From the point of view just developed, the investigation of the grammatical forms of primitive language is of particular importance for the psychology of primitive man. True, as has already been remarked, the languages of the most primitive tribes have not been preserved to us in their original form. And yet it is in this very realm of grammatical forms, far more even than in that of sound pictures and onomatopoetic words, that the Soudan languages possess characteristics which mark them as the expression of processes of thought that have remained on a relatively primitive level. This is indicated primarily by the fact that these languages lack what we would call grammatical categories. As regards this point, Westermann's grammar of the Ewe language is in entire agreement with the much earlier results which Steinthal reached in his investigation of the Manda language, which is also of the Soudan region. These languages consist of monosyllabic words which follow one another in direct succession without any intermediate inflectional elements to modify their meaning. Philologists usually call such languages 'root-languages,' because a sound complex that carries the essential meaning of a word, apart from all modifying elements, is called by their science a verbal root. In the Latin word _fero, fer_, meaning 'to bear,' is the root from which all modifications of the verb _ferre_ (to carry) are formed by means of suffixal elements. If, therefore, a language consists of sound complexes having the nature of roots, it is called a root-language. As a matter of fact, however, the languages under discussion consist purely of detached, monosyllabic _words_; the conception 'root,' which itself represents the product of a grammatical analysis of our flectional languages, may only improperly be applied to them. Such a language is composed of detached monosyllabic words, each of which has a meaning, yet none of which falls under any particular grammatical category. One and the same monosyllabic word may denote an object, an act, or a quality, just as in gesture-language the gesture of striking may denote the verb 'to strike' and also the noun 'blow.' From this it is evident to what extent the expressions 'root' and 'root-language' carry over into this primitive language a grammatical abstraction which is entirely inappropriate in case they suggest the image of a root. This image originated among grammarians at a time when the view was current that, just as the stem and branches of a plant grow out of its root, so also in the development of a language does a word always arise out of a group of either simple or composite sounds that embody the main idea. But the component parts of a language are certainly not roots in this sense; every simple monosyllabic word combines with others, and from this combination there result, in part, modifications in meaning, and, in part, sentences. Language, thus, does not develop by sprouting and growing, but by agglomeration