In all cases, no doubt, were our means of inquiry equally abundant, we should be brought to the conclusion that the formation of derivatives is but the agglutination of significant syllables to significant syllables, or, in simpler language, of roots to roots. This assertion is founded on various considerations : first, the mere fact that in a large and rapidly Increasing number of examples what were unintelligible affixes have received their duo explanation ; secondly, the d priori argument, already employed to show that roots obtained their significancy not by arbitrary compact, but by natural association ; and, thirdly, the consideration that as affixes are attached to the end of words, and also sustain for the time an inferior office, they are doubly liable to a careless—in other words, shortened—pronunciation. On ship-board, for example, time is often of the first moment. Thus, when the writer once heard the French steersman of a French boat give orders to a " mousse" for dropping the foresail, In the words " Down with her," unconsciously speaking English, his inquiry as to the use of these words was met by the remark that " Faites-le-demeedre" would be too lung. No wonder, then, that sounds such as bosen and foxd should be preferred to boatswain and forecastle. Again, our own word wanton, at first sight of sufficiently obscure formation, has been analysed as repre senting wan-tow-en," ill brought-up," where man is the prefix seen in the old English wan-hope," main apes," and still abundantly flourishing in the Dutch language ; while tow-en is the old participle of the verb One, the equivalent of the German rich-en and Latin duc-ere ; so that wanton is substantially the same as the German ungezogen, "unedu cated, untrained." Our term workhouse is self-explained ; but this word is beginning to obscure the second clement when pronounced as workhres. Again, take our word MOP or ma'ani—who beforehand would have supposed that the final liquid in this word represented the Latin word domino, as is the fact : tam dernina, madonna, madame, ma'am, mum, and in one of Dickens'e novels, mini.
here we have seen the second part of the lei:it'd shorn of its fair proportions. The blow may however fall on the prefix. Thus, ava, of the Greek language, wheu used before verbs, was reduced in the iEolian and Doric dialects to av and av, and under special circum stances to a mere a or a. No wonder, then, that the Latin a-onose-ere exhibits the same prefix as a mere vowel, or that the A. S. ou-enaw an, and Shaksperian a-eknow (part. acknown, ' iii. 3), as well as our existing a-eknow-ledge, have taken similar liberties. A still more violent corruption of what might be proved to be the same pronoun is seen in our c-lopc, which represents the German ent-lanfen, Dutch oat-loop-en. Cases oven occur where the original root is wholly absorbed in some derived form. Thus, as Bopp has noticed, tho German im, corrupted from in-de-m, has preserved the preposition and the case-ending, but wholly lost that to which they were at first attached, the element de (our the). Another example of similar extreme violence has been seen in I'd for I mould, the essential syl lable tee/ of the verb having utterly disappeared. As compression, then, is a very common occurrence in language, it behoves an inquirer not hastily to assume the radical character of all monosyllabic verbs. If lie find two consonants either commencing or terminating the syl lable, the strong probability in that a really disyllabic word is before him. Thus, know, pluck (rk is but one consonant), cam, hulk, should be regarded as corruptions of sumo such form as ken-we or con-mr, pulatek, cal-nm, hal-ark or hut-oek, and no only secondary formations from the simple irh or con, pull, coal, hull. Nay, in many cases where the iunitial consonant is a simple r, /, or a, a truncation of the commencing letters may be suspected. For example, the Latin noac.rre is histori cally known to be a corruption of gnose-ere, where the first syllable represents our know, end thus gee or gun, corresponding to our ken or con, is the base. The related Greek °roue has preserved the radical vowel, and only lust the guttural. So, again, when our laugh (Germ. forh.eln) is placed beside the Greek -yeaa-srv, -yelax-cra, (yelatw, Dori() wo see that the base of the words is gel. For an example of a deceitful r, we may quote our verb ran, which in the Dorset dialect in hint, and here the n is but the remnant of a common suffix, as seen in op-cn,heark-en,barn, mourn, turn. The base, then, is him, corresponding to the Latin cur of cum, the interchange of 4 and c being the law between these languages.
In the formation of secondary words, by such agglutinations as wo have spoken of, due attention must be paid to those very natural lawn which require an adaptation of discordant elements, a subject treated in all its integrity in Sanskrit grammars, under tho title Sandhi, "com position." The main principles are the saute for all languages. Thus, when medial or thick coimonauta, so to say, are brought into contact with thin consonants, the one or other must change its character. For example, in Greek you may prefer the thin consonant and say inra, OKTW, or you may prefer the thick and say et3Souos, o7Soos ; but adapta tion, at any rate for the voice, is required. So again labials and den tals refuse to coalesce. A German says lends, 'loins,' a Roman said Iambi ; in Greek we have or for an lEolian au-Sev-etv," to wet for the first time ;" in Latin im-eu-ere is preferred. But the must important, yet most neglected, law of assimilation in the composition of words, is that which concerns the vowels. No doubt the Sanscrit grammarian dwells at sufficient length, on the effect produced by the con tact of two vowels, as when he tells us that the meeting of a final a with an initial i leads to the substitution of the intermediate vowel e, and similarly that the junction of a with u produces the intermediate vowel o. But a far more material point is the adaptation of vowels,
more or less, when they are separated by consonants. In all dealings with vowels, whether in this Sanscrit habit of combining a and i into e, or a and u into o, or in the larger question now opened, the main guide of the philologer is the vowel law, established by the experiments of Professor Willis, namely,—that the kind of vowel-sound produced by the vibration of a reed, depends solely on the length of the tube that intervenes between the reed and the point of opening into the outer air ; and that, as the reed is lengthened, we have a succession of sounds in the same invariable order, i, e, a, o, u, as pronounced by an Italian, not that the sound jumps from one of them to the next, but passes by imperceptible degrees through other sounds intermediate between them, which our alphabet is utterly unequal to denote, and indeed no alphabet could in perfection. But for the full comprehen sion of the question, the paper, or rather two papers, should be read in extenso, as given in the Transactions of the Cambridge Philosophical Society for Nov. 24, 1828, and March I6, 1829. Now as the character of the vowel depends, for the human voice, solely on the distance between the chords vocales in the larynx, which act as the musical reed, and the margin of the lips, and Is otherwise in no way dependent on the form taken by the organs of speech ; and as on the other hand the character of the consonants depends solely on the form and relative position for the time of these organs, the length of the vocal tube being immaterial, it follows that it is au easier matter to produce the sound of a polysyllable,all the vowels of which are the same or similar ; and in this, as in all the labours of man, the saving of trouble is a paramount conside ration: Hence, in savage nations, the abundant supply of such geogra phical terms as Mississippi, Tennessee, Alabama, Oronoco, in America, or of Kamtchatka, Japan, Palawan, Sarawak, Sambawa, Samarang, Balambangon, Banca, Jara, Malacca, Andaman, Madagascar, Loochoo, Siou-siou, Sooloo, Comoro, &c., in Asia. Hence too in the formation of secondary words there is a constant tendency to assimilate the vowels of united syllables, when they happen to be discordant. This may be effected in several ways, by the adaptation of the first vowel to that which follows or the converse, or thirdly by an approach of one or both to some intermediate sound. It is commonly by a modification of the first syllable that the object is attained in the German, Scandi navian, and Keltic languages, and so far as the principle prevails in Greek and Latin, all of which are members of the great Indo-European family. On the other hand, in the languages of Tartary, Turkey, and Hungary, as well as those in northern Europe and Asia, spoken by the Finns, Lapps, &c., it is the suffixed syllable that is compelled to take a vowel more or less similar to the vowel of the preceding syllable. I fence,.in Languages of this class, for all that have been just enumerated belong to one family, we find suffixes to a great extent running in pairs, which, with a common power, have no other difference in form than the interchange of a strong and weak voweL Thus, in Turkish, kalpak, a cap, has nom. pl. kalpak-lar : but er, a house, has a nom. pl. fe-ler ; and again the datives pl. of these nouns are respectively kalpak lar-ah and Erler-eh. Or, to take an example from the Hungarian, the verbs rdr, " wait," and ismer, " know," form the following persons :— Nay to such an extent Is this law of assimilation carried out in the Mongolian, that It is turned to account in economizing the number of alphabetical characters. As the first occurring vowel decides the character of those that follow, a common symbol is used in all syllables after the first for a and e, a second common symbol for o and 6, But in the other division of languagee, as was just said, it is the first syllable that adapts itself to those which follow. If we look to the German languages, the familiar modification called " umlaut " is for the moat part made in the direction of exchanging strong for weaker vowels. Thus a, o, a, if followed by a syllable containing either i or e, are apt to become 8, ti respectively, where the two dots represent, and indeed have grown out of, an e, and ae, oe, ue, severally denote sounds weaker than those which they displace. One of the most interesting cases of the umlaut to be found in German,. is seen in the second and third persona of many so-called irregular verbs, as sehlaken "to sleep," du schlafest, cr schllift; "to push," du Missed, er ddsd. But it may be asked why the same modification is not found in the other persons, ich schlafe, wir schlafen, &c., and the answer is only to be found in the formation of the old German, where the suffixes of the several persons are : 1.—u, 2.—is, 8.—it ; p1. 1.—anies ; 2.—at, 3.—ant; so that the only persons which by the weak vowel of their suffixes were originally entitled to the influence, are precisely those for which it is claimed. But it is not merely the a, o, u, which are subject to the influence of the umlaut ; to the same cause must be attributed the modification which occurs in the second and third persons of brechen, " to break," whence clu briehst, er bricla ; essen," to eat," dm issest, er iest ; geschehen," to happen," es geschieht.