and whether elementa stands for alimenta, in the sense of feeders, or for olementa, in the sense of sources of growth (cf. adolere, sub-oles, &c.), it may have been intended originally as a rendering of rizómata.

49

From an historical point of view, letters are not the stoicheia or rizómata of language. The simplest parts into which language can be resolved are the roots, and these themselves cannot be further reduced without destroying the nature of language, which is not mere sound, but always significant sound. There may be roots consisting of one vowel, such as i, to go, in Sanskrit, or 'i, one, in Chinese; but this would only show that a root may be a letter, not that a letter may be a root. If we attempted to divide roots like the Sk. ki, to collect, or the Chinese tehi, many, into tch and i, we should find that we had left the precincts of language, and entered upon the science of phonetics.

Before we do this-before we proceed to dissect the phonetic skeleton of human speech, it may be well to say a few words about roots. In my former Lectures I said, intentionally, very little about roots; at least very little about the nature or the origin of roots, because I believed, and still believe, that in the science of language we must accept roots simply as ultimate facts, leaving to the physiologist and the psychologist the question as to the possible sympathetic or reflective action of the five organs of sensuous perception upon the motory nerves of the organs of speech. It was for that reason that I gave a negative rather than a positive definition of roots, stating 50 that, for my own immediate purposes, I called root or radical "Corssen, Aussprache, 2nd ed. i. p. 530. su Vol. i. p. 292.

whatever, in the words of any language or family of languages, cannot be reduced to a simpler or more original form.

It has been pointed out, however, with great logical acuteness, that if this definition were true, roots would be mere abstractions, and as such unfit to explain the realities of language. Now, it is perfectly true that, from one point of view, a root may be considered as a mere abstraction. A root is a cause, and every cause, in the logical acceptation of the word, is an abstraction. As a cause it can claim no reality, no vulgar reality; if we call real that only which can become the object of sensuous perception. In real language, we never hear a root; we only meet with their effects, namely, with words, whether nouns, adjectives, verbs, or particles. This is the view which the native grammarians of India have taken of Sanskrit roots; and they have taken the greatest pains to show that a root, as such, can never emerge to the surface of real speech; that there it is always a word, an effect, a substance clothed in the garment of grammatical derivatives. The Hindus call a root dhâtu, which is derived from the root dhâ,51 to support or nourish. They apply the same word to their five elements, which shows that, like the Greeks, they looked upon these elements (earth, water, fire, air,

Unâdi Sûtras, i. 70: dudhañ dharanaposhanayoh. Hetú, the Sanskrit word for cause, cannot be referred to the same root from which dhâtu is derived; for though dha forms the participle hita, the i of hi-ta would not be liable to guna before tu. Hetú (Unâdi Sutras, i. 73) is derived from hi, which Bopp identifies with Kiw (Bopp, Glossarium, s. v. hi.) This xiw and kivéw are referred by Curtius to the Latin cio, cieo, citus, excito, not however to the Sanskrit hi, but to root si, to sharpen. Cf. Curtius, G. E. i. p. 118.

ether), and upon the elements of language, as the supporters and feeders of real things and real words. It is known that, in the fourth century B.C., the Hindus possessed complete lists, not only of their roots, but likewise of all the formative elements, which, by being attached to them, raise the roots into real words.

Thus from a root vid, to know, they would form by means of the suffix ghan, Veda, i.e. knowledge; by means of the suffix trik, vettar, a knower, Greek histor and istor. Again, by affixing to the root certain verbal derivatives, they would arrive at vedmi, I know, viveda, I have known, or veda, I know. Besides these derivatives, however, we likewise find in Sanskrit the mere vid, used, particularly in compounds, in the sense of knowing; for instance, dharmavid, a knower of the law. Here then the root itself might seem to appear as a word. But such is the logical consistency of Sanskrit grammarians, that they have actually imagined a class of derivative suffixes, the object of which is to be added to a root for the sole purpose of being rejected again. Thus only could the logical conscience of Panini be satisfied. When we should say that a root is used as a noun without any change except those that are necessitated by phonetic laws (as, for instance, dharmavit, instead of dharmavid), Pânini says

* In earlier works the meaning of dhâtu is not yet so strictly defined. In the Pratis&khya of the Rigveda, xii. 5, a noun is defined as that which signifies a being, a verb as that which signifies being, and as such the verb is identified with the root (Tan nama yenâbhidadhâti sattvam, tad åkhyâtam yena bhavam, sa dhâtuh). In the Nirukta, too, verbs with different verbal terminations are spoken of as dhâtus. Nighantu, i. 20.

(iii. 3, 68), that a suffix (namely, vit) is added to the root vid. But if we come to inquire what this suffix means, and why it is called vit, we find (vi. 1, 67) that a lopa, i. e. a lopping off, is to carry away the v of vit; that the final t is only meant to indicate certain phonetic changes that take place if a root ends in a nasal (vi. 4, 41); and that the vowel i serves merely to connect these two algebraic symbols. So that the suffix vit is in reality nought. This is certainly strict logic, but it is rather cumbersome grammar, and, from an historical point of view, we are justified in dropping these circumlocutions, and looking upon roots as real words.

With us, speaking inflectional and highly refined languages, roots are primarily what remains as the last residuum after a complete analysis of our own dialects, or of all the dialects that form together the great Aryan mass of speech. But if our analysis is properly made, what is to us a mere residuum must originally, in the natural course of events, have been a real germ; and these germinal forms would have answered every purpose in an early stage of language. We must not forget that there are languages which have remained in that germinal state, and in which there is to the present day no outward distinction between a root and a word. In Chinese,53 for instance, ly means to plough, a plough, and an ox, i. e. a plougher; ta means to be great, greatness, greatly. Whether a word is intended as a noun, or a verb, or a particle, depends chiefly on the position which it occupies in a sentence. In the Polynesian 54 dialects,

Endlicher, Chinesische Grammatik, § 123. 54 Cf. Hale, p. 263.

almost every verb may, without any change of for be used as a noun or an adjective; whether it is meant for the one or the other must be learnt from certain particles, which are called particles of affirmation (kua), and the particles of the agent (ko). In Egyptian, as Bunsen states, there is no formal dis tinction between noun, verb, adjective, and particl and a word like an'h might mean life, to live, living, lively.55 What does this show? I think it shows that there was a stage in the growth of language, in which that sharp distinction which we make between the different parts of speech had not yet been fixel, and when even that fundamental distinction between subject and predicate, on which all the parts of speech are based, had not yet been realised in its fulness, and had not yet received a correspondi: g outward expression.

A slightly different view is propounded by Professor Pott, when he says: "Roots, it should be observed, as such, lack the stamp of words, and therefore their real value in the currency of speech. There is no inward necessity why they should first have enterel into the reality of language, naked and formless; it suffices, that, unpronounced, they fluttered before the soul like small images, continually clothed in the mouth, now with this, now with that form, and surrendered to the air to be drafted off in hundred-fold cases and combinations.' 56

It might be said, that as soon as a root is pronounced-as soon as it forms part of a sentence-it ceases to be a root, and is either a subject or a pre35 Bunsen's Aegypten, i. 324.