Page images
PDF
EPUB

after Latin had once been shown to be more primitive on many points than Greek, it was easy to see that the Teutonic, the Celtic, and the Slavonic languages also, contained each a number of formations which it was impossible to derive from Sanskrit, Greek, or Latin. It was perceived that all had to be treated as co-ordinate members of one and the same class.

The first great step in advance, therefore, which was made in the classification of languages, chiefly through the discovery of Sanskrit, was this, that scholars were no longer satisfied with the idea of a general relationship, but began to inquire for the different degrees of relationship in which each member of a class stood to another. Instead of mere classes, we hear now for the first time of well regulated families of language.

A second step in advance followed naturally from the first. Whereas, for establishing in a general way the common origin of certain languages, a comparison of numerals, pronouns, prepositions, adverbs, and the most essential nouns and verbs, had been sufficient, it was soon found that a more accurate standard was required for measuring the more minute degrees of relationship. Such a standard was supplied by Comparative Grammar; that is to say, by an intercomparison of the grammatical forms of languages supposed to be related to each other; such intercomparison being carried out according to certain laws which regulate the phonetic changes of letters.

A glance at the modern history of language will make this clearer. There could never be any doubt that the so-called Romance languages, Italian, Wallachian, Provençal, French, Spanish, and Portuguese, were closely related to each other. Everybody could

see that they were all derived from Latin. But one of the most distinguished French scholars, Raynouard, who has done more for the history of the Romance languages and literature than any one else, maintained that Provençal only was the daughter of Latin; whereas French, Italian, Spanish, and Portuguese were the daughters of Provençal. He maintained that Latin passed, from the seventh to the ninth century, through an intermediate stage, which he called Langue Romane, and which he endeavored to prove was the same as the Provençal of Southern France, the language of the Troubadours. According to him, it was only after Latin had passed through this uniform metamorphosis, represented by the Langue Romane or Provençal, that it became broken up into the various Romance dialects of Italy, France, Spain, and Portugal. This theory, which was vigorously attacked by August Wilhelm von Schlegel, and afterwards minutely criticised by Sir Cornewall Lewis, can only be refuted by a comparison of the Provençal grammar with that of the other Romance dialects. And here, if you take the auxiliary verb to be, and compare its forms in Provençal and French, you will see at once that, on several points, French has preserved the original Latin forms in a more primitive state than Provençal, and that, therefore, it is impossible to classify French as the daughter of Provençal, and as the granddaughter of Latin. We have in Provençal : —

sem, corresponding to the French nous sommes,

[blocks in formation]

and it would be a grammatical miracle if crippled forms, such as sem, etz, and son, had been changed

back again into the more healthy, more primitive, more Latin, sommes, êtes, sont; sumus, estis, sunt.

Let us apply the same test to Sanskrit, Greek, and Latin; and we shall see how their mutual genealogical position is equally determined by a comparison of their grammatical forms. It is as impossible to derive Latin from Greek, or Greek from Sanskrit, as it is to treat French as a modification of Provençal. Keeping to the auxiliary verb to be, we find that I am

[blocks in formation]

Now, the termination of the second person is si, which, together with as, or es, would make,

[blocks in formation]

But here Sanskrit, as far back as its history can be traced, has reduced assi to asi; and it would be impossible to suppose that the perfect, or, as they are sometimes called, organic, forms in Greek and Lithuanian, es-si, could first have passed through the mutilated state of the Sanskrit asi.

The third person is the same in Sanskrit, Greek, and Lithuanian, as-ti or es-ti; and, with the loss of the final i, we recognize the Latin est, Gothic ist, and Russian est'.

The same auxiliary verb can be made to furnish sufficient proof that Latin never could have passed through the Greek, or what used to be called the Pelasgic stage, but that both are independent modifications of the same original language. In the singular, Latin is less primitive than Greek; for sum stands for es-um, es for es-is, est for es-ti. In the first

person plural, too, sumus stands for es-umus, the Greek es-mes, the Sanskrit 'smas. The second person es-tis, is equal to Greek es-te, and more primitive than Sanskrit stha. But in the third person plural Latin is more primitive than Greek. The regular form would be as-anti; this, in Sanskrit, is changed into santi. In Greek, the initial s is dropped, and the Æolic enti, is finally reduced to eisi. The Latin, on the contrary, has kept the radicals, and it would be perfectly impossible to derive the Latin sunt from the Greek eisi.

I need hardly say that the modern English, I am, thou art, he is, are only secondary modifications of the same primitive verb. We find in Gothic

[blocks in formation]

The Anglo-Saxon changes the 8 into r, thus giv

[blocks in formation]

By applying this test to all languages, the founders of comparative philology soon reduced the principal dialects of Europe and Asia to certain families, and they were able in each family to distinguish different branches, each consisting again of numerous dialects, both ancient and modern.

There are many languages, however, which as yet have not been reduced to families, and though there is no reason to doubt that some of them will hereafter be comprehended in a system of genealogical classification, it is right to guard from the beginning

against the common, but altogether gratuitous supposition, that the principle of genealogical classification must be applicable to all. Genealogical classification is no doubt the most perfect of all classifications, but there are but few branches of physical science in which it can be carried out, except very partially. In the science of language, genealogical classification must rest chiefly on the formal or grammatical elements, which, after they have been affected by phonetic change, can be kept up only by a continuous tradition. We know that French, Italian, Spanish, and Portuguese must be derived from a common source, because they share grammatical forms in common, which none of these dialects could have supplied from their own resources, and which have no meaning, or, so to say, no life, in any one of them. The termination of the imperfect ba in Spanish, va in Italian, by which canto, I sing, is changed into cantaba and cantava, has no separate existence, and no independent meaning in either of these modern dialects. It could not have been formed with the materials supplied by Spanish and Italian. It must have been handed down from an earlier generation in which this ba had a meaning. We trace it back to Latin bam, in cantabam, and here it can be proved that bam was originally an independent auxiliary verb, the same which exists in Sanskrit bhavami, and in the Anglo-Saxon beom, I am. Genealogical classification, therefore, applies properly only to decaying languages, to languages in which grammatical growth has been arrested, through the influence of literary cultivation; in which little new is added, everything old is retained as long as possible, and where what we call growth

« PreviousContinue »