Against Grammatical Computation of Scalar Implicatures

Recently, Chierchia (to appear) (henceforth C) has proposed a radical departure from the Gricean theory: SIs, he argues, are not the product of rational behavior between cooperative conversants as in (1), but rather are computed automatically in the grammar by means of special semantic composition rules and lexical scales. He argues that reformulating the process of SI computation in this way has two advantages: it accounts for the generation of a class of SIs that the Gricean theory doesn’t; and negative polarity item licensing, a grammatical process, is sensitive to SIs, which are available to grammatical processes under his theory but not under a Gricean theory. I will argue that, despite the ingenuity of C’s formalism, neither of his arguments for adopting it is valid. I show that a Gricean theory can account for each of the implicatures C identifies, and that negative polarity item licensing is not affected by SIs. Thus, the empirical predictions of C’s theory, which requires the stipulation of a secondary semantic composition system, do not surpass the empirical predictions of the Gricean theory, which assumes only that speakers are cooperative.