V N ^ «^ ,«• ' Cornell University Library P 543.P52 3 1924 026 454 862 . Date Due ■\ ' ""Im 1 1 19»^ Djrfu m nirnm /V Ibid. pp. 5-9. 2 Ibid. p. ix. SOUNDS. 3 Suffice it to mention the judgTuents of Benfey^ and of Schweizer-Sidler/ who consented to revise the German translation of Ascoli' s book by Bazzigher : ' an unwonted and deserved compliment. And to these judgments we might add that of Whitney, who regarded as much to be deplored the delay in the promised continuation of the work of Ascoli;' and that of the French "Academic des inscriptions et belles lettres," which,, on the 29th of July 1870, honoured with a prize the LecUtres (Corsi), although as yet incomplete, of the Italian philologist.' And indeed the portion of them published is indubitably, as Schweizer- S idler well remarked, one of the most important works which have appeared during the last few years in the course of the historical investigation of the word. The results of the preceding researches are there seen not only collected and expounded with diligence and exactness, butalso subjected to a prudent and independent criticism, and augmented by the researches of the author, at whose uncommon breadth of learning and rare power of analysis and synthesis we are again and again forced to marvel. The exposition of Ascoli puts before us not only the results of the in- vestigation, but the entire progress of this investigation itself, portraying with a fidelity which we might call photographic, all the intellectual labour of the author, and training us to scientific research. The style and language 4 of Ascoli have been frequently, at least in private con- versations of Italian scholars, made the subject of vigorous 1 Gottingische gelehrte ameigen, " The researches into the Romance 1870 i. 793-98. dialects more recently given to the " Zeilschrift fur vergleichende public by Ascoli in the ArcUvio rpraohforschung, etc., xxi. 257-66. glottologioo italiano, established and 3 Vorlesungeii iiber die vergleioh- conducted by him, won him two more ende Uutlehre des sansTcrit, des prizes, viz., the Bopp prize in griecMscken. tend des lateiniscTien, 1874, and in the following year the etc. I. Halle, 1872. first prize of the Society for the * Mivista Uuropea, anno 4°, I. study of the Romance languages at 640. Montpellicr. 4 PART I. CHAP. I. § 1. criticisms. It is not our intention to maintain that in the book under discussion, and in the other writings of the distinguished philologist, the form is always both clear and pure, and genuinely Italian : but it appears to us right and proper to observe that, besides the merits of exactness and thoroughness which no-one could deny, it must be especially commended as far as regards Italian linguistic nomenclature, which Ascoli has enriched by some technical terms almost all chosen and employed with the happiest daring. We cannot, and we will not, disguise the fact that Ascoli^s method does not seem to us the most fitted to initiate the profane into the first studies of philology : but, when any one has begun to read the pages of Ascoli con amore, we believe it may with reason be said that he has learnt much. Among the subjects treated by the author, one appears to us particularly worthy of attentive considera- tion — the history of the guttural tenuis (Lectures 2nd and 3rd, pp. 27 — 95). In the exposition we will follow as far as possible Ascoli himself, availing ourselves frequently of his words.' The most notable transformations of this sound in the Aryan languages may be represented by three phonological equations, of which the first is the follow- ing : Sanscrit and Zend f = Lithu-Slavonic sz, s (Lithu- anian sz, Slavonic s) = Greek and Latin k. Example : Sanscr. Zd. gata- [cento] = Lith. szim-ia-s, Bulgarian suto = Gr. k-KHTO-v, Lat. ceivtu-m. This equation shows us the Proto-Aryan h weakened to a sibilant in Indian, Iranic, Lithuanian, and Slavonic, but preserved exempt from such 5 alteration in the other languages of our stock. As Ascoli ' For the physiological analysis of cie«, Leipzig, 1876, pp. 59-62. We this sound, and of the cognate sounds, invite the attention of students also see the quite recent essay of Sie- to the PAoMeiiseie ifreiyVoj'era, pub- ■VRXSiQ-rundziige der lauipjiysiologie lished by Hofforg in the Zeitschr. iur eitifiihrung in das studium der f. vgl. sprachforsch,, xxiii. 525-58. lautlehre der indogermanishen spra- SOUNDS. 6 teaches us, the Indo- Irano- Lithu- Slavonic agreement is general both with regard 'to the examples in which the ancient k has been reduced to a sibilant sound, and with regard to those in which it has been preserved, while "to the Italic, the Greek, the Keltic and the Teutonic groups, all pro-ethnic coincidence of any one at all of their sibilants with the Indo-Iranian sibilant {g) for an original h is foreign. The coincidences which nevertheless do exist, are here manifestly accidental, due, that is, to patho- logical congruence {congruenza patologiea), and not to his- torical continuity/^ ' This " special resemblance between the Indo-Iranian and the Lithu-Slavonic, which it is alto- gether impossible to call fortuitous," can be explained, writes Ascoli, only in two ways; either by supposing the corruption to have taken place in a period of pre-historie Indo-Irano-Lithu-Slavonic unity (a hypothesis, which cer- tain phonetic and lexical facts seem to support, but sub- ject to the most grave objections) ; or else by imagining that the original h, slightly affected by a parasitic fricative' "in a definite number of instances, even from the Proto- Aryan period, afterwards freed itself, in some languages, from this corruption, and in others on the contrary by consistent development of the ancient affection, underwent consistent changes, which would represent effects similar but inde- pendent one of the other of the same cause. On this hy- pothesis, the word for died (Ital.), for example, would have ' Corsi di glottologia, etc., I. 50. position of the mouth, which is re- ' " One of the most frequent af- quired for the production of a given fections of the original consonants consonant, to the different position is, in the Aryan system, the striking which is necessary for the utterance root after some of them of a pa- of the sound which immediately rasitic fricative, and especially j follows, and is, as a rule, a vowel, (jnj, Ij, Jcj, etc. ..)... meanwhile there is touched upon or brought we will here make the general re- about that position by which is pro- mark that the origin of these sounds, duced the fricative which we call which we call parasitic, really lies parasitic. . . .'' Ibid. p. 43. in the fact that in passing from the 6 FART I. CHAP. T % \. been, in the period of unity, with a slight corruption of the h, da¥a; whence, on the one hand, the type daha, the 6 restored type as it were, to which Greek, Italic, Keltic, and Teutonic would ascend again ; and, on the other hand, the type dalcya, with the intruding parasite, to which would revert the two words with the sibilant, the Lithu- Slavonic and the Indo-Iranian."' And this seems to the author a " safer hypothesis than the other," although he by no means disguises an objection which may be raised against it. — The and equation is as follows : Sanscr. and Zd. ft' = Gr., Lat., Lithu-Slav. h ; as appears from Sanscr. and Zd. ruk' = Gr. XeuK(o-s), Lat. luc(-s), from the Sanscr. M¥a [hair] = Bulg. Mku, etc. The complete harmony between San- scrit and ^end in the series of the examples for F = primi- tive k shows us that such ¥ points back to a pre-Indian period, or it may be to the Indo-Iranian age. " To set off against this, there is not, with respect to the phenomenon of Indo-Iranian i' for original k, any European agreement, which can be imagined to stand in genealogical connexion with this phenomenon; in other words, there is not a single fact which may induce us to believe this alteration to have been effected in an epoch anterior to the complete severance of the European branch of Aryan from the Asiatic, although there are remarkable quantitative (not qualitative) coincidences .... which lead us to be- lieve that the original i, afterwards becoming Indo-Iranian ¥, was corrupted and damaged in a definite number of examples, even from periods far more remote than the Indo- Iranian, but that it was not, nevertheless, as yet, in these periods distinctly altered."^ And now we come to the 3rd equation: Sanscr. and Zd. ^'=Gr.-Lat. ^«)= Lithu- Slav, y^y 1 Ibid. pp. 56-7. Latin, and Lithu-Slavonio an nn- 3 Ibid. pp. 48-9. Afterwards, in altered k corresponds, is referred to the table given on p. 193, the In- a later type k'. do-Iraniau k', to which in Greek, SOUNDS. 7 we may take as an example the Sanser. k'ttivar-, Zd. Math- war-, comp. Gr. Tetrcra/se? {*kgethvar-, *TZeOFap-y =Lat. quatiior, MoXva Gr. nretyavpe'i (^^pethvor-, *'7re6Fvp-) Kymr. r joetwar, Osc. petor-a,'[Jrahr.petur- = Litb. ketun, Irish cetliir? In this example^ and in four others, " we do not find, on the one hand, any certain trace of the v in the Asiatic words, nor have we, on the other hand, any reason which may lead us to assert, or at least render us inclined to believe, that the v is an etymological element, that is to say an original constituent of the word. Hence the v will here be a parasite, in kind not unlike the parasitic /, which in its proper section (§ 14) we saw to be developed in like manner after the original guttural tenuis; nevertheless it, too, will be a « of very ancient origin, and what should sufHee to make us abundantly convinced of the fact is the agreement which several European languages exhibit in re- verting in these same examples to an ancient lev. To this we subjoin the very remarkable fact that they all show, in the Indo-Iranian equivalent, not the pure h, nor the r, which is the most frequent Indo-Iranian representative, as against the European representatives, of the original guttural tenuis, but in fact the k' alone, the most un- usual sound (§§ 11, 1£). This coincidence, supported also by other parallels . . . . , convinces us that here we have to deal with . . . original /^-sounds which were affected even from the Indo-European age, but in an indis- tinct manner, so that the development of the affection was determined afterwards in the successive ages in various ways. If, therefore, in the consideration of the sibilant . which in the Indo-Iranian and in the Lithu-Slavonic branch is held as the successor of the original guttural tenuis (p. 56), we were to propose the typical example da¥a, ([ten], ' InAsooli's work the 2 repre- Italian «c [Engl. «i], e.g. 'vo.scemo. sents a sound identical with that of See pp. 13 and 22. tho French/ : the J answers to the ^ Ibid. p. 92 : cf. 77, 73, 53. 8 PART I. CHAP. I. § 1. whence dalija dalcza dam daqa), then, for the examples in question, we should have to figure to ourselves a typical example which might be written k^atvar [four], the indefinite parasite of which (something like a Greek v) came to as- sume among the Indo-Iranians, in a period relatively 3 modern, the palatal pronunciation [kjatvar-, whence h'at- var-, c'atvar-, see p. 44), and among the Europeans, on the contrary, or at least among those whose dialects reflected an ancient Jcv, halted as a rule (see § 21) at a labial or labio-dental pronunciation {hvaivar- Jcvatvar-, whence qii,a- tuor and *^hator, etc)'. In this jvay we should have in the Indo-Iranian branch the full development, but certainly not co-temporaneous, of both of the afiections (dah'a daga; l^atvar It'atvar), which would be resolved into one and the same affection with twofold result ; and the development kjatvar hjatvar would come to coincide with the hj [h' from h) which sprang from the unimpaired stem in the Indo- Iranian period . . . . ; while in the European section we should have the type dah'a restored everywhere else but in the Lithu- Slavonic branch and the not very numerous examples of the type T^atvar, on the other hand, restored precisely in the Lithu- Slavonic branch (e.g. Lith. Tceturl . . . ), as in a different way they are restored besides in Ireland . . , and sometimes . . . also elsewhere.''" Not at all unlike is the history which Ascoli traces of ' The phonetic group Jcv, by the the Greek t = 7cj from Jc, beside ir progressive change of the continuous = hv from kr : " . , . the product sound » into the explosive labial surd of Jcj, when it has reached that stage under the influence of the preceding in which the guttural tennis is re- explosive surd sound, which it duced to such conditions that it is eclipsed, became transformed gradu- hardly distinguished from the dental ally into *6, *p, pp, p : hence, beside tenuis (Teg' tg', pp. 44-5), might have the Latin qv, we have the Greek, Os. rested at the latter sound, and little can, TJmbrian and Kymric p. See by little the palatal or lingual ad- Ibid. pp. 71-8. dition would have vanished from it, 2 Ibid. pp. 84-5. — The best proof so that t remained in place of the of the h" = European kv and original k " (p. 92). Asiatic kj {k') is, iu Ascoli's view, SOUNDS. 9 the guttural media ig) and of th aspirate {gh) : the various changes of which in the Aryan languages he explains with the help of the hypothesis just referred to. We may, however, pass over them in silence in this brief treatise and proceed at once to critical considerations touching the doctrine of Ascoli which we have set forth. Let us begin with the following words' from the pen of that learned philologist and mythologist, M. Breal: ""We do not know, in phonology, an instance of a sound which after having been changed has reverted to its primitive purity;' i moreover the hypothesis of Ascoli only serves to shift the ground of the problem, because, though it points out for what reason thechangeisfound in the same words in Slavonic and Sanscrit, it does not enable us to understand the prin- ciple on which the restoration takes place uniformly in Latin, in Greek, in Gothic, in Keltic/'^ Another objection is started by Schweizer-Sidler himself, who remarks how ill the theory of Ascoli under discussion can be reconciled with the doctrine of a special affinity of Slavo- Lithuanian with Teutonic j a doctrine maintained by A. Schleicher and his most learned pupils.* Jolly, too, finds fault with Ascoli's hypothesis, deeming it too complicated : " not only the symbols, k' and h", selected by him to denote the two affections which he attributes to the primitive k, but also the hypothesis itself of a mere affection instead of a primordial duality of the ancient h, are artificial, and this last supposition led him further to the opinion, still more unlikely, that the impaired k had in some languages been restored,healed." "Besides," continues Jolly, "why should the It have developed after itself a para- sitic sound ?"^ Windisch admits the transformation of ¥ ' See, however, Ascoli, Studi pp. 357-61. Critici, ii. Eoma-Torino-Firenze, ' Zeitschrifi f. vergl. spraohfor- p. 28. schung, xxi. 257-66. 2 Eevue critique d'histoire et de * Noch einmal der sfammicmm liiterature, 5th year, 1st semester, der indogermanischen sprachen 10 PART I. CHAP. I. § 2. into li, at least as an expression of a change of W into fc (these symbols will be remarked on shortly) : but believes such a phenomenon to have come about without parasites. " Physiologically considered it consists only in a slight alteration : the enclosure formed by the back of the tongue with the palate in the production of the guttural is gradually , forced more and morfe forward from the soft hinder parts of it. 1" Hence there results at last a position, just where the palate and the gum touch each other, in which no longer even a ¥ can be pronounced, but only a ;!-sound' and the so-called palatal //. A nd he denies that the Kymric p always ap- pears regularly where, according to Ascoli's hypothesis, we should have the right to expect it."^ As the reader will have observed in the foregoing remarks, the objection, which assails most strongly the hypothesis of the Italian philologist, is derived from that great phonetic law, which teaches us that a sound when corrupted, far from reverting to its primitive entirety, tends to become constantly more corrupted. And here we are indeed in that part of the domain of language in which the inexorable fatality of the phonetic laws rules with absolute power. To Italians it recalls Manzoni's simile of the rock, which will lie immovable in its sluggish mass where it fell head- long, unless a friendly power comes to raise it aloft. And we seek and seek again, but ever idly, the friendly power to restore the impaired sound in Ascoli's hypothesis. § 3. The obstacles, which oppose themselves to the deriva- tion of the various sounds referred to from a single Proto- Aryan h, induced other philologists, and, so far as we know, first among them Pick,^ whose important lexical labours we shall mention later on as they deserve, to suppose a (ZeitschriftfwrvSncerpscy'hologieund ziur vergl. sprachforschung, viii. spracliv>issenscTiaft,m\. 190-205). 1-48). 1 The dorsal <, Briioke's i'. ' Die ehemalige spracTieinheit der ' Verlust und auftreten des p in Indogermanen, Huropas, Gbttingen, den celtischen sprachen, {Beitrage 1873, pp. 2-34. SOUNDS. 11 double primitive guttural tenuis. He hopes to be able to demonstrate that our linguistic stock, both in its entire pro- ethnic period, and partly also in the several languages, possessed two surd guttural sounds, completely distinct from one another (like the Semitic), of which two sounds the n one is represented in Indo-Iranian by k and by c,' the other by g, and between these almost no contact took place, while partly in Greek and Italic, almost completely in German, they became fused into one sound. For the sake of brevity he denotes these two sounds by the symbols k and k^ and he puts before us their changes in the various families of the Aryan dialects. He throws into relief, in the Indo- Iranian, the affinity existing between Jc and c and the differ- ence between Jc and f, considering c as a successor of k and observing that there is not, on the other hand, an assured instance of q derived from k and used in place of it, and that herein, with very rare exceptions, Slavo-Lithuanian also agrees with Indo-Iranian. The various ways in which the two sections of the Keltic languages represent the Proto- Aryan k, which in Old Irish is regularly reflected by c {ck) , while in Welsh it is refracted sometimes into c, sometimes into JB, lead Pick to the opinion that in the primitive Keltic there existed two ^-sounds, which in Irish became fused into a single sound (e), in Welsh maintained themselves distinct and became c and p. Hence the two equations : 1st. O. Ir. c=Welsh j9 = Indo-Iran. k and c; 2nd. 0. Ir. c = Welsh c = Indo-Ir an. f =Lith.«2=Sl. «. The first sound, which, becoming c in Irish and j» in Welsh, must have had a power intermediate and wavering between c and p, may be expressed, according to Fick, by kv : the power of the second can only have been k. In Greek and in Italic the primitive k appears represented by kv (and by the sounds which this group originates) and also by k (cor- 1 The of Pick corresponds to other linguists in the transcription of the h', used by Ascoli and many the Indo-Iranian languages. 12 PART I. CHAP. I. § 2. responding- to an older hxi) : of the Proto-Aryan h, (Indo- Iran. ^=Lith. ««=Church-Sl. «=Ir. c=Welsh c) the suc- cessor is Jc. In Teutonic the primordial difference between the two guttural surd sounds of the fundamental Aryan is 13 for the most part obscured by the ' lautverschiebung : " the one and the other we find represented by k, while this aspi- rate does not discover to us its origin from h or from Ac . Only in a few instances does initial or final /iv show us that, in this family also of Aryan dialects, the i: corresponding to the primitive k undergoes the change to iv. In Slavonic the Proto-Aryan k appears well marked only in the group sk. Pick's hypothesis of the double primitive^ was received with favour by several philologists, among whom we would first mention G. Curtius, who, to remove all doubt with respect to the genealogical tree of the Aryan languages, considers himself bound " with Pick to suppose for the Indo- Germanic period a double i, or, to be brief, a guttural k and a palatal i.'" Ha vet, too,' believes in the existence of the two Proto-Aryan guttural surds, which he represents by the symbols yi, and k^ and to which he attri- butes in the primitive and fundamental Aryan the same sound which they had in Latin, pronouncing k^ (=Jfc* of Ascoli, i of Pick) as yJw, ^j (=^ of Ascoli, 4 of Pick) as A. But he sees in the development of a parasite after the explosive the effect, not the cause, of the original change of the consonant in question. The change of i into k' is, in his opinion, prior to all formation of a parasitic sound. ' See the Deutsche grammatik 1870, pp. 99-104. (Part i. Book i. Gottingen, 1822) '' Oriechisch t und sicr. k' (Sta- and the OeschicMe der deutschen dien zur griechisehen und lateini- spraclie (Leipzig, 1848, pp. 392-434) schen grammatik, vii. 265-72 : see of J. Grimm. — See also M. p. 267). Miiller, Lectures, etc., 2nd series, 3 L'unite Unguistique europiene. London, 1864, Lect. v.; Helfen- Laquestiondesdeiuciiarioeu/ropeens stein, A comparative grammar of (Memoires de la SoeiHe de Unguis- the Teutonic languages, London, tique de Paris, ii. 261-77). SOUNDS. 13 He then proceeds to show how, by means of successive cor- ruptions, the two Proto-Aryan >^-sounds became changed in is such a way that it was possible for them to be confounded. The change of the primitive explosive guttural surd into a sibilant, a change which we see in several Indo-Iranic words and in the Slavo-Lithuanian words corresponding to them (as we have just now seen), took place, if we believe Havet, separately in each of the two sections of the languages mentioned, just as, e.g., the c (Lat. k) of centum was sibilised in the c of the French word cent quite independently of the (^ of the Old Indian and Zend qata- ' . Hence he proceeds to demonstrate the existence of the double k in the Proto- Aryan period by the following equations : 1st. ^■2=Graeco- Italo- Keltic k, Teutonic h {k) = Indo-Iranian s (corresponding to our c), SI. s, Lith. « {k) ; in all these languages in the •most ancient form the sound was k : hence it was k before the separation. 2nd. ^i = Gaelic k, Kymrie ^=Pan-Keltic kw; /c, = Lat. to, Osco-Umbrianj9 = Pan-Italic y^K^; ki=Tr, Ion. /c= Pan -Hellenic kv): hence /^i=Graeco-Italo-Keltic kw — /£i = Teutonic hv {fcw),f [p, kw), h (y^) =Proto-Teutonic kto; ;Ji = Lithu-Slav. k, sometimes kw, p {ho) : hence ^i = Teutono-Lithu-Slavonic kw — hence ^i=Europeanto — ki= Indo-Iranic k,t' [=k' oi the more usual transcription), some- times p (Jew), kw. thence ^i = Indo-Iranian primitive kw. Now, if ki is=kw of Indo-Iranian and of the fundamental European, there results this last equation : ^i = Proto-Aryan kw. The limits prescribed for our treatise do not allow us to follow Havet in the replies which he makes to several objections, and in the exposition of the advantages which he believes may be derived from his theory. Jolly himself ' ". . La rencontre ario-slave est the term 'Aryan' Havet under- aussi fortuite que la rencontre ario- stands ludo-Iranian. In the belief francaise et . . . . nous n'avons pas that this wonderful coincidence is plus h, detacher le lettoslave du accidental the French philologist will groupe europeen que le fran9ais du not, we think, have many students of groupe roman."— Note that under language on his side. U PART I. CHAP. L § 2. also, in the monograph quoted above, admits in Proto- 14 Aryan two ^-sounds quite distinct from each other, the true physiological value of which we can with difficulty deter- mine, because they have reached us only in one series of representatives. He, therefore, willingly accepts Havet^'s symbols : \, h^. The fact, observes Jolly, that the written language had only one letter for the guttural tenuis, con- tributed unquestionably to confuse two sounds originally distinct. Nevertheless it should be remarked that the written language, whence the written letters used by the Aryan peoples took their origin, offered them two characters for the primitive sound le, by which they might well have indicated with accuracy its two different values. Fichu's hypothesis, well received and defended by the philologists mentioned, found a formidable opponent in that learned and acute inquirer Johann Schmidt, who subjected it to a severe examination, in his review of Pick^s work on the ancient linguistic unity of the Indo-Germans of Europe.' Against Fick' s theory of the primitive double h J. Schmidt observes, in the first place, that, by the confes- sion of Fick himself, these two supposed Proto- Aryan gut- tural sounds coalesced in Teutonic almost always into Ti ; frequently into h on Greek and Latin ground ; in Irish they are not distinguished at all, nor are they always clearly discernible in Kymric. He goes on to quote examples of Indo-Iran. ^ = SI. s, Lith. &z reflected by Lat. qv, and of descendants of Iv, corresponding to Sanscr. ^, against the as- sertions of Fick. Further "the South-European languages and the German not only have often the simple h where Fick^s rule requires ^w, but also Iv when the rule forbids it, i.e. the distinction between the two sounds in these lan- 16 guages is not generally complete." Fromtheieci!Mre«of As- coli he learns that there is not always a well marked difference between the two ^-sounds even in Indo-Iranian and in 1 Jenaer Uteraiurzeitung, 1874, pp. 201-4. SOUNDS. 15 Slavo-Lithuanian. Therefore the development of h into kj (Sanscr. ^, SI. s, Lith. sz) was still incomplete when a rela- tion of continuity existed between Indo-Iranian and Lithu- Slavonic : much less complete must it evidently have been during the far more ancient period of the primitive Aryan unity. Moreover^ observes our critic, every Proto-Aryan tenuis has side by side with it a media with an aspirate : thus we have t, d, dli^p, h, bh. Hence if we were bound to ad- mit a primitive double h, we should have to expect also a Proto-Aryan double g and double gh, especially as the Indo- Iranic and the Slavo-Lithuanian dialects have sounds which we might look upon as descendants of the six sounds above named, i. e. Sanscr. h, g, gh, h, and p, /, h ; Old Bulgarian Ic, g, and sz ; Lith. 7c, g, and sz, z. Now Fick is far from wishing to demonstrate in the primitive and fundamental Aryan the existence of a double media and a double aspirate corresponding to the supposed double tenuis k. Lastly, if the ^j of Fick ( = Indo-Iran. f, SI. «,Lith. «z) corresponded to the simple k (not kv) of the other European languages, the logi- cal result would be that the Indo-Iranic and Lithu-Slavonic mediae and aspirates of this tenuis ought to be represented in the other European languages by g, gh, not changed into gv,gJiv : and this cannot be positively affirmed. Bezzenberger, in his critical remarks on the second part, recently published, of J. ^Schmidt's work, ^w gescJiichte des indogermanischen vokalismus,^ has given expression to certain opinions concerning the present argument, which we do not think it right to pass by in silence. " In the fundamental Lithu-Slavonic language,'" he writes, " there le was not from the very beginning a sibilant as a substitute for the primitive k^, or a corruption of it, such as the Sanscr. f is, or presupposes, but it was reflected by a simple k. This results — 1st from its being represented by a simple k in the other European languages, 2ndly from the fact that this k 1 GSttingiscTte gelehrie anzeigen, 1875, pp. 1313-44. 10 PART I. CHAP. I. § 2. has been preserved in some eases.^' ^ " If in some eases/^ he observes, " the development of the sibilant from Ic^ is a phe- nomenon of less ancient origin, it may be such in all the casesj" and accidental. Therefore the agreement of the Sanscr. and Old Bactr. j? with the Slav, s and with the Lith. S3 is quite unimportant.'' He then proceeds to examine some etymological views held by J. Schmidt in opposition to Fick, which seem to him of doubtful value. He thinks it very doubtful also that every tenuis must have side by side with it a media and an aspirate, as Schmidt supposes, and he quotes, by way of example, the labial media which is so rare, and has an existence so scantily demonstrated in the fundamental Aryan.' It is clear from the foregoing exposition that, in spite of the efforts of Ascoli, of Fick, and of some other philolo- gists, the history of the Proto-Aryan h has not yet been explained in such a way as to dispel all obscurity. For the final solution of the problem we still need fresh studies on the descendants of the sound in question. And the results of the fresh researches will be far more useful to philology than many people think, in that the problem, which we have dis- cussed up to this point, is intimatelj' connected, as will be seen later, with that of the special affinities which are generally thought to exist between the various families of the Aryan linguistic stock.* • Windiscli, Beitrdge, etc. viii. but just exigencies of modern pliilo- 29. logy. ^ The second argument may be left '" Schleicher, ComperadiMm, etc. to the judgment of the most authori- Weimar, 1871, § 117. tative students of the Lithu-Slavonie * Our account of the most recent dialects. But, so far as concerns the studies on the Indo-European h has first, we may be allowed to observe been lately charged (who would have that it can hold good only on the predicted it?) with violated pers- hypothesis of a fundamental Euro- pective (Ascoli, Stvdi Criiici, ii. pean language, the existence of 29) ! The charge, however, is not which, as we shall see in the second accompanied by any explanation or pMvt of this book, has not been yet proof, so that we do not even know liro\ cd in a way to satisfy the strict whether linear perspective or aerial SOUNDS. 17 § 3. And it is for this reason that our attention is drawn to another consonant, the existence of which in the primitive and fundamental Aryan is still doubtful. We perspective is meant ! For the rest, while awaiting strong and dear ar- guments and fresh criticisms, espe- cially from philologists who have not taken an active part in the dis- cussion of this difBcult suhject, we leave unaltered for the present the order of our account, as >¥e know no reason for changing it in any part or in any way. [The subject of the Indo-European /fc has recently been exhaustively dis- cussed by T. Le Marchant Douse (Gfrimm's Law, a Study or Mints towards an explanation of the so- called " Lautverschiebung," Sfo., London and Strassburg, 1876, Triibner & Co.). See esp. pp. 134-75 in which the author de- scribes and examines the doctrines of J. Schmidt, Fiek, and Havet on this sound, and proposes a new explanation of the phenomena which it presents. He remarks (p. 138) that Schmidt (Jlje verwand- scliaftsverhdltnisse der I. Gr. Spra- chen, Weimar, 1872) first applied "the phenomena exhibited by the primitive h to the denial of any such clear and decisive original separa- tion as the 'period' theory (of Pick) asserts. Schmidt urges that the Li.-Sl. really agrees in some important points (as e.g. in the splitting or radiation of a to a, e and 0, and in the evolution of I from r) with the European division ; but in its treatment of Tc it agrees just as completely with the Aryan di- vision. Li-Sl. therefore belongs to both at once ; and we are no longer justified in imagining any such broad separation between the two as the 'period' theory requires." Pick makes an elaborate reply. " But," says Douse, "his treatment of the purely phonetic question is affected by two antecedent considerations : first, his determination to vindicate the ' period ' theory in its most un- compromising form, so that his pho- netic hypothesis holds a place com- pletely subordinate thereto; and, se- condly, his assumption throughout thatthe guttural peculiarities in ques- tion, although they did not originate, were yet developed, subsequently to the original Separation." And again, to account for two important facts, viz., "1st that the characteristic af- fection of h^ (Havet's notation) has disappeared in the other dialects of Europe; and 2nd that a great majority of the h\ which, in these dialects, correspond to i,, and should, therefore, on Fick's hypothesis, ex- hibit the labial affection, actually exhibit no affection at all, but are, in fact, like the A's representing Ic^, purek'i." Fick invokes summary processes, Verwischuncf or ' Oblitera- tion,' and Verschmelzurtg or 'Fu- sion,' in virtue whereof the Labial and Sibilant affections were cleared away in certain cases, and the result was the pare Tc. Douse objects that this Verwischung would be a violation of the 'Principle of Least Effort:' "It means that all the Europeans, except the Lithu-Slaves, on no liftuted scale, and for no ap- parent reason, raised a weaker sound to a stronger.'' Moreover one (among several) of the main ob- C IS PART I. CHAP. I. § 3. mean the sound l.^ Lottner* is of opinion that it was developed in that language from which sprang, according to his view, as so many distinct forms, but nevertheless in a particular way akin to each other, the Aryan languages of Europe. Schleicher would not include it in his table of the sounds of the Indo-Germanic mother-language, as he jections to Pick's hypothesis is ' that it would leave the Holethnic speech without a pure A." The merit of Havet's hypothesis, says Douse, consists in the respect it ap- pears to pay to the Principle of Least Effort. But he disagrees with Ha vet as to the relative value of Tew and Ic pure, the latter of which Havet considers to be a debilita- tion from the former. Havet re- lies mainly on the history of hm (gw) in the Romance languages. Douse considers the doctrine unsafe. " On the whole, then, M. Havet's view of the relative strength of Tew and 1c pure seems to me to be incon- sistent both with the comparative physiology of the two sounds, with the analogy supplied by the rela- tionship between ley and /u, and with the tendency of Tew to become p." To Douse' s own theory it is im- possible to do justice within the short limits of a note. We must content ourselves with quoting the summiiry of his argument in his own words. He claims to have shown that " 1st there is originally a single language (the Holethnic) employing a single soxiud of a cer- tain character (/u) ; 2nd, this lan- guage divides, or tends to divide into (for our present purpose) two dialects, an Asiatic and a European; 3rd, in one of these (the Asiatic) a debilitation (7i-;/) of that sound springs up and spreads; 4th, the other dialect (the European) at first resists that debilitation ; but 5th, the two dia- lects continue in presence of each other ; hence, 6th, by the habit of answering to ley by Ic pure a per- ception of incongruity and the Dis- similating sentiment are at last awakened among the Europeans; and 7th, under the influence of the former, this people proceed to adjust (as they suppose) their sounds to those of the commingled dialect ; but, diverted by the latter, their efforts only 'result in a counter- balancing corruption of such of their own pure h's, as correspond to the unaffected Asiatic i's— the sound they actually produce, however, not being an exact reproduction of the Asiatic hy, but differing from it in being a stage nearer to Tew (say leu), from which it ultimately descended or advanced to Tew {qu)." We strongly recommend the reader to master the whole argument of pp. 134-175, which does not yield to the rest of the book in lucidity and close reasoning. The last section (§ 64) contains some clever suggestions on the evolution of » and u from a. — 2V.] ' For r and I physiologically con- sidered see Si e vers, Grundziige der lautpTiysiologie, etc., pp. 50-6. 2 Uler die stellung der Italer innerlialh des indoeuropaischen stammes (Zeitschr. f. vgl. sprach- forschung, vii. 18-49, 161-93). SOUNDS. 19 called it. ' Fick^ moreover, in the second edition of his Ver- gleichendes worterluch der indogermanischen sprachen (section I. 1870) marked under I six among the words and roots which he attributed to the primitive and fundamental Aryan •? but afterwards, in the corrections and additions with which the second half of the second section closes, he substituted r for l,^ thus giving us to understand that he had gone over to the opinion of the two philologists above named. And in the work already mentioned, Die ehemalige sprachein- heit der Indogermanen Europas (Gottingen, 1873), he pro- ceeded decisively to defend the position of Lottner^ con- sidering the I as one of the characteristics which, according to him, lead us to believe in the existence of a European linguistic unity and distinguish it from the Indo-Iranic i' this view we shall have to discuss in the second part of the pre- sent work. The 'consonantismus,.^ for so he expresses himself, is of the Aryan languages of Europe is distinguished from' that of the cognate languages of Asia by the copious develop- ment of the I common to all the former) whila the Aryan mother-language, and the Indo-Iranic period, do not yet know this sound, and in place of it offer in every case r, whence we must suppose the European I has sprung. The less ancient Sanscrit exhibits, with tolerable frequency, the sound I for the most part in the same roots and words which possess it in the European languages : nor, continues the author, is the ^ less diffused in Iranic,, but it is only in epochs considerably later. To the languages of this family in the most ancient period of them which is known to us, that is, to the languages of the Avesta and of the Cuneiform Inscriptions, the l^ according to Eick, is alto- gether unknown. In order to admit that it existed in the * Compendium, etc., pp. 10 and edition of the Vergleichendes wor- 163. terluch, etc. ' See pp. 175-76. * v.\\, Biegemeinsam-europaische ' See p. 1066 ; so also in the third entwicMung des 1, pp. 201-61. 20 PART I. CHAP. I. § 3. period of the Tndo-IraDic unity it would be necessary to suppose that it was lost in the neighbourhood of the Iranians as soon as these were separated from the Indians : a hypothesis certainly not absurd, but in the highest degree improbable and impossible to prove, because in all else the phonetic systems of the old Indian and the old Iranian dialects are closely cognate, and do not differ from each other in the total loss of primitive and common sounds, but only in developments and transformations of some among them ; moreover languages, instead of losing their ancient sounds have a tendency to develope new modifications of them to be capable of expressing, by these, differences of meaning. Let us add that in the language of the Vedas, or the most ancient form of Indian known to us, the I seems to be only at the commencement of its development, and many roots which later in Sanscrit have I, in Vedic are still written with r. As it cannot be supposed that, from the very outset, there have existed double forms, one group with r, the other with I, for the same roots, and as it may be shown in every one of such forms that the ^ is a 19 transformation of r, the latter, and not the former, should be considered as the primitive sound in all of them. And hence Fick proceeds to note several words in which he thinks we must attribute to chance alone the agreement of Sanscrit and the Aryan dialects of Europe in the substitution of the sound I for r, which he deems the primi- tive. He admits the existence of seven words which, on both Indian and European ground, agree in the I of the suffix without our being able to point to the more ancient r beside the I: but, though we recognised them as primi- tive forms, we should still not be bound to consider them provided with the I as early as the Proto-Aryan stage. We might with better' reason suppose that, in these cases, only the less ancient Sanscrit forms with I have reached us, while the archaic forms with r were accidentally lost. No SOUNDS. %\ one, concludes Pick, will deduce from these words a proof of the primitive nature of the sound I. We may at the most allow that in the Proto- Aryan the / was pronounced not always uniformly, but in some cases with a sound approaching I, especially at the end of a root and in the suffixes. But assuredly the I, as a sound quite distinct from r, cannot be assigned either to the great Aryan unity or to the Indo-Iranic unity : it was developed sepa- rately in Sanscrit, in the less ancient Iranic languages, in the fundamental European language. All the Aryan dialects of Europe agree in the change of r to I; but the Greek and the Slavonic sometimes have I where in the other languages the r is preserved unaltered : among the numerous examples quoted by Fick we will notice only laghu (light), li (Lat. linere), Kk (to leave), Ugh (to lick), lip (to anoint), luh (to shine), lug (to break), hlu (to hear). Moreover it should be observed that the Europeans availed themselves of the change of r to iJ to denote new ideas, akin to those represented by the more ancient forms with r : or, if such forms had a widely extended sense, it was so distri- buted that part of it was left to the older forms with r, part 20 was derived from the later with I. There follows a third series of roots which have I in the European languages, and to which there do not exist corresponding Indo-Iranic forms with r. Are they new roots which have arisen on European ground, or do they represent older roots with r which have been accidentally lost in the Indo-Iranian ? The author does not venture to propose a solution of this problem, and contents himself with observing that without doubt also the languages of India and Irania lost a considerable part of their oldest store of roots, nor, perhaps, should we uncon- ditionally deny to a linguistic period so remote from us as that of the European unity the power of creating roots. But, continues Fick, roots of this kind, whatever be their origin, attest by their form a common European activity. 22 PART I. CHAP. Ti § 4. It is attested also by the I of several suffixes, since in tlie European languages new formations of words appear with derivative elements, the characteristic of which is the sound I : and among these formations especial mention should be made of the diminutives, which, while very rare in Indo- Iranic, abound in the European dialects.' § 4. To the results of Fick''s investigations into the history of the sound I stand in point blank opposition those which Heymann arrived at in his researches, and which he ofifered, furnished with as many proofs as he could collect, L in a recent monograph." He thinks that the agreement of Sanscrit with the European languages in the develop- ment of the Hn a series of examples cannot but lead us, as in similar cases, to admit the Proto-Aryan nature of this sound. In a large number of roots and words, undoubtedly primitive, I appears as the symbol of a well marked modifi- cation of the original sense, as opposed to older forms with r ; and of this modification, no less than of the power of I which expressed it, those who spoke the most ancient Proto- Aryan tongue must have been conscious. Among the twenty -five examples quoted by Heymann it must suffice to mention ruh (to shine) and luh (to see), n (to flow) and li (to adhere). Nor can Old Bactrian stand in forcible opposition to the claim of I to be original, because, observes ' Among the characteristics of the pean, is found especially: (1) in a European mother-language, the ex- considerable series of old and im- istence of which he endeavours to portant nominal forms (about 30) ; prove. Pick enumerates also the de- (2) in present-tense-stems (40 or velopment of the vowel sound e from more), the e of which sometimes per- a. Such development, he says, is vades all the other forms of the verb, common to all the European dialects. See ibid., v. 176-200. With respect and was begun, and in great part to this argument we shall see later completed, in the period of the unity the opinion of J. Schmidt (see § 31). of the European languages and peo- '- Das 1 der indogermanischen pies. This e, common to all the Ar- sprachen gehort der indogerman- yan tongues of Europe, and asoe;id- ischen grundspraohe, Gottingen, ing, therefore, in all probability to 1873. the primitive and fundamental Euro- SOUNDS. 23 the author, whenever the forms with r and those with I are distinct in meaning', Old Baetrian appears always to have rejected the latter, while the former appear for the most part abundant. It is known, and proved by examples, that a language can be so powerfully averse from certain sounds as to lose them altogether. A grave objection to the existence of the Pro to- Aryan I certainly cannot be derived from the Old Persian, in which, Heymann observes, with the exception of two proper names, no form with r cor- responds to a primitive form with I. Lastly, if the Sanscrit I had been developed from r, independently of the European I, it is clear we ought to find examples of Sanscrit ^= European r, since it is evidently quite possible, and even probable, that the original r was preserved, at least in some cases, in the European mother-language and became, on the contrary, I in Old Indian, which does not always exhibit the primitive sound unchanged nor always preserve them intact more faithfully than the cognate languages. The lack of such examples is, in Heymann's view, a new proof of the change 22 of r to I even in the primitive and fundamental language of the Aryans. But just as in many European roots the change of r to I is complete, while the former sound remained unaltered in the corresponding Indo-Iranic roots, so the greater frequency in the development of the I may be con- sidered as a characteristic of the Aryan dialects of Europe : certainly not the change of r to ^ as a phenomenon arising in every case separately on Asiatic and on European ground. The opinion of the author is supported by that most important prosecutor of Iranic studies, Spiegel,^ who however acknowledges, on the other hand, the weight of the contrary arguments. That the ancient Iranians possessed a letter indicating the sound I, and that it is a mere accident (as Lepsius and Oppert think) that • Beitrdge, etc., viii. 121-28. 24 PART I. CHAP. I. % 4. such a written symbol is not found in documents which have come down to us, seems to Spiegel, on account of the number and magnitude of such documents, by no means probable. Possibly the ancient Iranic dialects knew the I not less than the Old Indian: but it may be that there was not a clear consciousness of the dif- ference existing between I and r, and hence such difference was not represented by a written symbol. The experience of the readers compensated for the want of a special sign. Spiegel believes, however, that the absence of a letter expressing the I in the ancient Iranic forms no grave obstacle to the admission of the Proto- Aryan existence of the I. J. Schmidt, in the severe criticism which he passed upon Heymann's^ brochure, throws upon him the reproach of having undertaken the investigation without sufficient preparation, of not having consistently followed the same method, of having sometimes arbitrarily derived meaning from meaning; while he affirms the only result of such labour to have been to prove that, as a rule, there do not appear in the Old Iranian those stems of words which in Sanscrit and in the European languages have 23 I. Schmidt thinks that in order to solve the proposed problem it is necessary to investigate the less ancient Iranic languages, all of which, he says, have the I. Such a sound occurs also in Persian and Scythian names handed down to us by the Greeks, and in Zend alphabets. This fact Heymann should have brought forward, and he should have availed himself of the authority of Lepsius and of Oppert, who has made it appear, if not certain, at all events very probable, that a symbol used twice in proper names on the Cuneiform Persian Inscriptions has the power of^. ^ Bezzeuberger, in his critical remarks on Hey m ann^s monograph, declared himself less favourable ' Jenaer literaiurzeilung , 1874, ' Mevue de linguistique, iii. 459, pp. 204-5. sqq. ; iv. 207, sqq. SOUNDS. 25 to the hypothesis of a Proto- Aryan l.^ He does not believe in the primitive nature of the Bound I except when there are not Iranic forms v?ith r arrayed against forms with I of the other Aryan languages : but, when the former appear, we ought to recognise in their r the primitive sound. That even in the oldest and fundamental Aryan forms with I were developed, with a meaning more or less distinct from the primitive forms with r, this critic is not very much inclined to believe, because, in his opinion, he is pi'evented from doing so by several words drawn from the less ancient Iranic dialects, which Heymann ought to have taken into account. It may appear strange that Sanscrit should have given to forms with I, which have been developed independently of the European forms, the same sense as we find in the latter : but it would appear more strange still that the Iranic languages should in every ease have lost the forms with I with their definite meanings or, from an inconceivable dislike of I, should have substituted for them new forms. To us the arguments adduced in favour of the claims of this sound to be Proto-Aryan, appear both in number and weight to be so superior to those of the opposite side, that we think it reasonable to add to the catalogue of the primitive phonetic elements of our linguistic stock the 34 sound I, though we readily admit the possibility, that, both in the last periods of the Aryan unity, and also imme- diately before its division, the I and the r were not yet always quite distinct from each other, and that of the difference which separates them our most ancient fathers had not as yet full consciousness. § 5. Passing now from the study of the consonants to that of the vowels, and of the various and remarkable relations which we see to exist between these two classes of the phonetic elements, we hasten to mention, in the most ' Zeitschrift f. vgl. spraohforschvng, xxii. 356-61. 26 PART I. CHAP. I. § 5. complimentary terms, the important work of J. Schmidt, entitled, Zur gescJdcUe des indogermanischen vokalismus (Weimar, 1. 1871 ; ii. 1875). According- to the intentions of the author it should consist of three parts, or of three monographs, distinct, but nevertheless closely allied to each other. The first two have already seen the light, and one of them investigates the action exercised by the nasals on the preceding vowels, the other that of r, I on the neighbouring vowels ; the third, the publication of which does not seem to be close at hand, will attempt to solve the problem whether, in the Proto- Aryan mother-language, there existed roots of like meaning with different vowels, one set of roots beside the other, and, if that shall appear to be the fact, in what way such diversity of vowels has originated.' This is one of the most solid, rich and risefiil works which have been given to the public in the last few years in the field of Aryan philology, in that it throws light upon a series of important facts, not yet sufficiently examined, with a rare diligence and learning which is extended in a wonder- ful way to all the families of our linguistic stock, and with an uncommon acuteness of skill; arriving at results exceed- ingly important both for the history of the Indo-European 25 vowel-system generally, and for that of the individual languages. We regret that the limits of this work constrain us to notice only the chief among such results, and that the reader cannot form an adequate conception of the minute disquisitions in which consists the value of the work under discussion : but we are consoled by the hope that a book of such worth will be read and reflected upon by all who give their attention to philological studies.^ After some considerations which need not be noticed here, ' Ibid. sect. 2nd, p. iv. 73-92)and of Bezzenberger (Go^ ' See on tliis work the two cr'iti- tingischegelehrte anzeigen,lS75,-pp. cal articles of Delbriick {Zeits- 1313-44). chriftf. vgl. sprachforsclmng, xxi. SOUNDS. 27 Schmidt, in the first of his monographs^ proceeds to treat of the lengthening and increase of vowels caused by follow- ing nasals. And starting from the origin of the nasals in radical syllables he observes that all the inserted nasal elements are not simply phonetic and devoid of all ety- mological value whatever, but arise partly, as Kuhn remarked, from nasal sufiixes (cf. Lat. pango and Gr. irrjrf- vv-fM, Sianscv. jungmas and Jwnag'mi) . This passing of the nasal from the suffix into the root will have come about just as in epenthesis or metathesis, which consists in the passing of an i or / into the preceding syllable, and of which Old Bactrian offers us so many examples ; the nasal, added as a suffix, will have given by assimilation a nasal sound to the preceding syllable, and will have then sometimes disappeared : in Greek, e. g., from the stem and root Xa/3- we should have the series *Xa^-vco, *\afi^-va), Xafi^dvo), or *Xa^-vo}, ^Xa^-avw, *Xa/j,^-dva}, in which, as in the preceding, the nasal suffix has been preserved. Afterwards Schmidt proceeds to discuss the lengthening of vowels owing to following nasals in Indo-Iranic, in Teutonic, in Lithuanian, in Old Bulgarian, in the three Northern European families taken together, in Keltic, in Latin, in Greek, in Graeco- Italic, in European : from among the very numerous 26 examples quoted by him we select the Indo-Iranie (mas) (= European mans) [month], the Graeeo-Italic vlianti (Dor. fiKan, Lat. viginti, cf. Sanscr. 'vi(^ati ) and the European sUbh (from stimbh, stambh). Some of the forms quoted show us that the lengthening took place not after the disappearance of the nasal (as compensation), but while it still existed. The increase also may take its rise from the influence of a nasal, so that we have, in the same or in different languages, a form with a nasal beside another with increase : as instances our author adduces junakti and jogate (Ved.) from the root jug, Ihinatti and ihedati (Ved.) from the root bJiid; -Trvvddvofiai {irev6ofj,ai,) and bodhdmi, 28 PART I. CHAP. I. § 5. mingo{mejo) a.ni me /t ami, etc. ^ The cause of these phenomena is made plain in another which we may now mention, the change of the nasal sound into a vowel : hence the action which a nasal exercises on the preceding vowel may be not only quantitative, but also qualitative ; the latter however takes place within limits considerably more restricted than the former, because it is only the primitive a and its successors which are subject to it. The vowel sound in the nasals is closely akin to u, in the opinion of Helmholtz. Hence it happens that before m, nt, a becomes «, or that, as the nasal element gradually changes to o, u, the result is ao, au : an intermediate form in both cases is the vowel changed to a nasal vowel (a) . Examples common to the Aryan dialects, so that they seem to be Proto-Aryan, are the roots stubh (cf. 27 stambh), clhu (cf. dhan) ; Old Indian offers final -us from ^-ant, OM from am ; the fundamental European language supplies the roots rub, lub (from ramb, lamb) , etc. ; the Slavonic dialects are rich in such changes.^ The second monograph is devoted to the examination of the action exercised by the sounds r, I on the neighbouring vowels. In the majority of the languages of our stock the vowel sound inherent in r and I manifests itself with such force that between the liquid and the neighbouring consonant it may be developed into an independent vowel. This vowel is called by Schmidt, by a name borrowed from the Indian ' It is not however certain, ob- fall upon it (Sievers, Orundzuge serves Delbriict, in some Vedio derlautphiisiologie,etc.,-^.2,i,s,(\(i): presents that the increase takes its hence it appears that -m, -nti, and origin from the nasal : the present -am -anti, for example, are only dif- with a nasal and that with an in- ferent forms of one and the same crease may be parallel forms proceed- sufBx; hence are explained differ- ing from the same root, as in Vedic ences of vowels in the several lan- we see so often to be the case, guages and other phenomena of a 3 Osthoff and Brugman as- likenature. See B rug man, iftwaiis sign to the primitive Aryan a sonant sonans in der indogermanischen nasal sound, or one of such a kind grundsprciche (Studien z. gr. a. lat. that the accent of the syllable could gramm., ix. 285-338). SOUNDS. 29 grammarians, svarabhaldi, a term which he thinks more exact than the Greek eTrev6eai,<;, avd'mv^i'i, etc. The quantity of such vowel is^ according to circumstances, and as the grammarians just mentioned teach, various : qualitatively considered it appears identical, e. g., with the e of the Old Bactrian in ddda/rega (= Sanscr., daclarga). The field on which this phenomenon is most frequently exhibited to per- fection is the Slavonic and the Bactrian, so much so that the svarabhaMi may be considered a most important criterion of distinction among the different Slavonic dialects and between these and the languages most nearly related ; but it appears also in the other families of Aryan languages. Sanscrit ex- hibits many examples of vowels qualitatively altered by the influence of the following liquid {ir, il from ar, al — ilt, %l from ar, al — 'thence, by metathesis, ri, ru) ; whence it comes that roots with a change their vowel, interchanging thus with the roots which have i and u (root Sanscr. tul = orig. tal [tollere] — krijate from *ki.rijate, *kirjate, ^karjate etc.). Similar examples are furnished in various ways and in various proportion also by the cognate languages. Re- markable again is the quantitative change of the vowels due to a following liquid : it may suffice to quote as instances the Sanscr. piirna (comp. Old Bactr. perena-, Old Russ. pulvMU, Proto-Aryan parna - [full]) and the Italo-Greek suf- fix -tor- (from the orig. -tar-'). The theory of svarahhakti explains also the positio dehilis : a problem of which W. 28 Corssen and other philologists have in vain attempted the solution.^ § 6. In the preceding remarks we have had occasion to ' See my Grammatica storico- donbtedly that of the author men- comparativa delta lingua latina, etc. tioned in the text, who discovers the Roma-Torino-Firenze, IS^B, pp. reason of the positio debilis in a 102-7, and the Sivista di Jilologia vowel developed between the mute e d'isiruzione classica, ii. 226-39. and the liquid so as to offer an obstacle The best of all the illustrations to the true position, which have been attempted is un- 30 PART I. CHAP. I. § 6. make mention of a phonetic fact of no light moment in the history of the vowel sounds, the increase. It will be well now to acquaint our readers with the results of the researches on this subject made by Priedrich Miiller,' who opposed to the theory of Schleicher three propositions strengthened by proofs. And in the first place he aflBrmed that the in- crease in its origin was peculiar only to the two vowels i, u, and that it was not till later that there was gradually deve- loped an increase of a. To such an opinion the learned phi- lologist is led principally by the consideration that in all the languages of Aryan stock we see to correspond with each other only the increases of i and of u : with respect to the increases of a uncertainty prevails sometimes even in the individual languages. And he notes cases in which in Old Indian we find a where we should expect the first increase of this vowel, while in others, quite similar, a is found j instances quoted by him are qravas (from qtu) , teg' -as (from tig'), gan-as (from gan), vds-as (from vas), etc. In Greek o corresponds sometimes to the fundamental vowel a,sometimes to the first increase of it. Secondly, he asserted that the increase takes place only in the radical vowels^ observing that, in his view, the nominal bases in i and u, which in some cases exhibit in their termination an aj or an av, are 29 not at all against his doctrine, because the stem which is commonly thought to be affected by the increase is, accord- ing to his opinion, the stem in a primitive form : stems in aja, ava, became gradually stems in aj, av, then on the one hand in i, w, on the other in a ; the stems in aj, av, were preserved before certain case-suffixes. In like manner he discovers more ancient forms in the stems of presents in -nau, au, than in the corresponding stems in -nu, u? Third 1 Die vocalsteigerung der indo- ^ Compendium,etc.,-p-p. 11-12, etc. germanischen sprachen (Siizungs- ^ ^^s Asco\i, Sludiario-semitici, berichte der K. Academie der wis- second article (JHemorie del S. Zsti- senschaften, philosophisch-histori- tuto lombardo, x. 13-36). sche classe, Ixvi. 213-24.) SOUNDS. 31 and last comes the hypothesis that the Proto- Aryan increase of the vowels was restricted to that which Schleicher terms first {gunas of the Indian grammarians), by which from a arose aa, from i ai, from u ait : the second increase [vriddhis), whence in Sanscrit we have a { = dd) from a, di from i, du from u, did not, according to Muller, belong to the primi- tive and fundamental Aryan, but was developed separately in the several languages of our stock. And indeed — in the first place — Old Bactrian exhibits but very slight traces of the second increase ; Old Persian offers only the two diph- thongs ai and au ; 3ndly, Old Indian in certain cases of vriddhi presents forms phonetically decayed to such a degree that we cannot assign them to the primitive Aryan ; 3rdly, the Greek diphthongs ot and ov are wrongly given by Schleicher as the representatives of the di and the du of the second increase, because in the former the short o cannot be considered as representing a long a of the fundamental lan- guage ; 4thly, Latin, moreover, Gothic, Lithuanian and Old Slavonic, according to the author, are opposed to thedoctrine of Schleicher. He himself in fact, as appears from note 1 to § 2 of the work quoted, did not consider as quite certain, but indeed only in the highest degree probable, the existence of the second increase in the mother-language which he called Indo- Germanic, and in the scientific reconstruction of which he so successfully co-operated, nor did he disguise the fact that in the use of the vriddhi the individual languages 30 frequently do not agree. The various hypotheses proposed by philologists to ex- plain the origin and the cause of the increase are critically discussed by W. Corssen, in a few pages, which students will certainly not read without advantage.' According to this learned investigator the increase, like the acute accent, tends to emphasize the syllable which, owing to its peculiar • Ifler aussprache, volcalismus und heionung der lafeinischen sprache, Leipzig, 1868-70, i. 620-2. 32 PART I. CHAP. 1. § 6. meaning, is the most important for the speaker : two such modes of strengthening were naturally often united^ but without being necessarily connected. The Aryans would not, and perhaps could not, in certain forms be contented with the simple lengthening of i, u to I, u, but felt also the need of uniting such sounds to the strongest and fullest of the vowel sounds, that is, to a, as far as it was possible without changing entirely the special phonetic character of i and u, in other words, without abandoning entirely the position of the organs of speech in which i, u, are pro- nounced. Opposed to this is the view of W. Scherer:' according to him it was not emphasis, but ease of pronun- ciation that changed I, u into ai, au. Among the primitive vowel sounds the most easy to pronounce was a : u and i re- quire an effort. If, therefore, the most natural position of the organs of speech in articulating the Proto-Aryan vowels was that which produced a ("indifferenzlaut''^), in order to make easier the action of the muscles necessary for pro- nouncing * and u, the duration of these last two sounds kept on diminishing and there was developed before them an indeterminate vowel, which finally resulted in a, in such a manner that it was only by degrees that the transition was made from the normal state to the extremes of articulation.^ ^1 Against this hypothesis of Scherer there are, in our opinion, formidable objections in two important facts, con- firmed by a large number of examples : first, the frequent coincidence of the increase with the acute accent, from which is shown the need the speakers felt of strengthen- ing certain syllables furnished with special significance with reference to the sense, as was well observed by Corssen and other philologers ; secondly, the decay of a to i and «, a phenomenon of which historico-comparative grammar offer's 1 Zur geschicUe der deuischen physiologically considered the above spraohe, Berlin, 1868, pp. 21-9. quoted work of Sievers, pp. 34-50, 2 With reference to the vowels will be consulted with profit. SOUNDS. 33 a large store of examples, and from which we perceive that that instinctive inclination to ease, to laziness, which led the Aryans to strive after expressing phonetically their thoughts with the least possible tension of the vocal organs, and was the cause and the supreme law of so many weak- enings and vanishings of sounds, was much less hostile to i and u than to a. And of this weakening we even find not a few examples in that dialect, by means of which we can best ascend to Proto-Aryan, that is to say the most ancient language of the Indian Aryans.' § 7. The discussion on the increase in the Indo-European languages necessitated a mention here and there of the accent. It is our intention to make some allusion to this most important phonetic fact, both because Schleicher , did not think it a proper subject for treatment,^ and because to the labours devoted to this subject towards the middle of this century by Benfey, by Benloew, 32 by Weil, by Bopp, have been added in the last few years those of Corssen;' and, thanks to him and Baudry,* important conclusions have been drawn which ' ^cU.e\.eTaeT, Compendium, etc., gesichtspunkte fast auf altindisch pp. 20-3. t""i griechiseit zu beschranken hat, 2 " Die betonung der worte obgleioh der echte accent nns eine scheint zwar besonders wegen der in veranderung der vocale zu sein diser beziehung zwischen griechisch scheint, die, der steigerung ver- und altindisch obwaltendeniiberein- gleiohbar, zum zwecke der stamm- stimmung schon in der indogerman- und wortbildung dient " ischen ursprache in bestimter weise {Compendium, § 13, Anm. 2.) fest geworden zu sein, die vor ligen- ^ Uber aussprache, etc., 2nd ed., den sprachen (die beiden genanten ii. 794-1000 : Uber die sfracJie der aus genommen) gehen aber in irem Mruslcer, Leipzig, 1874-5, ii. worttone so stark aus einander, dass 364-83. eine ermittelung irer urspriinglichen * Grammaire comparie des lan- betonungsweise unmoglich ist. Wir gues classigues, Paris, 1866, i. 14-30. scUiesseu deshalb die lere von der See also Misteli, tfber griechische betonung aus, da sich eine vergleich- betonung, sprachvergleichend-pTiilo- ende zusammenstellung der indo- logische abliandlungen, i. Paderborn, germanischen sprachen unter diesem 1875. 1) 34 PART /.. CHAP. I. § 7. cannot but prove acceptable to students of tbe Aryan lan- guages. Tbe bistorico-comparative investigation of tbe tonic laws in tbe ancient dialects of India, of Germany, of Greece and of Italy, revealed two principles of accentuation substantially different, tbe logical principle and tbe pbonetic : tbe former prevailed in tbe first two, the latter in tbe remaining two of tbese languages. The Sanscrit and the German accent appear to be altogether independent of tbe quantity whether of tbe entire word, or of any one syllable whatever. Tbe first especially is seen to be endowed with wonderful freedom, as being able to fall on any syllable from tbe beginning to the end of a word, so that its position is not at all governed by phonetic laws, but, as we shall see, by logical laws which determine tbe syllable on which in the individual words tbe tonic elevation ought to rest. In the examination of these logical laws worked Benfey, Benloew, and Weil, and tbe founder himself of Indo- European comparative grammar, Franz Bopp. But, while Benfey regarded as primitive only that accent which falls on tbe prefixes and on tbe suffixes j while Benloew de- veloped this principle in his doctrine of tbe " determinant final," according to which of several formal elements added to one root to determine its meaning, that one is held to have originally received the accent which was attached last, as representing the last idea which makes a more powerful im- 33 pression, especially on crude minds of very ancient ages ; Bopp, moved by other considerations, would not recognise an original accent except in the initial radical syllables of words, regarding as an indication of decay the accent of tbe final syllables. Tbe examination, which was on both sides instituted, of tbe opposite theory showed that neither the first nor the second doctrine answers exactly to reality : it showed that tbe two principles mentioned are both true, but only within limits considerably less extensive than their sup- SOUNDS. 35 porters have imagined^ both false beyond such limits ; it showed that far from being mutually and inexorably exclu- sive, the two opposed views can and should be reconciled, and only on this condition is a scientific understanding of the facts possible. And, indeed, as the most ancient Indians were inclined to pronounce with the acute accent that syllable which appeared to them more important with respect to the sense of the word, and as this seemed to be sometimes the root-syllable, expressing the fundamental idea of a word, sometimes a syllable belonging to an afi&x representing a vividly conceived determination of that idea, it naturally could not but happen that they attracted to themselves the accent, with varying alternation, like two opposite poles, now the one and. now the other remaining victor in an even contest. In any case it is clearly seen that the supreme law of the accent in Sanscrit is not phonetic but logical, at least as far as results from the investigations above mentioned. The fundamental principle too of the German accentuation is logical : the acute accent falls on radical syllables, deno- ting the substantial meaning of the words, or on syllables of prefixes which limit it. The basis, on the other hand, of the Hellenic and the Italic accent, and especially of the latter, is phonetic and not logical. The severe laws which, in the classical ages, do not allow it either to pass the limits of the antepenultimate syllable in retreating from the end of the word, or to rebel against the infl.uence of the last syllable when long in Greek, and the penultimate when long in Latin, were, we may say .perhaps with considerable 3* probability, foreign to the prehistoric epochs of the languages in question : nevertheless he would make a bad guess who should attribute to the Italic and the Greek accent of the most ancient times not merely greater freedom, but an abso- lute independence of phonetic influences, an independence which nothing gives us the right to assume, and which we should not know how to reconcile with the principles which 36 PART I. CHAP. I. § 7. govern the Greek and the Latin accent in the historical periods of their existence. But, since we have noticed the accent in the two classical languages without as yet distin- S'uishino' the tonic characteristics of the one from those of the other, because both in the latter and in the former the phonetic principle prevails, we must now observe that the Greek accent moved within the limits which were assigned to it, as well as to the Latin, with considerably more freedom than the other ; not shunning the last syllable (except only in the verbal forms for the most part and in the jEolic dialect) and allowing the. length of the final syllable to regulate it to a considerably less extent than the long penultimate does the Latin accent, of which the quantity appears to have been the vital principle. And in many cases the tonic system of the Greek finds its counterpart in that of the Sanscrit, and seems to stand, we should almost say, intermediate be- tween that and the Latin. And now it would be well to consider whether the two principles brought into notice, the logical, which governs the Indian and the German accent, and the phonetic, by which the Greek and the Italic accent is guided, are both equally ancient, or whether the one of them is more nearly original, the other developed in a later age, "V^Tien we consider that in the Old Indian there still rings more clearly than in the other dialects of the same stock the echo of the primitive word of the Aryans ; when we reflect that it is quite reasonable to suppose that they made use of the tonic elevation, as of every other element of the language, to express their ideas ; when we think of the high signifi- cant value which the accent has in other languages ; when, 36 finally, we observe that the elevated tone, by which a syllable in a word composed of several elements is affected, may be often rightly compared, as Baudry has done, with the rhetorical accent, which among several words of a sentence or a period throws into relief one of them from its peculiar SOUNDS. 37 importance with regard to the sense : the hypothesis will appear undoubtedly quite natural and well-founded, that the primordial cause of the accent was the instinctive inclina- tion to raise with a higher tone than the remaining syllables the one representing the idea, the importance of which relatively to the meaning of the whole word seemed to the speaker greatest. Hence the tonic system of Old Indian and Teutonic, a system in which sometimes the radical syllables, sometimes those of the affixes appear strengthened by the accent, according as the first kind or the second made stronger impressions on the minds of the speakers, appears to us considerably more original than the Greek and the Italic, in which, and especially in the latter, the logical principle was forced to surrender the field to the phonetic. And the cause of this fact is probably to be sought in the different transparency, so to speak, of the word in the different dialects which we are discussing. For — while in Old Indian the significance of each of the several elements of which the words were made up was still clearly seen in many cases, and hence the accent marked with different elevation of voice the different importance of such elements with reference to the meaning — in Greece and in Italy the recol- lection of the internal, primordial constitution of the word was less and less present to the speakers. The word ceased to appear as a whole composed of several parts : men no longer saw clearly the multiplicity of the molecules, so to speak, which united to form it, but only the unity of the whole. The meanings of the several parts were more and more confused in the synthetic sense of the whole : consequently the primitive tone indicating the different relative value of the different elements of a word was succeeded by an accent the only function of which is to mark the individuality, the se independence of a word. The struggle, if we may so ex- press ourselves, between the two tonic systems described reveals itself also here and there in Greek, where, espe- 38 PART I. CHAP. I. % 7- cially in the nouns^ we still find not a few traces of an accentuation closely akin to the Sanscrit^ while in the verbs there prevail tonic tendencies less ancient^ and the ^olic dialect with its dislike of oxytones agrees with Latin, as has already been aptly observed by Priscian. The preva- lence of the phonetic principle is greatest in Latin, where the quantity exercises a much more powerful influence than in Greek. And what we have said of Latin we must, if we would not refuse credit to the results of Corssen-'s acute investigations, affirm also of Umbrian, Oscan and the other Italic dialects which are more closely connected with them, and even of Etruscan, in which the constant absence of the acute accent in the final syllables must have been, according to the eminent investigator, the reason why such syllables have so frequently been weakened and vanished : a fact which united with others to impress on Etruscan word that strange character so well known to all, and to obscure all appearance of relationship between the Etruscan and the Italic dialects of which Cor s sen believes and proclaims it a brother.' But from the study ot the later popular Latin to the third century of our era,^ and of the Neo-Latin dialects,' we perceive how the accent, already victim of the quantity in Latin, has withdrawn itself from the dominion of quantity and subjected this to its own power, in such a way that the acute accent, having 1 Seemy CennisopTaGuglielmo comparison with Latin to discover Corssen, e la lingua etrusca, Fi- the laws of the accent in those dia- reuze, 1876 (extracted from the Si- lects, laws which appeared to him visia Mwropea). As is known, the exactly similar to those of the Latin learned philologist, not being able to accent. derive either from written symbols, ^ Schuchardt, Der voJcalismus or from testimonies of ancient gram- des vulgarlateins, Leipzig, 1866-8, marians, any ideas with reference to passim. the tonic system of the Italic dialects 3 Diez, &rammatiJc der Moman- rolated to Latin, availed himself of ischen sprachen, Bonn, 1870-2, i. an accurate and acute observation of 500-12. the phonetic corruptions, and of the SOUNDS. 39 again become the absolute master of tbe word, by strengthen- 37 ing above all the rest the syllable which it raised, not only kept it long if it was so already, but lengthened it if it was short, while the length of the syllables with a grave accent became diminished. And not only in Latin, but also in the Vulgar Greek of the decadence, quantity was forced to surrender to accent.' Therefore, while quantity, as long as it exercised its influence on accent, was the principle which shaped the old classic verse, accent having become free again, and having made itself master of quantity, began to govern the formation of the Latin and the Greek verse^ with a power which became greater and greater in process of time. § 8. Hitherto we have considered the Aryan sounds in themselves, without paying any attention at all to their pos- sible relations of afiinity with the phonetic elements of other languages, and especially of the so-ealled Semitic family. Whether the origin of Proto-Semitic and Proto- Aryan can be said to be common, so that these two mother-languages ought to be considered as two forms of a very ancient speech which might have contained them in germ, perhaps with others, is one of the most difficult problems on which philologists have laboured, which they have not yet been able to bring to such a solution as would suffice to put an end to the long dispute between believers and non-believers , in the primitive unity of the Aryan and the Semitic stock in a single more ancient stock which may well be called, and has been called, Aryo- Semitic' To the school of ' Mullacli, CframmaUk der people.' grieohisehen vulgarspraehe in his- » For the attempts to discover and torischer entwicklung, Berlin, 1856, to demonstrate community of origin pp. 70-3. — Sophocles, A glossary between the one and the other class of later and Syxantine Oreeh, Lon- of dialects and the discordant views don, 1860, pp. 37, 64, seqq. of eminent philologists see De- 2 Hence the so-called verse iroKi- litzsch (Priedrich), Studien, vber TtKoe, or ' common, in use with the indogermanisch-semitische wurzel- 40 PART 1. CHAP. I. § 8. 38 Renan, of Schleiclier, and of the others who deny the pre-historic existence of a parent language of the two great linguistic systems, belongs Friedrieh Miiller, who in a very brief but copious monograph/ set himself to demon- strate that " Indo-Germanic and Semitic are two linguistic stocks fundamentally distinct, each of which pre-supposes an origin independent of that of the other/'' And in his analysis of the elements of the two stocks, beginning with the consonantal sounds, he observes : first, that while in Aryan the aspirates gh, dh, hh correspond to the sonants g, d, h, on the contrary in Semitic the surds Tc, t, p, appear with aspiration in h, s, i,ff secondly, in Aryan the sound I has been developed from r, in Semitic I seems to be the original ; thirdly, in the latter stock we have, beside the guttural and dental surds, emphatic sounds which the Aryan stock altogether lacks. And, with respect to the vowel sounds, he observes that the diphthongs ai, au, which we find in Proto- Aryan, do not belong to Proto- Semitic. More- over, in the constitution of the syllable, he perceives some diversity between the two linguistic systems alluded to. These arguments, the worth of which is certainly not very great, are followed in his treatise by others of far greater force : but the order which we must adhere to obliges us to defer the exposition of them. We now proceed to speak briefly of the supporters of the Aryo-Semitic unity. "Within the narro\v limits of time assigned to our treatise, that is to say within the last decade, we have certain monographs of Rudolf von verwandscTiaft, Leipzig, 1873, pp. senschaften, phil.-hist. cl., Ixr. 5- 3-21, and my Introduction cl I'litude 20). de la science du langage, transl. ... ^ It should be observed, however, by V. Nourrissou, Paris, 1875, pp. that also in some Aryan languages 134-41. surd sounds are aspirated, as e.g., in 1 IndogermaniscTi und semitiscTi, Old Indian which possesses surd and ein leitrag znr wiirdigung dieser sonant aspirates, and in Greek, which beiden sprachstdmme {Sitzungsbe- has only surd aspirates. richte der K. Academie der Wis- SOUNDS. 41 Raumer in continuation of the writings previously pub- 3" lished on this subject/ and the book already quoted of Priedrich Delitzsch. Starting from the first, it will be well to examine attentively his method of investigation before setting forth the results of it. The proofs of kinship of the Aryan languages with one another, so our author begins, are : 1st, the primitive identity of flexion ; 2nd, the extremely definite phonetic laws common to all the languages mentioned ; this second proof is of greater im- portance than the first. The same indications of affinity should be sought between the Aryan and the Semitic stock. Nay, just as flexion evidently originated in great part after the separation of the two stocks,^ so we cannot expect many indications of original identity in this class of facts, and we ought rather to direct our attention to the affinity of the Semitic sounds with the Aryan, regarding as a chief, and almost only mark of common origin, the regular corres- pondence of the former with the latter. Of the objection which others may found, and which has been so often founded upon the different constitution of the roots in the two linguistic systems (because the comparison of ■ the ' G-esammelte spraokwissenschaft- no stock of languages is flexion a UoTie sehriften, Frankfurt a. M. primitive fact, but in reality a form 1863, pp. 460-539. — Rr. prof, developed from others more simple : Schleicher, in. lena mid die uner- a position maintained especially by wandtschaft dersemit.u.indoeurop. Schleicher and by M. Miiller, sprachen, ib., 1864. Fortsetzung fiercely assailed by Pott, by Be- der untersttchimgen iiler die urver- nan and by other philologists (see wandtschaft der semit. u. indoevarop. my Introduction, etc., pp. 120-6, sprachen, ib., 1867. Zweite fortset- and the 3rd chapter, § 17 of this ZMB^, etc., ib., 1868. Driite fortset- book). Hence the illustrious French xmg, etc., ib., 1871. Vierte fort- Semitic scholar denies the possibility tetzung, etc., ib., 1873. Die urver- of an extremely ancient ante-gram- wandtschaft der semit. u. indoeurop. matical affinity between the two sprachen (in the Zeitschr. f. vgl. stocks, because, in his opinion, every sprachforsch., xxli. 235-50). language comes into existence with 2 This assertion, as is clearly its grammatical system already com- evident, cannot be well received ex- pletely formed in its essential parts, cept by those who believe that in 42 PART I. CHAP. I. § 8. sounds is not possible without that of the roots), we shall have to speak towards the end of the following chapter. 40 Here, on the other hand, we must give some hint of the method on which Raumer thinks the Semitic and the Aryan sounds ought to be compared together. He holds that it is necessary, with respect both to the one and the other stock, to ascend to the most ancient period to which the investigation can reach, for example, far beyond the Hebrew known to us : but he does not think that the Proto-Semitic and the Proto-Aryan must needs be recon- structed in order to compare them together, being of opinion that such reconstruction cannot be completely achieved, and that hence it would happen that some primitive elements contained in one, or in some of the languages of each stock, would not be made good use of. But, we observe^ how is it possible, without that reconstruction, to distinguish the Proto-Aryan and Proto-Semitic elements from those which were only developed later in the individual languages? And without this necessary criterion do we not perhaps hourly run the risk of regarding as Aryo- Semitic an element which only accidentally presents itself to us in the same form, or in forms apparently cognate, in two or more languages of different stock ? In the study of the phonetic facts and in the research into their laws, Raumer limits for the most part the investigation to Hebrew, Greek and Latin, and considers it not impossible to discover regular Aryo- Semitic correspondences of sounds even without extending the limits of the comparison beyond the field described, while he quotes in proof the discovery of the Teutonic ' lautverschiebung.^ But the proof will not appear to all, we believe, quite appropriate to the present case, and, as F. Delitzsch well remarked, a reference to Sanscrit and Arabic would have been of the greatest advantage to him. It remains now to see what fruits the author has gathered from his comparative investigations, what phonetic laws he SOUNDS. 43 has discovered, or at least believed himself to have dis- covered. The phonological results of his researches may be epitomized in the two following propositions : 1st, the hard Semitic explosives or mutes are represented etymo- logically by the corresponding Aryan sounds; 2ndj the soft Semitic explosives for the most part find their counter- *8 part in the hard Aryan homorganic sounds (e.g. an Aryan JO would correspond to the Semitic i).' This last proposi- tion is not, in the opinion of Delitzsch, either demonstrated or demonstrable by any certain example. Raumer, con- vinced of having not only affirmed, but also furnished with strong proofs his phonological equations, goes on to observe that such constant correspondences of Aryan sounds with Semitic sounds cannot be considered either as accidental, or as ptoduced by the passage of words from the one to the other stock, or as due to a natural afiinity between sounds and meanings. For the Aryan languages not only exhibit in great abundance sounds equivalent to the Semitic, but also offer us in no less number sounds different from those, which they represent according to laws I quite definite and constant : the first case is found with respect to the hard, the second to the soft Semitic explosives or mutes. He who would seek the cause of similar facts in the natural power of the sounds might by arguments altogether identical, deny the common origin of Greek and German, of German and Swedish, and even of High and Low German. In the last pages of his book above quoted Friedrich ' These phonological equations " aspiration," by which there would proposed by Raumer for the first correspond to the Sanscr. i, <, ^, the time in 1863 were succeeded, in the Semit. M, th,pTi; that which af- following year, by the seven laws on firms the afiinity of the Sanscrit which stress was laid by Ascoli in aspirate medials with the Semitic the letter to A. Kuhn, Del nesso unaspirated medials; lastly, that ario-semitico (PoUtecnico, xxi. which establishes the primitive 190-216;, among which we may be identity of the Sanscr. g with the allowed to notice at least that of Sem. j. 44 PART I. CHAP. I. § 9. Delitzsch. gives us, as the result of the comparative investigations made by him with respect to a large number of Semitic and Aryan roots, a table of Indo-European consonants with the sounds corresponding to them in Arabic, in Ethiopic, in Hebrew, in Chaldaic, in Syriae. Particularly worthy of remark in this catalogue seem to us the equations of Aryan p with Semitic f (especially in it Arabic and in Ethiopic), and of the Aryan v with the Semitic v and / ; among other things also, it appeared to us remarkable to see represented in Semitic the Aryan aspirates dh and hh no less than the unaspirated sonant homorganic d and h. In many of his phonological com- parisons Delitzsch agrees with Raumer: but there stands a barrier between them in the second of the two laws of Raumer, which is absolutely denied by Delitzsch: who, as will appear better from the following chapter, regulated himself in his comparisons in a way to deserve praises for judgement of no common order. § 9. The comparison was extended to a wider field by Schultze in his short dissertation entitled Indo- germanisch, semitisch und hamitisch (Berlin, 1873). Later on, when we come to discuss roots, stems and words, the reader will have an opportunity of discerning adequately what is the worth of the comparison which the author instituted with reference to the three stocks of language mentioned. Suffice it now to make some allusion to his phonological comparisons. Starting from the vowels Schultze holds that in all the languages of the Noachidae (as he calls them) a is the fundamental vowel, i and u alterations of it, which arose by means of a lingual {i) or labial (w) constriction of the vocal tube. These three simplest shades of vowel sound reveal themselves to us originally not quite distinct from one another in reference to meaning, as appears especially from the Semitic writing- system. We must call secondary, chiefly in Hamitic, the SOUNDS. 45 long vowels : and secondary the diphthongs, the complete development of which only took place in Aryan after its separation from Hamito-Semitic^ but before its division into several languages. Proceeding to discuss the consonants Schultze notes as peculiar to the Semites alone the use of the weak faucal sound (^|rlXov 'jrvevfjua, aleph) without a vowel. There came afterwards, but before the rest, the consonants b [p), d (t), g (Jc) : their greater antiquity 43 appears to Schultze sufficiently demonstrated by the great simplicity of the hieroglyphic symbols which representthem, and by their position in the Semitic alphabets immediately after aleph. The difference between the media and the tenuis is not primitive, as is clear from the written symbols, and from the frequent alternating of the one with the other. After some remarks on the origin of the nasal sounds the author bids us attend to the appearance on the scene of r and s, the affinity of which he attempts to prove by a scant number of examples. In his opinion the more ancient of these two sounds is r, which exists in languages that have no s : r produces I, whose existence does not go back beyond the period in which the Hamito-Semitic stock was sub- divided into two stocks. All the other sounds are relatively more recent. The combinations of sounds employed in all the languages of the Noaehidae in the constitution of the syllable are: 1st, spiritus lenis + vowel {'a, etc.) ; 2nd, con- sonant + vowel {&a, etc.); 3rd, spiritus lenis + vowel + con- sonant {'ai, etc.) ; 4th, consonant + voweH- consonant {6ai, etc.), as to which last form of syllable it is not known for certain whether it already existed before the first separation. Those forms which begin or end with several consonants are expanded forms of more simple syllables. The mention which we shall have to make, in the following chapters, of the other parts of this little work will serve, as we have just said, to make us estimate rightly also the worth of that which we have just been epitomizing. 46 PART I. CHAP. I. § 9. It is chiefly from its strange novelty that our attention is drawn to the recent work of Eeinisch on the unity of origin of the languages of the ancient world.' Reinisch considers the races of the ancient world as descending from 44 one family whose primitive seats were near the equatorial lakes of Africa. Hence the Southern and Central African dialects, the Erythraean (Semito-Hamitic)' and the Aryan. Among the speeches of Interior Africa a very remarkable one is the Teda, the language of the people known under the name of Tibbu or Tebo, which for quite 3000 years has inhabited the eastern margin of High Libya from the desert of Kufara down to Lake Tsad. The study of this dialect; which Reinisch believes to be related to the others above mentioned, and in which most uncommon interchanges take place of the dentals with each other, and with many other sounds, and the investigation of the speech of children, led the author to imagine the following genesis of the sounds. In his opinion the first sound of the fundamental language was t {d) : from this grew the gutturals k {g) and the labials^ {b) ; then the semivowels _;, I, r, n; then h, w, and afterwards from these the vowels %, a, u. It is hardly necessary to observe that this book which is, to begin with, hardly to be commended for exactness in the quotations of Semitic and Aryan words, attempts to propagate hypotheses not founded on any solid base, and contrary to the most certain results of philological investigation into the dialects of the Aryans and the Semites. ' Die einheitlicJce tn-spmng der ' This name " does not seem to spraclien der alien, welt, nachgewie- us appropriate, because by the name sen durch vergleichung der qfrioan- Mare Erythraeum the oldest and ischen, erytJirdischen und indoger- most accredited authors of antiquity manischen sprachen mit zugrundeU- uiiderstood not the Eed Sea of to- gung des Teda, i. Wien, 1873. In day, but the Perso-Indian Sea." our remarks on this work we shall (F. Miiller, Grundriss der spraeh- make use of the opinion passed upon wissenschofl, i. Wien, 1876, p. 135, it by the ZiterariscJies centralblatt, note). 1874, pp. 636-8. SOUNDS. 47 To these remarks we might still add a word or two oa the comparisons, not very conformable to the principles of the new science of language, made by Edkins in a recent book of his ' between the sounds of the Aryan languages of Europe and those of Chinese and of Turanian languages, as some still call them. But we shall find a fitter place to discuss the comparisons of E dkins in the following chapter, in which we propose to treat of the recent studies on the subject of the Aryan roots considered in themselves, and in their possible relations of common origin with those of other languages. 1 China's plaee in philology, etc., Loudon, 1871, pp. 321-49. OHAPTEE II. Koots. § 10. The results of the great etymological labours pro- secuted according to the severe rules of the new comparative method on the subject of the languages of Aryan stocky and especially of the wonderfully wide researches of Pott are presented to us collected and set forth with a useful novelty of arrangement in Fick's Vergleichendes worterhuch der indogermanischen sprachen, a work of which we have already the third edition with some very important additions of the author (Gottingen, 1874 — 6). It is divided into seven parts contained in three volumes : the 1st (I. 1 — 258) is devoted to the words of the Indo-Germanic mother -language ; as Fick still terms it ; the 2nd (I. 359 — 468) to the words peculiar to the Indo-Iranian (Aryan, according to Fick^s nomenclature) linguistic unity; the 3rd (I. 469 — 843) to the words peculiar to the European linguistic unity ; the 4th (II. 1 — 288) to the words peculiar to the Slavo- Teutonic linguistic unity; the 5th (II. 289 — 508) to the words peculiar to the Slavo-Teutonic linguistic unity ; the 6th (II. 508 — 701) to the words peculiar to the Lithu-Slavonic linguistic unity, with an appendix (II. 703—84) on the Prusso-Lettic words; lastly, the 7th, which comprises almost the whole of the third volume, gives the words peculiar to the German linguistic unity ; the fourth and last volume contains an Appendix (3 — 120) andnumerous indices compiled by Dr. A. Fiibrer. It is certainly to be regretted that Fick has not extended his investigations also to the Keltic family, in order to be able to assign them in his lexicon the place which up- doubtedly belongs to them, and that he has contented himself with adding to the third edition just published of ROOTS. 49 the Vergleickendes worterinch the Keltic words with which, ^ thanks to Windiseh, the fourth edition of the Grundziige der griechiscJien etymologie of G. Curtius (Leipzig, 1873) has been enriched. This lacuna, we repeat, is to be regretted, both as far as concerns the lexicon considered in itself, and as far as relates to the division and subdivisions of the primitive and fundamental language of the Aryans, and the historical problems which are so closely connected with them. Nevertheless, in spite of this defect which the wonderful industry of the author will, we trust, remove from a new edition of his work, we may with Windiseh' regard it as one of the most important works which have in the last few years been given to the public on the subject of the Science of Language. We invite the attention of our readers to the Appendix above mentioned, entitled Roots and root determinaiwes, which appears to us of no slight moment for the subject to which the present chapter is devoted.^ The author begins by distinguishing two classes of roots : 1st, roots expressing ideas which only a being conscious of himself can conceive and represent phonetically ; 2nd, roots which do not presup- pose self-consciousness (interjections, imitations of sounds, children's words), the influence of which on the formation of the Proto-Aryan language Pick reduces to its due limits. And turning his attention to the first and far more im- portant kind, he observes that it is only with the distinction between pronominal roots and verbal roots that the real human language, and the possibility of its development, commence : " the thought which is founded on the self-con- sciousness begins with the capacityfor dividing anyperception 1 Zeitsehr. f. vgl. sprachforscTi, single volume in which this lexicon xxi. 385-434. ' is contained). In our criticisms -< This appendix is found also in and quotations it is this edition to the 2nd edition of the Vergleichendes which we have adhered. worterhuch (pp. 927-1044. of the 50 PART I. CHAP. II. § 10. whatever into its two fundamental elements, of distinguish- ing the author of the action from the action itself, and *7 reuniting the former with the latter." Such a distinction had its phonetic expression in the distinct, but contem- poraneous, creation of syllables denoting only the subjects (pronominal roots), and of syllables representing the ac- tivities put in operation by the same (verbal roots) . From the more and more close union of the one kind with the other kind of roots was produced the Aryan word, a verb or a noun according to the prevalence of the verbal or of the pronominal element. And here the author proposes to himself one of the most arduous, but most seductive tasks which a philologist can set himself: the analysis of the constitution of the roots. Many of them are considered by the most eminent students of the Science of Language, as some of the so-called simple bodies are by chemists, rather as not yet decomposed than as not decomposable. Now if the portion common to two or more stems, both in their phonetic matter (we ask pardon for the expression) and in their meaning, gives us the right, nay, imposes on us the obligation, of ascending to the root whence both spring, why shall it not be possible and obligatory for us to institute a similar comparison between two or more roots which are related to one another as those stems are, and, by means of the comparison, to discover the most simple root of which they appear to us to be expanded forms ? And this analysis would not only serve to furnish us with more exact notions about the first significative elements in the Aryan languages, and in their mother-language, but also, as BreaP excellently remarked, to bring to light new relations b€rt;ween the ideas of our most ancient forefathers, and perhaps also to reveal new affinities between linguistic stocks. The 1 Bopp, Grammaire comparie trad. . . par M. M. Sreal, Fava, des langues indo-europiennes .... 1866-74, ii. xxii. ROOTS. 51 difficult enterprise has already been attempted here and there, on various principles and methods, with varying extent of investigation and with various results, by several philologists, among whom it may suffice to quote Pott and G. Curtius.' The chief result of Fick's investigations, 4 which succeeded to the researches alluded to, is the following proposition : there are in Proto-Aryan and in the dialects which spring from it primary roots, that is to say, no longer decomposable, and secondary roots derived from the primary; these we findformed, 1st, only fromsimple vowels («,«,M),3ndly, from the vowel a + a consonant [ad, ap, as etc.), 3rdly, from a consonant (simple or double) + the vowel a {da, sta, etc.). Every root otherwise formed derived its origin from one of the primary roots just described, and this came about either by means of an alteration of some sound, or by way of reduplication:, or owing to the addition of some final element, which Fick calls, with Curtius, 'a root-deter- minative.' The proof of Fiek's theory rests in the fact, certain in his opinion, that all, or nearly all the roots of a structure not agreeing either with the first or the second 1 Pott, MtymologiscTie forschun- attention, when they should have gen auf dem geUete der indo-ger- done, to the weakening of tlie manischen spraohen, etc., Lemgo- vowels. According to Ho velacque Detmold, 1859-73, Part 2,. Section 1, every really simple element, verbal p. 265. sq. C\irtius,G., 6-rund- or pronominal, of the Indo-European ziige der griechischen etymologie,. language consists either of a, vowel, Leipzig, 1869, pp. 32-71. We or of a consonant and a vowel, or of should like to notice also a mono- two consonants followed by a vowel, graph of Hovelacque with the so that the root is in every case an title Racines et iUments simples open syllable. Every root in or is dans le sist^me Unguistique indo- an expanded form of a root in r-. europien, Paris, 1869. — The author in any Indian root whatever, ter- censures modern linguists for fol- minating in a consonant, this is the lowing too closely the teaching of initial sound of a derivative element, the Indian grammarians with respect These assertions would, in my to roots, which those grammarians opinion, need more numerous and did not ■ know how to extract with stronger proofs than are those ad- rigour of method, as not having paid duoed by the author. 52 PART I. CHAP. II. § 10. or the third class of the primary roots may be reduced, without violence either in respect to form or to meaning, to roots comprehended in one of the three classes men- tioned. We had better now see, by an examination as B detailed as the nature of this work admits, what are the characteristics which, according to Fick, distinguish the secondary from the primary roots. In the first place we observe the changes of the sounds of the primary roots, and, first of all, the weakenings of vowels. We see a weakened to i, whether initial or medial or final; we see it also obscured to w, especially when final : i and u are never primitive in the roots, according to Fick, but spring from a (m sometimes from vd)} — A phenomenon equally worthy of remark is the strengthening of vowels which appears in their reduplication and increase, but with much greater frequency in the former than in the latter : even before the division of Proto- Aryan it appears that the lengthening of the final a of roots was widely extended ;^ that of * appears alto- gether foreign to Proto-Aryan; in two cases at least we must believe that of final u to be primitive. The study of the consonantal sounds in the roots discloses: 1st, vanishing of initial s before h, t, p, n, in a certain number of examples; 2nd, metathesis of r, and perhaps also iWindisch, in the review or less closely connected with each quoted of Pick's work, considers other. This mode of expression, ob- this assertion not proven, and in par- serves Windisch, would find its ticular the attempt to derive « from counterpart in the languages of the va appears to him strange. In the Semitic stock. roots related in meaning, and only 2 The Indian grammarians, as we distinguished from one another by know, followed in this by many of the quality of the vowel, he believes our philologists, do not admit verbal that it may be assumed that this roots in short u,. Schleicher quality was originally not at all de- manifested a contrary opinion ( TJ'ar-- terminate, and that, according to the teln avf a im indogermanischen in different colouring of the vowel, men the Beitrdge, etc., ii. 92-99). wished to express different ideas more ROOTS. 68 sporadically of n, in some very doubtful eases ; we do not find in roots additions of initial or medial consonants in such a way that the phenomenon may be referred to the primitive and fundamental Aryan. As far as concerns the reduplication it will be sufficient to observe that some so-called roots were produced by means of it from true roots, and should be considered as present-stems, and 50 as intensives which have become general stems. They are for the most part words denoting sounds, as ^«& (to laugh) from hi-lia, a reduplicated form of the root ha (to utter a sound).' We shall have to be less brief with the 'determinatives' of Pick, or the elements added, as suffixes,' to the primary roots which, by means of them, ' With regard to the reduplication of the Aryan roots Brugman's monograph deserves notice XTber die sogenannte gebrochene redapUoaiion in den indogermanischeri spraalien, (in the Studien sur griechisohen und lateinischen grammatik, edited hy G. Curtius, vii. 185-216, 273-368). The author notices first of all how- rare is complete reduplication, as, e.g., in the Sansc. gar-gar-a-s, in the Gr. /Mep-fiep-o-s, in the \jsA.fwr-fwr : often the first element of the reduplicated form undergoes a phonetic decay by which it becomes a mere prefix (for instance, in the reduplications of the present and the perfect), because the reduplicative syllable had not its own accent, owing both to dissimila- tion and generally to desire of easier pronunciation. To the two reduplica- tions noted, the complete and the in- complete resembling a prefix, is added a third, an incomplete reduplication in the shape of a suffix, which is found when not the first, but the second element of the reduplicated form ex- periences a loss either of the initial sound or of the final: in this last case we have the so-called broken reduplication (e.g., Ved. dudh-ra-t, in which du-dh. comes from du- dhu; and this from dhu-dhv. ; the dku of the second syllable has lost its final vowel «, preserving only its initial sound dh, in order, Brng- m a u believes, to maintain the equili- brium in the word, the first syllable of which was weakened by dissimila- tion, byabandoning its primitive as- pirate). Thence came several muti. lated forms which were afterwards considered and treated as true roots (e.g. da-d from da-da). And here the author sets himself the task of studying the reduplication which we are discussing, both in the roots end- ing in a vowel and in those which have a final consonant. Some of the broken reduplications seem to B r u g- m a n to reach farther back than the beginning of the existence of the individual Aryan languages as sepa- rated from one another. ^ Pick rejects absolutely Po tt's famous doctrine of the secondary 54, PART I. CHAP. U. § 10. 61 are transformed and multiplied into secondary roots. These elements have been divided by our author into two classes, the first comprising those which for the most part do not alter the sense of the roots, while to the second belong all those others which generally give rise to secondary roots, distinct also in meaning from the primary roots from which they spring. The determinatives of the first class are a, n, m. The a we find, e. g., in the secondary Proto- Aryan roots gna (to know) from gan, ja (to go) from i, true present-stems which have become general stems ; the n in the primitive gan from ga (to beget), and in several other secondary roots which have come from present-stems in -na-, -nu-; the m in gam from ga (to go), and this suggests the -ma- used in the formation of old participles and infinitives, from which it is probably derived. The determinatives of the second class, which Fick enumerates according to their phonological order, are more numerous. 1st, Determinative h in 60 Proto- Aryan roots (e. g. vale [to speak] from va = u [to utter sound], dalt [to bite] cf. da [to mangle]): probably identical with the suflS.x -sk- {-ska-) which forms presents. 2nd, Determinative g in 24 primitive roots (e. g. in iig—vag [Lat. augere, vigere] roots formed by means of mutilated amples of the phenomenon; 3rd, no prefixes, a doctrine already vigorously one lias proved that certain elements, assailed by G. Curtius (Orundz., used commonly as prefixes in Sans- etc, pp. 38 — 52 of Bng. Ed.), the chief crit, existed already before the separa- objections to which may be reduced to tion of the languages as prefixes, the following : 1st, that such a doc- and exactly in the Sanscrit form j trine is not founded on facts ; 2nd, 4th, nothing gives us the right to that we cannot refer to a linguistic suppose that the union of preposi- period so ancient as that which pre- tions with verbal roots was even in ceded the division of the Aryan the earliest ages so close that new stock into several families of Ian- radical forms could be easily derived guages, a phenomenon which appears from them. See, however, Scherer, in epochs relatively very late, and ZurgescMchtederdeutscJiensprache, only for the reason that we find in Berlin, 1868, pp. 327, sqq. them a considerable number of ex- HOOTS. 55 from u-=zav [to be content], /m^ from ,;m [tojoin]): possibly this g came by weakening from h, sh, as it seems open to us to perceive from the fact that of the 24 above noticed secondary roots in scribed the comparison within too narrow limits, and espe- cially that he did not go back to those forms of the Semitic and the Indo-European word, which modern philo- logy has demonstrated to be the most ancient. Availing himself of the results obtained by the pre- ceding investigators, and especially of the studies of Raumer, Delitzsch also attempted the solution of the difficult problem. He distinguishes triconsonantism from polysyl- labism which he denies to the Semitic roots, because all the vowels in this stock of languages have the power of grammatical determinations and yet are not elements of the roots; with what vocal sound the consonants of the primitive Semitic roots were pronounced remains unknown to us.'' After having himself, too, attempted his reduction of Semitic stems to more simple forms, the author holds that the Semitic roots, when restricted to their essential ' It ia hardly necessary to state ' It is precisely in this that we that Fick's investigations on the find one of the strongest arguments subject of the Aryan roots have given adduced in proof of the profound results unfavourable to this assertion divergence between the roots of the of Eaumer. two stocks. 66 PART I. CHAP. II. § 11. elements^ appear to be formed from one^ two^ or three con- sonants, just like the Aryan roots.' Thus he establishes the possibility of a primitive radical relationship between the two great stocks, and as a sample of a more extensive work, he exhibits a goodly number of comparisons between Aryan and Semitic roots with Z; and with g : and the result of these comparisons is, according to Delitzsch, the disco- very of 100 Aryo-Semitie roots :^ another hundred will, he ^ What we said just now, with reference to Itaumer's doctrine, of the last researches on the subject of the Indo-European roots, may hold good also for this assertion of De- litzsch. ^ We think it will be not unwel- come to our readers if we enumerate them in the present note. a6A, to swell ; arTc {rak), to order, marshal, arm ; av, to aid, to desire ; «, to cry, to roar \ ud {vad), to gush forth, to wet ; kan, to sound ; leap, to curve, to bend ; har, to be cold ; kar, Tcarlcar, Icarlc, to call; Jcar (Jcvar), to turn; Tear, Teal, to burn, to cook ; Tcart, to cut ; Icarl, to con- tract; Tcard, Tcrad, to brandish, to shake ; harp, to pluck, to snatch ; kal, to put in motion ; kal, to em- brace, to bide ; has, to scratch ; hi, to honour ; hu, to cry ; leu, to burn ; leu, to be hollow, to swell ; hubli, to he made round ; leus, to embrace, to surround; hrt, to buy; hru, to knock against ; gan, to bend ; gdbh, not to shut close, to be deep ; gam, to be full ; gar, gargar, to swallow ; gar, gar gar, to cry; gar, to rub, to grind; gar, to move one's self; gal, gul, to be round, to wallow; gu, to cry; grahh, to grasp; gras, to swallow, to eat; gras, gars, to sound ; gliad, to take, to grasp ; gliar, to be red-hot ; ghart, to flow over anything ; gliars, to be rough to scratch : ghu, to call ; talc (tva/e), to arrange, to spread ; tar, to trem- ble ; tarp, to satisfy ; tal, tul, to raise ; trap, to turn ; trud, to push ; trup, to break in pieces ; dar, to burst, to tear ; dhic, to move violently; nu, to incline; nud, to push ; pat, to be open, to be spread ; park, to cleave ; iha, to shine ; bhag, to glitter ; hhar, bhal, to be clear ; ihar, to cut, to pierce; hharg, to shine; bhal (bldu), to bubble, ta flow; bMd, to divide, to cleave; bhrak, to sparkle ; ma, to measure ; male, to press, to knead; mad, to extend, to measure ; mad (mand) to be quiet, to delay; mar, to bind fast; (mal to wither) ; marg, to rub, to blot out ; mard, to crumble, to soften ; mardli, to be loosened ; mu, to wet, to stain ; rile, to pour out, to empty ; ru, to roar ; vag, to be of great weight; sale, to be attached, to follow; sale, slea, to cleave, to cut ; sad, to sit ; sar, to go, to has- ten ; sar, to tie ; sarlc, to throw, to strike ; sarg, Lat. dimittere ; sarbh, to swallow; sile, to moisten; ««, to shine ; su, to put in violent motion ; slea (slci), to settle down, to dwell; slcap, to scrape ; slcarp {harp), to be sharp, to be whetted, to cut ; sku, to look ; sla, to stand ; stah, to curdle; stag, to cover ; star, to spread on the ROOTS. 67 hopeSj be the fruit of new researches. The book of the learned Semitic scholar has, in our opinion, at any rate the double merit of not having been written without taking account of preceding labours, and of not abandoning those wise methodical principles against which an investi- gator should never rebel, especially in so difficult a subject. We ought, however, to notice the opinion expressed with reference to this task by that learned and intelligent student of philology, and particularly Semitic philology, Sayce, in a work which we shall certainly have another opportunity of discussing.' " If," he writes, "Aryan and Semitic are to be compared, we must commence with the structure and the grammar, not with the lexicon. Moreover, Assyrian and old Egyptian are deliberately ignored — indispensable as they would seem to be if we would find the oldest ob- tainable forms of the radicals ; and the roots selected for comparison are all on the one hand, more or less of an ea onomatopoeic nature;' and, on the other hand, contain three consonants, two of which may be pronounced together without the intervention of a vowel." What would De- litzsch do with certain other roots? Besides, there are not wanting objections of less weight which Sayce passes by in silence ; one of which is, to take an example, the great importance of the vowels in Semitic, an importance which we should hardly know how to harmonise with a theory in which necessarily no great weight is attached to them. ground; «teZ, to stand firm ; sfa,iQ> {Vergl. wdrterhuch, etc., 2nd. ed., stretch one's self ; spar^, to bud, to p. 932): Geiger (Tier uraprung, shoot; spal, to precipitate, to fall; der spraclie, Stuttgart, 1869, p. 26) smak, to rub, to smooth ; smar, to affirms positively that no certain remember. example is as yet known of imitation ' The principles of Comparative of sounds in language. See also PMoZo5y,2ndedition, London, 1875, Benloew, De I'onomatopee (in the p. 76, note. Apergu gindral de la science com- '' It should be observed that F i c k parative des langv.es, Paris, 1872, does not admit onomatopasa in Proto- pp. 93-119). Aryan except in very rare cases 68 PART I. CHAP. U. § 11. Before closing these remarks on the radical union of Aryan and Semitic we may be allowed to mention a monograph published four years ago by Grill.' He starts from the principle that the physiological analysis of the roots is a necessary preliminary to an etymological comparison of two linguistic stocks. After some introductory observations on the idea and the nature of the root^ and on its fornp, he proceeds to discuss the relations existing between the Aryan and the Semitic roots considered in their phonetic constitution. A carefully made comparison discovers as a " specific form " of the Aryan roots the vocalism, of the Semitic the consonantism : because the Indo-European vowel is (in contrast with the Semitic) altogether inde- pendentj stable, and a peculiar element of the root, so that 13 it also contributes to express the idea signified by the root. The author then goes on to treat of such relations between the roots of the two stocks with reference to their general type. He opposes the Aryan " formalism " to the Semitic " materialism :" because, " while Indo-Germanic has directed its power of production and its formative capacity with greater zest to that part of the language which aims at representing the form of the idea, on the contrary the creative instinct in Semitic has laboured, with an efficacy altogether superior, in the formation of the roots themselves, or in the matter of the idea." A first proof of his assertion he discovers in the variety which is apparent in the phonetic constitution of the Aryan roots, and in the uniformity of the Semitic : it would therefore be absurd to derive the first kind from the second or vice versa. But it is possible, remarks Grill, to conceive a primordial type, from which springs a biliteral Semitic root (to use the ordinary phrase) and an Aryan root, a consonantic root ' ifher das verhdltniss der indo- pTiysiologie der sprache {ZeitscJirift germanischen una der semiiischeu der deutscTien morgenldndischen ge- sprachwurxeln. Bin beitrag xur sellschaft, xxvii. 425-f)0). ROOTS. 69 (if it might be so termed) and a vocalic. "To sucli an original form we can ascend directly by assuming that the primitive roots have all, if the truth must be told, had a vowel as a material, essential element, and have consequently been formed according to the law of monosyllabism, but that the radical vowel has everywhere been the same," that is to say a (monophonic period, alphar-language). This a, peculiar to all the roots, would therefore be a material element of them, but lacking any specific value. Hence " the linguistic spirit could either increase the importance of the radical vowel or destroy it altogether : in the first case was produced a vocalism, in the second a consonantism." The primitive type common to the two great stocks he would like to be ascribed to the class of the Radical or Isolating languages, without however affirming that there has existed a single language of such a nature (" alpha- sprache ") whence all the others have derived their origin. The two opposed principles' of the Semitic and the Aryan were developed in the period of Agglutination. A second si argument in favour of his position Grill derives from the wealth of Aryan and the poverty of Semitic in the formation of the stems and the words ; a third from the varying productiveness of the roots ; for there are found perhaps " in Hebrew relatively at least ten times more roots and ten times fewer words than in English." And lastly, noting the ethnographic importance of this problem the author declares as a characteristic of the Aryan spirit formalism, of the Semitic materialism : in the synthesis of Indo-Germanic formalism and Semitic materialism con- sists, according to Grill, the eminently human character of Christian civilisation. — It is hardly necessary to observe that the doctrine put forward in this treatise (which the philologist reads not without pleasure) cannot be said to be grounded on facts to such an extent as to be received as a theorem of science. Such also is the opinion manifested 70 PART I. CHAP. II. §§ 12, 13. with respect to this monograph in an article of the Gottin- gische gelehrte anzeigen} § 13. In the pamphlet above mentioned of Schultze, InAogermaniscJi, semiiisck und hamitisch, we find a compa- rison instituted 1, between roots; 2, between expansions of roots; 3, between combinations of roots in these three linguistic stocks. A few considerations will sufficiently enable the cautious reader to form a fair idea of the worth of such comparisons. As regards the first, that of the roots, we will observe, to begin with, that the primitive radical elements, the various forms of which in the Aryan, Semitic and Hamitic languages are compared together, do not exceed thirteen in number. We will add that it is not always without great efiforts and ado that we pass from one meaning to another;^ that he has very little regard for phonology who considers as having a common 65 origin two roots which are related to one another, for example, as the Aryan dJia and da -^ lastly, that the existence of an Indo-European root a [to make, to happen, to be] is very far from being proved.^ The other two comparisons, between expansions and between combinations of roots, we do not even think it necessary to notice : for we believe that no valid proof can be derived from them in support of the argument to which Schultze's pamphlet is devoted. § 13. Still less in harmony with the severe but just exigencies of modern philology are the etymological com- ^ 1874, pp. 119-23 : the article is words adduced in confirmation and signed with the two initials H.E. exemplification of those meanings : well known to philologists. " ma, mu, mi, stumm, sein, daher, 1. 2 We will only quote in proof the denken, sinnen, minnen (liehend ge- senses attributed to the root ma, mm, denken) .... 2. stumm sein, todt mi, and all derived from the first sein .... 3. stumm und doch leben- which we will point out : in order to dig, heweglich sein, wie wasser, daher remove all risk of inexactness we feucht sein, netzen " (pp. will not translate, but quote the 14-5). actual words of the author, regret- '' See p. 13. ting that we cannot quote also all the * See p. 11. ROOTS. 71 parisons attempted, three times at least in the last few- years, between Aryan languages and Chinese. First in order of time comes Chalmers with a book on the origin of the Chinese and their relations with the western nations.' In the 3rd chapter^ he proceeds to compare 300 Chinese words with words of other languages. On the value of the results obtained let us hear himself : " To pronounce sentence on all the individual cases adduced in the following table it would be necessary to have an extensive acquaintance with languages and with the principles of comparative philology, an acquaintance which I do not at all pretend to possess. I oflFer it as raw material, from which others inore capable than I may extract the grains of gold. But, such as it is now, it discovers at least a little more than fortuitous resemblance." The complete absence of true scientific analysis, of fully demonstrated phonetic equations, ( forbids us to agree with the opinion expressed by the author himself at pp. 37-8 on the importance of his comparisons. Nor are we inspired with greater confidence by the strange comparisons of Edkins' between Greek and Sino- Mongolie words, between Latin and Sino-Mongolie, and between English and Chinese words : from which com- parisons the author is induced to believe that the Sino- Aryan civilisation may be recognised as the primitive Aryan ! We shall content ourselves with regretting that Edkins has not understood the necessity of a scientific 1 The origin of the Chinese : an in the Chinese with those of other attempt to trace the connection of languages (pp. 43-55). the Chinese with western nations in ^ China's place in philology : an their religion, superstitions, arts, attempt to show that the languages language and traditions, London, of JiJurope and Asia have a common 1868. origin, London, 1871 : see especially 2 Chap, iii, : The Chinese spoken chap. xiii. pp. 361, 363, 375-83 ; language (pp. 35-55). See in par- chap. xiv. p. 386. ticular A comparison of 300 words 72 PART 1. CHAP. II. § 13. method, and that a French review, generally highly commendable for its critical power, has passed a judgment on this work, which errs undoubtedly in the direction of excessive indulgence. Nor did the result of the comparative investigations of Chalmers and Edkins, appear very deserving of credit to Gr. Schlegel, who set himself the task of making anew similar researches by following the rigorous principles of the German linguistic school, and availing himself of all the few books which he could consult in the country where he lived, unpropitious as it was to such studies. As a product of his researches we have the book published by him on the affinity of the primitive Chinese roots with the Aryan.' The author commences with the results of the studies prosecuted towards the beginning of this century by Toan-ta-ling on the ancient Chinese pro- nunciation. And he observes, in opposition to the opinion of a great philologist, that in Chinese too there are words on which analysis may be practised. Then, after some introductory remarks, he gives a table of roots which he believes to be Sino- Aryan, among which he notes especially 67 roots of verbs and of pronouns. This exposition is followed by a semasiology, in which he attempts, by means of Chinese, to ascend to the primitive meaning of certain Aryan words. In all these investigations criticism cannot but discover two great defects which we have been before compelled to censure in several of the works noticed : 1st, comparisons grounded not on the base of an exact phonology, but on the too often fallacious resemblance of sense and sound ; 2nd, fantastic derivations of meanings from meanings. These defects spring assuredly from no other source than the imperfect scientific education of the author.^ 1 Sinico-Aryaca er Eigveda oder die darstellung des verh'altnisses der a- heiligen hymnen der Sr&hmana zum nomina zu den derivierten verhalfor- ersten male vollstdndig ins deutsche men : ein heitrag zur geschichte der uiersetzt mit commentar und ein- worthildung im indogermanischen leitung) is discussed very favourably (Sitzungsberiohte d. K. Acad. d. wis- by A. De Gubernatis in the Ist semchaften,pUl.-hist.cl.,\y.li\-M). number of the Sollettino italiano Der injinitiv im veda mit einer sus- degli atudi orientali which is under tematik der litauischen und slavis- his direction, and which we heartily chert verbs, Pi'ag, 1871. — Agglutina- reeommend to our readers, especially tion oder adaptation ? eine sprach- those of Italy, as a work which does wiszenschaftliche streitfrage mit honour to our studies, and deserves nacMr'agen zu des verfaszers ' In- to be praised and forwarded. finitiv im veda,' Prag, 1873. 2 Die entstehung der a-decliHa- STEMS AND WORDS. 99 by notes, in which we shall point to the most important critical observations on the theories of Ludwig made by distinguished students of philology, among whom we may mention Delbriick/ Bergaigne/ Jolly.^ We must see in the first place what is the opinion which > Zeiischriftf. vgl. sprachforsch, XX. 212-40 : this review of the work Der inflnitiv im veda etc. was met by Ludwig with a vigorous reply in the pamphlet AgglutmaUon oder adaptation ? in which he under- took to defend and develope certain opinions set forth in the preceding monograph. ^ Revue critigue d'histoire et de litie'rature, 7th year, 1st semester pp. 385-93. » Zeitschr. f. vSVcerpsychologie, etc., viii. 62-73. We regret that we have been absolutely unable here in Turin to read the remarks made by Benfey on the work D. inf. im v. in the iforth Sritish Review (Jan. — March, 1871), which were such as to deserve the thanks of Ludw ig him- self. [The writer (Benfey' s name is not appended) of the article in the North Brit. Eev., after giving ex- tracts to shew the nature of Lud- wlg's theory, proceeds to offer a brief criticism, and concludes with quoting a few passages from the Veda in which he thinks Lud wig's interpretation is wrong. On the theory itself of the origin of Indo- Germanic inflection he remarks that it is " almost entirely based on the deviations from ordinary Sanscrit, relating to inflective forms and grammatical relations which are found in the traditional texts of the Veda." "The justification of de- ductions from the traditional Vedic texts manifestly depends on a correct estimate of their history, their ori- gin, &c." But these texts, it is clear, " must have been exposed to all kinds of corruption" The cir- cumstances attending the tradition of the texts "lead to explanations of the abnormal phenomena . . . widely different from those of Pro- fessor Ludwig. Many of these phenomena may prove to be bye- forms of phonetic origin As to the date assigned by Ludwig to the completion of the grammar (seebelow,p.lll)thereviewerobserveB " Surely if the grammar, which by the hypothesis was completed only about 2000 B.C., was still, 500 years later, among the Indians, in such an unsettled condition that a maltitude of forms could express all relations indefinitely, then the other tribes could not possibly have attained to a. grammatical form in such har- monious conformity with Sanscrit and Zend." Though Ludwig "en- deavours to point out analogous phenomena in other languages . . . scarcely any of his hypothesis can be admitted by careful critics." But " in that part of the work which deals exclusively with Vedic forms without regard to the theory of in. flection .... the preponderance of good is so great that the work takes a very high place in the field of Vedic research." — 2V.] 100 PART I. CHAP. III. § 16. Ludwig holds of modern historical and comparative philology ; the reasons will thus appear which separate him from the linguistic school, the doctrines of which on the origin and the changes of Indo-European inflexion were developed and set forth with fidelity, vigour and pre-eminent coherence of principles and methods by A. Schleicher. Recent investigators have not, so Ludwig thinks, made good and sufficient use of the historical method. The impor- 93 tance oftheVedic language with reference to the study of the Aryan dialects has been theoretically recognised, but ^' in practice, as we must admit, the base of the philological comparison of to-day is Greek -^ only it has been rendered intelligible by means of Sanscrit. As long as the common phonetic laws sufiice to show in the Greek form a modified Sanscrit form, Sanscrit is welcome, and then the Sanscrit trumpet is allowed to speak : when Sanscrit is irreconcilable with Greek, the former is deprived of all power of harm by explaining it, that is to say, by maintaining a priori a a doctrine with regard to it, instead of allowing ourselves to be taught by Sanscrit. Nay, it would be very easy, starting from principles of logic, to demonstrate that the method which now prevails in the science of language is false and to be rejected. It consists in nothing else than changing comparison into a historical process. From a series of forms mutually connected one is arbitrarily taken, 1 We gladly take this opportunity tomed, and to wliich even at present of observing that, if we are not de- the young are accustomed by a ceived, one of the most serious oh- method of teaching often incom- Btacles in the way of free and truly patible with contemporary science, scientific philological investigation — As far as Vedic grammar is con- is the influence still exercised upon cerned much is with reason expected our minds by the ideas and even the from the studies of Benfey. Mean- technical terms of the Greek and while, we are very glad to notice Latin grammarians: ideas often er- Delbriick's book Has altindische roneous, terms not unfrequently in- verhum aus den ht/mnen des Sigveda exact (especially in Latin), to which seinem baue naeh dargestellt, Halle, we have been only too much accus- 1874-. STEMS AND WORDS. 101 the originality of which neither is nor can be proved, this is compared with the rest, and the expression of the relative difference is insensibly transformed into a historical process/-" Nor is it of any use to assume, as the school of Bopp is in the habit of doing, stems which did not in reality exist or the existence of which cannot be proved : mere hypotheses, results of an a priori study which cannot form a solid base for a truly scientific knowledge.^ Not m only in its progress, but from the very starting point, must philological investigation be severely historical. " The scientific treatment of the languages of the Aryans must be founded, in the truest sense of the word, on the Veda, as far as it extends."' It must "seek for stems the reality of which can be proved, the meaning of which is clearly revealed by the syntactical relation in which they are presented to us.^^'' A stem is, according to Ludwig, in the historical process of language, every word-form which is considered as separated from the sentence, or as not exercising a function in it nor placed in strict relation with the others of which the sentence is made up : in the contrary case we have a word-form declined or conjugated (" flectiert " ) . Hence it is clear that historical grammar (with practical grammar we have not to concern ourselves in this work) ought to perceive in inflexion a syntactical fact, and what and how extensive • JD. inf. im e., p. 83, and on p. sqq.) professes with respect to the 87 he writes : " .... we cannot accuracy of the Vedic text a faith help considering almost as harm- which is not shared hy all Vedio ful, as would be a positive ne- scholars, among whom Delbriick gleet, the manner in which not un- declared his doubts in his review of frequently at the present time use of Ludwig's book, D, inf. im v. is made of the Veda.'' See Agglu- It is a problem of Vedic scholar- tination, etc., pp. 39-40 (§ 20). ship with which neither the nature of 2 D. inf. im v., p. 70. oar special studies, nor that of the 3 D. inf. im v., p. 87. Here we present book allows us to concern should not omit to mention that ourselves. Ludwig {Agglutination, etc., p. 82 ■■ 2). inf. im v., p. 70. 102 PART I. CHAP. III. § 16. are the relations which unite syntax with morphology in Ludwig's system.' Let us now proceed to examine carefully the charac- teristics which the author attributes to his stemsj so different as they are from those of ordinary modern phi- lology. In the first place^ he teaches^ there is neither stem nor root which originally ended in a consonant; the final sound of every stem is a vowel (generally i) : and 95 this position is considered by Ludwig as so fundamental, that, until his opponents demonstrate its falsity, his doctrines will always remain unshaken.^ Secondly, he denies that the suffixes, in which we have been accustomed to perceive the signs of inflexion, had originally the function of denoting those definitions of ideas which or- ' Agglutination, etc., pp. 107 and 111 ; see also p. 29. ^ Die enstehung der a-decUna- tion, etc. : see especially § 14. — Ag- glutination, etc., pp. 113-5. On pp. 117-8 of this book he considers the -i of the locative not as a suffix, but as the final element of the stem : nor does he explain otherwise the -i of the potential (J), inf. im v. p. 118) and the verbal termination 4 (pp. 138-40). A propos of the many roots and stems which are gene- rally thought to end in a consonant, and to which Ludwig assigns the vowel i as a final sound, let us note the observation of Bergaigne, ac- cording to whom the new hypo- thesis frees us, it is true, from the difficulty caused by the copulative vowel in several forms, but forces us to admit the disappearance of i in a much greater number of forms. Ludwig, continues the critic quoted, always recognises the primi- tive form in that which is richest in phonetic elements ia order not to be obliged to suppose in the others anything but phonetic decay : but in that case it is necessary to have recourse to very grave phonetic changes, the possibility of which, as they are supposed to have taken place in a pre-historic age, we cannot directly prove, and which, as they are found, according to the hypo- thesis in question, in profoundly dif- ferent ways even in one and the same language, and with respect to the same primitive sound, seem a priori almost impossible. — After this it will perhaps appear to the reader a little strange that Ludwig proves him- self, in his criticisms on modern lin- guistic science, so severe in the matter of phonetic laws (Agglutination, etc., p. 30, etc.). Let Indian scholars con- sider whether it is possible to regard as proved, e.g., the derivation of the -us of the 3rd plur. act. "from -arus = -aru-s," whether phonology can accept, as a proven thesis, this very important disappearance of sounds (D. inf. im v., p. 126 sqq.). STEMS AND WORDS. 103 dinary philology believes to have been represented by them from the very beginnings of their use in the Aryan languages. According to Ludvfig they were by no means at first furnished with such capacity and their changes did not take place in a period of independent existence, as others think and assert without being able to prove, but in reality in the words to which they were joined and in them alone.^ Among the arguments which the German philologist adduces against the doctrine of the modern school of phi- lology with regard to the original significance of the suffixes and to their relations with the stems, a doctrine as he calls it of " agglutination,'' we will cite first of all the following dilemma : " either the Indo-European languages are agglu- i tinating, and in this case the supposed difference between these as inflecting and the agglutinating languages is false; or they are not agglutinating, and therefore the suffixes of word-formation and of inflexion cannot be agglutinate. But as these suffixes appear nevertheless attached to the root, and all recognise the fact, so it follows with abso- lute necessity that they were not united to the root and the stem in that sense which we see they carry with them.'" 1 Agglutination, etc., p. 27. syllabic, it certainly was not so ' Agglutination, etc., pp. 24-5. like the Chinese; if it was agglu- In his critical remarks, above quoted, tinating in a very mitigated form, it on this book. Jolly observes that was not so like the Tatar, that, in Ludwig was wrong in making the fine, if Indo-European can, super- difference between agglutination and ficially considered, appear in those inflexion to consist in an eminently two first periods of its proper exis- fundamental characteristic, while teuce, like the languages mentioned, most philologists consider the agglu- there was, nevertheless, always in it tinative form as an intermediate another germ, and of higher capacity, stage between the isolating and the (Zeitsehriftf. volJcerpsychologie, etc. inflexional. "We shall have to speak ii. 238-9). See also P e z z i's Introduc- later on this very important subject: tion, etc., pp. 120-6; Sayoe, The meanwhile we think it useful to re- principles of comparative philology, mind the reader of Steinthal's pp. 127-64; Miiller, F., (?»-MMdms profoundly wise observation that, der sprachwissenschaft, vol. i. § 1. if indeed the Indo-European mother- pp. 139-40. language was undoubtedly mono- 104 PART I. CHAP. HI. § 16. " It is an opinion/' so writes Ludwig, "the truth of which cannot be doubted, that all the material portion of the forms at one time did not exist, that it grew during a period of time probably immense, that the state of the forms as com- pared with the grammar of the entirely developed language was imperfect. Nevertheless men spoke and the speakers were undoubtedly understood. To what must the gram- matical ideas have been attached ? Naturally to what we now call stems. The stems, the groundwork of the subse- quent grammatical forms, are by no means mere abstractions, they appeared in syntactical usage A stem-form, \vhichcannot be demonstrated to have been actually employed, is a chimera, an absurdity which has done considerable 87 harm to linguistic study and quite as much to special re- searches on the subject of syntax.'^' And here the author proceeds to quote a series of passages from the Veda, in which he discerns stems employed in the function of cases without the usual endings (genitives plural without -dm etc.^) and cases used one in place of another (e.g., the locative instead of the dative^) and the numbers not denoted by suffixes exclusively peculiar to each of them (for instance the -su of the locative, the -Ihis of the instrumental were not employed in the most ancient period as plurals so ex- clusively as they were later, nor have the oldest genitives plural without -am a mark of number) :* so that, according to Ludwig, "it is impossible to speak of the original meaning of a case ; we can only talk of different uses of a stem," to which at a later time were added new distinctions/ He adduces other examples to establish the point that no particular relation either of person or number was originally represented by the middle endings -e, -se, -te, and hence not even by the active -mi, -si, -ti, from which the middle ' D. inf. im v., p. 4. * Ihid., pp. 20, 24-5. 2 Ibid., pp. 5-8. 5 iiid_^ p. 20. •' Ibid., pp. 11-13. STEMS AND WORDS. 105 endings by common consent must be thought to have sprung.^ Nor is the opinion different which he expresses with regard to the moods. " It would certainly be superfluous to make even a single remark on the original confusion of the con- junctive with the optative in syntactical usage : but it is easy to see that (unfortunately !) not even the indicative^ the conjunctive and the imperative are separated from each other by any solid barrier. Unfortunately^ we repeat, for, as is known (^nd was known — a sufficient proof of it is Say ana, etc.), it is often impossible to discover whether a verbal form in the Veda is to be understood in an indicative i or in a conjunctive-imperative sense No mood has a stem-sign belonging to it exclusively. It was only by means of the prevalence, which took place gradually, of the forms with e-, a-, a- that the potential, the conjunctive and the indicative became distinguished one from the other. The imperative, though undoubtedly itapproachesinmeaning more nearly to the conjunctive than to the indicative, has not, as all know, a stem-sign of any sort whatever.'" Ac- 1 Ibid., pp. 71-82, 145-6. suffixes have been used, not only in ' Ibid., pp. 111-3. To, state sum- composition but also apart from it, raarily the observations of that with the force of cases : he does not, learned philologist and Vedio scholar however, think that we can draw Delbruck, on the preceding state- from this the conclusion that the in- ments and quotations of examples, flexional suffixes had not originally we will say that the critic whom we the function of denoting those re- have just commended rejects some lations which at a later period they of them, because be perceives in certainly represented, but only that tbem errors of the Vedio text; as to- thelanguage, in a very ancient period, others he cannot feel satisfied, inas- of which he still perceives traces in much as he does not approve the the compounds, expressed the rela- sense attributed to them by Lud- tions existing between the ideas only wig, or considers it uncertain; by the varying position of the words, lastly, in some forms, which the The French critic shows himself more latter believes to be primitive, Del- inclined to admit the confusion, or briick thinks that disappearance of at least the imperfect primitive dis- final sounds has taken place. Ber- tinetion, of the nominal and verbal gaigne by no means denies that forms, although he considers it pos- some stems without declensional sible that, from the very beginnings 106 PART I. CHAP. III. § 16. cordingly he is constantly in error who seeks in the sufExes the meaning of the forms. A cause of such error was the tendency to place the sound in close relation with the thing signified.' The principle then of agglutination, Ludwig afiirms, which has been made use of by modem > philologists, is a principle neither demonstrated nor demon- strable, contrary to the nature of the Aryan dialects, contrary to the historical method.^ The conclusion there- fore is that the suffix had never originally, the power of modifying the sense of the stem, but on the contrary derived its meaning from the stem itself, after having lost its own which was ' demonstrative,'^ corresponding to a " want, which, intellectually subordinate, but practically very active, and one which ever and again makes its own claims good, produced the material, which, transformed by a higher intellectual want, was rendered obedient to this last,"^ in other words to the expression of the relations which we now see represented by inflexion. That the reader may perceive the manner in which, according to Ludwig, it came about that the suffix lost its own original simply demonstrative force and acquired that meaning of case, person, number etc., with which we see it provided, we shall here quote some passages found in the two works of the author which we have most frequently cited. " It took place owing to a certain change of equilibrium, since of their nse, the suffixes had each, (pp. 91-1) Geiger had taaght that BO to speak, in germ the force with " any sound may denote any idea, which we see them afterwards pro- any idea may be denoted by any vided. Such a question of language sonnd " (see also pp. 47-8, 51-2, of is evidently connected, at least in the larger work of Geiger, entitled great part, with Vedic philology and Vrsprung und entwickelung der scholarship : to solve it definitely, menschlichlen spracTie und vemunft, if that will ever be possible, we need Stuttgart, 1868-73, and onr Intro- farther new studies of the most an- duction, etc., pp. 201-3). cient surviving records of the Indo- ' D. inf. im v., pp. 1-2. European linguistic stock. s JUd,^ p. 4. Agglutination, etc. ' D. inf. im »., p. 8. Similarly pp. 26-7. inhis hook Derursprun^ der sprache * D. inf. im c, p. 85. STEMS AND WORDS. 107 naturally it was not possible to conceive the word with the suflBx as an indivisible whole, nor the meaning as consisting without distinction in the entire combination of the sounds tut it was also impossible to conceive as absolutely void of jense that which nevertheless was felt necessarily to be the subordinate element. Thus was effected in a spontaneous way, in which the reflexion played naturally only an en- tirely subordinate part, a new division of meaning, in which the root was subjected to a certain process of abstraction which exercised an immense influence upon the development of the language."' The process of word-formation by k* means of the demonstrative sufiixes noticed above "gradu- ally came to a stop and side by side with it there grew up a new tendency to take advantage of the forms which had diminished in value. And though at first the special indication of the ' agens,'' the ' actio,' the ' actum ■" was neglected and men were satisfied with the ' demonstration ' then evidently employed with great frequency, language afterwards proceeded step by step, it could hardly dispose of favourable material, to express such distinction, adapted as it was for increasing extraordinarily the intelligibility of speech, in which task nevertheless it laboured without any coherence. Having arrived by means of such dif- ferentiation at a certain stage it went on to a second differentiation, in marking number and case-relation, but for this purpose, too, were used only the elements which were already at hand, and we must not dream of the creation of a grammar."^ Elsewhere Ludwig regards as causes of the phenomenon in question the forgetfulness of the original meaning of the forms and the want which the speakers must have felt of explaining to themselves the variety of them, or of understanding them. " For,'' he writes, " it was undoubtedly by the meanings which were ^ 1 Ibid., pp. 4-5. s Ibid., pp. 15-16. 108 PART I. CHAP. III. § 16. attributed to the forms that they were believed to be under- stood."' " It is evident that the analysis of the thought and of the object of thought precedes the formal distinction, but this was called forth and maintained by the already ex- isting variety of the forms.^''^ And "the repeated trans- formation which the mind eiFects of the phonetic material, a transformation which closely connects the elements of word-formation and of inflexion with the root and the stem and causes them as they grow to be fused with these, which 101 constantly gives a new intellectual stamp to the old sounds, this transformation is what constitutes the real diflFerence between agglutination and inflexion.^'" Variety of forms obtained by means of suffixes^ which lost their original demonstrative significance, differentiation and adaptation of these forms to the various definitions the need of expressing which made itself more and more felt, are the two factors of inflexion according to Ludwig.* Hence it is clearly seen, 1st, why Ludwig has called his doctrine "theory of adap- tation (adaptationstheorie) /■" rejecting the denomination "stem-theory (stammtheorie)" proposed by Delbriick;° ' Hid.., p. 24. This idea of the phonetic changes he has recourse, f orgetfulness of the original meaning See Die entsteTmng der a-decUnation, plays a large part also in the system etc. (especially the first six para- of interpretation of the myths pro- graphs) and D. inf. im v., passim, posed by M. M tiller. * It should be noted how in the * Agglutination, etc., p. 112. doctrine of ' differentiation ' (' differ- 3 JJjrf., p. 28. enzierung') Ludwig agrees with ■• We regret that we are unable, Scherer and Geiger: the first of without increasing the already ex- whom reduces the entire intellectual cessive length of this paragraph, to life of language to two processes add to these remarks on the force of which are constantly renewed, meta- the suffixes some idea concerning the phor (" ubertragung ") and dififer- primitive form and the metamor- entiation {Z. gesch. d. deutsch. spr., phoses of them accoi-ding to Lud- p. x.); the second regards as the two wig, pointing out, e.g., to our principles which govern the develop- readers, what use he makes of the ment of the nature and the intellect suffix -ati, which has a prominent differentiation and chance, position in his system, and to what ^ Agglutination, etc., p. 107. STEMS AND WORDS. 109 Zndly-j thatj according to Ludwig's morphological system, it is impossible that there should be any absolute original difference between word-formation and inflexion, while, on the contrary, inflexion is regarded by him as nothing else than a development of the formation previously noticed.' There are two periods in the life of the Aryan languages which are not so completely distinct one from the other that we cannot still perceive, in an intermediate age of which not a few traces appear in the Vedic dialect and in the compounds, the linguistic forms peculiar to the one side by side with those of the other.^ We have also to dis- tinguish several periods in the development of inflexion. " Remnants of an ancient usage show that certain verbal suffixes did not originally stand to the several grammatical persons in that constant relation in which we afterwards find them. It happens at the same time, on the other hand, that these elements present themselves as signs of the infinitive,'"''^ the wonderfully various forms of which Lud- wig refers to nominal flexion,* The infinitive is in the organism of the verb that member which represents in the greatest original simplicity the verbal idea and there are still found infinitives in the bare form of roots. The defi- nition of person does not suffice to constitute the verb and is not inseparable from it ; hence it is not the essential characteristic of the verb. The true original verb was the infinitive.* And here the learned Prague professor pro- ^ Ibid., p. 115. ■* -D. inf. im v., pp. 45-6, etc. * D. inf. im v., pp. 25-9. For ^ Ibid., pp. 44-6. ourselves, as we consider the origin etc., pp. 104-5. We could wish from and primitive significance of many Ludwig a clear explanation of the suffixes to be still very doubtful, we ", verbale auffaszung," and the " ver- cannot accept as theorems of science bale gedanke." The critic of the the inferences which our author Litera/risches centrablatt (1873, draws from his tundamental princi- pp. 20-2) censures Ludwig forhav- ple of the original simply demonstra- ing assigned to the primitive infini- tive significance of such elements. tive functions which it only acquired ' Agglutination, etc., p. 104. later. Jolly, too, believes that our 110 , PART I. CHAP. in. § 16. ceeds to cite examples of the use of infinitives as of finite verbs in Latin^ Greek and Old Indian.' The verbal idea is represented in its fulness by the infinitive, incompletely 103 by the participle.^ Hence it appears that, according to Ludwig^s morphological principles, the verbal forms de- rive their origin from nominal forms and are consequently less ancient than the latter : that the verbal are bound to the nominal by close relations, in such a way that while by means of the participle the verb is lost in the noun, on the other hand by means of the infinitive the noun is lost in the verb.^ Thus it is seen how the verbal construction encroaches more than is generally thought, upon the field of the noun.'' Hence it appears, lastly, how that the work Ber infinitw im Veda is not only, as might be supposed from the title, a monograph, on the form and meaning of the Vedic infinitive, but in fact the exposition of a new system of Aryan morphology. We have already seen what relations exist, according to this system, between the infi- nitive and the persons of the verb in the several numbers. After the remarks which we have just made on the personal endings it will be sufficient for our object to quote the words with which Ludwig replies to a critical observation of F. Miiller:* "co dpicrre, you mistake if you aver that I deny to the personal suffixes any original meaning whatever. I attri- bute to them the demonstrative as their original significance author has fallen into mistakes 67-8. Delbruck and Jolly recog- through not having rightly con- nise in such facts nothing more than ceived the force of the Vedic infini- natural consequences of the force of tive. Bound unconsciously to the the cases in which the infinitives traditional idea of the infinitive he present themselves, not any proof of attempted in vain to open up new Ludwig's doctrine, according to roads for modern philology. We which the infinitive was originally shall have to speak of the infinitive the sole expression of verbality. at greater length in one of the para- 2 Agglutination, etc., p. 104. graphs of this chapter, and we will a 2). inf. im 0. pp. 44-5. therefore refrain from further re- * Ibid., pp. 46-50. mark upon it in the present note. 6 Agglutination, etc., pp. 62-3. JD. inf. im v., pp. 50-1, 65-7, STEMS AND WORDS. Ill which afterwards gives place to another function, the forma- tion of words ; afterwards they assumed a general verbal sense," (that is, of infinitives) ; " and lastly, when the number of such elements had increased, they were placed, according to accidental analogies and frequently even without any analogy at all, in connexion and relation with the cate- gories of the grammatical persons, ^categories which had meanwhile been formed into the pronoun of person." Such distinctions of person and number were attached to the several infinitives by the aid of numerous secondary for- im mations : hence the so-called personal suffixes.' The original force of these suffixes must have been changed therefore, according to Ludwig, by means of three meta- morphoses into that which ordinary philology regards as originally characteristic of them. — Let us conclude our exposition with a chronological note. Ludwig has been induced by the study of the language of the Vedas to believe that the grammatical formation of the Proto- Aryan did not attain its perfection (" always only relative,") until about five centuries before the most ancient Vedic period, or about 2000 years before our era. The westward migra- tions of the Aryans cannot be referred back beyond that epoch, because they all undoubtedly carried with them a language already grammatically developed in a perfect manner.^ Hence we have : 1st, the existence of roots and stems originally ending without exception in a vowel and actually employed in speech in such a way as to express, without the aid of suffixes, not only the ideas, but also their defini- tions; 2nd, expansion of these roots and stems by means of suffixes originally furnished with demonstrative force which, in their intimate union with the stems, was lost ; hence a large number of nominal forms; 3rd, a diflPerentiation 1 See also D. inf. im v., p. 145, etc. ' D. inf. im v., p. 148. Agglutination, etc. pp. 115-7. .112 PARTI. CHAP. III. §16. gradually introduced between such forms to represent with ever increasing accuracy the relations existing between the ideas ; hence, for example, the eases ; 4th, use of nominal forms to denote verbal conceptions in their widest extent, that is, origin of the infinitives ; 5th, genesis of the finite verb from the infinite by means of a difierentiation of the numerous forms peculiar to the latter. Such are the fundamental conceptions of Ludwig's Aryan morphology. We have described them by selecting 105 from among the very numerous notes which we collected in our perusal of the works of the learned Vedic scholar and philologist those which seemed to us adapted to give an adequate conception of his system, and by arranging them in the manner which we deemed most eflfective for rendering the development of his thought as far as we could intel- ligible to students. Nevertheless we are very far from presuming that we have made our readers fully acquainted with Alfred Ludwig as a philologist. Besides that his method of exposition rendered it very diflacult for us to follow its search through all the windings which it traversed with rapid and daring step, the limits which we are bound not to transgress in this short review of the most recent works of Aryan philology inexorably forbade us to penetrate into those detailed discussions of special facts which form so large a part of Ludwig's books. Nor can the reader learn from our remarks what sentiments and in what forms his polemic displays against this one or that one of his critics. Nevertheless from what we have said on the doc- trines professed by this author it seems to us evident with sufiicient clearness that, in starting from facts of which to a great extent either the existence is not certain or the value is doubtful and in availing himself of a phonology which cannot appear to all as rigorous as he thinks it to be, he has arrived at very daring conclusions which would need more numerous and stronger proofs and which have STEMS AND WORDS. 113 not found for hinij as far as we know, and probably will not find a large number of adherents.* We can easily understand therefore how it is that the opinions of Ludwig on the origin of the personal endings seemed to G, Curtius so "subjective" that he did not think fit even to submit io6 them to examination.'' He however, would make no slight mistakCj in our opinion, who should deem the labours of the learned Prague professor of no use to the science of the Aryan languages. Nor do we think that great advantage may be derived from them merely for the study of the Vedic infinitives, but it is our opinion that they will help just as much towards the investigation of the entire structure of the Indo-European languages. And in fact, in the first place, books like that of our author perform in science that ofiice which iu civil and political life the so-called opposition parties take upon themselves, that is, they subject to a severe examination the acts of opponents whose authority is great, they weigh their principles and methods, they bring to light their defects, they keep awake the attention both of them and of the public, they assail blind beliefs, t.hey render complete triumph, as well as the absolute and abiding domination of error, almost impossible. Moreover, Ludwig has the remarkable merit of having brought into prominence, as far as was possible, the importance of Vedic studies with respect to the historico-comparative grammar ' Bergaigne himself, who to a true that the verb derived its origin greater extent than the other critics from the noun hy means of the in- of L u d w i g linown to us agrees with finitive. liis ideas, and considers at least as [A brief and (as far as it goes) very probable the theory of the favourablecriticismof Bergaigne's adaptation of the suffixes in conjuga- views on ' Adaptation ' as against tion, in declension and in nominal ' Agglutination ' will be found in formation, does not give as hearty a Appendix II. of the 2nd edn. of welcome, as we have remarked above, Sayce's JPrinciples of Comparative to other important parts of Lnd- FUlology. — 2V.] wig's morphological system : thus. ^ Das verium der griechisehen for instance, he does not think it sprache, etc., i. 19 (note). I 114 PART r. CHAP. III. § 17. of the Aryan dialects and of having' invited the attention of philologists to that principle of differentiation ("differen- zierung") the action of which is perhaps, more than is generally believed, frequent and effective in the develop- ment of languages. For these reasons we regret that the few and short, but learned and very bold writings of the distinguished German Vedic scholar and philologist are not read and reflected upon by a greater number of students. § 17. In the works on Indo-European morphology which we have discussed in the three foregoing paragraphs, and especially in the books of Ludwig, we perceive here and there attempts at a chronological arrangement of the forms of Aryan speech. To this sole object is devoted the weU- M7 known monograph of G. Curtius entitled Zur chronologie der indo-germanischen spracliforsehung :^ a not very accurate title, which should be understood as though it ran Chronology in the formation of the Indo- Germanic languages, or The division into periods of the history of the Indo- Germanic stoch, as has been well observed by Bergaigne and Steiathal. Among the writings published in the last decade on the entire form-system of the Aryan languages this work of the eminent Greek scholar and philologist is the last with which we have to concern ourselves. Of the few touches drawn by a master hand, of which this most exquisite sketch consists, we will note only the most fundamental, referring those readers who are desirous of more extensive and detailed ideas to the monograph itself of Curtius to the French translation by Bergaigne and to the Italian 1 Of this work of Curtius we gv.es indo-germaniques, in tlie first linve been able to read only the ISrst part of the admirable BihliotUque edition (Leipzig, 1867) : a second de I'icole des hautes 4t»des, Paris, with some additions was published 1869, pp. 37-117) : it was also care- in 1873. It was translated into fully epitomised by Giussani in French by Bergaigne (La chro. his review, which we shall quote nologie dans la formation des Ian- shortly. STMMS AND WORDS. 115 abridgment by Giussani. But we will call to mind the critical observations made on this important little work by Justi/ Schweizer-Sidler/ Steinthal/ Giussani, Diintzer^ and especially by M. Miiller.' i03 The author begins by observing that, if there is a ' history' of language, there ought also to be a 'chronology' of language. In order to construct such chronology, as far as concerns the periods anterior to the most ancient literary records which have come down to us, we can only avail ourselves of internal criteria, which consist in the diligent analysis of the language itself. And here Curtius pro- ceeds to show by some examples how the sounds, the forms and the constructions carefully examined reveal to us here and there a chronological order of formation and development. In the life of the Aryan language ethno- graphically considered, he distinguishes two principal periods : 1st, that of unity ; 2nd, that of plurality of dialects gradually developed from the Proto-Aryan. He discerns two periods also, with Wilhelm von Humboldt, in the history of Aryan, regarded only from a philological point of view, or in its genesis, viz. : 1st, a period of organisation, in which the language acquires its essential form ; 2nd, a period of cultivation (" ausbildung ") or increase, in which the meanings become perfected^ while the sounds decay. It may be almost affirmed that the period of organisation coincides with that of the unity, the period of cultivation with that of the plurality; but, for the sake of accuracy, it should nevertheless be observed that probably the period in which the forms were coming to 1 Mevue critique d!Usioire et de * I>ie ursprunglichen casus im literature, 2nd year, 2nd semester, griechischen und lateiniachen (Zeit- pp. 273-8. schr. f. vgl, sprachforsch, xvii. 33- ' Zatsch/r.f. vgl. sprachforach., 53). xvii. 292-9. ° Chips from, a German WorTc- 3 ZeifscJir. f. volierpsycTiologie, shop : iv. ILssays chiefli/ on thescience etc., V. 340-58. of language, London, 1875, pp. 117- * Rivista orieniale, single volume, 44. DD. llfiO.72. 12fi.S-S4, 116 PARTI. CHAP. in. §17. perfection began before that in wbich the Proto- Aryan was divided into the various families of languages which sprang from itj as appears from the solidity of the essential forms. In the monograph of Curtius account is taken only of the purely philological division : thus in the formation and development of the Proto-Aryan are distinguished seven periodsj the denominations and most striking characteristics of all of which we will indicate with extreme brevity. I. Period of roots (in their simplest form), monosyllabic, not very numerous, already distinguished into verbal and 109 pronominal, employed in that oldest epoch as real words (primitive words) .' II. Period of tbe ' determinatives ' (in the sense in which we have already seen this word used by Fick, who borrowed it from G. Curtius) : these ele- ments, whose origin is still very obscure, by becoming suffixed to the primitive roots notably increase their number and render them fitted for expressing many difierences existing between the ideas which they represented.^ ^ It is a fact worthy of remark so-called determinatives : these roots that M ii 1 1 e r now doubts the neees- expressing different forms of the sary monosyllabism of the primitive same actions might in process of roots. He is afraid of mistaking for time have lost hy elimination the actual fact the last result of our in- features which distinguished them tellectual labour which induces us from one another, preserving only to consider as primitive all that ap- the element common to them all as pears more simple, while, he observes, then: most simple form, correspond- the nearer we approach the sources ing to the most general sense, of the languages, the more frequent Hence, concludes Miiller, the line do we find the words which repre- of separation is quite valueless by sent the most minute diflferences, the which Curtius divides the first more rare the general expressions. from the second of his periods. — But 2 M. Miiller observes that these it should be observed that M filler's " determinatives" have not been in- theory does not explain to us the vestigated hy scientific analysis with existence of the common element great success, nor do they exhibit in just mentioned, which seems very every case the same meaning. He clear on the other hypothesis, and thinks that, instead of supposing that the latter, as we have already to be primitive the simple roots to remarked, (§ 10, pp. 50 and 59), is which tlie determinatives might have much more in keeping with the con- been added, we might with equal stant process of philological iuvesti- riglit regard as primitive the roots gatiou. STEMS AND WORDS. 117 III. Primary verbal period. To verbal roots are indis- solubly attached pronominal roots as signs of the subject : between the former and the latter there is a predicative relation, the precise conception of which is the characteristic of the Indo-Germanie verbal structure ; from this bond is produced a small sentence, the germ of the larger sentences. The active forms appear to be anterior to the middle. That furthermore the simplest of the verbal forms came into existence before the nominal stems constructed with special suffixes and before the cases G-. Curtius attempts to demonstrate by four arguments : 1st, the primary verbal forms are not numerous ; 2nd, they would be denominative, and would show themselves as such, if the manifold nominal i^" forms had preceded them ; 3rd, the primary verbal forms were better preserved than the others ; 4th, a varied nominal formation anterior to the primary verbal is improbable ; before this the formation of the cases would be quite in- conceivable. To the period which we are discussing, mkI which might be variously subdivided, G. Curtius refers re- duplication and augment. Hence two forms of present and past, a form without strengthening and a strengthened form, in the active and the middle : no indication of moods. The noun has not yet assumed a form of its own in antithesis to the verb : gender is not yet marked at all.' IV. Period of stem-formation. The absence of equilibrium between the verb and the noun, rendered necessary the expansion lu of roots by means of nominal suffixes increasingly complex, ' TLe illustrious critic above quoted first period were not yet fixed iu the does not regard as very forcible third. The absence of certain suf- tho proofs adduced by Curtius for fixes of nominal flexion from the supporting his assertion of primary verbal forms is not sufficient evi- verbal formation anterior to any be- dence that the former was not de- ginning of nominal formation. In veloped contemporaneously with order to admit the verbal forms as these latter, because there is a re- they were proposed by Curtius, it markable difference between the would be necessary to assume that flexion of the verb and that of the the phonetic laws existing even in the noun, nor does language always 118 FART J. CHAP. III. § 17. and indicating' slight differences, other than those which were denoted by intensification and accent. It is probable that the nominal stems were originally very numerous^ and that afterwards usage distinguished the one kind from the other. To this age seems also to belong the distinction of adopt in like cases the same means. It is incomprehensible that, while the first forms of conjugation were already in existence, some form of declension could not yet have arisen. Such also is Steinthal's opinion. Still more fiercely does the doc- trine here taught by Curtius come into collision with that which had a powerful champion in Ascoli {I)el nesso drio-semiticOj lettera ad A. Kuhn. — Studj drio-semitid, article 2ud) and before him, as far as we know, in F. Miiller (Tier veriaU ausdruch im drisch - semitischen spraehJcreise, in the SiizungshericMe der phil.-hist. cl. der K. Academie der wissenschqften, xxv. 379-415) : a doctrine according to which the verbal stem is a ' nomen agentis.' " We then," writes Ascoli (Studj, etc., ii. 33), " should consider that the noun existed In Aryan speech before the verb. For us, the Aryan verb would reveal a vast and continuous system of nominal formations, of ap- pellatives of the agent; all these forms, from that which can hardly be called an expansion of the original monosyllable, to the full trisyllables, should be traced from derivative elements which still continue with like functions in periods relatively modern." Ascoli, therefore, as Giussani observes, would hold that the period of the original mo- nosyllables was immediately suc- ceeded by the period of the nominal stems, the fourth according to Cur. t i u B 's chronological order. Nor does the necessity of admitting serious phonetic corruptions in so ancient a period of the Aryan language suffice to deter Giussani from giving a hearty welcome to Ascoli's hypo- thesis : to him it appears quite natural for such corruptions to have taken place in that epoch in which, while Aryan from being monosyllabic became polysyllabic, without losing at once the primitive dress of mono- syllabism, this latter must have exercised its influence on several of the new polysyllabic words, shorten- ing them into a single syllable, to which afterwards were united new suffixes. This is a difficult problejn which we can hardly hint at in the present note. Let us add only that the desire to discover in every verbal stem a nominal stem may appear to many an exaggerated induction, pro- ceeding from excessively systematic intellectual tendencies, and which, perhaps, cannot be maintained by unassailable arguments. Hence Sch we izer-Sidler agrees with C u r t i u s in admitting the existence of a series of verbs which have no denominative character. Such seems to be also Justi's opinion, and Steinthal, as we shall see shortly, regards as contemporaneously de- veloped vei'bal stems and nominal stems, and considers these to have been in the same period used botl in conjugation and in declension. STEMS AND WORDS. 119 the masculine from the feminine gender. As the difference which we now see did not yet exist between verb and noun, it may well have happened that nominal stems were, like simple roots, used for verbal stems, e. g. [bhar-a-) : thus continuous action, expressed first in the noun by the addition of the suflSx -a- to the root, would have been denoted in the verb as well. Other suffixes of the present may themselves too be considered as nominal suffixes. The nominal stem having in this manner become a verbal stem, may be conceived as a ' nomen agentis.'' The number of the verbal forms was thus remarkably increased : to the foregoing were added, 1st, stems in -a-, with or without intensification of the radical vowel ; 2ndly, stems in -nu- and stems in -na-. The act devoid of extension in time was naturally represented by the shortest form : the longest, on the other hand, were used, sometimes one of them, some- times another, to indicate the action conceived as extended or protracted. From this same principle was developed, ii3 according to Curtius, also the first distinction of mood: through the relation which is seen to exist between pro- tracted action and action intended to be completed (the original force of the conjunctive), in so far as such actions are both opposed to the act which is rapidly achieved, the author, following Steinthal, discerns in the conjunctive a present denoting duration. At first the short -a- as we have already seen sufficed to indicate this mood : when this vowel was for phonetic reasons introduced into forms to which it was originally foreign, then the conjunctive was expressed by lengthening it. Evidently the conjunctive bhard-ti bears the same relation to the indicative bhara-ti as the conjunctive hana-ti to the indicative han-U. Thus mood became completely separated from time. Under the influence of the conjunctive arose a difference between the verbal forms with thematic -a- and with the pure radical vowel and the verbal forms with a strengthening of this 120 PART I. CHAP. III. % 17. vowel : the former expressed the momentary action, the second the continuous action; a difference preserved in the conjunctive and in the imperative, which seems to have been developed in this epoch. V. Period of the compounded verbal forms (compounded with verbs which gradually lost in them their own meaning, and are related to the other verbs something like the article to the pronoun). The compounded verbal forms are divided by Curtius into two classes. A. Tense stems compounded from nominal stems without stem-sufExes with auxiliary verbs. In this function we see : 1st, the verbal root as denoting momentary past action (compound aorist); 2nd, the verbal moi ja, jd [to go], according to Bopp and Curtius, from which fundamental meaning come the following : 1st, ' to go about,'' etc. (present of duration) ; 2nd, ' to fall into anything ' (passive) ; 3rd, 'to tend to anything' (optative). From the optative is distinguished in the endings by its fulness the compound 113 future, which consist of a verbal stem and the future of the root as. To this first class belong various compound forms exclusively peculiar to the several families of the Aryan languages : the germs, however, of such forms are not per- haps posterior to the division of Proto-Aryan. B. Tense- stems compounded from nominal stems, which are already provided with stem-suffixes, and auxiliary verbs (the verbs in -a-ja-mi may serve as examples) : this composition extends, observes the author, to all the forms of the verb. Nominal flexion cannot have been developed contemporane- ously either with the first or with the second class of such compound verbal forms : a compound aorist must have been formed in an epoch in which nominal stems had not yet an ending indicative of number; the presents in -a-jd-mi would have been impossible in a period in which men were already accustomed to express by case-suffixes the relation of the noun to the verb. The place of declension STEMS AND WORDS. 121 which had not yet come into existence was supplied by- several means, and of such we even now see how great is the efiBcacy in languages of imperfect structure.' VI. Period of case-formation, in which the author distinguishes two strata. To the first belong, in his opinion, the vocative, the accusative and the nominative. The close afiinity existing between these three cases appears from their identity in the n^ neuter, and from their not interchanging with the rest. The vocative, which in the singular is generally the pure stem, should be considered as a remnant of the preceding period : the formation of the accusative and the nominative appears to Curtius a continuation, a new development of the for- mation of stems. The case did not appear in its true force until the suffix which represented it, at first not clearly distinguished from the elements of stem-formation, was regarded as movable, and until there became attached to the same stem, without changing the extent of the idea, sometimes one ending, sometimes another, sometimes none. Case-formation was in its origin so intimately connected with stem-formation, that the suffixes -vi of the accusative, -s of the nominative expressed not only the relation of a ' Here M. Muller notes that, have felt the need for clearly dis- even after the development of tinguishing the pingular from the nominal flexion, were formed com- plural and the nominative from the pounds, the first element of which accusative, before the need for denot- appears in the form of a stem and not ing the differences existing between of a case. It is not the fact, observes the three persona. Steinthal re- the distinguished Vedic scholar and jects the argument which Curtius philologist, that composition could drew in favour of his position from have taken place only in times an- the compound verbal forms, which terior to flexion because after this exhibit nominal stems without suf- latter the speakers were ignorant of fixes of case and number : this the stem-forms: these stem-forms comes about, according to Stein- showed themselves in certain cases of thai, because forms like these are declension, and moreover the sense really compounded and not peri- cf analogy was the guide for the phrastic. Moreover composition, he formation of new compounds. Aiid teaches us, has no meaning except in it appears to Muller that those antithesis to flexion. very ancient ancestors of ours must 122 PART I. CHAP. in. § 17. noun to other words, but also its gender, masculine, feminine or neuter. And perhaps this was the original function of these two suffixes, and the indication of gender preceded and gave birth to that of case : when the thing as living was denoted by means of the suffix -«, and the thing as not living by the suffix -m, it is natural that the former should appear as subject, the latter as object. The singular was probably followed by the plural and afterwards by the dual. That the Proto-Aiyan language was for a long period content with this first stratum of cases we see from the extensive use which was made of the accusative. The second stratum comprises all the other cases, among which Curtius selects as the subject of his most exquisite analysis especially the genitive.^ VII. Ad- 115 verbial period. Even before the Aryan unity was split into 1 This relatively very late origin of the cases does not seem at all probable to Miiller and Stein- thal, who opposes to Curtlus's theory the following objections : 1st, there is not, according to it, a proper equilibrium between verbal and nominal flexion in the epochs of their generation, because the forms of the verb would have de- veloped themselves considerably be- fore those of the noun j 2udly, in order to establish the necessary equilibrium between the latter and the former, the formation of nominal stems is not enough, for these can- not correspond (at least In the order of the ideas) to verbal flexion ; Srdly, it can hardly be understood how a nominal stem without case-suffix was employed side by side with a verbal form with personal ending, if nominal stems were used as verbal stems. Moreover, the very obscurity in which, even according to Cur- tius's opinion, the origin of the cases is wrapped, the serious pho- netic corruptions of their suS^es, the petrified forms in which they not unfrequently present them- selves, and the agreement of the Aryan languages in several of them prove their high antiquity. Stein- thal, however, in his much com- mended Characteristih der hawp^ sdc/Uichsten iypen des sprachbaues (Berlin, 1860, p. 300), supposed that, while the nominal forms in -a- by means of the addition of personal suffixes were changed into conjuga- tional forms, there were added to the demonstrative -a- of those nominal forms other demonstrative elements (and especially -« from sa, as mark of the nominative), thus giving origin to declension. With regard to the resemblance which appears between vocative, accusative and nominative, and which was emphasized by Curtius, STEMS AND WORDS. 123 the various families to wliich it gave origirij some word- forms belonging to nominal flexion became fixed in the form now of this, now of that case, that is to say, they lost the capacity for being completely declined like the rest : from this phenomenon arose adverbs and prepositions, the number of which was afterwards so remarkably increased in the several Indo-European languages. In this function at first were employed especially certain words not very full of meaning, and therefore principally used to express relations of place and time and other closer relations. From the oldest records of the various Aryan dialects, for example, the Homeric poems, we learn that the use of such petrified forms was in Proto-Aiyan exclusively adverbial. It was only by degrees that the habit grew of connecting them more closely with verbs and nouns (hence the prefixes) ; it was only little by little that the relations were formed which are now seen between prepositions and cases ; it was only us in this last stage of their development that some of the forms in question became post-positions. The infinitives also, observes Curtius, are isolated forms of declined nouns of action : but the wonderful variety which we observe in them, the differences existing between the individual Aryan dialects in the selection of the suffixes V7hich form these infinitives, affi)rd us almost certain evidence that the infini- tive was developed not before the division of Proto-Aryan, Diintzer attempts to give an ex- tinguished from one another. With planation of it such that their this hypothesis, which, to tell the primitive affinity matters not. The trath, does not seem to us to be suffix -m of the neuter is probably a founded on a very solid base, and sign of gender in the nouns (as t is with some other observations, in the pronouns). The fortuitous Diintzer opposes the doctrine of coincidence of the accusative singu- Curtius, remarking that there is lar in -»j with the nominative in -m the most marked antithesis between of neuter stems in -a- brought nominative and accusative, and about the result that, by analogy, denying to the first of them, no in all the other neuters, and in all less than to the vocative, the force the numbers, the nominative and of a real case, the accusative were no longer dis- 124 PART I. CHAP. III. § 17. and independently in the various languages in which it underwent the transformation.' Such is the chronology of the formation of the Proto- Aryan language according to G. Curtius. Justi well re- marked that in these researches so novel and so attractive the distinguished Greek scholar and philologist was able to give proof no less of prudence than of boldness^ and to collect a large number of facts and place them in the light most suited for them. And Giussani well observed that it is precisely in the attempted arrangement of the facts that the greatest and most alluring merit of this work of Curtius consists. Of the individual ideas which we note in it many are not at all new, others are not of unquestion- able value ; the fundamental conception of a chronological order in the formation of the original Aryan is certainly anterior, as we have just seen, to this monograph: but its worth and attractiveness consist in the complete realisation, although within very narrow limits, of such conception, in such a manner as to comprehend the entire existence of Proto-Aryan and all the most important phenomena which it presents. Nor, in the opinion of several critics whom we have mentioned, has the heroic daring failed to attain a welcome success, if we except the doctrine concerning the 117 late origin of nominal flexion. We should not, however, pass over in silence the fact that M. Miiller declared him- self opposed to any chronological division, in a strict sense, of the life of language (because none of the forces operating in it is wont suddenly to cease from the exercise of its action, and every period is continued in the following)/ and affirmed that in the development of the original and 1 M. Muller doubts whether ^ We are sure that G. Curtius this period can reasonably be dis- does not attribute to the word tinguished from the preceding, be- ' chronology' in this work too strict cause several adverbs exhibit the a sense, nor does he intend to draw most ancient forms of declension. lines absolutely separating, with STEMS AND WORDS. 125 fundamental Aryan we can rightly distinguish only th^ three phases which we have been accustomed to call the isolating, the agglutinating and the inflexional, and which he makes to consist in the successive prevalence of three diflPerent tendencies, each of which impresses a particular character on a period of language (understood in a wide sense) without vanishing altogether in that which follows it, and each predominates in certain classes of languages without being entirely wanting to the others. We have in another place' set forth the grounds on which is based the theoiy propounded by Schleicher," M. Miiller,' Whitney,'' and other philologists, a theory generally received as a demonstrated truth, according to which language must have arrived at the inflexional form only by passing through those of isolation and aggluti- nation : and these grounds consist especially in their con- ception of the original force of the formal elements and in the marked tendency which is here and there displayed by isolating languages towards agglutination, and by agglu- tinating languages towards inflexion. We have there stated simply the objections to this doctrine raised by Pott^ arid \>j Renan:° that the change of an inorganic us into an organic language is quite inconsistent with the laws of the human mind; that no proof of such change can be derived from the monosyllabism of the Aryan roots, nor geometrical exactness, from each re-editcd in the Chips from a German other the seven periods distinguished Workshop, iv. 65-116) . by Lim in the formation of Proto- ■* Language and its study, etc., Aryan. London, 1876, lecture 7, pp. 249-87. ' 8eoVe7,z\'s Introduction, eta., ^ M.Miiller unddiekennzeichen pp. 120-6. '^fi'' sprachverwandtsohaft {Zeitschr. 2 Die deutsche sprache, Stutt- d, deulschen morgenliindischen ge- gart, 1869, p. 45, etc. — Die Dar- sellsohaft, ix. 405-64 : see especially winsche theorie und die sprachwis- p. 412. - jEtymologischenforschungen censohaft, Weimar, 1873. etc., 2nd ed., part 2, section 1, p. 95. 3 On the stratification of lanr- ^ De I'origine du langage, Paris, guage, London, 1868 (a monograpli 1858, p. 10 sqq,, 103-17 151-68. 126 PART I. CHAP. III. § 17. can it be believed tbat such roots in that their simplest original form were used with the force of words in speech ; the farther we go back towards their origin the more synthetic do the dialects appear ; we ought not to recognize in the isolating structure of a language a certain indication of a civilisation inferior to that of peoples who speak lan- guages grammatically richer ; no example is supplied by the history of languages of the supposed transformation of a system of speech, nor does such hypothesis seem allow- able when we reflect how many centuries of undivided life it would be necessary to admit with regard to the original Aryans and Semites in order to be able in this way to ex- plain the development of Proto-Aryan and Proto-Semitic. We have, lastly, stated the very remarkable opinion of SteinthaP who thinks that, if the original and funda- mental Aryan resembled at all the isolating and the agglu- tinating dialects, such resemblance was only superficial, and that there was always in its substance another germ endowed with greater efficacy. SteinthaPs view seems to be supported by the Italian Lignana/ who thinks it " very probable, that the Aryan language before definitely fixing its organism, and becoming established in what we call the first epoch passed through three phases analocrous to the three types pointed out by Wilhelm Humboldt," that is, to the isolating type, the agglutinating and the in- flexional. " But these phases,'' adds the learned professor 119 of the Ateneo romano, " are pre-historic and em- bryonic phases, and the first epoch is that of the completed and definite type." And he calls them "pre-historic, or embryonic, precisely because it is in their nature to pass necessarily the one into the other, and not to stop until the whole evolution in the direction of the type has been com- 1 Zeiischr. f. volherpsychologie, e le tre epochs delle lingue e let- ete. ii. 238-9. terature indo-europee, Eoma, 1871, ^ Le irasformazioni delle specie pp. 22-4. STEMS AND WORDS. 127 pleted." The Aryan type may have had various periods, each with its own characteristic of structure : " but that does not prevent the types, which have afterwards realised in a permanent manner one of these morphological criteria from being the historical antecedents of the Aryan" type." — Sayce has declared himself totally opposed to the hypothesis of the evolution alluded to.^ It is absolutely impossible that the language of a people should pass from one to another of the three forms of speech in question without there being effected what is little short of "a radical metamorphosis of the mind." Moreover, in the theory of the three phases no account has been taken of the polysynthetic and the incorporating languages.^ It may be added that verbal flexion is not effected entirely by pronouns or by verbs added to roots, but also by vowel- change : nominal flexion is not explained at all, he thinks, by the hypothesis of suffixes of pronominal origin. " Can we suppose that the same people who so distinctly marked out the meaning of mi in the verb can have employed it to express the sense of the accusative ?" Nor is it of any use to reply that the pronouns, all possessed of indefinite signification, might be attached to the roots at 120 haphazard to express the various relations existing between them in the sentence, out of which the different cases gra- dually grew in a mysterious way, each appropriating as its suffix one of the pronominal roots in question : to.such hypo- 1 TTie principles of comparative that several philologists do not regard philology (a work quoted before, in these as constituting a fourth and a which this Semitic scholar and phi- fifth classoflangnages,butonlya sec- lologist, with vastness of learning tion of the second class, which com- and acuteness of analysis, undertook prisesall theagglutinating languages to examine the principles, methods (in a wide sense). See Pezzi's and most important results of recent Introduction, eUi-f^. 116-8; Sayce, linguistic investigations) : see chap. 4 ibid., p. 146; Hovelacque, La (The theory of three stages of de- Unguistique, Paris, 1876, pp. 112-5 velopment in the history of Ian- [now translated under the title guage), pp. 132-174 (ed. 2). Science of Language, by A. H. 2 It should, however, be observed Keane, London, 1877. — 2V.]. 128 PARTI. CHAP. III. §17. thesis it might well be answered^ 1st, that it is not founded on facts ; 2nd, that with such machinery mutual under- standing would have been impossible; 3rd, that the supposed merely accidental terminations, far from leading the mind towards analysis, would have confused it ; 4th, that certain distinctions which it was most important to make are not seen to be marked at all by means of difference of suffixes ; 5th, that, even if we accepted the proposed hypothesis, we should be no nearer finding the primitive language of the Aryans in a condition resembling that of the agglutinating languages, since these do not form their cases by means of pronouns, but use postpositions, or rather nominal and verbal roots. If, in historical times, a word with a meaning defined and independent of any other gradually degenerated into a mere element of flexion, this phenomenon proves only that the tendency towards inflexional structure already swayed the language, nor can what took place in a period relatively late be believed without doubt to have taken place in an epoch far more ancient.' How can we know whether the suffixes were once independent words, while their meaning, such as it appears to us in the surviving records of the languages, is certainly not what is attributed to such words ? How is it possible that the Aryans, when hardly conscious of the relations of case, represented them by words endowed with an independent existence, suffering 121 them afterwards, when such consciousness had become more clear, to be changed into mere suffixes?^ Phonetic ' Let us, however, draw attention possibly reply to him, that in the to the continuity which is observed words originally denoting the cases in the life of languages, and reflect a change took place analogous to that it is only from the study of that' which happened in the concep- linguistic facts belonging to the his- tions of them : the former and tlie torical epochs that we can learn the latter may have acquired in pro- art of investigating the pre-historio cess of time a signification growing forms of human speech. constantly less material and more 2 The upholders of the theory so formal, fiercely assailed by Sayce might STEMS AND WORDS. 129 changes, facts of a purely external and material nature, were not sufficient to produce, as they are not sufficient to explain, such internal and formal revolution of language as would be the change of an agglutinating to an inflexional structure. Sayce regards as even less probable the hypothesis of an isolating condition in the formation of our fundamental language : in that first condition the Aryan root would have been, owing to its peculiar inde- finiteness, incapable of denoting an idea limited by another. — Lastly, the opinion undoubtedly deserves notice which has lately been declared on this subject by that learned and profound philologist, Friedrich Miiller.' Starting from Steinthal's fundamental principles with regard to the various nature and significance of the linguistic types '^ he considers the isolating and formless languages as embryo- nieally related to all the languages which are agglutinat- ing and unendowed with form, while the Chinese, which is isolating, but adapted for distinguishing matter from form, would be embryonically related to the formal inflexional languages. An agglutinating language, which is not adapted for the distinction just noticed, cannot possibly be changed into an inflexional language in which the form appears clearly conceived and represented : it would have to undergo the strangest metamorphosis, which could not be brought about by any cause that we can see. This lengthy discussion of a problem which appears to us of no slight importance, both for the history of the Indo-European languages, and for the affinities of philology 122 with the Darwinian hypothesis on the changes of species,' ' Qrundriss der sprachwissen- Lign ana, see Ferri^re, ie -Da?'- scTiaft, i. 139-40. winisme, Paris, 1872, part 2 (La s6- ' See ibid., pp. 77-82; Stein- Uction dans les langues), pp. 107- tlial, Characteristic, etc., sect. 4; 39j Muller, M., Mr. Darwin's Pezzi's Introduction, etc., pp. Philosophy of Language (la Fraser' s 129-32. Magazine, May — July, 1873) ; Id., > On this subject, besides the works My reply to Mr. Darwin (in the already quoted of Schleicher and Chips, etc., iv. 433-72), etc. K ISO PART I. CHAP. III. § 18. has led us, if we mistake not, to the following conclusion : although not all the Aryan forms which are generally con- sidered as grown from agglutination are to be regarded undoubtedly as such, we nevertheless find several which, not to have recourse to very rash hypotheses, we are forced to assume to have arrived slowly at the inflexional con- dition from that of isolation through an intermediate period of agglutination ; but, from their very origin, such forms, or rather their rudiments were powerfully attracted towards the inflexional structure by a linguistic tendency peculiar to those conjugating and declining languages which Stein- thal calls endowed with form : a tendency which seems to separate them, even in their germs, from all the other lan- guages. With this conclusion, which appears to us the most probable in a subject so difficult, we bring to an end the discussion of the most recent works on the origin and development of the Indo-European forms of speech as a whole and we proceed to treat, with greater brevity (as is evidently necessary), some works of less extent on the stem-structure and inflexional structure of the Aryan lan- guages. § 18. Among works of this kind the first that meets us is the Commentatio of Weihrich Be gradibus comparationis linguarum sanscritae graecae latinae gothicae (Gissae, 1869), which was ah amplissimo philosophorum ordine in Academia Ludoviciana praemio publico oniata : and not undeservedly, since the author in the ample development of his subject gives proof of accurate observation and subtle examination 123 of facts. Weihrich's monograph is divided into two books : in the first is discussed the meaning and usage of comparison ; in the second the formation of the degrees. It is not our business to follow the author in his analyses of the various kinds of comparison, of its syntactical value, of the suffixes and the other means by which we find it ex- pressed in the four languages mentioned. But our task is STEMS AND WORDS. 131 to make clear the eonceptiou which Weihrich holds of com- parison. Seceding from the ancients and from many moderns^ who, like them, understood in too narrow a sense the use of the suffixes of comparison, and proceeding by the new route opened up by Corssen, he sought to reach the original meaning of those elements, which, according to him, must be considered ' local.' He discovers proofs of his affirmation in several forms, undoubtedly very ancient, as appears from comparing together the Aryan languages, which denote relations of place by suffixes which we see later becoming signs of comparison. How from this ori- ginal sense its later significance was developed let us hear from the author himself : " Aliunde vero novimus linguam in antiqua ilia intuendi ratione, qua omnes conditiones ad loci analogiam perciperet, non acquievisse, sed eas res, quae ad animum pertinerentj etiam animo i. e. sensu interiore comprehendisse. Quare lingua, cum vellet in aliis sub- stantiarum accidentibus gradus quosdam distinguere, easdem illas antiquissimas rationes in qualitates proprias, quae animo percipiuntur, per metaphoram quandam transtulit et ex horum accidentium nominibus suffixorum comparativorum auxilio nova adiectiva derivavit, quae comparativus et super- lativus vocantur. Hinc prima ilia significatio compara- tionis e loquentium memoria sensim evanescere coepit, ut posterioris aetatis homines res pure cogitatas non amplius ad loci analogiam intuerentur, sed sicut mente compre- hendebant, ita etiam ad animum referrent.'"' The cases 121. which, in the ordinary parlance of grammar, are said to be governed by the comparatives, themselves also declare the most ancient significance of comparison, being those which denote the motion by which a man approaches or withdraws from anything (ablative, genitive, instrumental in Old Indian, genitive in Greek, ablative in Latin, dative in ' See the whole of chapter 1 ( Q,mi gradus eomparationis), pp. 1-13. sit comparatio et quid signiflcent 132 PART I. CHAP. III. § 19. Gothic) : hence a new argument in favour of Weihrich's position with regard to the original sense of the suffixes of comparison.' § 19. To stem-formation belong also two other works which we propose to notice briefly^ one of which concerns the formative suffixes of the various present-stems, the other the element which is generally regarded as a sign of the optative mood. The first of these is the dissertation of Brugman en- titled, Zw geschicJite der praesenstammbildenden suffixe? The author begins by distinguishing the present-stems which are simple roots from the denominative present- stems, which he considers to be later than the former. And as the radical presents do not seem to be at all different in meaning from the nominal presents, as he terms them; as, further, we very frequently see various present-suffixes used without any kind of difference in the same verb, while the force of such suffixes in the no- 126 minal stems does not seem clear, accordingly Brugman does not believe that such elements were originally adapted for representing important definitions of the verbal con- ception, for example, as some think, that of 'duration.' He regards as unoriginal, and brought about only by phonetic causes, the limitation of these suffixes to the present-stem : in proof of this affirmation he adduces the most ancient use of the suffixes -ta-, -na-, -nu- and -ja-, 1 "... . Cum res, quacnm com- feoto rei, quacum comparatur, no- pai-atur, ita cogitari nobis videatur, men earn declinationis formam in- ut componatur vel collocetur cum ea duere videmus, qua vel componendi re, quae comparatur, et separetur ab et consooiandi vel separandi et dis- ea, sententiam nostram de significa- cernendi uotlo signiflcari solebat " tione primaria optime probatam (ibid., see pp. 31-5). gaudebimus, si comparationis, uti 2 Sprachwissenschaftliche ab- perceptio ipsa e loeorum intuitiono Tiandlungen hervorgegangen aus O. profectaest, strucfcuram syntacticam Curtius' grammatischer gesell- eadem cogitandi dicendique ratione schaft zu Leipzig, Leipzig, 1874, pp. coustare deprelieuderimus. Ac pro- 153-75. STEMS AND WORDS. 133 a use wbieli in fact does not appear to be connected with the idea of ' continued action/ The second of the two works alluded to is a monograph of that most learned Indian scholar and philologistj Theodor Benfey.' His aim is to prove that the opta- tive in the original Aryan was formed by means of a composition of verbal stems with the indicative and with the conjunctive of the present and imperfect of a frequen- tative or intensive verb, * { = ii,a reduplication of i [to go]) also expanded to ia, with the sense ' to have recourse to, supplicate, desire, wish ' derived from that of ' going often or anxiously.' Thence several forms which Benfey re- cognises, more or less faithfully preserved, as formative elements of the optative mood in the various families of the Indo-European languages. Benfey's reasoning failed however to convince Bergaigne, who does not think it proved that the above mentioned i had the force of a suffix in conjugation ; it is no proof of this that it was used as an auxiliary in Latin and in Sanscrit ; the very multipli- city, observes the French critic, of the roots meaning ' to go,' used in Vedio periphrases shows us that " the category was as yet more in the spirit than in the lan- guage." Besides it is one thing to be used even frequently 126 in periphrases, another thing as a regular suffix. Lastly the illustrious German philologist has recourse not to a real root, but to a verbal stem which probably belongs only to Old Indian.^ > Uher die entstehtmg und die 2 With the optative is closely con- formen, des indogermanischen opta- nected the future compounded with tiv (potential) so wie iiber das fu- -s-ja- (— as-ja =), which also is turum amf sanshritiscTi a^imi u.sM., referred to Proto-Aryan. Of the Gottingen, 1871. See the observa- various means used by the Indo- tions made on this work of Benfey Europeans to express future action by the learned Bergaigne in the we have spoken in a Dissertazione Revue critique d'Ustowe et de lit- storico.comparativa_ on the Forma- tiraiure, 6th year, 1st, semester, zione del fuiuro aiiivo negP idiomi pp. 194-8. iialici ed ellenici (Torino, 1871), 134 PART I. CBAP. III. § 20. § 20. From these researches on the subject of tense- and mood-suffixes we pass on to those which concern the personal endings, a subject which gave Priedrich Miiller an opportunity of putting forward, in two very short dissertations/ some ideas incompatible with the doctrine ordinarily followed/ which have been subjected to a minute examination by G. Curtius.' The fullest forms of the personal suffixes, observes Miiller, are generally regarded as primitive, and as derived from these by means of phonetic corruptions the others which are less rich' in sounds : hence the division of such suffixes into primary and secondary. But what, he asks, could possibly be the cause of these supposed phonetic corruptions ? May it be, perhaps, that the augment, by drawing upon itself the acute accent, weakened the final syllable ? It is known that in the oldest condition of Indian, Iranian and Greek, the augment was very far from being as common and as 127 regularly used as in the classical age of these languages :^ it may be noticed, moreover, that the optative and the imperative had no augment. May it be, perhaps, that there was in languages like these a tendency to mutilate the last syllable of the words, especially the vowels ? If laying stress on the feet that the phil.-hist. el. xxxiv. 8-16 and Ixvi. Aryan langoages do not exhibit a 193-212). verbal form exclusively belonging to ^ Schleicher, Compendium, etc., the future, and distinct from those pp. 663-706. Curtius Das verium of the present : whence it appears d. gr. spr. etc., i. 34-103. that the difference between that ' Zur erklSening der personalen- which is being completed and that dungen [Studien z. gr. u. lat. gram., which is to be completed was not iv. 211-23). The critical notes with understood by the primitive Aryans which we shall furnish our descrip- and translated into language so tion of Muller's theory will be precisely as the antithesis between drawn from this work of the re- the completed and the incomplete nowned Greek scholar and philo- (p. 42). legist. 1 Zur suffixlehre des indogerman- ■> Schleicher, Co»»pe»(Z»«»j, etc. iscJien verbums ( Sitzungslerichte d. pp. 749-61. K. Academie der wissenschaflen. STEMS AND WORDS. 135 that were the case, the weakening would appear in all the suffixes : it is not, in fine, a phenomenon which belongs to the most ancient period of our languages. Supposing a weakening of -ma to -mi to have taken place, there would have heen intermediate forms between the first and the second : as it is, no trace of them appears in Aryan languages." Hence F. Miiller cannot recognise in the so-called secondary personal suffixes forms less ancient than the primary. Nor can he bring himself to believe that the plural and dual endings are due to the additional com- position with two pronoun -stems, 1st, for phonological reasons; 2ndly, because in no language of high organisation is there found a plural pronoun formed in a similar manner (as is shown also by Semitic and Turkish examples) ; 3rdly, because in several of the suffixes to be explained w6 fail to find the supposed composition ; 4thly, beeaiuse, in fine, the compounds of the class mentioned are almost ex- clusively peculiar to Indian and Iranic.^ He pronounces a judgement no less severe on the ordinary theory of the middle 128 endings, which are generally thought to have been pro- ^ Here, however, we should notice very important the usage of lan- with Curtius that it is a funda- guages which are of a different mental idea of historico-comparative stock, and which frequently diverge grammar that the fullest forms have from Aryan. He admits that there preceded the others. For the rest, do not exist, in pronominal flexion, even in Proto-Aryan we find exam- plural forms constructed by means pies of wealiening of a to i, e.g. to of the addition of two stems, but lei from Tea, dvi from dva, etc. It is considers declension to be developed not conceivable that several grades subsequently to conjugation. The of phonetic decay must be admitted laws of the composition of nominal -between a and i : probably there was stems already fully formed are only an S, which we shall have to very different tiom those which notice again. There is no need to governed the beginnings of all wonder that there have not come flexion. Of additional composition down to us intermediate forms, we have examples in a class of words which are so often wanting, especi- which have several relations with ally when we treat of forms anterior personal pronouns, that is in the to the division of Proto-Aryan. -numerals. 2 Curtius does not regard as 136 PART I. CHAP. III. § 20. duced by the reduplication of the corresponding active endings. First of all, it seems to him very strange that an m, an h, a t, between vowels, should have vanished in the primitive Aryan; secondly there are some endings which cannot be explained at all by such hypothesis; in the third place we must not regard as object either the first or the second of the two supposed suffixed pronouns ; not the first, because in Aryan compounds the dependent member (with a single exception for forms of no great antiquity) precedes the member on which it depends (hence the suffixed pronoun denoting the object should precede the verbal stem) ; not the second, because such suffixes re- presenting the object,and added at the end to a form and afterwards fused with it (as in Semitic) are not a characteristic of the Indo-European linguistic stock ; nor, lastly, are we justified in recognising, e.g. in the -ma- (from -md- ?) o£ tudamai (from tiulamdi, tu-damdmi ?) an accusative of the pronominal stem ma- inserted between the verbal stem tuda- and the active ending -7ni, both because insertions are foreign to Aryan and for other reasons which it would take too much time to state here.^ Since, therefore, the distinguished professor of the Vienna Atheneum cannot give his adherence to the doctrine generally professed on the development of the personal endings, he reconstructs 129 the history of them in the following manner, dividing it into five periods all belonging to Proto- Aryan : we will notice their fundamental characteristics. Period I : addition of personal pronominal stems to stems of verbs without exact determination of number and tense (ex. iuda-ma). 1 F. Mviller, according to this facts might be brought into doabt. critic, is wrong in proceeding here Thus, for example, if there are not to judge of very ancient formations, in Aryan objective suffixes apart taking for the guide of his own from verbal flexion, it is only in this opinions linguistic tendencies which that we find subjective suffixes : shall prevailed only in subsequent epochs, we be obliged ou this ground to deny By similar arguments most assured the latter also ? STEMS AND WORDS. 137 Period II: intimate cohesion of the two parts of the verbal expression with weakening to e and afterwards vanishing of the fi^al unaccented a- sound of the suffix (tuda-me, hence tuda-my Period III: indication of number {fuda-m-as, plural, formed from tuda-m by the suffix -as which appears in this function also in nominal flexion — tuda-v-as, dual, in which the -vas comes, by a process of differentiation, ixova.-mas). Period IV: expression of the reflected action by means of an -a- (pronominal stem of the third, person, parallel to -sva"), which only in course of time became fused with the verbal expression ; hence the middle endings {tuda-m-af Period V: denotation of the present by means of the suffix -i (pronominal stem signifying that which is more near at hand), contemporaneous with that of the imperfect, the aorist etc. by means of a prefixed a- (which itself, too, is a pronominal stem, but denoting remoteness) ; in the latter the tense-sign precedes, in the former it follows {tnda-m-i, tuda-tiz-as-i, tuda-ma-i etc. — a-tuda-m etc.) .'' Such formations are naturally followed by certain iso phonetic corruptions (-wa from mas etc.). The description ' The hypothesis of this disap- minations in question reduplicated pearance does not seem easily recon- active suffixes, cilable with the opinion which we * It has been observed, not with- have just seen expressed by Miiller out reason, that verbal stems in a on the phonetic decay of the suffixes. consonant could not long preserve If we believe possible a weakening formswith the endings)»,«,< without of a to 2 why should we deny a weak- a vowel, before the -i of the present eningofgtoj? On what grounds was added to them. And it seems will one who admits a vanishing of strange that by such an addition a o refuse to admit a weakening of a tense should be indicated which least to i ? of all needs to be marked by a pe- 2 That such a may have a re- culiar suffix, since the combination flexive force has by no means been alone of an object with a predicate^ demonstrated. affirmed in the present, is sufficient 3 F. Miiller does not explain by to cause it to be conceived as hold- his hypothesis the secondary middle ing good for the present. The posi- forms and imperative forms on which tion also of such 4 may seem strange, especially the common theory is while other tense-marks (augment, based which perceives in the ter- reduplication) are initial. 138 PART I. CHAP. III. § 21. which we have given of the theory of this illustrious philologist with respect to the formation of the personal suffixes, and the critical observations of G. Curtius, which we have stated, appear to us sufficient to convince . the reader that a theory like this contains elements of very doubtful worth perhaps to a greater extent than the ordinary doctrine, especially if it is assumed in the form in which it was set forth by the learned author of the work above cited on the structure of the Greek verb. — And now, since we cannot, without being led into too long and minute disquisitions, examine certain other treatises on verbal flexion, treatises of a very special subject' and of very slight importance for our object, we proceed at once to notice several works on the subject of declen- sion. § 21. Let us start with three treatises which concern the origin of nominal flexion^ considered in all or almost all its forms. In the first of such treatises, a work of Chaignet,' only a few pages are devoted to the subject which we are discussing, but in these the author labours to combat the doctrine of Bopp with regard to the pro- nominal natui-e and the primitive independent existence of the elements which make up Indo-European declension, observing 1st, that owing to the profound difference there is between the endings of verbal flexion and those of nominal flexion, the sufilxes of the latter cannot be con- 131 sidered as personal pronouns ; 2ndly, that even admitting the local signification of the cases, it does not follow from 1 Among these it will be enough flexion, that is to say declension as to quote as an example Benfey's opposed, in all its extent, to conju- monograph Vher einige pluralbil- gation. dungen des indogermanischen tier- 3 La pUlosopUe de la science du hum,, Gottingen, 1876. langage itudiie dans la formation - Under this term we propose, as des mots, Paris, 1875 (see pp. 185- always, to -comprise also pronominal 94). STEMS AND WORDS. 139 this that they came from personal pronouns, which were certainly not (except those of the third person') mere demonstratives with such signification ; 3rdly, that if the impugned hypothesis be accepted, the cases of such pronouns, which are themselves declinable, can no longer be explained ; 4thly, that the elements of nominal flexion appear for the most part in too slight a form to admit of their being considered as words having originally an independent existence.'' To illustrate the genesis of the cases Chaignet attempts to resuscitate a doctrine which for a long time appeared to be dead and forgotten, accord- ing to which the so-called suffixes of declension would come, as a rule, from an " organic development," natural, necessary, multiplex in its forms : " we ought, therefore, to believe that the greater part of the cases consisted originally only in hardly sensible shades ('nuances^) of utterance, of which the mind, owing to the law of economy which is common to it with nature, and owing to the need of clearness which is peculiar to it, made itself master, causing to proceed therefrom, by the due development of them, the whole system of declension.'^ And this may suffice for Chaignet's book: a book in which the pro- cedure often starts from philosophical conceptions foreign to the science of language, or, from linguistic facts too wide consequences are deduced ; a book which is here and there wanting in that strict accuracy which is one of the most fundamental characteristics of every truly scientific work. The second of the treatises which we ought to notice is 132 ' It is precisely in these that the tain (at least apparent) importance philologists of Bopp's school (with of all those which the author raised some very rare exceptions) usually against the ordinary theory, he has seek the origin of the formative suf- recourse to examples drawn from fixes of the cases. epochs in which such elements had 2 To support this objection, the already become weakened by slow only one which seems to us of a eer- phonetic decay ! 140 PART I. CHAP. III. § 19. a monograph of Bergaigne,' in whieli that learned Prench Indianist and philologist, considering, with G. Curtius/ declension as a development of derivation (or stem-formation), proposes to investigate the sense of the stem suffixes, the combination of which constitutes nominal flexion, with the exception of the elements -s, -m and -t, to which alone he allows for the present the name of endings, without, however, ceasing to recognise in them three elements of derivation. First, he enumerates the various methods of final expansion of stems; secondly, he divides such expansions into two classes, one of which corresponds to the strong cases and the other to the weak cases; in the third place he propounds a hypothesis as to the function of the first and the second. We will describe briefly, using frequently his own words, the results obtained by the researches of the author. In nominal flexion Bergaigne recognises two kinds of derivation. The one is formed by the suffixes -as, -an-, -i-, -a, -Ja- (-«), which are attached to the strong form of the primitive stem without displacing its accent : it is nothing but a prolonged primary formation of feminine and neuter abstracts ; hence, owing to the affinity existing between abstract and plural,' it was used to denote this number and the dual. The other takes place by means of the elements -sma, -sja -sja and -Ja, -i, -an b/d, -su, -i — a and -a, which are united to the weak form of the fundamental stem and generally draw upon themselves the accent : a derivation which changes the primitive stem into an adjective which performs the function of a genitive and assumes, as an adjective used adverbially, the sense of the instrumental, ^DurSle de la derivation dans ^ For instance, Bergaigne quotes la diclinaison indo-europ4enne (Mi- the word humanity (Ital. umanitd) moires de la ifocieti de lingidstique which may be used in a sense de Paris, ii. 358-79). precisely equivalent to that of the 2 See above, p. 121 : see also, on plural human beings (Ital. uomini). pp. 108-9, Ludwig's opinion. STEMS AND WORDS. 141 the ablative, the locative,- and even the dative. The suffixes of each of these two classes of stem-formation are sometimes reduplicated, while, on the other hand, the 133 cases of the plural and the dual which should contain an element of the first and an element of the second class, sometimes lack that one which indicates the number. These derivatives are further, in certain cases, augmented by one of the three final elements -s, -t, -m : which, already distinguished from those previously noticed in that they are not followed by any other, are also the only elements which appear united to the fundamental stem in those cases which have no need either of the derivative suffixes expressing the number, or of those which denote the an- nominal ease or the adverbial cases, that is to say, in the nominative, accusative and vocative singular. The novelty and attractiveness of this work consist in the examination of the significance of the stem-expansions in declension. But, besides that the results of the researches of Ber- gaigne are extremely hypothetical, as the author himself admits, there would always remain the task, even for one who accepted them as demonstrated truths, of solving another and not less difficult problem, to discover the origin and the original meaning of those derivative elements which were used to denote numbers and cases, among which, the signs of the nominative-vocative and accusative singular, of which Bergaigne did not attempt even the analysis, are perhaps those which it would be of most value to us to see satisfactorily explained. We hope the illustrious author will continue his investigations : not a little may be expected from his talent and his learning. The idea of an intimate affinity between stem-formation and declension appears also in the third and last of the works above mentioned.' The object of the author is to 1 Meyer Gustav, 2«w^eso&'cAfe »»(i declination, Leipzig, 1875. der indogermanischen atammbildung i 142 PART I. CHAP. III. § 21. show how in Indo-European word-formation stems in -a-, -i- and -w- are interchanged. Hence he gives us a tolerably- long list of stems in -i- parallel to stems in -a- ; a second 134 list of much smaller dimensions of stems in -u- having beside them stems in -a- ; a third, very short, of stems in -(■'- corresponding to stems in -u-; lastly, a fourth with three forms distinguished from each other by the final vowel -a-, or -i-, or -w- (e.g. Sanpc. a^ra agri, Lith. asztru-s). He then comes to the relations existing between the stems in -a- and those in -i- in nominal flexion. The stems in -a-, he observes, exhibit in one part of their cases stem-forms in -ai- ; this happens also with the stems in -i- : hence the two series -a-, -ai-, -aja- and -i-, -ai-, -aja- exhibit, as far as regards the stems in -i- of the second, parallel forms in -a-. Forced as we are to leave unnoticed many other of the author's opinions we will only note how he explains the genitive plural, in which he recognises a stem expanded by means of a suffix {-sa, -a-, -na-) which does not indicate the case, and to which is added the ending -m, and this ending alone in such a formation would represent the genitive plural. — In this monograph of G. Meyer we agree with Bezzenbergher" in gladly praising the independence of judgment and method, the aeuteness and the carefulness, the conciseness and the exactness of the exposition : but we cannot refrain from observing that the novel analyses of certain forms of flexion should call for stronger proofs and that some of them seem to us almost arbitrary. Nor with regard to the results of the investigations above noticed does the conclusion of the critic whom we have quoted differ much from our own.^ ' Gattingische gelehrte anzeigen, chichte der stammabstvfenden de- 1875, pp. 1104-20. ' clinationen (i. die nomina auf. -ar 2 We think worthy of notice also nnd -tar ; in the Studien *. gr. «. the studies of Brugman Zur ges- lat. gramm., ix. 361-406) and Zwr STEMS AND WORDS. 143 § 32. T. Benfey in two quite recent dissertations attempted to illustrate the origin of the vocative and of a form of the genitive singular in Aryan.^ The distinguished philologist regards it as almost certain that in the primitive and fundamental language of the Indo -Europeans the nominative^ in all three numbers and without change of soundsj was used also in the function of vocative. But, owing to the special nature of this second use, it seems 135 that the accent was gradually drawn towards the first syllable. This tonic distinction, after becoming a rule, especially in the singular, constituted, so to speak, the in- dividuality of the vocative and occasioned the disappearance of the final -s of the nominative singular masculine and feminine of certain stems when it was used in the sense of a vocative. But, in our opinion, it is not easy to see why this displacement of the acute accent could not have taken place in the simple stem as far as concerns the vocative singular, which might belong to a more ancient formation than the plural and the dual. — No less worthy of remark is the explanation which Benfey proposed of the Proto- Aryan genitive singular in -lans, las, -la. In these termi- nations he discovers various forms of a well known suffix of comparison. "As the genitive denotes properly 'be- longing' and is therefore substantially a possessive, so it is quite natural that it might also have been formed by means of an exponent which serves to form possessives. That these moreover avail themselves of affixes of the com- geschiohte der nominaUuffixe -as-, added to it : a phenomenon which -jas-und-vas- (in the Zeitschr. f. is occasioned by the fact that such vgl. sprachforsch. xxiv. 1-99). Un- suffixes are partly accented, partly der the name of ' stain mabstufung ' without accent, and which Brug- or • thematic gradation 'Brugman man refers to Proto- Aryan, understands the different conforma- ' Uher die enstehwng des indo- tion which the same stem, verbal or germanischen, voJcativs, Gottingen, Nominal, with or without a suffix, 1872 Tiber die indogermaniacHen assumes when the different suffixes endungen des genitiv sing, ians, ias, of the persons and the cases are ia, ib., 1874 Ui PART I. CHAP. III. § 23. parative is aknownfact" (cf. Gr. ^/xe-TejOo-?,Lat.wos-fer etc.). The -sla, -sja, which in the singular corresponds to -sdm (beside -«m) in the plural^ is according to Benfey^ an ending originally merely pronominal^ which then was introduced also into the declension of nouns and the initial s of which is pro- bably (like that of -«a;w) aremnant ofapronominal stem united in composition to others in which we find these endings. § 23. The various infinitives which present themselves in the languages of Aryan stock are generally considered by the philologists of the historico-comparative school as petrified forms of nominal fiexion. We have already de- scribed Ludwig''s doctrine with respect to such forms' and 136 our readers know what importance he has attributed to them in the development of the primitive Tndo-European language. Whatever may be the judgement which others think they are bound to pronounce on that doctrine, with regard to which we have noticed the views of several critics and expressed our own opinion, no one certainly will ■ be inclined to deny the great value of the collections of linguistic facts furnished by Ludwig concerning the form and usage of the Vedic infinitives, collections which have been justly praised by Delbriick and by Jolly. Previ- ously to Ludwig, Wilhelm had deserved well of Indo- European philology, as far as concerns the study of the infinitive, by a monograph^ which, considerably added to some years after', was welcomed with praises by J. Benfey'' and by other linguists'" for the carefulness the learning and the critical power which we admire in it : it has however been observed that the author has not made sufiieiently 1 See § 16, pp. 108-10. ■• Gottingische geUTirte ameigen, * De infinitivl vi et naiura, Eise- 1873, pp. 869-72 and 1751. nach, 1869. » See the criticisms of G. Meyer a De ivfiniiivi linguarum san- {Zeitschr.f. vgl.sprachforseh., xxii. scritae lactricae persicae graecae 334-40) and of Holzm an (WteAj-! oscaeumlricaelaiinaegoticae forma f. x'olkerpsychologie, etc. viii. 361 6t usu, Isenaci, 1872. seqq.). STEMS AND WORDS. ' 145 ■clear the slow process by which the infinitive attained its peculiar force. The accurate investigation into this development constitutes the chief value of Jolly^s work entitled Geschichte des infinitivs im indogermanisehem (MiincheUj 1873). The necessity for being brief and for not wandering into too minute disquisitions, which concern the individual languages rather than the Aryan stock in its unity, sternly forbids us from following Jolly in the description which he gives, in critical fashion, of the various doctrines held with regard to the infinitive by the old grammatical schools (pp. 12 — 48) and by the new historico-eomparative and psychological school ,(pp. 49 — 76) and does not permit us to follow him in the analysis to which he subjects the infinitives peculiar to each of the Indo-European languages (pp. 96 — 328). Still less i37 must we notice in a manner at all detailed the observations of the critics.' Of the very numerous notes therefore which we have collected in reading Jolly^s book we will give only those which contain the final results of his researches.^ Among the very various terminations of the infinitives in the Aryan languages there is only one, -dhjai in Indo- Iranian =-o-0at in (jreek, which Jolly thinks can with certainty be referred to the primitive and fundamental lan- guage of the Aryans. This however might have possessed, not indeed infinitives, but rather tendencies to form them in different ways : thus the author agrees with the opinion expressed by G. Curtius in the Chronologie, etc., though he at the same time admits that even in the Proto-Aryan certain cases of verbal substantives may have become petri- ' Seethereviewsby Schweizer- critique d'histoire et de litUrature, Sidler {Neue jahrlucher fur pU- 8th year, Ist semester, pp. 337-41) \ologie und paedagogiTc, vol. cix., and by Bezzenbergher (Qot- sect. 1, pp. 1-6), by Holzman tingische gelehrte anzeigen, 1874, (ZHtsehr. f. volkerpst/ckologie, etc., pp. 1067-75). viii. 361-5), by Bergaigpe {Eevue = Ibid., pp. 228-37. L 146 PA^T I. CHAP. ///. § 23. fied, because between the infinitives and the prepositions (some of which belong to the period of Indo-European lin- guistic unity) there exists an affinity which cannot be denied. It is further indisputable that in the period alluded to the language became prepared in two ways for the for- mation of the infinitives, that is, by developing the category of the noun of action (to which a large number of suffixes cdntributed) and by the verbal construction of these nominal forms. Jolly then proceeds to distinguish five stages in the development of the infinitive. 1st stage r the old con- struction with the accusative (which was perhaps originally the only casna obliqtms) is preserved in a series of abstract verbal nouns, and for this usage certain suffixes are selected^ whence arose a more and more close affinity between the 338 nominal forms mentioned and the corresponding verbal forms ; it might be called the supine-stage. 2nd stage : while the Slavo-Lithuanian languages, the Classical Sanscrit and the Persian do not go beyond the first stage, in the three principal languages of the European civilisation the infinitive penetrates, even in pre-historic times, into the purely verbal sphere of tense-formation. 3rd stage : besides indicating various tense-relations the infinitive proceeds also to mark with exactness the diflference between the active and the middle and passive, especially in Greek. 4th stage : while in the preceding stages the infinitive became more and more verbal, in this we see it re-approaching the noun, assuming here and there the function of subject, allowing itself to be preceded by the neuter article and being declined like a substantive ; a phenomenon which reaches completion only in epochs decidedly historical, appearing in Attic Greek, in New High German and especially in the Romance languages. 5th stage : change of the old infinitive of object into accessory final clauses, or even those of another kind ; this transformation, occasioned by the ever increasing need of clearness, was completed only in ages relatively STEMS AND WORDS. 147 very late and we find frequent examples of it especially in Modern Greek. — Thus Jolly himself summarized his ac- count of the infinitive in the Aryan linguistic stock. It is hardly necessary to say that his accurate and acute researches have met with frank and cheerful acceptance at the hands of criticism. It has however been observed' that he did not show that he had a clear conception of the infinitive, in re- cognising infinitives in certain forms which have beside them other cases of the same stems. But adverbs also, ' which are closely related to infinitives, are petrified forms of declension and neyertheless very often other cases correspond to them, nay the very cases that are used adverbially are yet found here and there with a different meaning. What i39 really impresses upon a stem-form the character of an ad- verb or that of an infinitive is not therefore, Holzman concludes, the lack of other cases of the same stem, but in fact the non-recognition of the bond existing between the one and the other. The want of other cases is only an index, by no means the cause of the formation of the infinitives. From this fundamental conception Jolly ought to have started, Holzman thinks, in his investigations into the changes of the Aryan infinitive.^ § 24. The order which we have followed leads us now to notice the most recent studies on the subject of compo- sition. The work to which we wish especially to draw the attention of our readers is Tobler's book Uber die wortzu- sammensetzung , etc., and it is this which we shall discuss in the present paragraph. In order to understand what was, in Aryan philology, the doctrine of composition before Tobler^s work it will be useful to recur, not to the Com- of Schleicher, but to Justi^s treatise JJber die See Holzman's criticism senschaftliche abhandlungen aus Q. quoted above. Curtius' gra/m/matischer geselU- ^ See also Jolly's monograph cTiaft zu Leipzig, Leipzig, 1874, pp- ZuT lehre vom particip {Sprachms- 71-94) 'us PART I. CHAP. ni. § 24. ziisammensetzung der nomina in den indogermantschen spraoTien (Gottingeiij 1861). It is divided into two parts : in the first is discussed the form, in the second the meaning of compound nouns. With respect to the form Justi distin- guishes three stages in the development of composition : 1st, the simple primitive juxta-position of word with word; 2nd, complete composition ; 3rd, its decadence.' When fur- ther the meaning of the compounds is considered, we find 840 them divided into two great classes. The first or inferior comprehends all the examples of coordinant and of suhordi- nant composition, which is subdivided into determinative composition^ and composition determined by relation of ease.' The second or superior is either relative^ or adverbial.^ We now come to the work of Tobler above noticed." It is divided into three sections, in the first of which the 1 Of the first stage we find examples for the most part only in Old Indian and Bactrian. A cha- racteristic of the second, peculiar to the Aryan stock, is the purely thematic form of the determinant member, while the determinate ap- pears furnished with the forms of fiexion (e.g. KXuTfJ-^uai/Tts). A mark of the third grade is the appearance of the composition - vowel which unites the two members and which Jnsti discusses as well as the stems and the accent in compounds. ^ The determination may be ap- positional or numeral ; the former is further distinguished as comparative and purely appositional. 3 This " is found when the first or determinant member must be con- ceived as dependent on the second member in a case " (p. 102). * It is "a kind of composition ..which compresses a whole relative sentence into a single word, which, however, just as much as the whole sentence, has a relative meaning :" he adduces, as an instance of this, the sentence 4(f>dv7j 'iii>s ^oSoSaKTu- Aos=re. *H. ^Tivi ol da.KTvh.oi Sum ^6Sa €ia-tv (p. 117). While in the compounds previously noticed the subject was external to them, this kind of composition comprises the subject in itself (p. 118). 5 " Here too a whole small sen- tence, which may always be ex- plained by an 'is,' is reduced by composition to a single word, the second member of which is always a noun, the first generally an inde- clinable one " (p. 126). * ZTier die wortzusammensetzimg nehst einem anliange uber die ver- stdrJcenden zusammensetzungetit ^« heitrag zur pMlosophischen und vergleichenden sprachwissenscJiaft, Berlin, 1868. — See Steinthal's critical observations in the Zeitschr. f. volTcerpsychologie, etc., vi. 264-80. STEMS AND WOUDS. 149 author proceeds to distinguish composition from other formations apparently similar. And, first of all, he notes the difference which separates composition from flexion and from derivation : a difference consisting, he thinks, in this — that flexion and derivation (considered as a preparation for the flexional form) give origin to words, or to linguistic elements which have come to be parts of speech only by means of these, while composition presupposes them already existing as such ;^ in the products of conjugation and of declension the elements ui are much more closely united together than in compounds ; moreover the afiixes of flexion (like those of derivation) are as a rule attached to the final part of the stems, in composition on the contrary the determinative word-form precedes (in the form of a stem) the fundamental ; lastly, in composition the free and conscious activity of the speaker appears much more than in flexion and in stem- formation. Hence he goes on to the differences existing! between composition and reduplication, union of roots, incorporation, syntactical construction. The second section investigates " the internal differences of composition :" the latter is divided by To bier into legitimate and spurious;'^ ' But, as Steinthal appropri- But in order not to be forced to ately remarks, just as the author such conclusion it is advisable, con- does not by any means deny that the tinues Steinthal, to lay down elements also of flexion and of deriva- more clearly the fundamental ideas, tion may have been once, at any " A form of flexion consists of a rate partly, word-forms endowed stem and a suffix : a compound is with an independent existence, with a made up of two stems, to which re- meaning of their own, so the difler- garded as a unity a suffix is added." ence indicated would not be of great 2 Tobler is wrong, according importance: the products of stem- to Steinthal, in refusing to re- formation and of flexion might be gard the oo-ordinant or copulative only very ancient and petrified forms (dvandva) as legitimate composition, of a primitive composition, of which Who has proved that there is no what we generally call by that name real composition except where there would be only a continuation. This is a relation of subordination of one is certainly not Tobler's opinion, member to the other ? It is not the 150 PARTI. CHAP. III. §24 legitimate into proper and improper. But it is not easy to conceive with exactness and to mark in a few words the differences alluded to.' We are therefore constrained to pass them over in silence as also some brief disquisitions, of very slight importance for our object, with which the second section concludes. The subject of the third is composition logically and psychologically considered. Then come two fresh divisions, of which it will suffice to describe summarily the first. I. Relation of reciprocal complement, co-ordination : 1st, the two members are generally species of one and the same genus, and therefore there is between them 142 an antithesis, but they appear united together exceptionally to form a new unity (ex. dvSpoyvvrj'}) ; 2ndly, the two members are in a certain manner varieties of the same species, hence the one is not opposed to the other (a rarer case, ex. KoXoKajado';) . II. Relation of unilateral comple- ment, subordination : 1st, the second member stands to the first actually as genus to species (ex. KiTpofMJXxiv) ; 2ndly, the second member is considered as a genus relatively to the whole ; the grammatical relation of the second to the first element of these compounds may be (a) attributive (in a narrow sense), (b) a relation of case. We will pass over as foreign to the nature of this book the three psychological forms of composition which are noted by Tobler and to which he attempted, without complete success (as he himself was ready to admit), to make the logical and the grammatical forms of it correspond. Whence it appears, observes Steinthal, that in spite of the acute use of copulative composition, but means approves the method followed only the Indian abuse of it that we by Tobler in this subject : a better must condemn. It is the most sen- plan would have been to trace the sible, poetic, forcible form of com- type of composition from the study position : it was afterwards lost or of linguistic tendencies j then by corrupted by the increase of the comparing with it the various com- power of abstraction. pounds, to determine the degree of • The critic quoted above by no meaning peculiar to each of them. STEMS AND WORDS. 151 investigations and speculations of the author, the first attempt to found on a psychological basis the doctrine of composition has not met with great success.' § 25. Let us consider now whether in stem-formation, in inflexion, in composition, traces have been discovered of a common origin of the Aryan and the Semitic lan- guages. F. Miiller, in his work on this subject before quoted,^ attempts to demonstrate that both the constitu- tion of the word, and the various categories of it, and the structure of the compounds separate the Aryan stock from the Semitic. The former in word-formation employs only suffixes : the latter suffixes and prefixes. Aryan possesses m three categories of gender, Semitic only two, but the latter marks it also in the pronoun of the 2nd person sxA in the 2nd and 3rd person of the verbs ; in the oldest language of the Aryans we find eight cases, in that of the Semites not more than three ; with the full development of the Indo-European verb is contrasted, under the head of the expression of the tenses, the Semitic conjugation restricted to two forms only with a temporal force, that is, to signifying by suffixes completed action, by prefixes action in process of completion. Nor again can it be said that the variety of composition belongs to Semitic which we have seen in Aryan ; moreover in the former the determinant always follows the determinate, while on the contrary it ' We will &ere further mention in Vher die formelle unterscJieidung a note three works, which, although der redetheile im griecMschen und they do not concern the entire Ar- lateinisehen mit heruclcsiclitigung jan stock, may nevertheless be of no der nominaleomposiia, Leipzig, slight help to anyone who studies 1874. — Clemm, Die neusten for- eomposition in the Indo-European schungen, auf dem geliet der griech- linguistic unity. ischen composita (from the 7th Meunier, Lee compos4s syntac- volume of the Studien edited by G. iigues en grec, en, latin, en franfais Curtius). et misidiairement en zend et en ^ Indogermanisch und semitisch, indien, Paris, 1872. — Schroder, etc., pp. 11-5. 152 PART I. CHAP. III. % 25. constantly precedes in Indo-European compounds; lastly Semitic can attach the object, if it be a pronoun, immediately to a verbal form. — As F. Miiller brought out clearly the differences existing between the two stocks, so Ascoli both in the Letters to A. Kuhn and to F. Bopp' and in the Studj drio-semiticP with that vastness of knowledge and that acuteness of intellect which we admire in him, attempted to trace and set forth what appeared to him indications of primitive afEnity between the languages of the Aryans and the Semites. The grave differences observed by other philologists are not sufficient to shake his faith. He admits that symbolism is much more frequent in Semitic than in Aryan flexion, but does not regard it as such a characteristic of the former as to be able to separate it absolutely from the latter. He admits that the Aryan word is formed only by suffixes, while the Semitic word exhibits both suffixes and prefixes ; but he does not think himself bound, by reason of this 141 difference, to regard as impossible the affinity of the two stocks ; " the division must have taken place before the true verb had come into existence from the close union of the pronoun with the nomen agentis." Our readers are already familiar with Ascoli's comparison of the Aryan present-stems with the supposed Semitic radicals :' the one and the other we should regard as roots with suffixes of the agent, 'nomina agentis,'' from which nouns, by means of intimate connexion with pronouns, both the Aryan and the Semitic conjugation have derived their origin, but "the one independently of the other.^^ Ascoli also considers as common to the two great families of languages two suffixes of comparison and several of nominal flexion. These comparisons of Ascoli were criticised Politecnico, xxi. 190-216; xxii. hardo, etc., cl. di lettere, etc., a. 121-51. 1-12, 13-36. ' Memorie del B, Istituto lorn- ' See above, pp. 63-4. STEMS AND WORDS. 153 perhaps too severely by P. Delitzsch in the book of which we have spoken.' — After the works of Ascoli on the difficult subject which we are treating, we are very sorry to be obliged to mention a book by Raabe/ far inferior to them in scientific worth. The author teaches us that by his book "an affinity is proved between the two stocks of language; but an affinity like this has not yet been discovered between languages of any sort.'' In Raabe's work one might look in vain for a phonology and a syntax ; among the Aryan languages for the most part advantage is taken only of the languages of the Indo-Iranic section ; to numerous and considerable lacunae (especially in the theory of conjugation) are added incredible caprices : this work therefore by no means furnishes us with a demonstration, not even an exclusively morpho- logical one, of the supposed affinity between Aryan and Semitic. We think it therefore useless to speak of it i*5 at greater length, nor is it more worth our while to pay attention to the few and unimportant considerations of Schultze on words and principally on grammatical gender in the little work mentioned above, Indo-ger- manisch, semitiscJi und hamitiscJi. We may therefore proceed at once from morphological investigations to the syntactical researches made during the last decade by several philologists of the historico-comparative school with a success for the most part not unworthy of the noble laboriousness which they have devoted to it. § 26. In the introduction to a book which we shall have to notice again' Jolly sketched briefly the history of these studies, mentioning the researches of J. Grimm, of Micklosich, of Diez on the syntax of the Teutonic, I See above, pp. 39, 43-4, 65-7. sfrachm, etc., Leipzig, 1874. S Gemeinschaftliche grammatih ' Mn hapitel vergleichender syn- der arischen nnd der semitiscAen tax, etc., Miinchen, 1872, pp. 3 sqq. 154 PARI I. CHAP. III. h 26. the Slavonicj and the Neo-Latin families j describing how, among the foremost^ L. Lange as early as 185^ demon- strated the possibility and the necessity of a historico- comparative investigation of the syntactical phenomena over the whole field of the Aryan languages ; setting forth the reasons why, with the exception of two short monographs of Schweizer-Sidler on the ablative and the instrumental in the Rig-veda (1846-7) and the remarks of Regnier also on Vedic syntax (1855), no work of any importance on the subject of which we are speaking has been published until the last decade ; noticing, lastly, the very remarkable treatises of Delbriick and Windisch aud other philo- logists. And, as far as concerns especially the meaning of the forms of nominal flexion, Hiibschmann, in the first part of a very recent work of his^ of which we shall speak soon, traced in a detailed and critical exposition the develop- 1*6 ment of the syntax of the cases, studying it first in the ancient grammar (which in his opinion begins with the investigations of the Greek philosophers on the subject of language and reaches to G. Hermann^ inclusive), afterwards in the school of philology to which the powerful genius of Wilhelm von Humboldt gave impulse, lastly in the historico-comparative science of language. The opinions manifested by distinguished investigators on the subject of the original meaning of the cases have also been set forth and examined by Holzweissig' in a very recent work. But the zealous reader may learn from the two works above mentioned the history of the long intel- 1 Tmt casuslehre, Miioehen, 1875 ^ Wah/rheit nnd, irrthum der lo- (ersier theil : zur gescMchie der ca- calistischen casustTieorie . ein bei- suslehre, pp. ]-146). trag zur rationellen hehandlung der 2 With regard to this great griechischen und lateinischen caszts- scholar see also Frennd, Triennium syntax aufgrund der sicheren ergeb- philologicum oder grundzilge der nisse der vergleichenden sprach- pMlologischen wissemohaften, etc., forscTmng, Leipzig, 1877, pp. 1-24. i., Leipzig, 1874, pp. 80-81. STEMS AND WORDS. 155 lectual struggles which, in this department of the science also, the arduous acquisition of certain truths cost. It is our business, spurred on as we are by the length of the subject, to describe at once, with some critical notes, the most important results of the syntactical researches of which we have made mention, starting from the investi- gations into the meaning of the cases. We first come to two treatises by Delbriick concern- ing the use of four forms of declension.^ It is well at the outset to see what method the author has observed. In vain, he teaches us, has philosophy with its speculations, in vain has morphology with its analyses attempted to discover the primitive force of the eases : the only means is the historical examination of their usage. The "fundamental idea (Grund- begrifF) " of a case is the most ancient signification of it which it is possible to trace, whether it consists of only one or of more ideas : a fundamental idea, not perhaps absolutely, but certainly relatively to us, so that beyond it we cannot push our investigation. If therefore the families of the 147 Aryan languages had faithfully preserved the eight cases which Delbriick, with the school of philologists to which he belongs, regards as Proto- Aryan, the comparison of the senses of each of them in the individual families alluded to would be quite possible, and by it would be discovered the ' Ahlativ localis instrumentalis two works see tlie opinions of im altindischen lateinischen griech- Schweizer-Sidler (in the Zeit- isehen and deutschen, ein leitrag zur schrift quoted above, xvii. 301-2), of vergleichenden syntax der indoger- Thurot, (in the Mevue critique manischen sprachen,, Berlin, 1867. d'liistoire et de litterature, 4th year, — ijier den indogermanisehen, spe- 1st semester, pp. 114-16) and of eiell den vedischen dativ (in the Leskien in the Gott. gel. am. Zeitschr.f.vgl.sprachfors6h.,yiv\\\. 1868, pp. 475-80). On the 1st. ed. 81-106 ; 2nd and improved edition of the latter a somewhat unfavour- of the monograph De usu dativi in, able opinion has been expressed by carminibusS,igvedae,lia\\s,\%&l). Goldschmidt in the Oott. gel. With regard to the first of these anz., 1868, pp. 600-9. 156 PART I. CHAE. III. § 25. original sense of each of them.^ But of these eight cases some have been lost in European Aryan. Nevertheless just- as the comparison of several cases still existing in it with the corresponding Vedic eases shows the meaning of them to be common to the former and the latter, so we may believe that the forms of nominal flexion not preserved by the Aryan languages of Europe had a meaning not very different from that with which we see them endowed in the above mentioned most ancient records of the Indian family. Such is the fundamental principle of Delbriiek's method: it remains now to note the most important results of his researches. In comparing Slavonic with Teutonic, Greek with Latin, it is perceived that the loss of certain cases took place in the individual languages, that is after their separa- tion. What is, according to the author, the cause of such 148 loss ? The more vividly ever increasing culture caused to be felt the want of representing with exactness the relations existing between the ideas, the less adequate did the cases appear for this end : thus recourse was had, with constantly greater frequency, to prepositions. This usage diminished 1 This position, which will appear istic would not so much indicate to ouv readers bo consistent with the originality in this as affinity in the method followed by the linguistic- modes of conception peculiar to the students of our time in all the other speakers of those various languages, branches of the historico-compara- But one who admits the Proto-Aryan tive grammar of the Aryan lau- origin of a form of flexion and sees guages, is assailed by Thurot, who it used with a meaning always thinks that syntax cannot make use fundamentally identical in all or of such comparisons as they are nearly all the families of the Indo- made use of by the theories of the European languages, has not, we sounds, the roots, the stem suffixes think, any right to deny the origi- and flexional sufiSxes, because the nality of this meaning, especially phenomena of grammatical construe- when it is such that it can be easily tion might have taken place in the reconciled with the intellectual con- individual languages, or separately, dition of the pre-historio Aryans, in such a way that the not unf re- according to the conception of it to quent coincidence of two or more which we are led by the most certain languages in a syntactical character- results of linguistic studies. STEMS AND WORDS. 157 gradually the value of the declensional endings and inter- changes could readily take place among them : lastly, per- haps, after some period of uncertainty, language selected, out of two or more, one form which also took upon itself all the meanings of the others.' Thus the surviving cases made up for those which were lost : such is the position which Delbriick proposes to maintain, by showing what in Greek, in Latin, in Teutonic are the syntactical successors of the primitive cases which have not been preserved, in each of the meanings which belonged to such cases. But since our task is only to discuss the linguistic facts which are spread over the whole Aryan stock, we shall not follow Delbriick in his investigations of the cases which succeeded those that were lost in the languages mentioned, and we shall rest content with noticing the primitive meanings which he attributes to the locative, the instrumental, the dative. The fundamental conception expressed by the ablative is the idea of ' separation : ' to this are to be referred all the senses in which Delbriick points out this case to have been used ; by it is explained also the ablative of comparison. We have a locative properly so called and a locative of object : the first of them has a varying meaning of place and time, and is used also like the abso- lute cases in the classical languages f the second in many constructions is hardly distinguished from the first. The 149 ' being together ' is the conception originally expressed by 1 To the influence exercised by also in the locative case. To the the prepositions must be added, primitive locative and instrumen- Hiibschmann observes (Zwr ca- tal absolute syntactically corre- snslehre, etc., pp. 85-6), the action spends the ablative absolute of the of phonetic laws and of the accent. Latins : the original locative abso- ' The characteristic of this con- lute is represented in Greek by the struetion consists, according to the genitive absolute, which is not to be author, only in that the locative of compared with the Indian, this lat- a noun-substantive has added to it ter being posterior to the Vedic age. also a participle, itself naturally 158 PART I. CHAP. III. § 26. the instrumental, which Delbriick observes to denote sometimes concomitance^ sometimes the means by which (that is to say ' together with which ') an action is done (so- ciative or comitative instrumental, instrumental of means).' These cases appear frequently used with prepositions ; but, observes the learned philologistj it would be an old error to think the preposition governed the case, or the case the preposition ; it is rather an adverb which is added to the case to indicate with greater exactness in which of the various meanings belonging to it we are to understand it. Lastly^ the original function of the dative was to denote the 'tendency to something'' (thus agreeing with the locative of object) : from this fundamental meaning of ' inclination/ or ' motion ' of the body or of the mind spring all the other senses which belong to this case and the datives of advantage and disadvantage, to use the ordinary phrase- ology, are explained, as well as the dative of possession and also that of purpose. To any one who further carefully considers the use of this ease, especially in Sanscrit and in Latin, it will appear probable that in Proto- Aryan it was not joined with prepositions. Instead of proceeding from the form of the individual eases to the investigation of their uses, as his predecessors had done, Autenrieth in the dissertation entitled Ter- minus in quern, syntaxis comparativae particula (Erlangse, MDCOCLXViii) started from this terminus, inquiring what cases were used to denote it.^ A severe judgement has been ' Bat, observes Schweizer- the genitive, and how could the Sidler, if, for example, the abla- suffix -IM have been common to the tive had had, even in the most an. dative, the ablative and the instru- cient epoch, the clearly determined mental? Delb ruck's explanation function of expressing the idea of therefore does not seem to the illus- ,' separation,' how could its form, trious critic at all sufficient to solve even in the most archaic Sanscrit, the problem of the primitive meaning have been confounded in the singu- of the cases. lar for the most part with that of = "Ut alii congesta exemplorum STEMS AND WORDS. 159 passed by Holzman and Hiibschmann' on the method iso followed byAutenrieth: we do not think the importance and the attractiveness of his researches great enough to make us recommend them to the attention of our readers, although certainly this treatise is not wanting in the merit of scholar- ship nor in that of accuracy. — A. remark will suffice also with regard to the monograph of ^itcke De genetivi'' in lingua sanscrita imprimis vedica nsu (Berolini, 1869), in which the author investigated the various uses of this case in the Vedic Indian, paying attention also to the Epic and the cognate languages. An observation of Benfey on this work' appears to us noteworthy. The eminent Indian scholar and philologist cannot agree with Siecke in accept- ing the theory of Max Miiller on the genitive, which the latter regards as an adjective of relation in a thematic form copia varium singulorum casnum usum investigare student, ita opinor licet e contrario quaerere, qui casus adhibeantur ad exprimendam cer- tain aliquam notionum seriem. In his autem termini quos vocant, lo- cales et temporales, imprimis digni sunt quibus indagandis operam de- mus, et initium equidem faciam ab eo quern terminum in quern vocare consueverunt grammatici " (p. 5). 1 Holzman observes (in the Zdtsckr. f. vSlTcerpsychologie, etc., vi. 488-92) that by the method al- luded to is sought " the expression of ideas and forms of ideas whose existence itself has not yet been proved." For the conception of the 'terminus iuquem' may belong to Latin, Greek, Teutonic, without being Proto- Aryan. Moreover such method needs the knowledge, which we possess only in part, of the primi- tive meaning of the cases and of various senses which were developed from it. — Hubschmann {Zur caanslehre, p. 72) speaks of Ante n- rieth as follows: "He finds his 'terminus' expressed by all the casus oiliqui of Sanscrit, Zend, Old Persian, Greek and Latin, and rightly, to tell the truth, if the German translation is the standard by which we should judge the gram- matical forms of those languages. And as that is not the case, so Autenrieth's work, otherwise praiseworthy for its diligence, is a failure. 2 The 'yevixii ■mums' of the Stoics should probably be rendered in Latin by 'casus generalis' (a term which we find in Charisius), or the 'case which expresses the genus ' as opposed to. the species, etc. See Hiibschmann, ^urcfZfWfMre, pp. 12-4. 3 Gottingische gehhrte ameigen, 1869, pp. 1255-63. ■160 PART I. CHAP. III. § 26. 161 without expression of gender.' The fundamental conception common to the genitive and the ablative is, according to Ben fey, the idea of ' proceeding from/ with this difference that the ablative marks at the same time the ' detachment ' from the point of departure, while on the other band the genitive denotes the ' remaining united ' with it. Jolly's Geschichte des infinitivs im indogermanischen, a work of which we have spoken already in the foregoing morphological considerations, contains also a syntactical portion in which there is a discussion of the accusative and of the dative with the infinitive (pp. 243 — 70). After having set forth and examined the various hypotheses by which it was attempted to throw light upon those constructions in which the accusative is generally regarded as subject of the infinitive,^ Jolly lays stress upon the last one mentioned by him, according to which the accusative in the construc- tion in question should be understood as object of the verb of the principal clause.' But these hypotheses belong for the most part to a period of philological investigation anterior to that of which we are briefly narrating the his- tory and therefore it is not our business to occupy ourselves with them further. We ought rather to turn our attention to the use of the dative with the infinitive (e. g. in the Vedic sentence quoted by Jolly j)ibd vrtrdja hdntave [bibe Vritrae occidendo:]) a use in which Ludwig* was the first to notice the agreement of the Indo-Iranic with the Slavonic ' Comp. the genitive S^oio (from marks on the last two pages. *87;/io(rio) with the adjective-stem ^ yf^. gn^ tjjjg construction fre- hi\ii.o-aio-. See Hubs chmann, Zar quent especially in Greek posterior casuslehre, p. 104 sqq. to the Homeric age and in Latin : '■' On this subject see also Al- we find scanty traces of it in Sanscrit V re cht's monograph J)e ace««aopd, not vice versa, is clear not only from logical considerations, but from the historical study of the use of the pronouns. The anaphoric pronouns may be subdivided into demonstrative and simple : in the former there is still the Set^t? which is wanting altogether in the latter, whose task consists in 156 nothing else than representing a previously mentioned noun. From these general conceptions we pass on to the relative pronominal stem ja- to wh^ch we must pay par- ticular attention. Windisch brings out first of all two facts : 1st, that this stem ja- is in every case an ex- panded form of the pronominal root i ; 2nd, that it does not always appear with the meaning of a relative pronoun, but often here and there as a simple pronoun of the third 166 PART I. CHAP. III. § 27. person. From the course of the investigation the purely- demonstrative force originally belonging to the pronominal root i becomes more and more evident, and it is seen how, even before the fundamental Aryan became transformed into several languages, it was weakened to a simply ana- phorieal function. Nor does this conclusion result only from the comparison of the Aryan languages with one another, but only because, immediately after the division of Proto- Aryan, the stem ja- retained the force of a simply anaphorical pronoun even in those languages whose records no longer exhibit it except as a conjunctive relative (" satz- verbindendes relativum"). The development of the latter from the former is described by Windiseh as follows: " first the use of an ordinary anaphoric pronoun was limited to the case in which the two sentences belonging to it were very closely connected together in their ideas; secondly, the relative pronoun was placed first ; thirdly, there came about a change of the usual order also of the other words in the relative sentence." At first the bond between this and the principal sentence was close and necessary, so that the one could hardly be understood without the other : ' afterwards were developed relative sentences less tenaciously connected with the principal sentences to which they referred.' 1 To the results ofWindisch's among other examples, the German researches on the origin of the re- pronouns welch, wer, was, heside der, lative pronoun and the relative sen- das, the English which, who, what he- tence in the Aryan languages it side iAai, and remarks that the Greek ■will be useful to add some remarks relatives Sirov, 6iro7os, etc., are com- drawn from the above quoted re- pounds, the second member of which view of To bier. He observes is an interrogative. "This use," that the relative pronoun might he writes, "naturally can be ex- come not only from the demonstra- plained only by the primitive para- tive, but also from the interroga- tactic construction," that is to say, tive. This appears from Latin, by assuming with Aufrecht (Zei<- Teutonic, and also from Greek, Zend, schr. f. vgl. sprachforsch., i. 284) and Lithu-Slavonic : Toblerquotes, "that the relative sentence took its STEMS AND WORDS. 167 We should like now to be able to discuss, in a manner 157 befitting the importance of the subject and the worth of the book, the Syntahtische Forsehungen of Delbriick and Windisch/ a work which certainly deserves the foremost honours with respect to comparative syntax, in spite of the censures which have been passed upon it by some criticSj'^Sd which we are going to notice. But the limits which partly the authors and partly the subject-matter itself have marked out for these learned and profound researches,^ and the impossibility of following them without going into too minute considerations prevent us from treating of this work with that fullness which we could wish. We cannot, therefore, do more than set forth the fundamental ideas of the two learned philologists on the original force of the conjunctive and optative, and on the origin of secondary or dependent sentences. And these conclusions we find expressed as follows in the review cited of Thurot : ''the \m Indo-Germanic languages had a period in which men eon- versed only by means of independent sentences: co-ordination is anterior to subordination. Moreover affirmative sentences ' are anterior to negative and to interrogative sentences . . . origin from contraction of an inter- Greek exhibit a conjunctive and an rogative sentence with the relative optative quite distinct from each reply." other : to these languages therefore ' I. Der gehrauch des conjunC' must be limited the investigation tioi und optativs im Sanscrit tind into the use of these two moods. grieohischen, Halle, 1871 : the chief For personal reasons moreover the merit of this important work belongs two authors excluded from their re- to Delbriick. — See the criticisms searches Zend, a lacuna which was of Thurot and Bergaigne in the speedily filled by Jolly with his 2Jej)«e critique d'histoire et de lit- valuable monograph i?iffl i;a^JieZ cer- ttrature, 6th year, 2nd semester, pp. gleichender syntax : der conjunctiv 27-31, 129-34, and that of Holz- und optativ und die neiensatze im man in the Zeitschr. f. vSlkerpsy- zend und altpersisehen in vergleich chologie, etc., viii. 40-57. mit dem Sanskrit und griecUschen, 2 Among the Aryan languages, Munchen, 1872.— See on this Holz- Delbruc.k andWindisch (pp. 6-7) man's remarks in the Zeitschr. f. observe, only Sanscrit, Zend and vSlJcerpsyehologie, etc., viii. 57-62. 168 PART I. CHAP. III. § 27. Lastly, the primitive signification of the moods should be seen more clearly in the independent affirmative sentences in which the verb is in the first person singular." These positions the authors think are strengthened by the obser- vation "that the various uses of the conjunctive and the optative cannot be reduced to unity of meaning except by admitting that the primitive signification is ' willing ■* for the conjunctive, ' desire ' for the optative, a signification which is found pure only in independent affirmative proposi- tions in which the verb is the first person singular.^'' The J " Les hypotheses de M. D./' writes Thurot (article cited), "me semble contestables ^ deux points de Tue, d'abord il n'a pas tenu assez de compte des modifications que I'asso- ciation des mots apporte 4 leur sig- nifications ; ensuite il a confondu Tant^rioritelogiqae avec I'antSriorite ehronologique Si tons les mots conservaient leur sens propre et primitif dans toutes les conjitrue- tions, il n'y aurait aucun mojen de se faire entendre. On en pent dire autant des formes grammaticales. Si I'optatif signlfie proprement le voBU (ce que me parait fort dou- teux), il perd cette signification et il la perd an point qu'elle ne pent pas meme se presenter 4 I'esprit, quand il est employe au style indirect. II en est de meme du snbjonctif; quand il signifie ce que M. D. appelle I'attente {erwartnng), il ne signifie plus la volonte, et il est impossible de lui maintenir ce dernier sens." But even supposing to be very ancient these changes of the primitive mean- ing of the forms and' the words which have taken place in the various con- Hructions, we are nevertheless evi- dently forced to admit that in an epoch still more ancient forms and words were used in speech in their original meaning and to examine in what manner and owing to what causes such meaning underwent a change. " Je ne saurais admettre davantage," continues the French critic, " qu'on alt parle longtemps par propositions co-ordonuecs uuique- mentj avant d'employer des propo- sitions subordonnees. Quand la sub- ordination existe dans la pensee, et en beaucoup de cas elle ne pent pas ne pas exister, par exemple pour les circonstances de temps et de lieu relativeraent a I'action qu'elles ac- compagnent, les relatifs adverbiaux qui expriment le temps et le lieu ne peuvent pas ne pas exprimer la subordination de la proposition qu'ils precedent a la proposition principale." And in Bergaigne's article we read that Delbriick " denature compl^tement le sens de certaines propositions subordonnees par le parti-pris de les traduire comme de simples co-ordon^es. Dans les propositions dont la subordina- tion est reellement necessaire, la de- pendance a dii fetre sentie nou- seulement dSis I'epoque vedique ^ STEMS AND WORDS. 169 criticisms of ptilologists on the value of such assertions may- be various, but there can be no doubt of the diligence, the iss scholarship, the acumen of the two distinguished investi- gators, and of the importance of the problems to the solution of which by means of their researches they have so laquelle M. D. emprunte ses exem- ples, mais d^s le premier jour oil le langage s'est hasard^ il rendre une seule pens^e complexe au moyen de deux propositions." And here, perhaps, the two French philologists did not pay enough attention to the condition of the intellectual life in that epoch so remote and so dif- ferent from our own that we can hardly figure to ourselves the slow development of the thought and the word in their reciprocal relations. Not all that appears to us original is certainly such, not merely with re- ference to ourselves, but also in reality : in many cases it might be nothing else than the result of a long evolution. — Tburot adds to the above quoted observations the following: Ist, that the optative, morphologically considered, is shown to be akin to the historic tenses ; 2nd, that the meaning of ' desire ' •might perhaps be better derived from that of ' past' than vice versa ; 3rd, that in fine the original force of the conjunctive and the optative might perhaps be sought with more success in the dependent than in the independent sentences. Bergaigne attempted to prove that the forms of the conjunctive and the optative could not, originally and in them- selves, denote either ' willing ' or ' desire.' " M. D. dit lui lui-meme" (p. 17) : " Un des points de vue les plus importants et qu'on ne doit pas perdre de vue, c'est que le mouve- ment subjectif de la volenti ou da d^sir demeure toujours chez lameme peraonne, et ne pent pas passer a une seconde ou ^ une troisifeme. C'est par la que les desideratifs par exemple se distinguent des modes pour le sens." Mais si t/icpoijui sig- nifiait par lui-meme 'je desire porter,'