nations, express plurality by means of the words plurality, totality, or the like. Instead of the symbolical way in which past time is expressed by reduplication in Sanscrit and Greek and MaesoGothic, it is expressed metaphorically in the Yarura tongue by a word ri, which denotes distance.*

In languages which express formal relations by means of substantial words, it is plain that there can be no such thing as inflexion; the word which ought to be subordinate is too weighty and inflexible to admit of its being so treated by the tongue as to form a real affix to the word which it is to define. The meanings of both words stand equally independent before the mind, therefore they cannot be fused into that inseparable unity which constitutes the essence of inflexion. The mouth cannot amalgamate what the mind keeps apart. The formation is at best a compound word. But there is an ingenious process to produce unity of word used in the Tataric stock, especially in its nobler languages, belonging, as far as we know, to them alone, which, though Humboldt does not mention it, may be worthy of our attention. We mean the law of vocal harmony. The vowels of the formal syllables must, according to this law, harmonize with those belonging to the substantial part of the word. In this manner, the formal parts are subordinated to the latter and more important part, which is the more necessary, as in these languages, (for instance in the Turkish verb,) there may be a long string of formal syllables tending to obscure the root.t

But we must discuss more fully the subject of inflexion, and speak of the alteration which the word undergoes in the highest class of languages, in order that it may come under a particular category, or be fitted to take its place in a sentence. It may be altered in two ways: by inward change and external addition. Where internal change is practicable, and is even, as in the case of the Semitic languages, favoured by the structure of the words, the distinction between the indication of the category and the designation of the object is easily secured. The substantive identity of the word is preserved whilst its form is changed; just as in the conception the substance is unchanged, whilst the formal relation is added. On the other hand, addition from without is a kind of composition, but with this difference the simplicity of the word is not to be impaired, two conceptions are not to be combined into a third-one is to be viewed in a specific relation. Therefore, that part of the word

* It is scarcely necessary to mention that the representation of time under the form of space is very general.

Thus in Turkish aghâ= lord, has aghâlar in the plural; er = man, er-ler; in Magyarish, kert garden, kert-esz-nek, to the gardener. In Finnish, teräs or teräkse = steel, teräksetä, from the steel.

which indicates the relation is to be put on a different line, so to speak, from that which designates the conception; its clearness of sound is not to be impaired, but we are to be made to understand that its significance has departed and its independence is lost. Composition tends to preserve the integrity of the syllables, that their significance may not be lost; inflexion tends to destroy the significance of the part added. We are not, then, to regard inflexion as mechanical, as a conjunction of what was in itself separate, and a subsequent obliteration of the marks of union. The inflexion grows out of the word, and is as much one with it as the various parts of an opening bud. However plainly the pronouns may be recognised as forming the inflexional parts of the persons of the verb, they are not merely attached to it. The verb was presented to the mind in its individual forms, and each of these forms burst through the lips one and indivisible.

"There is this difference," says Humboldt, "between inflexion by inward change and by the suffix, that the former is always, whether or not we can enter into the feeling, symbolical; the latter has, in most cases, at some time, had an independent meaning.' The suffixes, we would add, may be symbolical too, and indeed contain a symbolism deeper than that which lies in the analogy of sounds. If the researches of Bopp and others have left it unquestionable that not only in Sanscrit, but also in other languages, the suffixed syllables are more or less derived from those roots which refer immediately to the speakerthe so-called subjective roots-then their symbolical meaning rests upon this very fact. The categories of thought and speech cannot be more fitly pointed out than by sounds which refer exclusively to the thinking speaker. The sounds, however, in some of these suffixes may be symbolical. Bopp has well remarked, that in the pronoun of the third person the clear sound of s is assigned to the living person-the dull sound of m to the neuter; the same sounds being likewise used to distinguish severally the active subject, (the nominative case,) and the passive object of the action, (the accusative.)

Humboldt points out well how a highly-organized language, such as the Sanscrit, from which he draws his examples, secures two things of great importance in speech-the unity of words in themselves, and their suitable distinction as definite members of a sentence. These two different, though nearly connected objects, the Sanscrit accomplishes by using different means, and by so doing distinguishes the objects themselves.

* On the original meanings of the Sanscrit, &c., suffixes, see some excellent observations in Dr. M. Müller's Paper in the Report of the British Association for 1847

Unity of Word and Unity of Sentence.

219

Letters at the end of words, and those in the interior of words, are, in fact, treated on different principles. By the first mode the word is kept separate from the other words, and made an individual member of the sentence; by the second it is knit together in itself-i.e., the inward unity of the signs of conception and relation is secured. Hence, though the Sanscrit marks the conjunction of the words more closely than any other language, there is no risk of the unity of the separate words being sacrificed the analysis is as complete as the synthesis. But, further, it points out the different degrees of unity in words by a different treatment of their component parts. No less than five degrees are marked. Compound words belong to the lowest class; the grammatical forms of declension and conjugation form the highest. The compound words are treated more according to the rules just alluded to, which hold in the case of separate words-the three highest classes by the rules which determine combinations of letters in a word. Besides the changes which take place in consonants when they come together, changes still more important for the unity of the word take place in the vowels. All the cases in which such changes occur are treated with the greatest attention to logical consistency and euphony. Thus, in order to maintain the root in its integrity, its vowel is not assimilated to the suffix-it is expanded or altered in a way not qualitative but quantitative; so that the radical vowel is easily traced, because it is amplified according to rule.* The difference between rude natural sounds and graduated tones is more apparent still in what is called reduplication, a process which tends greatly to ensure unity of word. The repetition of the first syllable of a word to denote its increased importance is variously used amongst uncivilized nations, to mark plurality, frequency, &c.; but in Sanscrit this is managed with such delicacy and variety that five or six forms of it might be mentioned.

We must now turn to the consideration of the various means which are in use in different nations, to secure the unity of the sentence. "This is a higher unity," Humboldt remarks, "not only because it embraces more, but because it is more intellectual, more independent of the element of sound." The origin of the sentence itself gives the reason for this unity. We may be sure that man from the first associated a proposition with every word he uttered.

* Thus, though the vowel is amplified not in one degree, but in two, (in what are called Guna and Wriddhi,) the original vowel is readily detected, and to the ear the effect seems to proceed from the depth of the original syllable, and conveys all the feeling of a development. This raising the vowel to a higher power, (to use a mathematical expression,) is only in one case symbolical in Guna, viz., in that of intensive verbs. Wriddhi, which is related to Guna as superlative to comparative, is more frequently so.

Every name implies a judgment on the object designated; and in the early stages of language, the reason why a particular name was given was probably manifest in the term itself. It expressed qualitative existence though not in a developed form. The next step would be the division of this whole into its parts by an act of analysis. But the perception of this, as a whole, would not be lost; the object produces a total impression upon the mind, and the sentence must have unity. Now this unity may be either that of singleness, as that of the word in its undeveloped state, or that of an aggregate, (i.e., a mechanical unity, implying an imperfect distinction of parts,) or that higher unity which presupposes the recognition of difference, the unity of the body and its members, all distinct, but all designed for each other, all co-operating. Such an unity as the last is the sentence of the inflecting languages, which is formed out of words, bearing mostly in themselves the marks of their relation to the sentence. In the Chinese language the parts are distinct enough, (monosyllables, without the slightest tendency to coalesce,) but there are no signs of their being intended for each other beyond their position, and the unity of the sentence is feebly marked, just as the unity of magnetized iron-filings is not so high as that of a body. The unity of the Mexican sentence is the inorganic unity of an aggregate; in order to mark the unity of the whole, which cannot be perceived from the adaptation of the parts, it heaps it up into a single word. The Mexican language, however, rightly considers the verb to be the centre of the sentence, and annexes to it the governed and governing parts, giving to this combination by vocal contrivances the stamp of a connected whole. Thus, ni-naca-qua means, I flesh eat. But how does this differ from such a word as xκрewpayéw? In this, that the Mexican is bent upon giving the scheme of a whole sentence in his word; and therefore, if the governed substantive is not incorporated, he inserts the pronoun of the third person. The substantive is placed afterwards, as we should say, in apposition -thus, ni-c-qua in nacatl, I it eat, the flesh. Even if no definite object is added, an indefinite is inserted, having a double form for persons and things, ni-te-tla-maca, I somebody something give. This language is interesting, because in its construction of the sentence it bears the type of an early stage of development. It is like a recollection of the time when men represented their meanings in a single word, but were beginning to feel that more definite expressions were necessary, and there

That the same word may either be substantive, adjective, or verb, is plain from the Chinese.

We select this illustration from a recollection of an expression of Bunsen's in his linguistic dissertation before quoted.

Synthetic Power of a Language.

221

fore added these, whilst they still clung to the original unit of expression, and put it first. Some languages, much like these in character, do not, however, go so far as to let the verb absorb the noun, but allow it to take this liberty with the pronouns both governing and governed. As the appetite may grow with what it feeds on, so here; for when this method has acquired full sway in a language, several pronominal objects are admitted into the conjugation of the verb. This is the case in some of the North American languages, and remarkably so in the singular language of that mysterious tribe which occupies its stronghold. at the very western extremity of Europe.* In all these so-called incorporating languages, the connexion has suffered from want of distinction of the parts. But though they are content to embrace their impressions as a whole, " because the different points of contact between the object and the feelings are not clearly represented to the mind, they shew not only ingenuity and freshness, but also a right view of relation." The personal pronouns are very prominent, and they make use of them symbolically to express the most general relations. This speaks plainly of an early but genuine mode of representation, of a stage in man's education when he was learning to understand things, by seeing how they were related to his own being. It is only at a more advanced period of culture that men learn to restrict this reference to cases in which it is necessary.†

But, after all, that which constitutes the chief characteristic of the noblest class of languages, and gives to the Sanscritic stock that permanence and fruitfulness which has distinguished it, is what Humboldt calls their synthetic power. It is this, employed on other materials, which enables the artist to wed together mind and matter in the painting or the statue, the lack of this attribute of genius being marked rather by the feebleness of the total impression than by defect in details. In language this power is manifested, in the formation of the word, in giving the impress of mind to the outward sound; but its importance is still more apparent, though its action is better felt than described, in the higher act of forming the sentence. It is well to have the parts clearly distinguished; it is well to have them shaped so that their correspondence may be at once perceived; but the principal thing is to knit them together into a whole. The presentiment of this organic whole is indeed the condition of the right construction of the parts.

* There are in the Basque no less than 206 conjugations arising from the cause above mentioned.

The Malay tongues scarcely belong to the incorporating languages, but bear a certain resemblance to them, as they heap a number of significant prefixes upon the verb, in order to supply the lack of inflexions and mark the course of the

sentence.