are the most immutable peculiarities?' of which the former appears easier to answer than the latter. Each is here requisite, because we need to know whether it is possible in a later stage to lose the primitive elements themselves.

Even without making a real induction from the languages of savages (a difficult and unpleasant task), we can judge pretty well by considering the wants and habits of men who approximate to the savage state, what elements are most essential to all language. These are, among substantives, (1) The names of family relations, as father, mother, son, wife, &c., and the names of persons who constitute a household, as man, woman, boy, girl, &c. (2) The names of the principal elements and materials of nature, as water, fire, earth, &c., and of simple astronomical conceptions, as sun, moon, sky, day, night, &c. (3) The names of domesticated or well known animals. (4) Names of parts of the body, as head, foot, eye, hand; other sets of words are equally essential, referring to the food, habitations, and occupations of a people; but they vary too much with climate and situation to allow us to expect great agreement even in kindred languages. But next, (5) there are verbs which no language can want, as live, die, eat, sleep, see, hear, know, go, &c. (6) Simple adjectives, as great, small, good, bad, white, black. (7) A set of miscellaneous words, expressive of number and quantity, as many, few, much, often, &c. (8) Others, that are vaguely called pronouns; relative, demonstrative, &c. (9) The personal or proper pronouns. (10) Prepositions, and a few particles. (11) The cardinal numbers, one, two, &c.

No Englishman need be told that a system of declensions for nouns is not essential to language; but it is not wholly needless to insist that no verbal conjugation is essential. The lingua franca of the Mediterranean, and the Negro-English of the West Indies and other parts, would show (if the language of the nursery did not) how well men can make known simple wants and simple tales to one another, by using only the root of the verb. For this reason, it is no proof whatever that two languages have not a common origin, because the system of declensions and conjugations have nothing in common: this will only show that they separated at a very early period, and afterwards completed their grammatical systems. Indeed, it has been above remarked how striking is the diversity of Welsh and Irish in this respect, while no one has yet doubted their primitive identity. On the contrary, the more is found to be common in the nominal and verbal formations of kindred languages, the later must have been their separation; and it is this which proves the close relationship of the Indo-Germanic tribe, since, indeed, some of their earliest types, as the Gothic, Sanskrit, and Doric Greek, are far more variously developed as to the verbs and nouns, than are the

modern languages deduced from them; yet the earlier they are traced up, the stronger are their similarities.

It being granted that we have laid down correctly the primary elements of speech, can we infer that any two languages which are from the same source will have all these elements in common? Surely not; for these elements are not immutable. A comparison even of Greek and Latin (the connexion of which was manifest even to those most miserable of etymologists, the classical writers), shows us singular diversities, where we might not expect it. The names for the very primitive notions, man, woman, husband, wife, son, daughter, brother, sister, are wholly different in these two languages; so likewise for heaven, earth, fire, water, sea, mountain, tree, grass, body, soul, breast, hand, nose, mouth, face, &c. We have never seen a computation made, but we are disposed to think, that the two vocabularies have more opposition than agreement, as to the words most essential to speech. Yet their striking similarity (which is often identity), as to the inner structure of their grammar, shows that they did not separate till after the primitive tongue was fully formed.

We are not wandering from our main subject in seeking to account for this. The activity of the human mind in inventing words, is probably greater in a less cultivated state of society, as also is greater in the early period of life, than afterwards. Witness the readiness of schoolboys to invent and to adopt slang terms of every kind. They often love to distort and elongate, or on the other hand to clip, the old names, when they do not bring in what is wholly new. So, when a rude tribe spreads itself over a wide continent, its scattered families gradually add on fresh and fresh words, mere duplicates of the old ones, and wholly useless; nevertheless, in an unwritten language, they are soon all of equal dignity, and their respective antiquity becomes untraceable. In this state a language obtains that which some call great copiousness, but which we should name cumbrous verbosity. At one time the Arabic had eighty names for honey, and 200 for a lion; which are but indications of the monstrous confusion produced when a heap of dialects talked by scattered illiterate tribes is collectively designated a language. Literature and intercourse soon casts off the greater part of this load, and if the nation has one principal centre of government and religion, the later language seems to a casual observer to be merely less copious.' But when physical geography parts them into two or more nations, they make different selections out of their vocabulary; and either two dialects, or two languages, are produced, accord

*

It is a useful distinction of terms to which some scholars adhere, that when two parties have a different speech, yet are able to understand one

ing to the degree in which intercourse has been broken off and diversity generated.

The fact being certain, that in process of time, the vocabulary does change most unaccountably, many lexilogists have been led to the opinion that our principal clue, for judging of the relations of languages, is to be sought in their grammatical peculiarities, chiefly in the mode of conjugating the verbs. This is looked upon as the vital point, the heart and soul of the language. It is allowed that in the later stages of life, a great disorganization may take place, a breaking up of the curious fabric, and the substitution of rude and cumbrous auxiliaries (for this is a fact quite on the surface of history): but it is denied that any language can ever borrow from another its principles of conjugation. Hence it is alleged that there is no similarity so decisively proving common origin, as similarity of inflexions; and an unwillingness has arisen to admit that the analogies of mere vocabulary can suffice to constitute a claim to family relationship, unaccompanied by the far more decisive proof drawn from grammatical structure.

Great as is the reputation of the names which may be quoted on this side, we are not able to concur in the opinion. We admit that it is easier to introduce a foreign word than a foreign inflexion; but we think it is not easier to borrow a large apparatus of elementary words in numerous branches, than to borrow the principal inflexions of verbs. No clear historical example, we believe, can be adduced of either: hence it is impossible to reject the evidence for the affinity of languages, founded on an extensive agreement of their vocabulary, merely because they have not also agreement in grammatical formations. But beside this, we think it has been too summarily concluded by some of the ablest etymologists, that foreign inflexions cannot be at all ingrafted on another language. Their argument is based on the fact, that nations who have a literature of their own, and a grammatical structure fully developed for every practical purpose, do not thus borrow; just as nothing would induce the English to conjugate their verbs on the French model. But it does not follow that a nation wholly illiterate and possessing an exceedingly defective grammatical system, might not learn from more cultivated neighbours at once greater accuracy of thought, and a more accurate mode of expression. It is certain that new methods of conjugating and declining do not arise where there is a fixed literature, and do arise, from time to time, where there is not: in the latter case, it is too much to presume that no help can possibly be sought from the foreigner, should any contiguity or

another, they are said to speak different dialects of the same language; but when they cannot understand each other, they speak different languages.

intermixture of nations facilitate it. Rude tribes of mankind are often remarked to possess great propensity to imitation; a halfformed language has a ductility, which it were as vain to expect in such as have already attained a high development, as to demand the pliability of the tender scion from the aged and knotted bough. We have heard, on most trustworthy authority, cases of English children reared on the coast of Syria, who would introduce Arabic inflexions into their English speech; and we see no reason why a few families of illiterate English, settling on the same spot and wholly cut off from their countrymen, might not permanently ingraft such foreign forms on our noun and verb; much more would this be credible, if our language were hitherto exceedingly ill provided with a grammatical apparatus for expressing the various shades of thought.

For these reasons we are not disposed to allow any exclusive importance to similarities of inflexion; but we think the affinity of two proposed languages is to be judged of by their total amount of resemblance. If the likeness is not confined to a part of the vocabulary which is ordinarily imported (namely, to articles of commerce, religion, and literature), if a large fraction of each division of its elementary words is found to be common; this appears to us to be in itself so very strong a proof of a single origin, that if they agreed in not even a single inflexion of the grammar, we could not reject the evidence. So harsh a hypothesis as that the common portion of the vocabulary had been imported from the one to the other, after each language was in existence, could at least be justified only by some very strong necessity; some utter want of analogy, and not merely of likeness, in the grammatical systems; some preposterous inversions of principle, and repugnance in their first conceptions of grammar. On this head we say no more, because we do not know that such a case exists.

We have dwelt thus upon the general theory which must guide our judgments, because our immediate question eminently depends upon it. The learned author of the book before us, is the first person who ever set about to investigate the affinity of the Celtic to the Indo-Germanic languages, on just philosophical principles. Alike rejecting the enthusiastic vagueness of the Welsh antiquarians, and the too sceptical rigor of those who would allow to the Indo-Germanic languages no kinsfolk more distant than first cousins, he addressed himself to the problem in the spirit of the great continental etymologists; and the result of his investigation is, that the affinities of the Celtic are such as decidedly prove a common origin with the Indo-Germanic. His conclusion has been disputed by some, on the ground that he has not accounted for the diversities: an objection which might seem to imply a misunderstanding of the whole inquiry. We might

VOL. VIII.

D

ask in turn, whether the objector can account for the diversity of Greek and Latin: but, indeed, it is manifest that it must depend on causes wholly inscrutable to us.

Hitherto we have named only the Irish and the Welsh, as the two grand branches of the Celtic. Beside these, there is the Gaelic or Highland Scotch, a dialect of the Irish; the Cornish and the Armorican (or language of Brittany), both dialects of the Welsh. Lastly, the Manks (spoken in the Isle of Man), is nearer to the Irish; from which, however, it deviates rather considerably. In Dr. Prichard's vocabularies, the two principal languages are assumed as adequate for exhibiting all facts important to the case; and it is only occasionally that he adduces words from the subordinate dialects.

It appears from his tables, that under almost every elementary portion of the vocabulary, one or other of the Celtic languages has words identical with one or other of the Indo-Germanic; and (since the family affinity of each class with one another is universally conceded) this is as useful for the argument as if he had shown the similarity of Welsh to Latin. Nor is this all; but by thus widening the investigation, we effectually obviate the objection which the half-informed so confidently urge; that the words may have been imported into Welsh during the dominion of the Romans. Such objectors either do not know what are our facilities for guarding against error from this source, or give the author little credit for knowing how to use them. The facilities are many. First, when words are imported direct from Latin or Greek to Welsh (which has certainly been the case in ecclesiastical appellations, and probably in the names for gold, silver, and lead), there is comparatively little change of the consonants in passing from one language to another: whereas the native words are generally related to the Indo-Germanic (as those of Latin or Greek to Sanskrit) according to certain characteristic laws of mutation. Again, it has in recent years become more known to scholars, that the Erse is often decidedly more like to Latin than is the Welsh; yet the Irish were never conquered by the Romans. Indeed, when we consider how large a portion of Latin etymology is wholly dark to us, and how inadequate the language is to its own elucidation, it is natural to look to Celtic as an important help in explaining the non-Hellenic portion of the Latin. Farther, the Welsh is often like to the Greek or some other Indo-Germanic language, where it differs from the Latin. Of this we will give some examples:

An interesting paper, which, we believe, has not been published, was read, a few years back, before the Philological Society of London, on the relationship of the Highland Scotch to the Latin: which is very considerable.