Words, I have just said, are the relatively smallest mechanical units in the sentence. There is probably nothing that is so characteristic of the thinking of a human species as the way in which these units are acquired by it. For the Bantu Negro a thing that he sees belongs first of all to a very large number of categories of comprehension. Correspondingly the word for it consists of a kernel or root and a number of monosyllabic prefixes. When he speaks of a woman in a field, his word is something like this: “living, one, big, old, female, outside, _human_”; this makes seven syllables, but it denotes a single, clear-headed, and to us quite alien act of comprehension.[191] There are languages in which the word is almost coextensive with the sentence. The gradual replacement of bodily or sonic by grammatical gestures is thus the decisive factor in the formation of sentences, but it has never been completed. There are no purely verbal languages. The activity of speaking, in words, as it emerges more and more precise, consists in this, that through word-sounds we awaken significance-feelings, which in turn through the sound of the word-connexions evoke further relation-feelings. Our schooling in speech trains us to understand in this abbreviated and indicative form not only light-things and light-relations, but also thought-things and thought-relations. Words are only named, not used definitively, and the hearer has to feel what the speaker means. This and this alone amounts to speech, and hence mien and tone play a much greater part than is generally admitted in the understanding of modern speech. Substantive signs may conceivably exist for many of the animals even, but verb-signs never.