By my life, all this pertains to those definitions that contain no falsehood. As for the logicians’ definitions—by which they claim to portray the very realities of things—they are false through and through: they conflate what is different and separate what is the same.

Proofs that their claim—that definitions convey a true picture of realities—is invalid:

We shall demonstrate that definitions do not, in fact, depict realities and that the so-called “real” definitions of the logicians corrupt the intellect. The evidence for this is presented in several ways.

First proof: whether a definition is cast as a composite expression or a single term, it does not impart knowledge of the thing defined.

One argument for this is that a definition is nothing but the definer’s own assertion. When he says, for example, “Man is the rational animal” or “the laughing animal,” he is merely making a declarative claim devoid of proof. The listener is either already aware of the claim or he is not. If he already knows it, then obviously he has gained no fresh understanding from the definition. If he is unaware of it, believing the claim solely on the word of a speaker who offers no evidence will not yield certain knowledge—especially since the listener knows the speaker is not infallible. Thus, on either supposition, the definition itself is not what imparts knowledge of the defined object.

If it is objected, “The definition is not the entire declarative sentence; it is merely your saying ‘rational animal,’ which is a single term rather than a sentence,” we reply: this objection arises from one of their two conventions. At times they regard the definition as a sentence, as al-Ghazālī and others state; at other times they treat it as a single meaningful term—like a noun—which they call “restrictive composition,” as al-Rāzī and the like mention. Yet uttering a lone term conveys nothing and cannot serve as an answer to the questioner, whether it is further qualified or not.