grammar is purely Tataric or Turanian. The Turks, however, possessed but a small literature and narrow civilisation before they were converted to Mohammedanism. Now, the language of Mohammed was Arabic, a branch of the Semitic family, closely allied to Hebrew and Syriac. Together with the Koran, and their law and religion, the Turks learned from the Arabs, their conquerors, many of the arts and sciences connected with a more advanced stage of civilisation. Arabic became to the Turks what Latin was to the Germans during the middle ages ; and there is hardly a word in the higher intellectual terminology of Arabic, that might not be used, more or less naturally, by a writer in Turkish. But the Arabs, again, at the very outset of their career of conquest and conversion, had been, in science, art, literature, and polite manners, the pupils of the Persians, whom they had conquered; they stood to them in the same relation as the Romans stood to the Greeks. Now, the Persians speak a language which is neither Semitic, like Arabic, nor Turanian, like Turkish; it is a branch of the Indo-European or Aryan family of speech. A large infusion of Persian words thus found its way into Arabic, and through Arabic into Turkish ; and the result is that at the present moment the Turkish language, as spoken by the higher ranks at Constantinople, is so entirely overgrown with Persian and Arabic words, that a common clod from the country understands but little of the so-called Osmanli, though its grammar is exactly the same as the grammar which he uses in his Tataric utterance.

There is, perhaps, no language so full of words

:

evidently derived from the most distant sources as English. Every country of the globe seems to have brought some of its verbal manufactures to the intellectual market of England. Latin, Greek, Hebrew, Celtic, Saxon, Danish, French, Spanish, Italian, German-nay, even Hindustani, Malay, and Chinese words, lie mixed together in the English dictionary. On the evidence of words alone it would be impossible to classify English with any other of the established stocks and stems of human speech. Leaving out of consideration the smaller ingredients, we find, on comparing the Teutonic with the Latin, or Neo-Latin or Norman elements in English, that the latter have a decided majority over the home-grown Saxon terms. This may seem incredible; and if we simply took a page of any English book, and counted therein the words of purely Saxon and Latin origin, the majority would be no doubt on the Saxon side. The articles, pronouns, prepositions, and auxiliary verbs, all of which are of Saxon growth, occur over and over again in one and the same page. Thus, Hickes maintained that nine-tenths of the English dictionary were Saxon, because there were only three words of Latin origin in the Lord's prayer. Sharon Turner, who extended his observations over a larger field, came to the conclusion that the relation of Norman to Saxon was as four to six. Another writer, who estimates the whole number of English words at 38,000, assigns 23,000 to a Saxon, and 15,000 to a classical source. On taking, however, a more accurate inventory, and counting every word in the dictionaries of Robertson and Webster, M.

Thommerel has established the fact that the number of Teutonic or Saxon words in English amounts to only 13,330, against 29,354 words which can either mediately or immediately be traced to a Latin source.* On the evidence of its dictionary, therefore, and treating English as a mixed language, it would have to be classified together with French, Italian, and Spanish, as one of the Romance or Neo-Latin dialects. Languages, however, though mixed in their dictionary, can never be mixed in their grammar. Hervas was told by missionaries that in the middle of the eighteenth century the Araucans hardly used a single word which was not Spanish, though they preserved both the grammar and the syntax of their own native speech.† This is the reason why grammar is made the criterion of the relationship and the base of the classification in almost all languages; and it follows, therefore, as a matter of course, that in the classification and in the science of language, it is impossible to admit the existence of a mixed idiom. We We may form whole sentences in English consisting entirely of Latin or

* Some excellent statistics on the exact proportion of Saxon and Latin in various English writers, are to be found in Marsh's Lectures on the English Language, p. 120, seq. and 181, seq.

+ " "En este estado, que es el primer paso que las naciones dan para mudar de lengua, estaba quarenta años ha la araucana en las islas de Chiloae (como he oido á los jesuitas sus misioneros), en donde los araucanos apénas proferian palabra que no fuese española; mas la proferian con el artificio y órden de su lengua nativa, llamada araucana."-Hervas, Catalago, t. i. p. 16. "Este artificio ha sido en mi observacion el principal medio de que me he valido para conocer la afinidad ó diferencia de las lenguas conocidas, y reducirlas á determinadas classes."- Ibid. p. 23.

Romance words; yet whatever there is left of grammar in English bears unmistakeable traces of Teutonic workmanship. What may now be called grammar in English is little more than the terminations of the genitive singular, and nominative plural of nouns, the degrees of comparison, and a few of the persons and tenses of the verb. Yet the single s, used as the exponent of the third person singular of the indicative present, is irrefragable evidence that in a scientific classification of languages, English, though it did not retain a single word of Saxon origin, would have to be classed as Saxon, and as a branch of the great Teutonic stem of the Aryan family of speech. In ancient and less matured languages, grammar, or the formal part of human speech, is far more abundantly developed than in English; and it is, therefore, a much safer guide for discovering a family likeness in scattered members of the same family. There are languages in which there is no trace of what we are accustomed to call grammar; for instance, ancient Chinese ; there are others in which we can still watch the growth of grammar, or, more correctly, the gradual lapse of material into merely formal elements. In these languages new principles of classification will have to be applied, such as are suggested by the study of natural history; and we shall have to be satisfied with the criteria of a morphological affinity, instead of those of a genealogical relationship.

I have thus answered, I hope, some of the objections which threatened to deprive the science of language of that place which she claims in the circle

76 THE SCIENCE OF LANGUAGE A PHYSICAL SCIENCE.

of the physical sciences. We shall see in our next lecture what the history of our science has been from its beginning to the present day, and how far it may be said to have passed through the three stages, the empirical, the classificatory, and the theoretical, which mark the childhood, the youth, and the manhood of every one of the natural sciences.