As its capacity for expression is no criterion of a language, still less is the abundance or regularity of its forms. For this very multiplicity, this excessive superfluity, is a burden and a drawback, and obscures the integration of the thought by attaching to it a quantity of needless qualifications. Thus, in the language of the Abipones, the pronoun is different as the person spoken of is conceived as present, absent, sitting, walking, lying, or running, all quite unnecessary specifications.[19-*]
In some languages much appears as form which, on close scrutiny, is nothing of the kind.
This misunderstanding has reigned almost universally in the treatment of American tongues. The grammars which have been written upon them proceed generally on the principles of Latin, and apply a series of grammatical names to the forms explained, entirely inappropriate to them and misleading. Our first duty in taking up such a grammar as, for instance, that of an American language, is to dismiss the whole of the arrangement of the “parts of speech,” and, by an analysis of words and phrases, to ascertain by what arrangement of elements they express logical, significant relations.[19-†]
For example, in the Carib tongue, the grammars give aveiridaco as the second person singular, subjunctive imperfect, “if thou wert.” Analyze this, and we discover that a is the possessive pronoun “thy;” veiri is “to be” or “being” (in a place); and daco is a particle of definite time. Hence, the literal rendering is “on the day of thy being.” The so-called imperfect subjunctive turns out to be a verbal noun with a preposition. In many American languages the hypothetical supposition expressed in the Latin subjunctive is indicated by the same circumlocution.
Again, the infinitive, in its classical sense, is unknown in most, probably in all, American languages. In the Tupi of Brazil and frequently elsewhere it is simply a noun; caru is both “to eat” and “food;” che caru ai-pota, “I wish to eat,” literally “my food I wish.”
In the Mexican, the infinitive is incorporated in the verb as an accusative, and the verb is put in the future of the person spoken of.
Many writers continue to maintain that a criterion of rank of a language is its lexicographical richness—the number of words it possesses. Even very recently, Prof. Max Müller has applied such a test to American languages, and, finding that one of the Fuegian dialects is reported to have nearly thirty thousand words, he maintains that this is a proof that these savages are a degenerate remnant of some much more highly developed ancestry. Founding his opinion largely on similar facts, Alexander von Humboldt applied the expression to the American nations that they are “des débris échappés à un naufrage commun.”
Such, however, was not the opinion of his brother Wilhelm. He sounded the depths of linguistic philosophy far more deeply than to accept mere abundance of words as proof of richness in a language. Many savage languages have twenty words signifying to eat particular things, but no word meaning “to eat” in general; the Eskimo language has different words for fishing for each kind of fish, but no word “to fish,” in a general sense. Such apparent richness is, in fact, actual poverty.
Humboldt taught that the quality, not merely the quantity, of words was the decisive measure of verbal wealth. Such quality depends on the relations of concrete words, on the one hand, to the primitive objective perceptions at their root, and, on the other, to the abstract general ideas of which they are particular representatives; and besides this, on the relations which the spoken word, the articulate sound, bears to the philosophic laws of the formation of language in general.[20-*]
In his letter to Abel-Remusat he discusses the theory that the American languages point to a once higher condition of civilization, and are the corrupted idioms of deteriorated races. He denies that there is linguistic evidence of any such theory. These languages, he says, possess a remarkable regularity of structure, and very few anomalies. Their grammar does not present any visible traces of corrupting intermixtures.[21-*]