It would be a mistake to suppose that the analogies of language are always uniform: there may be often a choice between several, and sometimes one and sometimes another will prevail. In Greek there are three declensions of nouns; the forms of cases in one of them may intrude upon another. Similarly verbs in -omega and -mu iota interchange forms of tenses, and the completed paradigm of the verb is often made up of both. The same nouns may be partly declinable and partly indeclinable, and in some of their cases may have fallen out of use. Here are rules with exceptions; they are not however really exceptions, but contain in themselves indications of other rules. Many of these interruptions or variations of analogy occur in pronouns or in the verb of existence of which the forms were too common and therefore too deeply imbedded in language entirely to drop out. The same verbs in the same meaning may sometimes take one case, sometimes another. The participle may also have the character of an adjective, the adverb either of an adjective or of a preposition. These exceptions are as regular as the rules, but the causes of them are seldom known to us.
Language, like the animal and vegetable worlds, is everywhere intersected by the lines of analogy. Like number from which it seems to be derived, the principle of analogy opens the eyes of men to discern the similarities and differences of things, and their relations to one another. At first these are such as lie on the surface only; after a time they are seen by men to reach farther down into the nature of things. Gradually in language they arrange themselves into a sort of imperfect system; groups of personal and case endings are placed side by side. The fertility of language produces many more than are wanted; and the superfluous ones are utilized by the assignment to them of new meanings. The vacuity and the superfluity are thus partially compensated by each other. It must be remembered that in all the languages which have a literature, certainly in Sanskrit, Greek, Latin, we are not at the beginning but almost at the end of the linguistic process; we have reached a time when the verb and the noun are nearly perfected, though in no language did they completely perfect themselves, because for some unknown reason the motive powers of languages seem to have ceased when they were on the eve of completion: they became fixed or crystallized in an imperfect form either from the influence of writing and literature, or because no further differentiation of them was required for the intelligibility of language. So not without admixture and confusion and displacement and contamination of sounds and the meanings of words, a lower stage of language passes into a higher. Thus far we can see and no further. When we ask the reason why this principle of analogy prevails in all the vast domain of language, there is no answer to the question; or no other answer but this, that there are innumerable ways in which, like number, analogy permeates, not only language, but the whole world, both visible and intellectual. We know from experience that it does not (a) arise from any conscious act of reflection that the accusative of a Latin noun in 'us' should end in 'um;' nor (b) from any necessity of being understood,—much less articulation would suffice for this; nor (c) from greater convenience or expressiveness of particular sounds. Such notions were certainly far enough away from the mind of primitive man. We may speak of a latent instinct, of a survival of the fittest, easiest, most euphonic, most economical of breath, in the case of one of two competing sounds; but these expressions do not add anything to our knowledge. We may try to grasp the infinity of language either under the figure of a limitless plain divided into countries and districts by natural boundaries, or of a vast river eternally flowing whose origin is concealed from us; we may apprehend partially the laws by which speech is regulated: but we do not know, and we seem as if we should never know, any more than in the parallel case of the origin of species, how vocal sounds received life and grew, and in the form of languages came to be distributed over the earth.
iii. Next in order to analogy in the formation of language or even prior to it comes the principle of onomatopea, which is itself a kind of analogy or similarity of sound and meaning. In by far the greater number of words it has become disguised and has disappeared; but in no stage of language is it entirely lost. It belongs chiefly to early language, in which words were few; and its influence grew less and less as time went on. To the ear which had a sense of harmony it became a barbarism which disturbed the flow and equilibrium of discourse; it was an excrescence which had to be cut out, a survival which needed to be got rid of, because it was out of keeping with the rest. It remained for the most part only as a formative principle, which used words and letters not as crude imitations of other natural sounds, but as symbols of ideas which were naturally associated with them. It received in another way a new character; it affected not so much single words, as larger portions of human speech. It regulated the juxtaposition of sounds and the cadence of sentences. It was the music, not of song, but of speech, in prose as well as verse. The old onomatopea of primitive language was refined into an onomatopea of a higher kind, in which it is no longer true to say that a particular sound corresponds to a motion or action of man or beast or movement of nature, but that in all the higher uses of language the sound is the echo of the sense, especially in poetry, in which beauty and expressiveness are given to human thoughts by the harmonious composition of the words, syllables, letters, accents, quantities, rhythms, rhymes, varieties and contrasts of all sorts. The poet with his 'Break, break, break' or his e pasin nekuessi kataphthimenoisin anassein or his 'longius ex altoque sinum trahit,' can produce a far finer music than any crude imitations of things or actions in sound, although a letter or two having this imitative power may be a lesser element of beauty in such passages. The same subtle sensibility, which adapts the word to the thing, adapts the sentence or cadence to the general meaning or spirit of the passage. This is the higher onomatopea which has banished the cruder sort as unworthy to have a place in great languages and literatures.
We can see clearly enough that letters or collocations of letters do by various degrees of strength or weakness, length or shortness, emphasis or pitch, become the natural expressions of the finer parts of human feeling or thought. And not only so, but letters themselves have a significance; as Plato observes that the letter rho accent is expressive of motion, the letters delta and tau of binding and rest, the letter lambda of smoothness, nu of inwardness, the letter eta of length, the letter omicron of roundness. These were often combined so as to form composite notions, as for example in tromos (trembling), trachus (rugged), thrauein (crush), krouein (strike), thruptein (break), pumbein (whirl),—in all which words we notice a parallel composition of sounds in their English equivalents. Plato also remarks, as we remark, that the onomatopoetic principle is far from prevailing uniformly, and further that no explanation of language consistently corresponds with any system of philosophy, however great may be the light which language throws upon the nature of the mind. Both in Greek and English we find groups of words such as string, swing, sling, spring, sting, which are parallel to one another and may be said to derive their vocal effect partly from contrast of letters, but in which it is impossible to assign a precise amount of meaning to each of the expressive and onomatopoetic letters. A few of them are directly imitative, as for example the omega in oon, which represents the round form of the egg by the figure of the mouth: or bronte (thunder), in which the fulness of the sound of the word corresponds to the thing signified by it; or bombos (buzzing), of which the first syllable, as in its English equivalent, has the meaning of a deep sound. We may observe also (as we see in the case of the poor stammerer) that speech has the co-operation of the whole body and may be often assisted or half expressed by gesticulation. A sound or word is not the work of the vocal organs only; nearly the whole of the upper part of the human frame, including head, chest, lungs, have a share in creating it; and it may be accompanied by a movement of the eyes, nose, fingers, hands, feet which contributes to the effect of it.
The principle of onomatopea has fallen into discredit, partly because it has been supposed to imply an actual manufacture of words out of syllables and letters, like a piece of joiner's work,—a theory of language which is more and more refuted by facts, and more and more going out of fashion with philologians; and partly also because the traces of onomatopea in separate words become almost obliterated in the course of ages. The poet of language cannot put in and pull out letters, as a painter might insert or blot out a shade of colour to give effect to his picture. It would be ridiculous for him to alter any received form of a word in order to render it more expressive of