Any theory of language – ancient or contemporary, philosophical or cognitive – faces the same problem, i.e. how to reconcile the unequivocally corporeal character of the speakers and the world they speak of with the somewhat ‘incorporeal’ character of the meanings of linguistic expressions. It is for this reason, for example, that direct-reference theories of language (Stroll 1999) seek to eliminate the Fregean notion of 'sense' (Sinn) from semantics. What is at stake is a completely corporeal account of language. However, such an attempt clashes with the fact that the vast majority of linguistic expressions do not refer either to any objects in the world or to the pre-scientific intuition that words have an autonomous 'meaning' (that is, that the 'sense' of a word does not coincide with the referent, Bedeutung). To solve such a problem, the Stoics introduced in their theory of language the notion of lekton, i.e. what is 'said' or is 'sayable'. Even if the lekton is, properly speaking, incorporeal, at the same time it is the corporeal product of what human speakers do when they utter a verbal utterance. In this paper I propose to compare the notion of lekton to the similar notion of 'use' (Gebrauch), much debated in Ludwig Wittgenstein’s Philosophical Investigations. The thesis of this paper is not that there exists a direct philological connection between the Stoic notion of lekton and the notion of linguistic 'use' in Wittgenstein (even if this cannot be excluded either). Instead, the idea is that when one wants to propose an adequate theory of language, one cannot but introduce a notion such as that of lekton or 'use'.