Page:Language and the Study of Language.djvu/265

VI.] Yet more contrary to sound method would it be, for example, to compare directly English, Portuguese, Persian and Bengali, four of the latest and most altered representatives of the four great branches of Indo-European speech to which they severally belong. Nothing, or almost nothing, that is peculiar to the Bengali as compared with the Sanskrit, to the Persian as compared with the ancient Avestan and Achæmenidan dialects, to the Portuguese as compared with the Latin, can be historically connected with what belongs to English or any other Germanic tongue. Their ties of mutual relationship run backward through those older representatives of the branches, and are to be sought and traced there.

But worst of all is the drawing out of alleged correspondences, and the fabrication of etymologies, between such languages as the English—or, indeed, any Indo-European dialect—on the ono hand, and the Hebrew, or the Finnish, or the Chinese, on the other. Each of these last is the fully recognized member of a well-established family of languages, distinct from the Indo-European. If there be genetic relation between either of them and an Indo-European language, it must lie back of the whole grammatical development of their respective families, and can only be brought to light by the reduction of each, though means of the most penetrating and exhaustive study of the dialects confessedly akin with it, to its primitive form, as cleared of all the growth and change wrought upon it by ages of separation. There may be scores, or hundreds, of apparent resemblances between them, but these are worthless as signs of relationship until an investigation not less profound than we have indicated shall show that they are not merely superficial and delusive.

Let it not be supposed that we are reasoning in a vicious circle, in thus requiring that two languages shall have been proved related before the correspondences which are to show their relationship shall be accepted as real. We are only setting forth the essentially cumulative nature of the evidences of linguistic connection. The first processes of comparison by which it is sought to establish the position and relations of a new language are tentative merely. No sound linguist