perceptibly, though by no means so rapidly, in English as in many barbarian dialects. We should have had such interrogative conjugations as havvi, havye, hazzy; imperatives like dōndo'ee and gimme; and our definite article, as in the cognate Danish and Swedish, would have lapsed into an inflectional modification in such words as thempress, thevangel, thapostle.*

These considerations enable us at last to deal with sundry phenomena presented by the Semitic languages, which have not hitherto been brought into conformity with the general principles of linguistic evolution. The efforts of Bunsen, Donaldson, and others, to explain the existing structure of these languages as resulting from the excessive grinding-down of a once complicated set of inflections,-whereby their career would be assimilated to that of the Aryan languages,- has resulted in failure. That the poorly inflectional Hebrew and Ethiopic represent a more primitive state of Semitic speech than the richly inflectional Arabic is, in view of the facts now to be cited, the only tenable conclusion. First, the transparency of Semitic formations is unlike anything which can be seen in languages, like English and French, which have once had a full system of inflections. Secondly, the process of modifying the meaning of words by internal vowel-changes cannot be explained, as in the English ring, rang, rung, as resulting from the phonetic influence exerted by inflections now lost. Thirdly, the Hebrew, and especially the Ethiopic, construct-genitive can hardly be explained as due to the effect of integration upon a previously developed genitive. It is rather a genitive case arrested in its development. Müller has acutely demonstrated that formations like the Hebrew Malk-i-zedek, or the Punic Hann-i-Baal, cannot be compared with a formation like the Italian fratelmo, where

*Marsh, I. 388. Compare the Wallachian domnul for dominus ille; Italian fratelmo = fratellus meus, patremo = pater meus; Hungarian atya-m, etc.; and the agglutinated articles in Albanian and Bulgarian. The influence of writing in retarding integration is noticed by Trench (English, Past and Present, p. 24), and by Whitney (pp. 472, 473), who thus explains the fact that English words anciently imported from Latin and Greck, as mint from moneta, church from kyriake, alms from eleemosyne, etc., are highly integrated, while words recently adopted have remained comparatively intact. See also Müller's remarks on Yes'r and Yes'm (I. 224); and compare A. W. Schlegel, Obs. sur la Langue et la Litt. Provençales, p. 18.

the integrated ending was once an adjectival pronoun equivalent to a true genitive. Fourthly, the much greater conspicuousness of onomatopoeia in the Semitic than in the Aryan languages shows that their words have better retained their primitive shape. Fifthly, the vast number of notions expressible by slight modifications of single Semitic root* is a feature characteristic, not of highly integrated languages, like French or English, but of languages, like Annamitic and Burmese, which have never arisen from their primitive meagreness.

In view of these facts, a comparison between the older Hebrew and the younger Arabic becomes very instructive. Let us bear in mind that the Hebrew became a written language at least one thousand two hundred years sooner than the Arabic, and that the influence of writing in retarding integration is acknowledged to be especially powerful in the case of the Semitic alphabets. (Renan, p. 437.) Now Arabic is far richer in inflections than Hebrew. The forms of the verb, which are five in Hebrew, and have been reduced to three in Aramæan, are nine in Arabic. In the latter language there are more compound tenses. In Syriac there are but few traces of a dual; in Hebrew we find it in some classes of nouns; in Arabic it is applied to adjectives, pronouns, and verbs. "The employment of certain conjunctions in regimen and with affixes, which is a characteristic feature of Arabic, is also found in Hebrew, but only in a rudimentary stage." (Renan, 424 – 426.) From all these data, which Renan has most skilfully generalized, he draws the incorrect conclusion that since the Semitic languages have become richer and richer in inflections with the progress of time, and since, moreover, they have not given birth to analytic derivative dialects, as the Latin gave birth to French, and the Anglo-Saxon to English - they are therefore the product of mental processes totally different from those which have produced the Aryan languages, and can in no wise be explained by the same formula of development.

* In Hebrew, for instance, " the word, in its various conjugations, means to mix, to exchange, to stand in the place of, to pledge, to interfere, to be familiar; and also to disappear, to set, and to do a thing in the evening; besides all this, with various vowel modifications, the same three letters mean to be sweet, a fly or beetle, an Arabian, a stranger, the weft of cloth, the evening, a willow, and a raven." Farrar, Chapters on Language, p. 229.

At the present stage of our discussion, the way out of this difficulty is short and easy. Although the mother-Semitic was a tertiary language, no less than the mother-Aryan, it would be very misleading to regard them as equally developed at the outset. The mother-Aryan was the product of a far more extensive antecedent process of agglutination than the mother-Semitic. While the former may be supposed, in its secondary stage, to have resembled such richly agglutinated languages as Telugu or Turkish, the latter must be supposed to have resembled such poorly agglutinated languages as Burmese or demotic Egyptian. Its progress, therefore, after attaining the tertiary stage, would be toward the further development of inflection. The Semitic vowel-changes show how meagrely supplied the mother-tongue must have been with determinative signs. In Chinese this meagreness is partly compensated by an elaborate system of tones. Probably the early speaker of Semitic similarly pronounced his vowels with an altered stress, which often leads to a vowel variation, as, in Hebrew, "pattach," lengthened into "qamets," may become "qamets-chatuph." Now in Hebrew the development of tertiary forms was early stopped by the introduction of writing and literary culture; * while the progress of Arabic for the next twelve centuries was not toward so-called "analysis," but toward a more extensive production of forms by amalgamation. The redactions of the Koran, under Othman and Abdalmalek, put a final stop to this.

Lastly, the progress of the Semitic languages, since they stopped amalgamating, has been toward further integration by the dropping of inflections and vocalization of consonants, after the manner of Aryan tongues, as we had occasion to observe at the beginning of this discussion. And let it be further noted that the Semitic dialect which was latest checked in its development of tertiary forms, and which therefore arrived at relatively great inflectional complexity, - namely, the Arabic,is just the one which most resembles the Aryan languages in its recent tendency towards "analysis." Even Renan compares the relation of new and old Arabic to the relation of new and old High German.

*See also, for the conservative influence of a priestly caste or fraternity, Whitney, pp. 148, 149, 159.

That the history of the Semitic languages presents no such striking case of change as the change from Latin to French, from Anglo-Saxon to English, or from Vedic Sanskrit to Hindustani, may readily be granted. But for a most intimate mixing of races, such changes would not have occurred in the Aryan domain. Except in the case of Amharic, no Semitic language is known to have been adopted as a vernacular speech by a non-Semitic race, as the Latin was adopted by Kelts, Iberians, and Germans. And in the Amharic we do find the nearest approach to parallelism with the Romanic dialects which the Semitic domain affords.*

To sum up this long argument: we have seen good reason for believing that by a universal process of integration primary languages tend to become secondary, and secondary languages tend to become tertiary. We have observed that tertiary languages are far from rejecting compounds and derivatives framed after the primary and secondary patterns. We have shown that lost inflections can be replaced only by a recurrence to the primary methods of juxtaposition; and we have seen the amalgamative process eliciting new tertiary forms from these juxtapositive compounds, save where it is checked by assignable retarding influences. Finally, we have exhibited the working of the same general processes in the Semitic languages, which have been wrongly claimed as constituting a separate domain of human speech, governed by dynamic laws peculiar to itself. Though the evidence here given is but a small fragment of what might be adduced, it is still quite sufficient to illustrate and confirm an inference which the results of grammatical analysis have long since rendered inevitable. Thirtythree years ago that incomparable philologist, Richard Garnett, saw clearly that if inflections were ever independently signifieant, the structure of all languages must once have resembled that of the Chinese. (Essays, p. 109.) Bopp and his fellowworkers, in proving the first point, proved also the second; and it is only by implicitly rejecting the one that Renan and his followers are enabled explicitly to reject the other.

The formation of the so-called "constructive mood" in Amharic may perhaps be compared to the formation of the Romanic futures. Isenberg's Amharic Grammar, p. 69, cited by Garnett, p. 297.

†M. Franck accepts Renan's theory, "parcequ'elle est la plus conforme à la

II.

We have thus arrived at a vantage-ground from which we may contemplate more nearly than of old the beginnings of human speech. Much of the ancient discussion upon this subject is at once seen to be no longer relevant. As long as inflections and modifying particles were believed to be arbitrary, they might well have been deemed the offspring of inventive contrivance; not so when they are known to be significant. The intricate grammatical system of Greek or Sanskrit might well have been thought a fit subject for miraculous revelation; not so the meagre and artless combinations of Chinese or Tungusian. Over the comparative claims of nouns, verbs, and prepositions, to be considered the primal elements of speech, we need no longer puzzle ourselves; for we know that in a primitive language there are neither nouns nor verbs, neither adjectives nor adverbs, but only naked roots, which may be used to denote either substance or attribute, position or action, as circumstances may chance to require. Let us consider how the Chinaman would say, "The sun is shining through the clouds." He would take the word for sun; a word meaning bright, brightly, brightness, or to be bright, according to the context; a word meaning passage, or to pass; and the word for cloud, coupled with some such expression as multitude, heap, or flock, to serve as a sign of plurality; and heaping together these raw materials, would make the statement, "Sun bright pass cloudheap." Crude as such a language undoubtedly is, we may rest assured that the winged words of our exquisite Aryan speech at first limped and halted with equal awkwardness. An analysis of the magnificent verse of Dante,

"Io non piangeva, si dentro impietrai,"

would show us that its wonderful tissue is woven of very humble stuff, -a word for "stone," a word imitating the sound of

dignité de notre nature et à sa mystérieuse grandeur." Études Orientales, p. 404 How long are we to see scholars, from whom better things might be expected, childishly accepting or rejecting theories, not because they are upheld or contradicted by established facts, but because they conform to, or conflict with, sundry preconceived notions of " dignity" or "fitness"?

* "Auf dieser urältesten Stufe sprachlichen Lebens gibt es also, lautlich unterschieden, weder Verba noch Nomina, weder Conjugation noch Declination." Schleicher, Die Darwinsche Theorie, p. 22.