will take the place of the modified general stem. In languages that possess a full equipment of adverbial and locative formative elements, generalized words of motion may be qualified by their use; while, wherever these elements are absent, new stems must take their place. The same is true of grammatical elements that designate form or substance. Where these occur, the languages may lack words expressing predicative ideas relating to objects of different form and consisting of different substances (like our words “to lie,” “to sit,” “to stand,” “to tear,” “to break”).
A lexicographic analysis based on these principles of classification promises important results, but requires a much more accurate knowledge of the meaning of stems than is available in most cases.
No less interesting are the categories of thought that find expression in grammatical form. The older grammars, although many of them contain excellent material, do not clearly present these points of difference, because they are modelled strictly on the Latin scheme, which obscures the characteristic psychological categories of Indian languages. Thus the idea of plurality is not often developed in the same sense as in Latin, but expresses rather the idea of distribution or of collectivity. The category of gender is rare, and nominal cases are not common. In the pronoun we find often a much more rigid adherence to the series of three persons than the one that we apply, in so far as the distinction is carried through in the pronominal plural and in the demonstrative. Furthermore, new ideas—such as visibility, or position in regard to the speaker in the six principal directions (up, down, right, left, front, back), or tense—are added to the concept of the demonstrative pronouns. In the numeral the varied bases of numeral systems find expression. In the verb the category of tense may be almost suppressed or may be exuberantly developed. Modes may include many ideas that we express by means of adverbs, or they may be absent. The distinction between verb and noun may be different from ours. In short, an enormous variety of forms illustrates the multifarious ways in which language seizes upon one or another feature as an essential of expression of thought.
Besides the greater or lesser development of categories that are parallel to our own, many new ones appear. The groups of ideas selected for expression by formative elements are quite distinctive, and they belong to the most important features in the characterization of each language. In some cases they are poorly developed, but most American languages possess an astonishing number of formative elements of this type.
In some cases their number is so great, that the very idea of subordination of one element of a word under another one loses its significance; and we are in doubt whether we shall designate one group as subordinate elements, or whether we shall speak of the composition of co-ordinate elements. While in some languages, as in Algonquin or Kutenai, this may be a matter of arbitrary definition, it involves a problem of great theoretical interest; namely, the question whether formative elements have developed from independent words, as has been proved to be the case with many formal suffixes of European languages.
The objectivating tendency of our mind makes the thought congenial, that part of a word the significance of which we can determine by analysis must also have objectively an independent existence; but there is certainly no a priori reason that compels us to make this assumption. It must be proved to be true by empirical evidence. Although the history of American languages is not known, and therefore cannot furnish any direct evidence for or against this theory, the study of the etymological processes will throw light upon this problem, because in many cases the very phonetic weakness of the constituent elements, their internal
