family. And if this common ancestry is an inference, it is obvious that there the quest must cease, so far as direct observation is concerned. It is possible to build inferences upon inferences, but only as an exercise of fancy, not as science. The notion of an original Indo-European parent speech is therefore a blind alley so far as these questions of primitive simplicity and complexity in speech are concerned.

The older theorists worked this conception of a simple unified parent speech very hard, and they thought they had found in it a plain solution for all the many difficulties of the history of language. Noah Webster, for example, writing in 1828, in the introduction to his American Dictionary, comments very seriously on the opinion of Dr. Alexander Murray, professor of Oriental languages in the University of Edinburgh, who maintained in his History of the European Languages that all the words in all the European languages could be carried back to nine root words. This simple system of nine root words Dr. Murray sets down explicitly and confidently. A learned philological treatise published at Zürich a hundred years later shows progress in that the nine primitive roots are here reduced to six, and from these six roots are derived not only the Indo-European languages, but all the languages of the human race!

But family relationships in languages can be studied without being carried back to the dim regions of remote antiquity. English has immediate relatives in German, in Dutch, in Danish, and in other Teutonic dialects. French we know to be close kin to Italian and Spanish, to name only the main divisions of this related group. All these Romanic languages are derived from Latin, and in、、

these languages we have perhaps the most manageable instance of a family group of languages. What can be inferred with respect to the notion of primal unity from these languages? But before this question can be discussed, both the term "derived" and the term "Latin " call for definition. French is not derived from Latin in the sense that it is the natural offspring of Latin. When the linguist speaks of a language as being derived from another, or when he speaks of a family of languages, he is using metaphors which must be understood only meta phorically. Languages have no generational existence, like animals sprung from animals, or plants generated from the seeds of other plants. Languages may be transferred or acquired, but strictly speaking cannot be derived or inherited.

But what was this Latin from which French is supposed to be descended? Can we think of Latin as providing a primitive unity so far as French is concerned? To this question the history of language answers decisively in the negative. In the first place the Latin of which French is a later form was in a great many respects different from the Latin of Vergil, Horace, and Cæsar. It was a popular speech, a vulgar Latin, so different from the literary speech that the literary speech is often quite useless when one attempts to account for the details of modern French. But the differentiation of popular and literary Latin is only the beginning of the distinctions that must be made in the meaning of the term "Latin." How many varieties or dialects of popular Latin were there? It is an obvious and a sufficient answer to this question to say that there must have been many, probably as many dialects of popular Latin in the days when Latin

they

was becoming French as there are of popular French today. In short, the quest for primitive unity by the historical route, as ever, ends only in disappointment. The farther it is carried, the more clearly one realizes that it is a fool's quest, a quest for an unreal, far-off world. What happened in the Romanic languages, undoubtedly happened also in the Germanic languages. The parent 'Germanic speech from which the various related Germanic languages are to be derived is not so definitely and minutely known as the parent Latin of the Romanic languages, but if it were, without question it would be found to be just as complex and variable.

A certain illusion of simplicity is sometimes produced when one looks at some of the languages of antiquity. The grammar, for example, of classic Latin seems remarkably simple and orderly. But one must remember that the usual descriptive Latin grammar is based upon a comparatively few texts, Cæsar, Horace, Vergil, and a few others, and that these texts have been highly regularized by copyists in transmission. The language represented by the Latin grammars is not a real and a spoken but a theoretical and literary language. Gothic is still more regular, is in fact almost a perfect system. But the doubt will not down that the Gothic language as we know it, and our knowledge is derived from a very small body of text, was made to order, was again a language of theory and not of reality. The farther back one goes in the study of a particular language, the easier it is to make a perfect system of it. This is true because the farther back we go, the less we can know about a language, and the less we know, the easier it is to reduce our knowledge to order. Only in the living present do we reach our

maximum of knowledge, and also in the living present do we realize the hopelessness of ever reducing a real language to an orderly statement.

If neither philosophy nor history provides any justification for belief in a far-off golden age of language, perhaps a third way of approach to the question may yield' results more encouraging. Genuinely primitive conditions are more readily reached through the study of the mind than by the examination of historical records. The most ancient of records of the past are but as the words of yesterday in the long perspective of time. Immature and undeveloped mind, however, may be extremely remote, for the standard by which we measure distance here is not one merely of years and centuries, which in themselves have no significance, but a standard of cultural development. Looking back then to the primitive psychology of human speech, what are the evidences with respect to this notion of original simplicity and orderliness?

First of all, one might examine the speech of infancy.、 The infant, it is true, does not learn its speech independently and separately, but largely by imitation and in relationships to various other individuals. In this respect, however, the sophisticated infant of modern days certainly does not differ from the infant of the stone age. Even in the beginnings of speech, human beings were social creatures. They had learned the playful trick of imitation, and by means of it were gradually building up their culture history.

Taking the infant of today, however, as the only one we know, and as representing the childhood of the race, no one would dream of finding in its early gurgles and babblings the simplicity and order which the adult speech

so lamentably lacks. When the infant reaches the age at which it is ripe to learn the speech of its race, it has at command an infinite quantity of the raw materials of speech, much of which it will never be able to use. And that adult speech is not merely a fragmentary recollection of these primitive elements of infant speech is proved by the fact that the teeth play an important part in all adult speech. Just when must one look for the infantile golden age of speech, before or after the child acquires teeth? But certainly the speech impulse comes before teeth are developed and it would seem that teeth therefore must mark the beginning of that downward career of corruption and disorganization in speech in which all the later days of the infant are to be spent.

These inferences with respect to the beginnings of speech derived from the observation of infants are duplicated when one contemplates the notion of the infancy of the race. No infants of the race can come under our direct gaze or be heard by our physical ears. It calls for no great stretching of the imagination, however, to hear them in their first experiments in speech, their chatterings, repetitions, squeaks, growls, lyric notes, shrieks of anger, mumblings of comfort and content, all the multitudinous assemblage of sounds in the great human orchestra tuning up and preparing for action. In these first experiments there was material for many languages. In speech as in other respects uncontrolled and undeveloped nature is prodigal and wasteful. But simplicity and order were not characteristics of these early experiments. orchestra tuning up can make sounds enough for a dozen symphonies, but the symphonies are not present, even potentially, in all these preliminary tootings and scrapings.

An