What Unarticulated Constituents Could Not Be

Lenny Clapp

Traditional semantic theories, as formulated by theorists such as Davidson and Montague, presuppose the Principle of Sentential Compositionality: the truth conditions of a declarative sentence are a function of (i) the logical form, or LF, of the sentence and (ii) the meanings, or semantic values, of the words and features in the sentence. There are, however, many prima facie counterexamples to this principle. I briefly consider four types of prima facie counterexample to this principle, and I then sketch the most common strategy defenders of traditional semantic theories have utilized in defending the principle from these prima facie counterexamples. This strategy involves positing phonetically unrealized lexical elements in the logical forms of the prima facie counterexamples, and appealing to the semantic values of these elements to account for the truth conditions of the sentences. The semantic values of such phonetically unrealized lexical elements are known as unarticulated constituents because they are constituents of the proposition expressed, but they are not the semantic value of an articulated lexical element. I argue, however, that this strategy fails for essentially the same reason that reductionism concerning numbers fails. Benacerraf (1965) argued that numbers cannot be sets because there are too many sorts of sets that would do the trick and no principled means of choosing between them. If we accept this argument against reductionism in mathematics, then we should agree that the truth conditions of the prima facie counterexamples cannot be explained by appeal to unarticulated constituents because in this case also there are too many entities that would do the trick and no principled means of choosing between them.