticality”, differently cut in different languages, and I postulate the existence of two discrete semantic concepts, lexicalized in English most clearly as if and if…(pluperfect) would.
The difference between conditionals and counterfactuals can, in a sense, be described as follows: conditionals allow us to imagine that something happens that we think can happen, whereas counterfactuals allow us to imagine that something happens that we think cannot happen. But a description of this kind cannot be regarded as semantic decomposition, because it assumes I think incorrectly that the concept of “imagine” is more basic than either IF or the (counterfactual) IF…WOULD.
In earlier work I proposed that “imagine” was indeed a universal semantic primitive , and I assumed that
both conditionals and counterfactuals could be somehow analyzed via “imagine”. With time, however, it became apparent that this assumption was not consistent with crosslinguistic evidence: it emerged that many languages do not have a word corresponding to imagine, and so “imagine” had to be crossed out from the list of universal semantic primitives. Currently, the growing body of crosslinguistic evidence appears to suggest that while “imagine” was indeed an ill-chosen candidate for this status, “if” and the
counterfactual “if…would” may be true lexicogrammatical universals. If it continues to be confirmed by further evidence, this finding will add considerable support to the hypothesis that both “if” and the counterfactual “if…would” are conceptual primitives.
