But, although the works which we have just examined are not such that their object can be said at least in part to be attained, nevertheless, for one who attentively considers the causes which will make them barren of results, they may be of no slight utility, in that by their fate the necessity of a severely scientific method seems to be more and more conspicuously proved. And the belief in such necessity will also be enhanced by some of the principal works which have been published in the last few years on the morphology of the Aryan dialects, and which we shall next discuss.

the Chinese lún and the Greek Xap-Bar (sic, p. 40) and the following words: "il est constant que

dans plusieurs mots aryens la lettre 7 primitive s'est indurcie en r" (p. 1-10).

(74)

69

CHAPTER III.

Stems and Words.

§14. The numerous and intimate relations which exist between stems and words, between words and their combinations, and still more the difficulty, and we would almost say the impossibility, of separating from each other, in some of the books which we are going to examine, the different portions in which these subjects are discussed, induce us to unite in this chapter what, according to an arrangement more strict, but too hard to follow, ought to form the matter of three chapters-we mean the results of the most recent studies on the form and meaning of stems and words considered independently of each other, and in their reciprocal relations in the unity of the compound and of the sentence, together with some notice of the relations existing between sentence and sentence. But, as far as may be possible, we shall, in treating all the subjects mentioned, make our critical exposition start with the form and origin of the constituent suffixes of stems and words, and then proceed to the synthetic use of them: the first part will be next subdivided so that the reader, to begin with, may acquire a knowledge of the morphological works of greater dimensions and of more audacious novelty, and these may be followed by the writings whose subject is more special. The number of the works of various magnitude which we shall have to consider, and the nature of some of them, impose upon us the obligation to be brief, to limit our discussion to fundamental conceptions, to abstain altogether from a too great par89 ticularity of treatment; descending to minute disquisitions.

only when it appears absolutely necessary in order to make the system or the method of an author understood. We shall be forced to feel only too well this deplorable necessity frequently even in the remarks which we proceed to make on the first work of which it is our business to speak in this chapter, that of Wilhelm Scherer.'

In this book, which, in spite of imperfections neither few nor slight, attracts to itself for its learning and originality the attention, the respect and often also the sympathy of the philologist, we must distinguish two parts which in the work and the intention of the author are fused together the part which relates to the special investigation of Teutonic phonology and morphology, and that which is made up of the general researches into the sources of Aryan flexion; researches to which Scherer felt himself committed by the nature of his genius and by his resolution to penetrate as far as might be possible into the inner constitution of the Germanic languages, and to discover their laws and causes, tracing in the formation of the language the formation of Teutonic nationality.3 It is clear that with the first part we need not concern ourselves in this book, in which account is taken only of studies having for their object the entire Aryan stock: of the second we shall speak in the present paragraph, confining ourselves principally to the already cited discussion of the personal pronoun, following the order of the author, and availing ourselves of the critical remarks which, independently, were published by A. Kuhn and Steinthal on 70 Scherer's work."

1 Zur geschichte der deutschen sprache, Berlin, 1868.

2 See especially pp. 213-361 (Das personalpronomem).

3 Read carefully the epistle dedicatory to Karl Müllenhoff (iii.-xiv).

4 Zeitschr. f. vgl. sprachforsch.,

xviii. 321-411.

5 Zeitschrift für völkerpsycholo gie, etc., v. 464-90.

6 See also the Revue critique d'histoire et de littérature, 3rd year, 2nd semester, pp. 354-7.

in

To investigate the various function, the various lot of the personal pronouns in the formation of the active and the middle, Scherer began with the following phonetic law: "the atonic a of monosyllables already independent, which were fused with their verbal or nominal stem in the unity of the word, is frequently lost without leaving a trace of itself." Hence it happened, according to the author, that, e. g., in Greek the final vowel a of the pronominal stem za vanished in the 2nd pers. singular of the active aorist e-0ŋ-s (from the Proto-Aryan á-dhā-sa, in which this a is not accented), while it was preserved, after passing to 0, the corresponding middle form, e-0e-o (from✶ e-0e-σo=orig. a-dha-sú, in which the a whence it is derived is accented). Another example adduced by Scherer is the Sanscrit drikšé = Proto-Aryan_dvik-sú-i2 = drik-trá [hates-self thee, thou art hated], an old passive which became confounded with the middle and combined in itself the one and the other meaning. Hence in the passive the expression of the person differed from that of the active only in the accent: the atonic pronoun denoted the active, the accented the passive; the final a of the first was lost, that of the second was preserved by the phonetic law above mentioned."-In

1 Kuhn rightly remarks that this law is not proved by Scherer.

Of the final i of this and the analogous forms notice will be taken shortly.

To this doctrine of Scherer the following objections have been raised by A. Kuhn: 1st, between E-On-s and -0ɛ-0, á-dhā-sa and a-dha-sá there is a difference, which ought to be explained, in the quantity of the radical vowel; 2nd, the Greek final o does not represent a pure original final a; 3rd, thesa in the passive form ought to be accusative, and this is in Greek σε, not σo; 4th, how could the

0

=orig. a have remained even after the accent had passed to another syllable? To these observations we must add three others: 1st, that it is very difficult to conceive why the last syllable in the form quoted, and in the others like it, should have lost the accent; 2nd, that Scherer's theory cannot be received without admitting in the medio-passive two formations quite different from each other; 3rd, that the middle meaning must in such case, according to the author, be derived from the passive, while we are accustomed to see the contrary phenomenon.

the final i which in the present active follows the a (e. g., in Proto-Aryan dvik-sá-i-Sanscr. dvik-sé) Scherer discerns, as Boller and F. Müller did before him, an "indicative adjunct (deiktischen zusatz)" designed to bring into prominence the person, or rather a locative particle, besides which Scherer marks also -am: the i would be added to the present and future active, and to the present and perfect passive after the disappearance of the final a had already taken place.' Hence the final ai of the middle presents us the a only as a real personal suffix. With this a ought clearly to be identical the final a of the "first principal conjugation" in the Aryan languages of the West, and in several forms of the Old Iranie: besides this a we find, in various forms and languages of our stock, as suffixes of the 1st pers. singular -i, -ma, -ân-, -am. Now, as -m leads us back to -ma, so -am does to -ama, as a primitive form, and we have, according to Scherer, the series -a, -ma, -ama, that is, the pronoun a (which in Sanscrit is a demonstrative of nearness and forms the augment), its superlative a-ma, and, by aphaeresis, ma, whence mi. A natural consequence of such premisses is the denial that the conjugation in a is derived from an older one in -mi, and the assertion that the difference between these two conjugations is original, in opposition to the teaching of the most important among recent philologers: an idea to which in fine Scherer had been already led by other considerations."

1 This hypothesis will be duly considered when we come to speak of the monograph of F. Müller, Zur suffixlehre des indogermanischen verbums.

2 But why, asks Kuhn, is not the entire and more convenient form ama? Why is the idea I to be expressed at one time with a positive, at another with a superlative? And can we admit as certain the existence

of a suffix of the 1st pers. singular consisting of an a?

See pp. 173 sqq. Against this assertion of Scherer it has been observed by Kuhn 1st, that a Greek - could not correspond to a primitive final a (1st pers. sing.); 2ndly, that we do not find even in the whole of the Aryan languages of Europe the two conjugations distinct from one another, for the