though in reality its operations are mostly subsequent to the operations of dialectic regeneration. I must ask you at present to take it for granted that everything in language had originally a meaning. As language can have no other object but to express our meaning, it might seem to follow almost by necessity that language should contain neither more nor less than what is required for that purpose. It would also seem to follow that if language contains no more than what is necessary for conveying a certain meaning, it would be impossible to modify any part of it without defeating its very purpose. This is really the case in some languages. In Chinese, for instance, ten is expressed by shi. It would be impossible to change shi in the slightest way without making it unfit to express ten. instead of shi we pronounced tsí, this would mean seven, but not ten. But now, suppose we wished to express double the quantity of ten, twice ten, or twenty. We should in Chinese take eúl, which is twc, put it before shi, and say eúl-shi, twenty. The same caution which applied to shi, applies again to eúl-shi. As soon as you change it, by adding or dropping a single letter, it is no longer twenty, but either something else or nothing. We find exactly the same in other languages which, like Chinese, are called monosyllabic. In Tibetan, chu is ten, nyi two; nyi-chu, twenty. In Burmese she is ten, nhit two; nhitIn she, twenty.

If

But how is it in English, or in Gothic, or in Greek and Latin, or in Sanskrit? We do not say two-ten in English, nor duo-decem in Latin, nor dvida'sa in Sanskrit.

Now here we see, first, that the Sankrit, Greek and Latin, are only local modifications of one and the same original word; whereas the English twenty is a new compound, the Gothic tvai tigjus (two decads), the Anglo-Saxon tuêntig, framed from Teutonic materials; a product, as we shall see, of dialectic regeneration.

We next observe that the first part of the Latin viginti and of the Sanskrit vin'sati contains the same number, which from dvi has been reduced to vi. This is not very extraordinary; for the Latin bis, twice, which you still hear at concerts, likewise stands for an original dvis, the English twice, the Greek dis. This dis appears again as a Latin preposition, meaning a-two; so that, for instance, discussion means, originally, striking a-two, different from percussion, which means striking through and through. Discussion is, in fact, the cracking of a nut in order to get at its kernel. Well, the same word, dvi or vi, we have in the Latin word for twenty, which is vi-ginti, the Sanskrit vin'sati.

It can likewise be proved that the second part of viginti is a corruption of the old word for ten. Ten, in Sanskrit, is dasan; from it is derived da'sati, a decad; and this dasati was again reduced to sati; thus giving us with vi for dvi, two, the Sanskrit visati or vinsati, twenty. The Latin viginti, the Greek eikati, owe their origin to the same process.

* Bopp, Comparative Grammar, § 320. Schleicher, Deutsche Sprache, s. 233.

Now consider the immense difference-I do not mean in sound, but in character between two such words as the Chinese eúl-shi, two-ten, or twenty, and those mere cripples of words which we meet with in Sanskrit, Greek, and Latin. In Chinese there is neither too much, nor too little. The word speaks for itself, and requires no commentary. In Sanskrit, on the contrary, the most essential parts of the two component elements are gone, and what remains is a kind of metamorphic agglomerate which cannot be understood without a most minute microscopic analysis. Here, then, you have an instance of what is meant by phonetic corruption; and you will perceive how, not only the form, but the whole nature of language is destroyed by it. As soon as phonetic corruption shows itself in a language, that language has lost what we considered to be the most essential character of all human speech, namely, that every part of it should have a meaning. The people who spoke Sanskrit were as little aware that vinsati meant twice ten as a Frenchman is that vingt contains the remains of deux and dix. Language, therefore, has entered into a new stage as soon as it submits to the attacks of phonetic change. The life of language has become benumbed and extinct in those words or portions of words which show the first traces of this phonetic mould. Henceforth those words or portions of words can be kept up only artificially or by tradition; and, what is important, a distinction is henceforth established between what is substantial or radical, and what is merely formal or grammatical in words.

For let us now take another instance, which will make it clearer, how phonetic corruption leads to the

first appearance of so-called grammatical forms. We are not in the habit of looking on twenty as the plural or dual of ten. But how was a plural originally formed? In Chinese, which from the first has guarded most carefully against the taint of phonetic corruption, the plural is formed in the most sensible manner. Thus, man in Chinese is ĝin; kiai means the whole or totality. This added to gin gives ginkiai, which is the plural of man. There are other words which are used for the same purpose in Chinese; for instance, péi, which means a class. Hence i, a stranger, followed by péi, class, gives i-péi, strangers. We have similar plurals in English, but we do not reckon them as grammatical forms. Thus, man-kind is formed exactly like i-péi, stranger-kind; Christendom is the same as all Christians, and clergy is synonymous with clerici. The same process is followed in other cognate languages. In Tibetan the plural is formed by the addition of such words as kun, all, and t'sogs, multitude.* Even the numerals, nine and hundred, are used for the same purpose. And here again, as long as these words are fully understood and kept alive, they resist phonetic corruption; but the moment they lose, so to say, their presence of mind, phonetic corruption sets in, and as soon as phonetic corruption has commenced its ravages, those portions of a word which it affects retain a merely artificial or conventional existence and dwindle down to grammatical terminations.

I am afraid I should tax your patience too much were I to enter here on an analysis of the grammatical terminations in Sanskrit, Greek, or Latin, in

* Foucaux, Grammaire Tibetaine, p. 27, and Preface, p. x.

order to show how these terminations arose out of independent words, which were slowly reduced to mere dust by the constant wear and tear of speech. But in order to explain how the principle of phonetic decay leads to the formation of grammatical terminations, let us look to languages with which we are more familiar. Let us take the French adverb. We are told by French grammarians* that in order to form adverbs we have to add the termination ment. Thus from bon, good, we form bonnement, from vrai, true, vraiment. This termination does not exist in Latin. But we meet in Latint with expressions such as bonâ mente, in good faith. We read in Ovid, 'Insistam forti mente,' I shall insist with a strong mind or will, I shall insist strongly; in French, 'J'insisterai fortement.' Therefore, what has happened in the growth of Latin, or in the change of Latin into French, is simply this: in phrases such as forti mente, the last word was no longer felt as a distinct word, and it lost at the same time its distinct pronunciation. Mente, the ablative of mens, was changed into ment, and was preserved as a merely formal element, as the termination of adverbs, even in cases where a recollection of the original meaning of mente (with a mind), would have rendered its employment perfectly impossible. If we say in French that a hammer falls lourdement, we little suspect that we ascribe to a piece of iron a heavy mind. In Italian, though the adverbial termination mente in claramente is no longer felt as a distinct word, it has not as yet been affected by

*Fuchs, Romanishe Sprachen, s. 355.

† Quint., v. 10, 52. Bonâ mente factum, ideo palam; malâ, ideo ex insidiis.