QVAERITVS^J^- THE SANSKRIT LANGUAGE, THE BASIS OF LINGUISTIC SCIENCE, ND THE LABOURS OF THE GERMAN SCHOOL m THAT FIELD— ARE THEY NOT OVERVALUED? BY T. HEWITT KEY, M.A., F.R.S., PROFESSOR OP COMPARATIVi; GRAMMAR IN UNIVERSITY COLLEGE, LONDON. BERLIN: A. ASHER (fc CO LONDON: D. NUTT. 1863. QVAERITVR. THE SANSKRIT LANGUAGE, AS THE BASIS OF LINGUISTIC SCIENCE, AND THE LABOURS OF THE GERMAN SCHOOL IN THAT FIELD— ARE THEY NOT OVERVALUED? BY T. HEWITT KEY, M,A., F.R.S., PROFESSOR OF COMPARATIVE GRAMMAR IN UNIVERSITY COLLEGE. LONDON. BERLIN : A. A S H E R k CO LONDON: D. NUTT. 1863^ QVAERITVR THE SANSKRIT LANGUAGE, AS THE BASIS OF LINGUISTIC SCIENCE, AND THE LABOURS OF THE GERMAN SCHOOL IN THAT FIELD — ARE THEY NOT OVERVALUED ?» T. HEWITT KEY, M.A., F.R.S., PBOFESSOR OF COMPABA.TIVE GEASIMAR IN DKIVERSITY COLLEGE , LONDON. I had thought at one time of placing at the head of this paper: "Doubts of a Non-Sanskritist". But on reflection it seemed desirable that the title should be more definite. If the words I have actually used, be thought by any one to savour of national ill-will, I must give the assurance that nothing could be more remote from my purpose or from my feelings. Those who have to deal with the classical languages, must be either blind or ungrateful, if they fail to acknowledge the deepest obligations to the Scholars of Germany. The editions of Greek and Latin authors that have appeared in England during the last half-century, have not been numerous, but even of these a large proportion * This paper, which grew out of an Introductory Lecture delivered at University College Nov. the 24*'^ 1862, -was read at the Philological Society of London January the 2^^ and 16*^; and was subsequently the subject of discussion at a special meeting on the 20^^, 1 2 OUR DEBT TO GERMAN SCHOLARS. have been simply reprints of German works. Again the Lexicons of the two languages that have issued from the English press during the same period, are for the most part so thoroughly of German material, that it would have been more creditable, if the title-pages had carried the words: "Translated from the German of with some few changes and additions". Again, if we turn our thoughts to the op- posite side of the English Channel, we find no great activity in the sphere of Classical, especially Greek Literature; yet what progress is visible there, is chiefly due to the energy of German, not French Scholarship, as witness the valuable collection of Greek Authors that has proceeded from the press of Didot. Nay, the high and indisputable reputation that Germany has won in this field, only renders the duty more imperative to watch lest failure or shortcomings on any side shoidd be kept from notice owing to that very prestige. Further I wish it to be observed that the term I have used, is 'overvalued', which is quite competible with an admission of great value; and again, I put what I have said in the form, not of a proposition, but of a question. It is only when that question is answered in the affirmative, or when the arguments put forward in this paper remain imanswered, that any the slightest damage can be done to the reputation of the philologers concerned. It would have been simply indecent, if the present writer had expressed his fears in the form of a direct proposition, conscious as he is that he comes to the enquiry wholly destitute of what may at first sight be deemed an essential requisite, a know- ledge of the Sanskrit language. Nay, he cannot pretend even to that smattering which may be obtained by a three weeks study of the language, and which has before now served to float a big book in the English market, a little sprinkling of the Devanagari character and a judicious use of the hard words 'Vriddhi, Anuswara, &c.', passing for profundity in the eyes of the uninitiated. Such little know- ledge as I have is that only which may be acquired in the perusal of grammars and glossaries and works of like nature. THE VEDAS. 3 The question here naturally suggests itself, how it is that I have taken upon myself to enter into a contest for which I am confessedly so ill-equipped; and my answer is that I find the same suspicions which have found a way into my own mind entertained by many others, and those too gentle- men whose position as scholars gives great weight to their opinions, though, like myself, they are wholly wanting in the special qualification, a knowledge of Sanskrit. In every point of view then it seems desirable that the question should be raised. If our fears are ill-founded, it is well that they should be removed, and the road more thoroughly cleared of aU obstruction for the Sanskritist. If otherwise, it is surely good for the progress of philological science, that the matter should be thoroughly sifted. I do not purpose to enter into the domain of Sanskrit history and chronology, a task for which I am wholly un- fitted, especially as those who have the best qualifications, admit that it is involved in the greatest obscurity, nor in- deed could one expect easily to find materials for accurate investigation in such a Literature as that of the Yedas. The 'Mantras' on the one hand, dealing for the most part with 'the Devotional', and the 'Brahmanas' on the other with 'the Ceremonial and Dogmatic', can scarcely be available for such a purpose. As to the Upanishads or the short ap- pended treatises I will be satisfied with a second-hand quo- tation from a work of a learned Hindu, that they "contain "some rude indications of philosophic thought, and like the "twinkling of the stars in a dark night may occasionally "serve as guides in a history of Hindu philosophy. They "do not however exhibit any great attempt at method, ar- "rangement, classification, or argument. Even there the "poetry predominates over the logic. Bold ideas abruptly "strike your fancy, but you find no clue to the associations "which called them forth in the author's mind, and search "in vain for the reasons on which they are based. Sublime "thoughts are not wanting, but they resemble sudden flashes, "at which you may gaze for a moment, but are immediately "after left in deeper darkness than ever. Nor are they free 1* 4 ANTIQUITY OF SANSKRIT. "from those irregular flights of the imagination in which "poets ^yith vitiated tastes delight to indulge, setting at "defiance all rules of decency and morality" (Banergea, Westminster Review, New Series, Vol. xxii, p. 463). An argument for the antiquity of the Sanskrit Language has recently been founded (Lectures on the Science of Lan- guage by Max Miiller, p. 204, third edition) upon certain passages in the Book of Kings and the Book of Job, but it is an argument which, as it appears to me, withers to the touch. All rests upon the statement that four articles im- ported from Judea in the days of Solomon, viz. the ape, the peacock, ivory, and sandal-wood, are called in the Hebrew text by names foreign to that language, but in- digenous in Sanskrit. But it is not an easy matter to prove that a word is indigenous in a language, and the Sanskrit- speaking race on their first entrance into the Indian pen- insula (for they are allowed on all hands to have been im- migi-ants) would naturally adopt the native, that is Non- Sanskrit terms for those objects which are peculiar to the country, provided indeed they had not already adopted them in the previous intercourse of commerce. Bat passing over this consideration, let us throw a glance at each of the four words, on which this important superstructure has been erected. Koph, the Hebrew for 'ape' is, we are told, "with- out an etymology in the Semitic languages, but nearly identical in sound with the Sanskrit hapV^ It is of course implied here, though not said, that the Sanskrit does supply a satisfactory etymology for its kapi. To supply the omission I turn to Bopp's Glossary, and there find that kapi 'ape' has for its root the Sansk. vb. kamp 'tremble', so that for some reason denied to us the ape was conceived by the Indian mind as 'the trembler'. Then ivory has for one of its Hebrew names shen hahbirn^ where as shen means 'tooth', habbim might well speak of the 'elephant', and this, it is said, "is most likely a corruption of the Sanskrit for ele- phant ihha, preceded by the Semitic article." If, as I sup- pose is the fact, Ma be a misprint for ibha, the resemblance is even then limited to the consonant, and we have nothing ANTIQUITY OF SANSKRIT. 5 offered in the way of proof that this name for the elephant is the original property of Sanskrit. Thirdly tukhi-im^ in Hebrew 'peacocks', bears no doubt a tolerably close re- semblance to the Malabar name togei; and this "in turn has been derived from the Sanskrit sikhin 'furnished with a crest'.". Lastly the Malabar and Sanskrit name for sandal- wood is valguka; and "this valgu(kay\ the Professor says, "is clearly the name which Jewish and Phoenician merchants corrupted into algum^ and which in Hebrew was still further changed into almugr I would submit that at any rate the word 'clearly' is somewhat out of place in an etymon which involves four assumptions, the aphaeresis of v, the apocope of ka^ a paragogic m, and the metathesis of gum to mug. Even if true, such derivations have scarcely strength enough to serve as the foundation of so large a theory ^ But the same writer has elsewhere (History of Ancient Sanskrit Literature, p. 524) contended that the Vedas have an antiquity far older than the knowledge of writing. "The collection of the (Vaidic) hymns and the immense mass of the Brahmana literature were preserved", he says, "by means of oral tradition only." In another passage of the same work (p. 507) he tells us that "before the time of Panini, nay even when he himself wrote (sic) his great work, writing for literary purposes was absolutely unknown." To understand the full force of this proposition, to form an adequate idea of the extent to which the Professor would tax the mnemonic powers of the Brahmans, we must re- member that Panini, according to his own authority, was preceded by whole generations of Grammarians. In his recent Lectures on Language (p. 110) he says: "Those valuable lists of words, irregular or in any other way re- markable, the Ganas, supplied that solid basis on which ^ I leave this as I wrote it, but I have since found that Prof. Max Miiller has borrowed the who^e argument from Lassen's 'Indische Alter- thumskunde, Vol. i, p. 538 (fee.', so that the setting alone is his own. It is true that he himself refers to this passage of Lassen; but his re- ference is so placed that a reader might well suppose the argument about 'ivory' alone to have been drawn from Lassen, ^ WRITING UNKNOWN TO PANINI? successive generations of scholars erected the astounding structure that reached its perfection in the Grammar of Panini.*' But if the structure be "astounding" and "the per- fection of a merely empirical analysis of language", it seems not to be possessed of much that would be interesting to the mere European scholar, for the Professor concludes his panegyric with the words: "Yet of the real nature and natural growth of language, it teaches us nothing." As regards the Yedas themselves, one can readily imagine that religious feeling and poetical feeling combined may do much to invigorate the powers of memory, while the mere rhythm of verse contributes to lighten the task; but intense indeed must have been the feeling of duty which could in- duce Brahmans to commit to memory and there retain a complete library of the driest Grammarians. The whole argument then carries with it, as it seems to me, its own refutation; and in truth the challenge implied in the words: "I maintain that there is not a single word in Panini's terminology which presupposes the existence of writing" — has already received a twofold answer from my colleague, Professor Goldstucker (Panini, his place in Sans- krit Literature, 1861); first a self-refutation, quoted from the Oxford Professor's own words: "This last word Upikara (a writer or engraver) is an important word, for it is the only word in the Sutras of Panini which can be legitimately adduced to prove that Panini was acquainted with the art of writing;" and as my colleague observes (p. 17): "It is obviously immaterial whether another similar word be dis- coverable in his Grammar or not; one word is clearly suf- ficient to establish the fact." But he further produces from Panini's own work an abundant supply of terms which could have no meaning whatever when writing was unknown. Let me quote one more passage from the same admirable book (p. 14): "As according to his, Max Miiller's, view, Panini lived in the middle of the fourth century B. C. (pp. 245, 301 if.), it would follow that, according to him, India was not yet in possession of the most useful of arts at the time when Plato died and Aristotle flourished." MIMETIC ORIGIN OF LANGUAGE. 7 I have entered into these details to show the unsatis- factory condition of the chronology of Sanskrit Literature, and at the same time I would suggest the question whether there should not be a little more caution in the acceptance of literary conclusions even from those to whom the English public has been accustomed to look as authorities above all controversy. But if we cannot have the advantage of a reliance on Literary history, we must be content to examine the in- ternal evidence supplied by the language itself and the dealings therewith alike of Indian and European authorities. As my own doubts, and I believe those of the friends to whom I have already alluded, were first raised by what appeared to us as most strange though generally sanctioned etymologies, I will proceed to produce some of these, limit- ing myself for the most part to a single class. Already Max Miiller (Lectures on the Science of Lan- guage, p. 369) himself quotes as an example of Indian etymology the derivation of the sb. kdka 'crow' from apa- hdlayitavya , i. e. 'a bird that is to be driven away', but adds that Yaska, another Grammarian, anterior to Panini, considered kdka to be an imitation of the bird's note. Whether the Professor himself adopts or rejects this mimetic origin of kdka^ his words do not enable one to say. But be this as it may be, in another Sanskrit noun, kdrava^ Lat. cor(o)vo- ' raven', he steadily refuses to see, what for one I must regard as a still more exact imitation of the bird's note, viz. cor cor. Had he included in his view the Greek y.oQ-aa- \ he might perhaps have assented to Pott's doctrine (E. F. ii, 506, 7) that crx in Greek substantives is a suflix of diminu- tival power, so that yioQ alone would be the root. He him- self, in his aversion to what he calls by way of disparage- ment the Bow- Wow theory, strives to deduce the whole family, kdrava, y.oQtov)], raven, (fee, from the Sanskrit verb rw, to which he ascribes "a general predicative power" as * We must not suppose the ancients in their nomenclature to have distinguished with modern accuracy the raven, the rook, and the crow, (See Mr. Wedgwood's paper on that subject.) 8 THE 'go' family. expressing sound, "from the harshest to the softest", and so applicable "to the nightingale as well as to the raven", nay even to "the barking of dogs" and "the mooing of cows". In a note however, he hesitates between this etymon and one from the Sansk. Mm 'singer'. To the special honour of this last derivation the raven seems to be about as well entitled as the parrot or the peacock; and the de- duction of kdrava from ru, a general term implying 'sound', would probably be regarded by lawyers as 'void for un- certainty'. The same objection of excessive generality applies to the whole class of etyma with which I now propose to deal, viz. those of words ascribed to roots of various forms, but with the one meaning 'to go'. Thus the S. go {gav)^ the equivalent in power and probably in form of Lat. hov-, Gr. /9oi;-, as also of our own cow^ is deduced by Sanskritists of all classes, Indian and European, from a S. vb. ga 'go' ; and that this explanation of the word may not suffer for want of company, I may add the S. ild 'cow', referred by Bopp (Gloss. ^ s. V.) to the vb. il 'go'. Now that animals like the 'hare' or 'stag' should receive a name from their marked power of locomotion, is, at any rate on the logical side, admissible, and thus we may perhaps be ready to assent to the current etymologies of hai^e (Germ, hase) the Lat. lepos- and the Gr. eXacpo-, But the cow is scarcely entitled to put in a claim for such distinctions as against any other living creature. Strangely enough the same pair of words, go and ila, also signify 'earth', and these also have the same origin ascribed to them (Bopp, Gl. s. vv.). So also the Gr. yaia passes with Bopp (V. G. § 123) as standing for yaFia, and so an adjectival offspring of a sb. cor- responding to the S. go 'earth' and eventually of such a vb. as ga 'go'. In the same section S. gmd 'a name for the earth in the Veda-dialect' is deduced from the S. vb. gam 'go'. Nay our own earthy though it comes immediately from ^ I have preferred to draw from this work, although now somewhat out of date and superseded by other works, simply because it comes from the founder of the science. THE ^GO' FAMILY. 9 our old English vb. ear 'plough', represented in Sanskrit by avj is traced ultimately to the S. r 'go' (Bopp, ibid., M. Mitller, Lectures on Language, p. 256, and Pott, E. F. i, 218). It would be an interesting fact, if such a series of at any rate consistent etymologies could be accepted as proofs that the Hindu mind had already discovered the motion of the earth, whether about its own axis or about the sun. But as it seems more probable that then as now there existed an inveterate tendency to treat the earth as the one fixed object to which all the movements around us are conveniently referred, we must look for some other ex- planation of the theory; and accordingly Bopp suggests that the movement of the earth must here be regarded as only 'passive', in other words, the earth (erde) is 'the betrodden one' ('die betretene'). Though it does not visibly move itself, man and beast would be in an awkward predicament for locomotion if there were no earth to move upon. Before leaving the earth I ought to notice that Prof. M. Miiller believes (p. 257) our word aroma to be another ramification of ar 'plough' and r 'go', for does not Jacob say (Gen. 28, 27) : "the smell of my son is as the smell of a field which the Lord has blessed"? From land I pass to water, but the same etymology pur- sues us. Thus the Ganges itself is the Sanskrit Gangd^ literally the "Go Go" (M. Miiller, ibid. p. 384). So we have S. salila 'water' from S. vb. sal 'go' (Bopp, Gloss, s. v.); S. ap 'water', the analogue of the Lat. aqu-a, from vb. ap 'go', sarit 'river' from vb. sr 'go' (M. M. ibid. p. 253) and saras 'water' from the same vb. (ibid.). This last noun is by Bopp translated by the Lat. lacus and declared to be one with the Gr. hlog (cr. form Heg-) 'a marsh', in which case the notion of 'going' seems to disappear. Yet after all Sanskritists may contend that marshland, being half water, half land, has a double claim to a derivation from roots which denote 'going'. Take next the class of worms and reptiles. Bopp, for example (V. G. § 86, 1), refers the Lat. vermi- (== quermi-') and S. krimi to the S. vb. hram 'go'; the Germ, schlange 10 THE 'go* family. 'snake' to S. vb. smng 'go'; and the Lat. serpem^ S. sarpa^ tirst to S. vb. srp 'go' and ultimately to sr 'go'. Had any of these verbs meant 'go by little and little', the derivation would have been satisfactory, but the meanings given by Bopp in his Glossary to these verbs as vs^ell as to all their compounds furnish no authority for such an assumption. No doubt in his comparison of the vb. srp with kindred languages he dwells much on the idea of slow movement in those kindred languages, and again Prof. Wilson in his grammar, though he adds the meanings 'creep or glide', gives precedence to the general term 'to go'. I cannot but think however that the suffix of sip as compared to sr and that of the Lat. ser{e)p- with the varieties of EQ{e)n' and L. r-ep- {repere)^ Eng. c{e)r-ep {crep-t^ creep) ^ Germ. kr-iech-en^ and with an additional suffix of diminution cr-aw-l, represent the idea paulatim, as it seems to do in the Lat. car-p)- (see my paper on the suffix agh &c.. Trans, for 1856, p. 336). When I wrote what is there seen, I expressly stated that I w^as at a loss for the root ser 'go'. This I now find in the S. sr, although I still believe a form ker^ as seen in creep and kriech-en^ to be more genuine than sr with the sibilant. I am not deterred from regarding the two roots as substantially one by the fact that as a rule the guttural k or c of western Europe is usually represented by the palatal 5 of Sanskrit, not by that which occurs in sr and sip. Another class of words, which Bopp is disposed generally to refer to roots significant of simple movement, are those which denote time (da iiberhaupt die Zeitbenennungen moi- stens von Wurzeln der Bewegung stammen, V. G. § 69). For instance our word year^ old Germ, yar, together with what he regards as an equivalent in form, the Gr. coQa 'season', is referred by him to the S. vb. ya 'go', but by Lassen and Burnouf it seems to S. vb. tr 'go'. Again the Goth, aics (crude form aiva) as well as its relatives Lat. aevum, Gr. ctuov , is deduced by Bopp, Graff, and Kuhn from S. vb. i 'go' (ibid.). The i might perhaps not have passed with the ill-informed as forming the kernel of these THE 'go' family. 11 words; but all is accounted for; the initial a, it seems, at- tains its position through 'Guna', and all that follows the i (or e) is to be regarded as a suffix. In spite of such a combination of authority I am still disposed to prefer my own etymology of aevicm from the Lat. aiige- (for the vowel- change compare the variety seen in the allied ae^-co and av^-av-to) with 'growth' for the original sense, as exhibited in the well-known line, Crescit occulto uelut arbor aeuo, 'grows like a tree with growth concealed'. The Lat. saeculum is also referred by some Sanskritist to a vb. signifying 'to go', but as I have mislaid my authority, I will proceed to an- other Latin noun. As Bopp (Gloss, s. v.) considers S. amati 'time' a derivative from am 'go', so he is also inclined to deduce from the same stem the Lat. annus (as standing for amnus) as well as the Gr. evog. That amniis was in fact the older form of annus is proved alike by the derived sol-emni-s and by the Oscan or Umbrian amno-; but to the derivation from a vb. 'to go' I would oppose that other de- rivation which connects it with the Lat. prep, am 'round', German um. The very idea of a year implies a circle, and the words annulus 'a ring' and the noun anus with a long vowel, seem to complete the proof. On the same principle the word year itself, like i/ar-d 'an enclosure' and gar-d-en &c., claims kindred with many words denoting a circle, as XOQ-TO, hor-to-, ZOQ-0-, cor-o-na-^ cir-co-. The initial change between a Gr. /, Lat. A, and a g (y) in German and English is in accordance with the usual law, as seen in xO^ag, heri^ hesterno-, gestern, yesterday. So much for the alleged deduction of substantives from Sanskrit verbs signifying 'to go'. But in the formation of secondary verbs also the roots i 'go' and yd 'go' are thought by Bopp well fitted to play important parts, as for example in furnishing suffixes by which verbs are converted into passives (§ 739) and causals (§ 740). As regards the former, if hord ydi^ to take Bopp's own example from the Bengali, have for a literal translation 'I am made' ('ich werde ge- macht'), as given by himself, then gemacht is by itself already a passive, just as verloren is in the modern German 12 THE 'go' family. cjehen verloren^ literally Ho go lost'. We too may say 'be- come detested' or 'become fascinating', where the distinction between the passive and the active idea turns upon the ac- companying participle, not upon the word 'become'. Again Bopp's illustration from the Latin amatum iri is surely not applicable. If the principle, for which he is contending, be valid, we ought already to have a passive in the in- dicatival phrase amatum eo 'I am going to love', but this is a mere future of the active. The introduction of a pas- sive of eo, whether in the indicative as amatum itur or in the infinitive as amatum in, is only a convenient mode of exhibiting an impersonal verb, equivalent to the French o?i va aimer. The examples of veneo and pereo, quoted by Bopp, are at first sight more to the purpose, and he would have done well to strengthen his case by comparing them with vendo and perdo. Yet after all venire, standing for venum ire, means probably 'to go into the window' and so 'be exhibited for sale' which certainly is more truly the meaning of the phrase than 'to be sold'. Again perire 'to come to an end', like the English go to the dogs or the Greek eqqe eg xoQaxag, contains no doubt what is virtually a passive idea, but this arises from the combination with the ;:>^r and the eg xoQaxag &c. That 'go' does not carry in itself the idea of a passive, is clear from our own phrases 'go to the Bar' or 'into the Church' or 'into business'. Curtius (Beytrage p. 329) goes still farther and conjectures that the O^r] which appears in the aorist and future of Greek passives, is connected with the S. vb. yd 'go', in which however all resemblance seems limited to the long vowel. I pass then from the passive. The causal mood of the Sanskrit verb, as well as the tenth conjugation in general, having for their distinguishing character the syllable ai/\ Bopp's mind is divided by a doubt whether this suffix should be referred to the vb. i 'go' or t 'wish'. The latter one would think is far better fitted for the formation of a desiderative mood, which, it seems, is > Of causals some make ay the suffix, some ya. THE 'go' family. 13 a general appendage to the Sanskrit verb. Nor does i 'go' at first sight appear a satisfactory element for the purpose of constituting a causal verb; but we are assured by Bopp (§ 740) that several Sanskrit words which denote 'motion' at the same time denote 'making'. Whether the particular verb i has this convenient privilege, he does not stop to tell us. Assuming however that it has, we have before us a strange combination, that roots expressive of 'going' are alike fitted to form passives and to form actives. But further although the causative idea is declared to be the character of the tenth conjugation, I find little proof of this in the list of 57 verbs quoted by Prof. Wilson in his Grammar, for of all these at the utmost one in five can be explained as containing the idea of 'to make'. Thus the first ten in the series are translated by the English verbs 'steal, disrespect, hurt, send, wink, speak, play, be feeble, be able, sound'. I am not then surprised to find in § 772 such a sentence as "It deserves however notice that in Sanskrit denominative verbs in ya * occasionally avail them- selves of the causal form without any causal meaning." My own feeling is that the original notion paulatim resides in «?/, and that it is the Sanskrit variety of that suffix which 1 have discussed at length in my paper on agh or ag^ the passage of a ^ between vowels {aydmi) into a y being a common occuiTence. On this theory the meaning may well pass into that of frequentative or continuous. But leaving this question open, if we accept that one of Bopp's two explanations which finds in the suffix of the so-called Sans- krit causals or tenth conjugation the root i 'go', we shall have to assign to this use of the word a somewhat vast domain in the Classical and German Languages, for Bopp connects with the same type all the vowel-verbs of the Latin, at any rate the first, second, and fourth conjugations of that language (§ 745 c.), all the Greek verbs in eco, aw, 00), a^co, Ltio (§§ 109 a. b., 749, and 762), together with the particular verbs /?aAAw; OTellco, takho, and ir^iit', and lastly ' See the note on the preceding page. /7 u all the weak verbs of the German stock (§109). A few of these verbs specially noticed by Bopp himself may claim a few words. We are assured that the Latin facio = S. Inw'dydmi^ literally 'I make to be'; iado — ydp-dydmi, 'I make to go' ; doceo = c'/hdp-dydmi 'I make to know' ; Q^apio = rdp-dydmi 'I make to give' (§ 747). It seems somewhat damaging to this theory that the suffixes (i or e) of the Latin contribute but little to the formation of the causative idea, seeing that fac-^ iac-^ doc-^ rap- already express the full notion of 'making, throwing, teaching, robbing' ; as may be seen in the ioxm^ fac-ere, iac-ere^ rap-ere^ and mfac-tus, iac-tus^ doc-tns, rap-tus. Yd-p-dydmi is thought to possess a second suffix of causation in its ^, so that yd 'go' is the real base of the verb ; and if this case be doubtful, a causal 77 is declared with greater certainty to be an element in (jhd-p-dydmi 'I make to know', ghd (or in English characters jnd) being what Bopp is pleased to call a root-verb, the equivalent of our hiow. But of this jnd more hereafter. To place Bopp's doctrine clearly before me, I throw aside the equivalent portions eo and dydmi., and there results the equation Lat. doc = S.J7idp. The palatal/ of the Sanskrit is with reason assumed to be a corruption of a medial guttural (/ or /. Then doc is to be proved equal to yndp. I make no difficulty about the final consonants, for a Lat. c habitually corresponds to a S. p. But there still remain three problems for solution, to identify the d with ^, the short with the long a, and to account for the appearance of n in the Sanskrit or its disappearance from the Latin. For the first Bopp simply quotes the instance J?]'jiit]T7](} == n^-^o^T,.^; on the difference of vowel he says nothing. The difficulty as to the nasal is disposed of by the assurance that for ghd-nd-mi 'I know' there occurs an actual gd- nd-mi^ and that in Persian there exists the form dd-ne-m * I know'. But surely the asserted loss of an n from gnd- nd-mi, when followed so closely by a second n, is but a poor justification for the disappearance of an n in doc for dnoc. For one then I must regard the doc of doceo as better explained within the limits of the Classical languages by THE 'go' family. 15 dec of delco (= dlco) and dstx-vufii^ by the day. of di-da^x)- Gxco, St-dax-Tog, and dctx-zvlog, by the die of di-dic-i and fZi]^ of dig-itus. But if 1 must look to the Sanskrit, here too I find a thoroughly admissible representative in the yb. dis 'show' with that palatal s which regularly corresponds to a Western Z:-sound; and indeed Bopp himself I find, in his Glossary, regards this root dis as one with the root of dtixvvf.u and the Lat. dico\ The 2> of ^'«i^- or rapi-, as also that of the S. rdp-aydmi^ is again treated by Bopp as of causal power, and he finds in his root rd 'give' only a variety of da 'give'. Thus 'to give' and 'to cause to give or rob' owe their marked dif- ference of meaning to the causal suffix, not that this is an essential matter with him, for this same root dd or rd is thought by him to be identical with the S. vb. 7a, to which simple form is ascribed the double meaning of 'to give' and 'to take', a mixture of ideas that might lead to inconvenient results ^ So much for the value to the Sanskritist of his roots signifying 'to go' in the way of etymology; and the stock is no small one. Taking of ihQ ten conjugations the first * As some friends well acquainted with Sanskrit could scarcely believe that a writer like Bopp could have published such 'extravagancies', I will quote his very words (§ 747): "Kami ich aber das c der genannten Form {facio) nicht mit dem skr. causalen p vermitteln, so glaube ich doch dem Lateinischen noch ein anderes Causale nachweisen zu konneu, worin c die Stelle eines skr. p vertritt, namlich doceo, welches ich im Sinne von ich niache wis sen auffasse und fur verwandt mit disco (eigentlich ich wiinsche zu wissen) und dem gr. itSdrjv, didctauM halte. 1st das d dieser Formen aus g entstanden (vgl. Jrjuijji^o aus rrjurjTTjn) , so fiihrfc doceo zum skr. gndp-dyd-mi, ich mache wissen (Ja-na-mi ich weiss fiir gna-nd-mi) und zum pers. dd-ne-m ich weiss. Als ein Beispiel eines lat. Causale, worin das urspriingliche p unver- andert geblieben ware, erwiese sich rapio, im Fall es dem skr. rdpdydmi ich mache geben entspricht, von der Wz. "^ rd geben, die, wie mir scheint, nichts anders als eine Schwachung von dd ist. Auch kommt, sowie neben dd eine erweiterte Form dds besteht, neben rd im Veda- Dialekt rds vor. Mit rd und dd scheint auch ihrem Ursprunge nach die Wz. Id identisch, welcher die Bedeutungen geben und nehmen zuge- schrieben werden." 16 THE 'go' family. alone, and again limiting myself to the series which Professor AVilson quotes in his grammar as 'the most useful verbs of this conjugation', I find just twenty, viz. 1. '^S[^^aj 'to go'; 2. "^ at 'to go' ; 3. -^ 2 'to go' ; 4, ig du 'to go' ; 5. "^^^ nkh 'to go'; 6. "^ r 'to go', 'to gain'; 7. "^paf ;:;' 'to be straight' or 'honest', 'to gain', 'to go', 'to live' ; 8. -gpj^^ kram 'to go', 'to walk'; 9. im^ gam 'to go'; 10. f^^? vichchh 'to go'; 11. ^ char 'to go'; 12. -^^^ dhauk^'to go'; 13. Tf^pat 'to go', 'to fall'; 14. -gj^ sad 'to wither' or 'decay'; 'to go', with this appended: When the verb means 'to go' the causal retains the final, -^jjirafTy sddayati 'he causes to go', or 'drives'; 15. ^ sad 'to decay', 'to be sad', 'to go'; 16. i^[^sas] 'to go';^ 17. ^\{^sidh 'to go'; 18. ^ 5f 'to go'; 19. ^srp 'to go', 'to creep' or 'glide'; 20. -^if^ skand 'to go' ''or 'approach'. 1 should have made some ai dition to this list, had I included those verbs which only express a more special or limited form of motion, as 'per- vade, jump, hasten, run, gallop, approach, wander'. With such an abundance of verbs to draw from, a philo- loger should the more hold himself bound to proceed with caution, and so take care that the logical connection between the root and the supposed derivative should be well-marked. Whether the examples I have quoted exhibit such caution, 1 leave to others to decide. Lastly I think it right to re- peat that by confining myself almost wholly to those in- stances of bold etymology which deal with verbs signifying 'to go', I avoid the charge of selecting instances favourable to my view. Indeed without some such limitation, it would be an easy matter to pick holes in any of the most care- fully elaborated philological works, for the most cautious etymologer is apt to be carried away at times by tempting theories. In the next section of my paper I purpose more particularly to consider Bopp's celebrated work, the 'Ver- gleichende Grammatik', in its general system. In the short discussion which followed the reading of the above, it was replied on one side that the idea of 'to go' was precisely that which was well adapted to denote an THE 'go' family. 17 active verb. To this I answer that a vb. 'to go' was equally claimed for the special formation of passives; but in truth the argument seems to me upset by its very generality. What is fitted to denote every form of action, is for that reason unfitted to denote any form of action. The very essence of language is distinction or difference. Accordingly the other answer to the difficulties I had raised was that although simple 'going' is commonly assigned as the mean- ing of the verbs I have quoted, yet in truth each of them originally denoted some special form of going. I will only reply to this that I took the verbs with the meaning at- tached to them by the several authorities from whom I was quoting. But over and above this, when the discussion was brought to the individual substantives, I found that the Sanskrit scholars who were present, employed in the defence of the Indian etymologies a vagueness as complete as that expressed in the general term 'going'. Thus go and ild 'the cow', and go and ild 'the earth' were said to be well entitled to such derivation, as being in the Indian mind the centres of activity most important to man. I take the opportunity of making a slight addition to the paper. As sr^ according to Wilson's Grammar (p. 200), at times signifies 'to go quickly' or 'run', I am the more justi- fied in attributing to the suffixed p of srp the power of paulatim. At any rate it has no causal power here. Further if the Sanskrit vocabulary could deduce from a verb signi- fying 'to run' by the addition of this suffix a secondary verb srp 'to creep', I am justified in connecting our own cr-ep (whence creep and crep-t)^ as regards its root, with the base of the Dorsetshire hir-n = A. Sax. yrn-an 'to run', and that base, /»V, corresponds of course to the Lat. cur- of curro. Again if the S. vb. sal 'go' is one with the vb. sr 'go', we have the analogue of this sal in the Greek alXo(-iai and Lat. salio, whence sal-tu-s 'a sheep or cattle rmi\ I am the more inclined to attach some value to this con- jecture, because as fal of fallere 'to cause to fall' seems to furnish the only root for fors fortis, so does saU for sors sortis 'that which leaps from the urn' (situla), a noun, from 2 18 TIIR VERGLEICHENDE GRAMMATIK. which has come the verb sortiri of tlie Latin and the verb mrtir (with a very different power, more akin to the original root) of the French. Lastly let me observe that if the Sanskritists had been contented to derive sarit 'a river' from a root sr 'go' or rather 'run', there could have been little objection, our owni terms 'current' and 'watercourse', Bull's 'Kun', and 'runlet' exhibiting a similar origin. Such terms as saras 'marsh or marshland' and ap 'water' have not the same justification. SECOND PART. It would be to shrink from the task I have undertaken, were I not to take into special consideration the great work of Bopp, who appears with something like general consent to be entitled the founder of Comparative Grammar as a science : and the claim upon my attention is only the stronger, that his 'Vergleichende Grammatik', the first portion of which was published in 1833, has been recently reprinted with some changes and considerable additions (1857-60). Here, as in wdiat I have already said, I shall without further apology for my temerity proceed to state unre- servedly the objections that have presented themselves to my mind, not expecting those objections to be accepted as valid, but desirous that they may attract the notice of scholars whose more intimate acquaintance with the subject will enable them to detect any errors I may have committed. The contest is happily one in which the victorious and the defeated must alike be gainers, the one object of both parties beiug to promote the cultivation of the science of language. First of all then 1 find in the very title of the commencing chaptei- (Schrift- und Laut-System) what appears to me un- philosophical, viz. the precedence given to writing over sound. Over a large portion of our globe there exist whole races WILLIS'S EXPERIMENTS NEGLECTED. 19 possessed of the faculty of speech, but without any know- ledge of written symbols; and indeed no small part of the population even of this country is in this position. But I should have passed over this matter, if the error, so to call it, had not told unfavourably on the arguments that follow. The very first paragraph in the chapter gives to tlfree of the vowels a special character, which, as it appears to me, is not due to them. Thus the title of original vowels (Urvocale) is assigned to «, i, tr, and this, I believe, on no other ground than that the Sanskrit alphabet had special characters for these, w^hen the sounds of e and o may have been denoted by combinations of the first three, much as the French language employs its diphthongs ai and au as simple vowels. Had the school of philology founded by Bopp looked upon the materials for oral language as belonging to the domain of physical science, and w^hoUy independent of those other forms of language which are addressed to the eye, such an error could not have occurred. In particular I must repeat the regret, which I already gave expression to in the year 1852 (Proceedings, Vol. 5, p. 192), that the valuable paper on Vowel-sounds which was read by Professor Willis before the Cambridge Philosophical Society (Nov. 28, 1828 and March 16, 1829), seems to have been wholly unnoticed by the leading Scholars of Germany. At any rate I for one ' have never yet come across the slightest allusion to this paper or to the principles established in it in any Gennan writer, while on the other hand I have read much from this quarter that would never have been written by any one acquainted with the results of Mr. Willis's experi- ^ I find I have not done justice to German scholars in this remark. In Dr. Bindseil's Abhaudlungen ziir allgemeinen vergleichenden Sprach- lehre (Hamburg 1838) p. 84 reference is made to Professor Willis's paper, and from Ihe appended note I learn that the paper itself was reproduced in the German language in Poggendorffs Annalen der Physik und Chemie. Still Dr. Bindseil himself seems to have been satisfied with a bare re- ference, making little or no use of the principle, nor does his work appear to have met with much notice among his countrymen. It stopped ab- ruptly with the first volume, although this contains only a general in- troduction and a treatise on gender. 2* 20 FORMATION OF VOWEL-SOUND. ments. Nay I do not recollect to have seen in any of their prominent works in the field of philology any reference to that physiological organ which may literally be called the pri/mim mobile of human speech, I mean the two chordae vocales. Now that Professor Czermak of Prague by his simple apparatus has enabled the enquirer to witness the action of these musical strings in the living man, we m^y hope that the study of oral language may be placed on its proper basis. It will then be laid down as the first dogma that as vowel-sounds constitute the substance of language (for brevity I drop the word 'oral', which is the only form here under consideration), so the character of any vowel depends almost wholly on the distance for the time Ijetweeu the chordae vocales and the margin of the lips, in other words on the length of the vocal pipe, the position of the tongue being of no moment so long as it does not close the passage of air. So thoroughly definite and mathematical is the character of the physical experiments, on which Pro- fessor Willis's results are founded, that he has given numerical values to the distances that belong to such of the vowels as are most familiar to English ears. At the same time as the number of points in a line is infinite, so the vowel- sounds pass by imperceptible gradations from the one ex- treme i (the sound in feet) to the other extreme u (or oo in boot). Thus it is wholly owing to the imperfection, yet necessary imperfection of alphabets, that there is but a limited set of symbols for vowel-sound. The number itself is essentially infinite; and it was therefore a subject of amusement as well as regret to hear some few years ago that a conclave of learned philologers was then sitting in London to determine, among other high matters, what was the full number of vowels. But the vowel-order 2, e, a, o, u (with the sounds which prevail on the continent), as i-esulting from Professor Willis's experiments, would have supplied the German philologers with a principle capable of solving pretty well all the problems that arise in connection with the "Vocalismus', not merely of the Indo-European family, but of language ASSIMILATION OF VOWELS. 21 in general. In the paper already referred to (Vol. V, pp. 191-204) I have shown in some detail that it explains the umlaut and ricck-uinlaut so-called of German philology, the formation of plurals in English &c. by what Grimm calls 'motion', that is an alteration of the root-vowel, as in geese from goose ^ and generally the assimilation of ad- joining vowels so familiar in all the Tatar languages and prevalent to a considerable extent in the Keltic, Teutonic, and Classical languages, to say nothing of others. In p. 203 of the paper, I gave from my colleague. Professor Maiden, a tabular view, showing the full development of the principle in the changes of Greek vowels and diphthongs. And I have little doubt that the mysterious Guiia and Vriddhi of Sanskrit are simply results of the same law. No doubt Bopp has allusions to the principle of vowel- assimilation, but these are altogether incidental. Thus it is only when he passes from the Sanskrit (§§41, 42) to deal with the Zend, that he notices some cases where the presence of a i/, i^ or e aifects the vowel of an adjoining syllable, and in § 46 mention is made of a similar euphonic influence belonging to a Zend v (w). But these are matters which should not be treated as peculiarities of the Zend. The philologer is bound to state the law of vowel-assimi- lation in its broad simplicity. But there is another point in which Sanskritists seem to have been misled by the habit of looking at language in its written aspect. They ascribe to the Sanskrit, in ac- cordance no doubt with Indian authority, two vowels, r and /;•, which at any rate do not present themselves in the vowel-series of the Cambridge Professor. Moreover it is admitted that this vowel /• is closely related to the ordinary liquid r. May I propose as the probable solution of the whole difficulty the following? It is well known that the two liquids r and I often lead to the disappearance of an adjoining vowel, most persons would say to a metathesis of the vowel, a doctrine which I hold to arise from an inaccurate view of the matter, though this for the present is not important. Our own thorough^ for example, appears 22 SANSKRIT ALPFIABET. in the two shapes through Eng. and durch Germ. Again in our provinces the form hrid is at times used, where the prevalent language prefers bird; so lyretty and perty coexist. The J.atin too has trifc- and toni-o-^ and the Greek O^Qaoog and xh(()U()g, with but little distinction of meaning and no distinction of origin. In such cases it is convenient to have a notation which will readily adapt itself to the two varieties of pronunciation; and on this principle it would not have been imwise to employ such a form as bi^d, 'prty, to re- present at once bird and brid, jjerty and pretty. The Slavic languages are not less given to such varieties than others; and accordingly words without any represented vow^el occur in the Bohemian vocabulary, as krt 'mole', krk 'neck', bib 'blockhead', wlk 'wolf. Yet Dobrowsky does not on this account class r and I with the vowels of the language. Possibly the habit of virtually dropping the letters r and /, as in the case of bird in the mouth of a Londoner (bbd)., and talk^ calm generally, as w^ell as the Fr. meilleur, may have had its counterpart in India, and so have lent some encouragement to the doctrine that they are vow^els. But to return to the ordinary vowels, if a language is limited to three symbols for their representation, it is a matter of course that a should have a first preference, be- cause lying in the middle of the series it is for that very reason the easiest to pronounce and consequently the mos't common; and after a the vowels i and u have the next claim, as occupying the two extremities. It has also been urged that the Sanskrit alphabet has a special claim to our consideration in its philosophic com- pleteness. But this claim is open to grave doubt, seeing that it appears to have been without any character for the sound, if indeed it possessed the sound itself, that is heard in the initial consonants of our English thin and thine, fat, vat, m the two consonants of the Fr. juge and the final of the German cinfach. On the other hand it appears super- Huously rich in its ten asperates distributed through the so-called gutturals, palatals, cerebrals, dentals, and labials; that IS, if our informants be right in pronouncing these BOPP ON ROOTS. 23 asperates as we pronounce the italic consonants of blocHouse, lo^/iouse, coa. 2, Anm., p. 239), so that the change of ken to ket is no difficulty for Bopp; and I confirm this by the parallel case of the Latin pronominal forms cit-ra^ cit-ro^ cit-erior^ cit-imus. I am further indebted to Bopp for a knowledge of three other analogues of my verb, quita or kita of the Philippines , the New-Zealand kitea, and Malagash Mta^ words also signifying 'to see' and identified by himself with the Sanskrit ket (§ 87. 2). Thus the area of the verb is as extensive as that of the pronoun. On the side of form then there remains nothing to desire; and as to meaning I would ask whether any idea could be in better keeping with pronominal de- monstratives than that of 'see', 'look'. The very word "demonstrative" which Prof. Miiller selects for his definition, suggests this interpretation; and he himself adds that their office is "to point" and so determine "locality". It would be more correct to say that it belongs to the finger to point and to the voice only to call attention to the finger's direction by uttering the word 'look'. It is with this feeling that the French has formed its void and voila^ and even in such a phrase as Terence's Luciscit hoc iam (Haut. Ill, 1) the full expression of the pronoun requires some such translation as: 'It is getting light, look, already'. When we ourselves utter the word this or that, we do little more than invite the person addressed to direct his eye to some object at which we are pointing, so that in real power these words arc equivalent to an imperative 'look'. No doubt the mind is not at once reconciled to the identification of a verb with an adjective, mucli less to the declension of a verb as though it were an adjective. Yet if the Latin ecce 'be- hohr IS a verb, and few will venture to deny it, we have a perfect parallel in such phrases as eccwn me, eccos video mccdcre patreni ct macjistnim, as used by Plautus. For the MULLER ON ORIGIN OF LANGUAGE. 27 full details of my argument I must of course refer to the paper itself. But whether my theory be right or wrong, I trust I have said enough to show that Prof. Muller's broad denial of the possibility of finding a suitable 'predicative' root is untenable. On the other hand let us look at the general theory of roots, whether 'verbal' or 'pronominal' as put forward by the German School. Bopp indeed puts aside for the most part the question of the origin of words as not falling within the scope of his work, but Prof. Miiller speaks somewhat more definitely on this subject. Yet his views, I think, will not be found satisfactory to others, and seem not altogether satisfactory to himself, for after touching on the topic at the beginning of his book, he practically postpones the question to his last chapter, pp. 349-399, and even then he nearly reaches the end of the chapter before he comes to the point. It is only in page 391 that he says: "And now I am afraid that I have but a few minutes left to explain the last question of all in our science — How can sound ex- press thought?" I find another reason for doubting wiiether he is a firm believer in his own theory. The said chapter begins with an admirable extract from a work of Dugald Stewart's w^hich spurns with contempt "that indolent philo- sophy which refers to a miracle w^hatever appearances both in the material and moral worlds it is unable to explain." I say then that when Max Miiller transcribed these w^ords, he had not yet given a thoroughly cordial assent to the view of language with which the chapter ends, for he him- self in his distress practically summons to his aid the cleus ex machina^ first telling us (p. 392) that "man in his primi- tive and perfect state possessed the faculty of giving ex- pression to the rational conceptions of the mind," and then adding that "that faculty was an instinct, an instinct of the mind as irresistible as any other instinct." Further in a note he says : "The faculty peculiar to man in his primitive state, by which every impression from without received its vocal expression from within, must be accepted as an ulti- mate fact." For myself I can only look upon this last 28 MULLER ON ORIGIN OF LANGUAGE. passage as a simple admission that he has no solution of the problem to offer, while the preceding assumption, that language is the result of instinct, seems to savour of that indolent ])hilosophy which the Scotch philosopher is quoted to condemn. Again the assertion that language first came into play, when man was "in his primitive and perfect state", seems hardly consistent with the tone of the first lecture, in which he led his hearers to anticipate a very different conclusion. That lecture begins with a justification of the phrase "Science of Language", and then refers the origin of every one of our sciences to the agency of man as stimulated by his "wants" when society was yet semi- barbarous or half-savage; and his argument further implies that all the sciences, including of course that of language, were things of gradual growth, beginning in what was humble and lowly. All this is surely at variance with his later theory that "the 400 or 500 roots" which are "the constituent elements" of language, are "phonetic types pro- duced by a power inherent in human nature", and "exist, as Plato would say, by nature; though with Plato we should add that when we say by nature, we mean by the hand of God." One cannot but think that such explanations must have been intended for the class of people, so w^ell described by Prof. Miiller himself (p. 364), those "who prefer the im- intelligible which they can admire to the intelligible which they can only understand." But probably the real inter- pretation of these inconsistencies is to refer them to some antagonism in his own mind, as between the principles which were nurtured in him while yet on German ground and the influence of an Oxford atmosphere. Yet after all there is an advantage in the publication of a self-contradictory book. The WTiter is as it were twice- armed. Let an opponent attack a position in the book at the same time putting forward his own view, and the author has his reply in a AVell but have not I myself said so? The advantage is parallel to what one so often finds in Giccro's letters. Habitually passing to and fro in his anti- cipation of the future, according as he was in an over- ORIGIN OF LANGUAGE. 29 sanguine or too desponding mood, the wavering statesman was able in the issue of events, no matter what that issue might have been, to boast of his wonderful foresight and say "Didn't I tell you so?" Or there may be yet another explanation. The nine lectures were severally delivered and so probably written with something like a week's interval between every two. In such cases it is less easy to secure consistency, and what was said in a review of some serial novel may apply here, that the author seems to have written the beginning without any foresight of the end and the end without any recollection of the beginning. But before I pass from his lectures I take the opportunity of commenting on two other kindred matters. In p. 352 having said that "man could not by his own power have acquired the faculty of speech which is the distinctive character of mankind, unattained and unattainable by the mute creation", he confirms his proposition by a reference to Wilhelm v. Humboldt's writings: "Man is only man through language, but to invent language he must already have been man." This is a taking argument, and one that would be thoroughly valid on the assumption that language must have been created so to say at one gush, like a metallic casting. But if we include in our view the pos- sibility of a gradual and slow development of the faculty, such as M. Mtiller himself in his first chapter assigns to the creation of all the sciences, including by implication the science of language itself, the whole difficulty is dis- pelled. On this theory the human mind and the faculty of speech react each on the other, and thus "the foundation- stone of what was to be one of the most glorious structures of human ingenuity in ages to come may have been supplied by the pressing wants of a semi-barbarous society" (Lec- tures p. 5). But there is another writer, and he not a German, who as agreeing in one of the two phases of the Oxford Pro- fessor's book claims our attention. The Dean of Westminster ('Study of Words' p. 16) says: "God gave man language, because he could not be man without it." This seems to 30 ORIGIN or LANGUAGE. imply that lan^ua^e was contemporary with man's creation. May 1 be permitted then to ask liow this doctrine is to be reconciled with what I suppose will be allowed on all hands as a fact, that the primitive language must have been wholly wanting in terms for spiritual and metaphysical ideas, seeing that the roots of language in their tirst meaning are very generally held to have a special reference to the material world. If the Dean's view then be right, at the very time that primitive man existed in the most perfect, the most spiritual condition, he was yet destitute it would seem of terms to correspond with all the sublimer elements of his mind. In saying that terms for spiritual ideas are generally traceable to a material origin I have in view such cases as the derivation of anima 'soul' from an 'to blow', of spirit from spirare 'to breathe', and of gJiost as connected with gust, with ga9y with yeast (Lectures on Language, p. 387); but I must demur to the Professors derivation of soul, Gothic saivala from saiv-s 'the sea', and still more to his explanation that "the soul was originally conceived bv the leutomc nations as a sea within heaving up and down\vith 7lLZ^^""^ Tr''''^ '^^^"^ "^^^ -^^h ^'^ the mirror coLerse VeorTof r^''""T ^P™^*''"^"' ''^^n with the consistent wrfhf ;.'%''«r!''^«o„, i„ other words more with the ir'd.'' '"* ^"'"-^ ''' ^- M«"-'^ "-'^ '"an Tlie 'instinctive' origin of language, as laid down in thp Lectures', might to some minds have suggested the infol that language ought then to bo the samp fl ., '"^'"^"Ce all countries, and that every infant at tl I ''" ^'^^''^^ "' l;.fc wouhl l,ave been possessed o,,!!/' '"*''' '^ '^« J""e "''."." of things so „„H.,, toTe 1 " ' ' • '^'''''' ''"' ^ <'0"- w'th fact. This difHcdtv l,o ev ! 'T'^'y ^' ^■'^^"00 f' onrc „,e„ts by a little corolhrv tl . •'"'"' '^ "" ""^ory '''■^'''"'"^-''--?tr:r^^^^^^ BOFP'S PRONOMINAL ROOTS, 31 Speaking of his "demonstrative" roots (p. 272) he seems to imply that the instinctive movement still retains its force. "The sound ta or scC\ says he, referring to the Sanskrit pronouns, "for 'this' or 'there' is" (note the present tense) "as involuntary, as natural, as independent an expression as any of the predicative roots." It must be due to some unhappy idiosyncracy I suppose that I myself feel not the slightest tendency to follow such an impulse, however natu- ral, however involuntary it ought to be. If I want to say 'this', 1 say t}d8\ if I want to say 'there', I say there. I certainly do not say either sa or ta. But admitting for the nonce the new doctrine of pro- nominal or demonstrative roots, let us consider the purposes to which they are applied by Bopp and his Oxford disciple. In the instances I am about to quote from these two writers, I wish special attention to be paid to the habitual, almost universal assumption, that if the conditions of outward foj;m be satisfied, it is unnecessary to enter into any logical proof of the appropriateness of the idea. As the references on this head to Max Miiller will be but few, those to Bopp numerous, it may be convenient to give precedence to the disciple over the master. In the index to the "Lectures" under the word declension I find the proposition that "most of the terminations of declension" are "demonstrative roots". Again in the text (p. 274) we are told that "the Latin word Zwc-s" is formed by "the addition of the pronominal element «" and signifies literally 'shining-there'; and he goes on to say that by adding "other pronominal derivatives" we get ^'lucidus, luculentus, lucerna &c." What these other pro- nominal elements are or how they are fitted for the purpose he deems it unnecessary to tell us. So in p. 221 he says that "the short i of the Sanskrit locative hridi 'in the heart' is a demonstrative root and in all probability the same root which in Latin produced the preposition W2." He goes on to deal with the formation of the genitive, dative, and accusative, but in a manner so misty to my comprehension that I fail to pick up a single idea and can solely refer to his book pp. 221-224. 32 BOPP ON CASE-ENDINGS. Bopp starts (§ 105) with the doctrine that the class of roots he calls pronominal "give origin to the pronouns, to original prepositions, to conjunctions, and particles." In § 11 5 he advances a step farther, claiming "the case-endings as at any rate for the most part of like origin." Looking upon the nouns of language as the Personae Dramatis of the World of Speech, he holds that "the original office of case-suffixes was to express the mutual relations between these 'Personae' in respect of place " ; and with this feeling he asks "what class of words could be better qualified to fulfil such an office , than those which at once express personality and the idea of place, whether nearer or more remote, whether on this side or on that." Accordingly (in § 134, p. 277) the s of the nominative is referred to the pronoun sa 'he, this, that', fern. sd\ (in § 156, p. 320) the m of accusatives masc. and fern, to the compound pronouns i-ma 'this', a-inu 'that', and the final t which presents itself in the neut. nom. and ace. of certain pronouns, as tat and kat of the Veda- dialect to the neut. pron. ta, Gr. to. Again in § 158 the suffix d of the instrumental case is "as he believes" but a lengthened variety of the pronoun a and one with the prep, a 'to' (Germ, an)^ a meaning however which, one might have thought, would be more in place in the accusative. In § 164, p. 329 the datival e is said probably to belong to the demonstrative ^, "which e how- ever is apparently only an extension of the stem a\ that is the very pronoun which has already done duty for the instrumental. In § 179 t we are told is the characteristic of the ablative and "no one (I quote his own words) who has once acknowledged the influence of prepositions or case- endings, can have any doubt in referring it to the demon- strative stem ta 'this' which has already in the neut. N. and Ace. put on the nature of a case-symbol and will presently be found supporting the character of a personal suffix in verbs"; so that Bopp seems to think that the fact of its employment in two duties is a reason for adding a third duty. Most people I think would have arrived at an op- posite conclusion. In § 184, p. 378 and § 194, p. 393 the BOPP ON CASE-ENDINGS. ^ 33 genitival suffix s is held to be one with that of the nom. and so the same as sa 'this', while the longer suffix sya of genitives is the Vaidic pron. si/a, that is, a compound of two pronouns sa 'this' and the relative ya. Lastly the i of the locative he identifies, like Prof. Miiller, with the demon- strative i. I might be charged with a want of fairness to Bopp if I omitted to report an argument by which he defends Ms theory as regards the nominatival s in the masc. and fem. In the declension of the simple pronoun sa 'this' he ob- serves that it is only the nom. m. and f. that present the s, the neut. nom. and all the oblique cases having an initial t^ just as in Greek we have o, rj with a mere asperate, but afterwards to, tov, Trjg, xov &c., so that there is a peculiar fitness in the employment of this pronoun for the two forms for which he claims it. However he subsequently damages his theory by admitting (§ 345) that originally the 5 may have been carried through all the cases and numbers excepting only the neuters, and quotes the Vaidic locative sasmin for tasmin, and the old Latin sum, sam &c. for eum^ earn &c. And even this persistance in excluding an s from the neuter is at variance with his own statement (ibid.) that the Greek orjzeg, Grjf,i£Qov stand for oo-sxeg^ ao~7]f.i£()ov, which oo he himself holds to be of the same stock with the Sansk. sa. Thus for all the case-endings it is enough with our author to find some pronoun signifying 'this' or 'that' or 'what', it matters little to him which, and to defend himself behind the position that case-endings are in their nature of a locative character. He fails to see that the pronouns in question are but pointers and define only position, and even then had no definite meaning in the outset of things, until aided by the pointing fingers. He himself indeed admits (§ 371, p. 180) that the same pronoun originally signified 'this' or 'that', 'nearer' or 'farther', the mind (he should have said the finger) supplying the necessary limitation. But while the demonstrative pronouns at most define only the 'here' or the 'there', it is the special office of case- 3 34 BOPP ON CASE-ENDINGS. endings to deal with motion as well as rest, to talk of the 'whence' and the 'whither as well as the 'where'. Nay if Bopp's system were valid, we might freely interchange all the case-endings. But I have yet two other objections to offer, which seem each of them fatal to his doctrine. In the first place the form he assigns to the case-endings is in most instances a very late and degraded form. For example the locative and dative, which I believe to have been of one origin, have assigned to them as suffixes nothing but the vowels e and i respectively. But the Latin in i-U^ ali'bi, utru-hi exhibits a h, and as the Greek habitually has ip as the representative of a Latin 5, there can be little doubt that the Homeric ovQavn-cpt presents the suffix in a more accurate shape than the ordinary Sanskrit locative. There is still another letter to reestablish in its proper position, a final n; and Bopp himself admits that ooQavocpiv is the older form whence ovQc^vocpi was derived. The Latin nobis, vobis by their long vowel also betray the loss of an w, and still more accurately defined is the suffix in the Old- Prussian dat. pi. in mans (§215, p. 424). Nay I cannot but suspect that the Sanskrit in its masc. loc. tas-min has also in the last three letters a satisfactory equivalent for the (fiv or bin. for on grounds independent of the present question (see Proceedings III, p. Q>^^ note §, and lY, p. 30) I should claim tas rather than ta for the root-syllable of the pronoun, and this view is confirmed by several other cases of the pronoun. So too the Umbrian locative appears to have had a suffix men or mem (§ 200, p. 400), and the Zend for the dative of the first personal pronoun has mai- byd, the long a of which w^ould have a satisfactory ex- planation in the disappearance of a nasal. But to take a more general survey of the question, I would object to the fragmentary manner in which the school of Bopp pursue the enquiry into the form of case-suffixes. Each case must originally have had a common form of its own, no matter to what declension a noun belonged, no matter what its gender; and again it is easy to see in nearly every case that the plural and the so-called dual forms (which in fact BOPP'S PRONOMINAL ROOTS. 85 are but varieties of plurals) contain in addition to the case- suffix of the singular a second suffix denoting plurality, either a nasal syllable as in our ox-en ^ or a sibilant as in our cows. Hence in our search for the full forms of case- suffixes we are entitled and therefore bound to include all the forms belonging to a given case without distinction of declension or gender or number. Then again on the other side Bopp appears to be unhappy in his dealings with his so-called pronominal roots. These also he has robbed, as it seems to me, of a final n, which readily interchanged as well with the liquid m as with members of its own dental class, t and s. Thus for the first syllable of the Latin is-to- I find a more satisfactory explanation of the s than Bopp's own theory (§ 343) that it results from "a petrifaction" of the nominatival s of the simple pronoun is. But I go further. In his zeal for pro- nominal roots he seems positively to invent them, as for example ma (§ 368), u (§ 1002), and above all his favourite sma (§ 165 &c.), of which he makes a most abundant, but I fear most unsatisfactory use. But it is a special office of Bopp's pronominal roots to supply a corps of prepositions, and accordingly he lays himself out for at least an easy solution of the problems likely to present themselves. The ideas of 'above' and 'below', of 'before' and 'behind', of 'in' and 'out' stand in the relation of opposite poles to each other. The metaphor is Bopp's own. Hence the demonstrative pronouns are ad- mirably suited to act as the needful symbols for these ideas, and so, what is particularly convenient, as they signify at once 'this' and 'that', 'on this side' and 'on that side', from one and the same pronoun we may deduce pre- positions of directly opposite powers (§ 995). Thus from the pronoun a, to take that first as exhibiting the most wonderful fertility, with the aid of various suffixes, whose meaning seems to be a matter of not the slightest moment, for he never stops to explain them, we have S. (i. e. Sans- krit) a-ti 'over', S. a-dhas 'under', Lith. a-ni 'up', Germ. ent, Lith. a-t 'to', 'back'; S. a-dhi 'over', 'up' (§ 997), with 3* 36 BOPP'S ORIGIN OF PREPOSITIONS. Lat. ad Ho' ; S. a-pi 'over', 'up' (§ 998), with em ; S. a-hhi 'to" (§ 999), with a^iq^L, Lat. amh or am 'round'. Germ, hei, and Lat. oh; S. a-pa 'from' (§998), with a-no ^ Lat. a-h^ Eng. o-f (the hyphens are Bopp's) — and (§ 1007) from a-pa itself, through an intermediate apara-s 'the other cut down to para, we have no less than live S. prepositions, viz. pra 'before', prati 'towards', 'parcl 'back', 'away', piiras^ and pari. Of these again 2J>m (insepar.) 'before' has for its cognates ttqo, Lat. pro. Germ. ver. Then prati (§ 1008) is represented by tiqoti and TjQog', while jp^m 'back', 'away' (§ 1009) gives us naQa; and through a second aphaeresis a prep, ra 'back' in some other language, which is one with the Lat. r^ 'back'. So much for one extensive family, all the progeny of the tiny pronoun a 'this' or 'that', in- cluding too at once arto and naQa, at once pro and re. To the S. pronoun u, if indeed such a pronoun exist, are to be referred it seems S. u-pa 'to', S. u-t 'up' as also the Gr. v-7io^ Lat. sub, and the adj. v-G-TSQo-g, together with Germ, aus^ Eng. out. To meet the little difficulty about the asperate of vtlo and the s of sub, Bopp proposes two theories: "The s is either a simple phonetic prefix or the remnant of a recently prefixed pronoun sa", which however, he adds, would be "here devoid of meaning". The S. pronoun ana gives birth to S. anu 'after', Old- Pruss. and Slav, na 'up', and ava 'up' ; also to S. ni 'down', Germ, nie-der; also to S. ni-s 'out' and perhaps to the Slav. i'su 'out', "which may possibly have lost an initial n"".— The loss is the more to be deplored, as we lose at the same time all resemblance between i-su and its parent ana. Thus Bopp has thoroughly fulfilled the promises he held out, as we have from the same sources words denoting 'above' and 'below', 'to' and 'from', 'backward' and 'forward', 'absence' and 'presence', 'up' and 'down'. And then how magical the changes. With this wonderful manufacture by the Bopp-school of prepositions and case-endings from pronominal roots it may be useful to contrast a few specimens which may show the possibility at least of deducing prepositions and case-endings BOPP'S ORIGIN OP NEGATIVES. 37 from verbs. Thus to commence with a quotation from one of Bopp's own followers, we find in the "Lectures" (p. 221): "The instrumental (in Chinese) is formed by the preposition y, which preposition is an old root meaning to useP So in a little paper of my own ("Proceedings" Vol. VI, p. 120) it is stated on Premare's authority that the syllable com- monly used in Chinese to denote the genitival relation (^tcV) is at times employed as a verb equivalent to the Latin pi'ojicisci. Again the Sanskrit inseparable preposition ni, Lith. nu 'down' is to be identified with the Lat. vb. nu, Gr. vev 'lower', 'hold down', and the Chinese ni 'descend'. In the French chez^ Ital. casa, and in our own through, Germ, durch and ckir we have prepositions formed from substantives, viz. the Lat. casa 'house', and Germ, thilr^ Eng. door, Gr. d^vQa. So little is it necessary to invent pronominal roots, as the source of prepositions. On Bopp's derivation of particles from pronominal roots I must be brief. That words denoting 'yes' should be de- rived from pronouns signifying 'this' can surprise no one. Thus we assent at once to such a derivation of the Lat. sic and ita and si of the French &c. But Bopp is bolder, he hesitates not to deduce the S. na 'not' and Lat. ne 'not' from his pronominal stem na 'this or that', the Greek f.it] 'not' from his stem ma\ and the Greek '« privativum' from a 'this' (§ 372. 1, p. 180). And here again he relies on his old doctrine that as such pronouns are qualified to denote alike 'this' and 'that' (dieses und jenes), in the second of these senses they may well represent negation, for what is there is not here. It is somewhat unfortunate that the pro- noun a has on his own showing a marked tendency to express presence (§ 366), as a-tra 'here', a-tas '•ixom here', a-dya 'to-day'. Nor is this to be set down as a late in- novation in the life of Sanskrit, for its position must have been already well established before the breaking up of the primeval language, seeing that (to use his own illustrations) it is found in the old Irish a-nochd 'to-night' of the far west, and in the Ossetic a-hon 'to-day' of the far east. But be this as it may, the same pronominal «, once firmly 38 BOPP'S ORIGIN OF THE AUGMENT. possessed of negative power, is deemed by Bopp a fitting symbol for past time. "I hold the augment", says he (the initial a in a-hhav-am 'I Avas' for example, and so corre- sponding to the syllabic augment £ of s-tvut-ov &c.), "to be in its origin identical with the a privativum, and look upon it as expressing the negation of present time." Nay even in such forms as leg-e-bam (the division is Bopp's) he once thought the long quantity of the middle vowel was referable to a suffixed augment, but his confidence in this theory has been shaken (§ 527). Even among the verbs he is inclined to think that his pronouns play a part over and above their use in the personal endings. Of the suffixed t in tvtt-t-co, v in dax- v-to and S€ix-v-v-/iu, av in Xafi^-av-co he speaks with the greatest hesitation, yet still (§§ 494, 5) "the most probable explanation" is that they are one and all of pronominal origin, their office being "to convert the abstract of the verbs in question into a concrete". Nay even the so-called connecting vowels, as in cpeq-o-f^iev , q^eQ-e-xe, must be ascribed he thinks to a similar origin (§ 500) , and indeed to our old friend «, for the o and e of the Greek verbs just quoted are represented in Sanskrit by an a. I now leave the pronominal roots with a strong impression on my mind that Bopp has failed to derive from his theory anything that adds to the value of his book. Even in his other division of roots I cannot divest myself of a fear that he has been wanting in caution. In § 109/? he gives us a list of thirty two root-verbs. In looking over these I find at least fourteen which I have little doubt are secondary, that is derivative verbs, and eight others that have been shorn of their fair proportions, having lost an initial or a final consonant or both. On the present occasion I cannot deal with more than a few of them, but to avoid all sus- picion of undue selection, I will take a batch that follow one another, those which stand 3rd, 4th, 5th, and 6th in his series. The verb gnd (ov jnd) is of course the Lat. gnosc-o, Eng. hiofc, but in these verbs all that follows the liquid constitutes a suffix, while our English vb. ke?i or BOPP'S ROOT-VERBS. 39 rather con exhibits the simple verb; and, as I have else- where noted, the Latin participles a-gn-itus and co-gn-itus are deduced from compounds, not of gnosc-o, but of a primary verb go7i^ corresponding to our con. From agnosco and cognosco we should have had agnotus, cognotus. The 4th in the series, vd 'blow', has suffered curtailment of its tinal consonant, and is really one with the 17th, an 'blow', which has lost its initial consonant, the two being truncated forms of a fuller van which appears scarcely altered in the Germ, wann-en^ and is the parent of our winn-ow^ loind^ and fan.) as also of the Latin vanntcs and ventus. This double corruption of van to vd and an would be exactly parallel to my assumption that the Lat. ce and m come from ken. The 5th sfd, Lat. sta^ though very generally set down as a root-verb, has a suffix or rather the remnant of a suffix in the a. The proof of this I find in the Latin sist-o as compared with gign-o, yiyv-of-iai, ^il(,lv-co^ nim-w, for as these are admitted to be reduplicated forms of yev, (.lev, ntx , so sist implies a primitive set or something like it. To this primitive I assign the idea of 'stop', a verb which is itself probably of the same stock, and I quote in support of this translation the familiar siste viator or better still s. aquam of Virgil, s. lacrimas of Ovid, s. alvom of Pliny, I say better because there is in these phrases no trace of the upright position, w^hich eventually attached itself to so many of the derivative forms. I may be asked here whether I propose to connect the assumed root set with the sed of Lat. sed-ere, sid-ere (for seid-ere) &c. and our own set, sit. My answer to this is at present neither yes nor no, but on the logical side I see no difficulty, as we ourselves have the phrase 'to set up', equivalent to the Lat. statuere. Again if I am asked to account for the fact that sta- and its derivatives eventually possessed as an im- portant part of their meaning that of standing or the upright position, I think I see two explanations. First the com- pound a-sta in Plautus has the simple notion of 'standing up' rather than that of 'standing near', so that the preposition is an (= ava)., as in an-hela-re 'to send up a blast of air'. 40 BOPP'S ROOT-VERBS. rather than the familiar ad Ho or near'. It should be noted too that it is precisely before an initial 5, that the Greek am, commonly reduced to av or rather ov in the Aeolic dialect, becomes further reduced to a or (Ahrens, de Dialectis 28, 1). The assumption that astare was in the end cut down to stare^ has its parallel in our own truncation of rise fi'om arise, for this is. the original form. This theory further explains in a thoroughly satisfactory manner the prefixed vowel of the Fr. etat^ etais^ etablir. But indepen- dently of this argument, if the original notion of stopping be considered in connection with man, and it is of man that we commonly speak, the first result of stopping is standing. The 6th verb t 'go', though found alike in Sanskrit, Greek, and Latin, I believe to be doubly corrupted. Already it-er^ com-it-ium, in-it-mni, ex-it-mm, comes (them. com-it-\ pedes (them, ped'tt-) claim a final t for the root, and the forms so familiar in Plautus per-btt-ere, inter-bit-ere ^ red-blt-ere, praeter-bit-ere, e-blt-ere (the last in Plant. Stic. 608, accord- ing to the palimpsest) exhibit an initial b. I have marked the i as long on the uniform authority of Plautus, though Forcellini hastily assigns a short i to these words. Then as regards the simple verb, Ribbeck has done well to follow the guidance of Fleckeisen in exhibiting baetere as the reading of Pacuvius in vv. 227 and 255. Thus bat, the root of baetere (as cad of caedere\ is the Latin analogue of (^av in fiaiv-co^ and so only a variety of vdd 'go', whence the imperfect tenses vad-ere &c. We have here an explanation of the apparent anomaly in the corresponding French verb which unites in the same conjugation, a stem va and a stem /, these, although wholly different in form, being in origin one, as je vais, tu vas, il va with firai &c. These two verbs sta and i may indeed be pointed to as containing the best evidence of the close intimacy between the Sanskrit, Greek, and Latin languages; but it is in the last of these three sisters, not in the Sanskrit, that w^e find the truest forms of the two roots. If it be replied to what I have here m-ged, that the Indian ORIGIN OF AMABAM, AMABO. 41 Grammarians, when they put forward a so-called 'dhatu', do not claim for it the honour of being an ultimate root, nay that they apply this term to the base of any verb though it be doubly or even trebly a derivative, I still con- tend that Bopp applies to his w^ords the very name 'roots' (Wurzeln), and that his whole argument implies that the verbs so called are ultimate forms. It would not be right to be wholly silent on his treatment of matters connected with the conjugation of verbs, but I must limit myself to the use he makes of the so-called verb substantive whether as or bJm 'be', though I may refer also to similar proceedings on the part of Prof. Max Miiller. That this verb is employed in the processes of conjugation I of course do not deny, for I have myself sought to ex- plain many forms by means of it. For example I contend that such phrases as 'I am a-dining', 'I am from dining', 'I am to dine' are found in many languages besides our own as formulae of presents imperfect or perfect and of future verbs; but then it is in the prepositions a (A.-Sax. a?i)^ from, and to, that I find the essential part of the tense- idea. Indeed the very fact of the verb 'to be' entering into all the three phrases is the best proof that it contributes but little to the notation. But Bopp and his devout wor- shipper proceed with far greater boldness. Thus the latter (Lectures p. 174) tells us: "bam in cantabam was originally an independent auxiliary verb, the same which exists in the Sanskrit bhavdmi and in the A.-Sax. beom 'I am'." Again (p. 234) he says: "In the Latin bo oi amabo we have the old auxiliary bhu 'to become', and in the Greek futures in G(o, the old auxiliary as 'to be'." (See also Bopp § 526 and §§ 648, 656.) This is to give to the past imperfect and the future of the Latin the very same origin, so that the Romans, it would seem, thought it no inconvenience to confound the two opposite ideas of time. Let me note too that the author of the Lectures by quoting in the one case the first person of the Sanskrit verb and in the other the mere base or 'dhatu' gives a deceptive plausibility to his argument, for one sees some resemblance to bam in 42 M. MflXER ON PERFECT-TENSES. bhavdmi and some resemblance to bo in bhu. My own views on the formation of the Latin tenses am-ab-a-m and am-ab-o are given elsewhere (Trans. 1856, pp. 308, 9). I will here merely say that 1 find the symbol of past time, not in 5a, but solely in the final a of am-ab-a-m^ just as I find it in the corresponding vowel of the Lat. er-a-m^ Gr. 7]v (= €av) or e-xid^e-a , and S. a-bhav-a-m. I have said that the two German Professors explain the g of Xe^co as the vb. subst.; but according to Bopp it is equally ap- plicable to the aorist els^a (§ 542) and to the perfect Terele-o-fiaL (§ 569). Nay even the x of edwxa and dsdcoxa is deduced from the same source (ibid.), a change which will prepare us in some measure for a still bolder doctrine, that the strange k which appears in the Lithuanian imperative duki 'give', is also a variety of the s of the vb. subst. (§ 680). As to the office it performs in this place, as in the others, not a word is vouchsafed. As a final specimen of the sort of reasoning which is allowed in the explanation of tense-forms, I may point to a passage in the oft-quoted Lectures (pp. 317, 8). From such phrases as 'I have loved', 'amatum habeo' it is inferred that the notion of 'habeo' is specially fitted to denote the past or perfect, the fact being that the essence of this idea lies in the suffixes of ama-tum and lov-ecl. And then, as something parallel, the writer quotes a Turkish phrase, which he tells us is literally "Paying belonging to me", but practically signifies 'I have paid'. I fear his knowledge of Turkish is not of the soundest, for at any rate the Latin phrase 'solvendum est mihi' and the English 'I have to pay' sound more like future than past tenses. I shall conclude my comments on the Vergleichende Gram- matik with a brief notice of the free use made by Bopp of Grammatical figures as they are called, and those too of the very class which the soberer philologers of late years have been disposed to reject as inadmissible except in rare cases, I mean the figures wiiich imply an extension of w^ords whether at the beginning or end or within the body. Bopp's much used terms vorschlac/, eimchiebung^ and zmatz, streng- BOPP'S USE OF PROTHETIC VOWELS. 43 thened occasionally by the epithet unorganiscTie ^ stand in the place of our old friends prosthesis (or prothesis)^ epenthesis, and paragoge. To the curtailment or compression of words, no reasonable objection can be made, as it is the general law of language that forms should be abbreviated. I propose to take the said figures in order. Prothesis. The initial vowels of the words avsQ- (V. G. 2nd Ed. Vol. i, p. 550, note), ovof-iaz- (ibid. 1st Ed. p. 311, note), oq^Qv- (ibid.), owx- (ibid.) are declared to be in- organic additions. The first of the set is further declared to represent the S. nr or nara; but unhappily for this doctrine the noun avaq- happens to be the example given by Dionysius of Halicarnassus, when he is speaking of Greek words that originally had the digamma; and as this letter 10 habitually interchanges with an m in many languages, there arises a strong suspicion that Fav-eQ- has its root in the first syllable and so is identical with our own man. This is further confirmed on the one side by the English corruption of man to one (pronounced with a digamma) in such forms as one says and no one^ compared with the German man sagt and nie-mand , and on the other by the Greek compounds TioL-^iavcoQ and ^va'^i-(.iavdQoq compared with GTvy-avcoQ and AXe^-avdQog. 2. As ovof.iaT- is always held to be one with the Lat. nomen, and as this, being a derivative from nosco, must originally have had an initial g (of. co-gnomen^ a-gnomen)^ we are driven to an older yov- ofiaT-, of which yov alone is radical. Indeed Bopp himself in his Glossary (s. v.) deduces the S. naman from jnd. 3. OrpQv- being compared with the S. hhru (gen. hhruv-as) is pronounced guilty of having in its first vowel something to which it is not entitled. But let us rather compare it with our own eye-brow to which eye contributes no small portion of the meaning. Surely then if a reasonable ex- planation can be given of the Greek word, such as shall include the idea of 'eye', we shall have what is more satis- factory. Now the most familiar root-syllable for 'eye' or 'seeing' is in Latin oc (ocidns) and in Greek with the usual letter-change on (orcTof-iai). But before an asperated letter 44 BOPP'S USE Ot AN EPENTHETIC N. an, will of course become orp, as in orp-d-al/nog. I suggest then that o(pQv- stands for o(p-(pQv-^ or I should myself prefer to say o(p-Qv-^ seeing that the Greek language habit- ually drops an initial labial when followed by q. Thus we have Qrjy-vvfiL rather than FQtjy-vv/tu, Eng. b?^eak, and Qay- rather than FQay-^ Eng. berr}/. 4. The noun ov-vx- I have little doubt is to be divided as here marked; and I say so partly on the evidence of the Latin unguis, ung-ula^ uncus, and the Irish ionga, partly because vx is a well established Greek suffix, as seen in oq-vx- 'dig' (oqvgg(o)^ the sb. di- (OQ-vx- 'a trench', and virtually in oQ-(vyx-o- 'a trench' (especially for vine-planting), and so closely related to the Lat. or-d-on which has precisely the same for its lirst and original meaning. Compare too for suflix ^oazQ-vy-, [:iooiQ- VX-0-, (SoTQ-vx-o- as w^ell as ^otq-v-. Indeed most nouns in u have lost a final guttural, as the Latin genu-^ metu-, anu- contrasted with genuc-ulum (Eng, knuck-le)^ metuc- ulosus, anic-ula. I might also have included the suffixes vy and v^ , of nt£Q-vy-, xal-vx-, as of the same origin with vx. I am myself too further moved by the long- established belief in my own breast that words with an initial n have generally suffered decapitation. Epenthesis. This doctrine is called in aid by Bopp not unfrequently, but especially when dealing wdth the genitive plural of certain vowel-ending Sanskrit nouns (§§ 246, 249) which he says "insert a euphonic n between the ending and the stem." ^ Among the instances he gives of this 'in- shoving' are asvd-n-dm 'equorum', trt-n-dm 'trium', simu- n-dm 'filiorum'. And he notes it as something very re- markable that the Zend, the Old German, Old-Saxon, and A.-Saxon exhibit a similar peculiarity. Surely then he ought to have asked himself whether this n may not be the sub- stantial part of a genitival suffix. Had he done so, he would have found I think abundant evidence in his own and other cognate languages. I have myself long been satisfied with this explanation of the en of the German compounds viond- cn-licht, has-cn-lage, and our own earth-en-ware, Ox-en-ford, Buck-en-ham and its equivalent Buck-ing-ham , as well as MANUFACTURE OF EPENTHETIC VOWELS. 45 the adjectives ivood-en^ lin-en^ silk-en &c. And then again we have in as a genitival suffix in Gaelic, as ho-in from ho 'cow'. It is the more remarkable that Bopp should have failed to hit this explanation, when he himself interprets (§ 248) the sdm of the S. te-sdyn 'horum', td-sam 'harum' as containing a double suffix, of which / represents the genitival element so familiar in the singular. Secondly in § 97 and again in § 727 note, he further teaches that while a final n in Greek has often originated in a final 5, such interchange is confirmed by the Prakrit. On this ^dew te- s-dm and asvd-n-dm would go well together. Again as an n is ever apt to become silent before an s (cf. eig^ yaQLEig, Tvcp'htg, cosol^ toties)^ it would have been more prudent perhaps, when dealing with the suffix of the dat. pi. in Old-Prussian mans , not to have considered the n as inorganic, on the sole ground that mas would agree better with the S. bhyas. His illustration too from the Latin ensis and 7ne?isis beside the Sanskrit asis and mdsas involves a similar assumption. But we need not hunt up particular instances, when we find a wholesale manufacture of epenthetic vowels established by A. Kirchhoff in the Zeitschrift (i. 37) and K. Walter (ibid. xi. 428). Thus eqs^iv^oq and oqo^oq and the Old- Germ, araweiz of like meaning are convicted of having stolen the vowel which follows r on the sole evidence that the Lat. ervum exhibits no such vowel. HIsxtqov cannot be entitled to the vowel £, because forsooth the S. ark 'shine' proves the original root to have been alk. Again the Greek having the two forms oQoyvia and oQyvia the former is declared to have a vowel that does not belong to it, in spite of the evidence of o^)ay-to. Nay even the long vowel of aX-co-Tiax is 'eingeschoben'. Walter's argument turns chiefly on the assumption that forms ending in rk^ Ik^ rg &c. are ultimate roots. Thus according to him co?mx-, Fwlax-, avlayi- , aXox- , all varieties of the same word signifying 'furrow', come from a root valk = Fel/.-. Now my own conviction, founded on a long and wide examination, is that such verbs are all of them secondary, I do not be- 46 BOPP'S USE OF METATHESIS. lieve in his suggested derivation of auXax- fi'om FeXx-, but if it were true, the Latin vel- (vello) exhibits the verb in a simpler form. But it is enough to place beside each- other such pairs of words as talk and tale, hark and hear^ l>luck and pull^ sparg- and otielq-, terg- and tblq-, calc- and heel^ stirk and stee)\ hoik and holl, both Scotch verbs signi- fying Ho dig', the latter of which is one wdth Lat. col- 'dig'. Paragoge, Bopp's instances of 'unorganische Zusatz' are numerous, but I shall be satisfied with quoting the Latin genctric' 'mother' and iunic- 'heifer', which are declared to have a c of this character, inasmuch as the ^.janitrt (§ 119) and yihii (§ 131) have no such letter. His use of Metathesis however is carried to the greatest extreme. Indeed the term 'Umstelluug', which is his name for this 'figure', incessantly presents itself to the eye. I am one of those who believe the doctrine implied in these words to be carried to an unjustifiable extent by even the more sober of philologers; but I will here confine myself to three examples selected from Bopp's book, w^hich I can- not but expect all persons will agree with me in condemning. In § 308, p. 60 he takes in hand the Gothic adj. hanfa (nom. hanf-s) 'one-handed', and first pronounces ha to represent the ka of the S. eka 'one'. This assumed, he holds the residue nfa to stand for nifa. By transposition of nifa he then gets fani which w^ould correspond no doubt with all accuracy to the Sanskrit pdni 'hand'. This taken altogether must be admitted to be a strong proceeding; and a German philologer in discussing a Gothic word w^ould have done well to cast an eye for a moment on the other Low-German and kindred dialects. Had Bopp done so, he would have found at home that for which he travels to the far East, viz. Old Norse hned 'fist' and Lowland Scotch, not to say Yorkshire, niece. Nay Walter Scott (Guy Mannering c. 24) has: 'q\va land-loupers .... knevelled me sair aneugh or I could gar my whip walk about their lugs"; and, to quote from a more Southern dialect, Shakspere has: "Give me your neif (Mids. N. Dr. iv. 1), and: "Sweet knight, I kiss thy 7ieif'' (Henry IV, pt. 2, 2. 4). ABBREVIATION THE LAW OF LANGUAGE. 47 In vol. i, p. 580 note, attention is drawn to an Armenian noun signifying 'man', of which the crude form is said to be aran. Of this the initial vowel is first discarded as a mere phonetic prefix, and then by 'Umstellung' ran is identified with S. nar or nr. Would it not be simpler and quite as justifiable to affirm that the Armenian aran was formed from the Sanskrit nara by reading it backward? Lastly in his Glossary s. v. nakha 'nail' we have the words: "hib. ionga fortasse litteris transpositis e nioga.'^ In terminating my remarks on Bopp's somewhat free and bold use of 'grammatical figures' I must be permitted to throw out the hint that if by any possibility the Sanskrit forms just compared with the classical have been advanced to a dignity which is beyond their due, in other words if they are after all the more degraded of the two, then all the difficulties which have presented themselves, disappear. From the objectionable figures prothesis, epenthesis, and paragoge we should pass respectively to aphaeresis, synae- resis or crasis, and apocope. In plainer English instead of assuming words to grow and extend themselves, we should have nothing but abbreviation, a principle which seems to recommend itself to the common sense of every one. A man need not be much of a philologer to account for the abbreviation of caravan, forecaBtle, -and cabriolet to van, foxel, and cab. In concluding these remarks, the length of which find their only excuse in the importance of the subject, I must be permitted to say that I have written in no spirit of hostility either to Comparative Grammar or to the Sanskrit language. On the contrary fully believing that the science must be benefitted, when the philologer extends his view over many languages, especially in the older varieties, but to the exclusion of none, I sincerely trust that some of our own classical scholars will apply themselves with indepen- dence and diligence to the study of Sanskrit. My chief object in the present paper has been to check that slavish sequacity which has long interfered with the advancement of linguistic science, and I lay down my pen with some- A 48 UNDUE ESTIMATE OF BOPP. thing like a conviction that my hearers or readers will not so readily give their assent to such propositions as the fol- lowing. Prof. Max Miiller tells us (Lectures p. 167) that "His (Bopp's) work will form for ever the safe and solid foundation of Comparative Philology." Again (p. 216) "Comparative Grammar has well nigh taught us all it has to teach." And another writer, if indeed it be another writer (Saturday Review, Jan. 10), speaks of Comparative Grammar as "a science which has always prided itself on the exactness and almost mathematical precision of its method." Berlin, priuted by Uuger Brothers, Printers to the King. k. r"