VOL. I.

PHILOLOGICAL SOCIETY.

MAY 24, 1844.

No. 23.

Professor WILSON in the Chair.

The following works were laid on the table :

On the Eastern Terminus of the Wall of Antoninus." By the Rev. Richard Garnett. Presented by the author.

Apposition St. Paul's School, 1844"; and Prolusiones Literariæ Præmiis Quotannis Propositis Dignatæ et in D. Pauli Schola Comitiis Majoribus Habitæ Die Maii viii. A. S. H. MDCCCXLIV." Presented by the Rev. H. Kynaston.

The following gentlemen were elected Members of the Society :— The Chevalier Bunsen.

Francis H. Dickenson, Esq., M.P.

A paper was then read :—

On the Origin and Import of the Augment in Sanscrit and Greek." By the Rev. Richard Garnett.

In

It has long been suspected that the vowel-prefix to certain past tenses (Sanscr. a, Gr. e) was originally a distinct element, potentially modifying the signification of the verb in its expression of the various relations of time. Not to dwell upon its restriction to particular tenses, it may be briefly observed, that in the older compositions in Sanscrit it is sometimes omitted, and sometimes separated from the verbal theme and placed between two prepositions. Greek, the Ionic and Æolic dialect frequently reject it altogether; and in certain verbs compounded with prepositions it is not unfrequently prefixed to the preposition instead of the radical portion of the verb. All these phænomena seem totally inconsistent with the idea of its being any integral part of the verbs to which it is joined ; as it is notorious, that though the constituent parts of compound terms may be disjoined by tmesis, the elements of truly simple words never are. Various theories have been advanced by grammarians to account for the origin and ascertain the precise force of this prefix. Some, confining their views to the Greek language, suppose it to have originated in the imperfect of the verb substantive, v or, was; an hypothesis involving a gross solecism, and subversive of all the established analogies of the Indo-European languages. Buttmann conjectures it to be nothing more than a mutilation of the reduplicate prefix of the perfect, so that eтUTTоv was originally TéTUTTоV. Though this idea might appear to derive some countenance from the epic forms of the second aorist, in which the syllabic augment and the reduplication appear to be employed almost indifferently, a slight comparison with the analogous forms in Sanscrit will show it to be to

6

=

tally untenable. Pott also regards the augment as a sort of imitation of the reduplication, but does not adduce any arguments in support of his position that appear of much cogency. Bopp has advanced an hypothesis, which has, at all events, the merit of originality-not to say singularity. He supposes the augment to be identical with the negation prefix a or an (Gr. å, åv), so that ëλeyov for instance is to be resolved into à-λeyov, I say no longer I said; the prefix not conveying a negation of the action, but simply of its present occurrence. In the last published part of his 'Vergleichende Grammatik,' Bopp labours to vindicate his theory against some severe and cogent remarks of Lassen in the Indische Bibliothek; but his defence is more remarkable for its learning and ingenuity than for its success in convincing the reader. He himself, indeed, seems to have some misgivings respecting the soundness of his hypothesis, since he admits that the prefix in question may be only collaterally related to the negative particle, as being derived from the same demonstrative pronominal root; and that, instead of denying the actual presence of the action, it may merely affirm its remoteness. He affects indeed to consider the two solutions as virtually identical; to which it is sufficient to reply, that the latter hypothesis is completely subversive of the former, and that the same element could hardly signify that, there, yonder, then, affirmatively, and express, vi termini, a negative proposition at the same moment. The object of the present paper is to show that the explanation which Bopp himself allows to be admissible,-namely, that the augment may be regarded as a demonstrative particle, primarily expressing remote place, and secondarily remote time,-is the one which unites the most probabilities in its favour. One cannot help feeling some surprise at the extremely limited view which has hitherto been taken of this question. Some have confined their investigations to the Greek, which gives absolutely no data for deciding the point: others have gone no further than the Sanscrit, which does not furnish any very satisfactory ones. The most rational and philosophical method of proceeding would have been to inquire how the same modification of time is expressed in other languages, especially those cognate with Greek and Sanscrit. If we find that any of those distinguish the past from the present by means of prefixes, and further, that those prefixes have a distinct meaning, suitable to the functions which they discharge, it is, à priori, very possible that the augment in Greek and Sanscrit may be of similar origin and similar import.

The Latin language will not afford us any assistance in this investigation; since, though it has partially retained the reduplication, it exhibits in its present state no distinct traces of a syllabic augment, or substitute for one. If we proceed to the Celtic, we shall find that all the dialects regularly form the preterite by the aid of prefixes, some of which are plainly significant. These preformatives are pretty numerous in Welsh, especially in ancient poetical compositions; but the one most commonly employed is a ; e. gr. canu, to sing; a ganodd, cecinit; caru, to love; a garodd, amavit. In old manuscripts the particle is regularly joined to the verb in writing;

aorug, he made, or did; aganodd, he sung; so that, had this orthography been persisted in, the prefix would have appeared as integral a part of the verb as a in Sanscr., atudat, or e in Gr., eruжTe.

The precise force of the Welsh element in this combination can only be inferred by analogical reasoning. As a pronoun, a denotes who, which, that; as a preposition, with; and as a conjunction, and. Reasons will be produced in the course of the present paper for believing that its original import was there, or then; denoting with greater precision the time of the action expressed by the verb. That it had a distinct meaning may be inferred from its changing the initial of the verb to which it is joined: a ganodd from canu; a dorrodd, broke, from torri. This phænomenon in the Celtic languages almost invariably denotes a grammatical or logical relation; namely, government, concord, composition or other modification of a word by something in immediate conjunction with it. It would be contrary to the analogy of the language to suppose that this effect could be produced by a verbum otiosum, or mere expletive.

Though a is sometimes used in Irish as a sign of the preterite tense, it is of comparatively unfrequent occurrence. The particle most commonly employed is do, which is seldom omitted, except when the verb precedes its subject. As a particle, do signifies to, and is employed as the sign of the dative and the infinitive. In ancient Irish we find greater variety of particles, and sometimes a combination of them, employed for the same purpose. At, ad, no, ro, ad no, do no, and do ro seem to be used indiscriminately, except that sometimes the compound forms may have the idea of greater precision or emphasis annexed to them. No and do no are interpreted by the lexicographers to signify then, which we believe to be the real import of most of those formative particles, the Greek and Sanscrit augment included. The prefix ro has its exact counterpart in the Welsh rhy, often used by old writers to form the perfect, pluperfect and future tenses. The element is significant in both languages as a particle implying excess, what is over and above, or further; and appears, when joined with verbs, to answer pretty accurately to our moreover. We may here remark the similarity of the Homeric particle pà, so frequently used in transitions. The common idea, that this word was formed by apheresis from apa, is both gratuitous and contrary to analogy. We believe the opinion of Mr. Donaldson in the New Cratylus,' who regards it as an independent term, implying addition, excess, remoteness, both alone and in composition, to be much better founded. It is indeed just as easy to affirm that apa is a compound, as that pà is an abbreviation.

The Gothic language exhibits a few instances of reduplication in preterites; but, with the exception of the particle ga, which will be noticed hereafter, neither it nor any other Germanic dialect has any thing formally corresponding to the augment. There is, however, a curious analogy in Upper German, which, if it had occurred to Bopp, might possibly have made him doubt the soundness of his theory respecting the negative import of the augment. Both in Old

and Middle High German we find the particles tho, do, ther, der, prefixed to verbs in the past tense, without any special reference to the idea of then or there, but simply, as it would seem, to denote the completion of the action. In translations from the Latin it frequently corresponds to the preterite in that language, unaccompanied by any particle. Thus in Tatian's Evangelical Harmony,' we find "quad tho Maria" dixit Maria; tho ward gitruobit" = turbatus est; "tho ther stigun sine bruoder" = ut autem ascenderunt fratres ejus. In the writings of the middle ages we find do and der employed rearly in the same manner. In the first edition of the Upper German Bible, a.d. 1462, are twenty examples of this construction in the first chapter of Genesis; as for example, ver. 3-4, "Un Got der sprach (dixit) liecht werde gemacht. Und das liecht ward gemacht; un Got der sache (vidit) daz liecht das es ward gut." So der rieff =vocavit; der macht = fecit; der beschuoff = creavit, &c. &c. Not only der, but also hin and her are frequently joined to verbs and participles in Middle High German to increase the emphasis and show that the action is done thoroughly.

The verb in the Slavonic languages presents some remarkable phænomena, well worthy the consideration of the philologist. It is known that in this class of tongues a regular, perfectly philosophical distinction is made between perfective and imperfective verbs, that is, between those expressing an action completed at once and not repeated, and those denoting continuance or reiteration. Thus to dig, implying a continued action, is regarded as imperfective; but to bury, which is done only once to the same subject, is a proper perfective. Sometimes this distinction is inherent in the form of the verb; but in many cases, verbs naturally imperfective, become perfective in the preterite by prefixing a preposition. What particular preposition may be employed with individual verbs depends on the custom of the language; those denoting out, from, by, with, after, are most commonly in use. Thus kropliù, I besprinkle, might denote a habit of so doing; to express a single definite act of it, already accomplished, would require po or na (after) to be prefixed to the simple preterite. The future perfect is formed in a similar manner by prefixing some one of the above-mentioned particles to the present tense. In all those compound phrases a sort of feeling appears to prevail, that the particle is necessary to convey the idea of completeness of action or precision of time, and in a great many cases the Slavonic preterite perfective would correspond pretty accurately with the Greek or Sanscrit aorist, used definitely.

The only remaining European language which seems to offer any analogy is the Albanian, which forms the future by prefixing do te to the present: e. gr. thom, I say, do te thom, I shall say. As a particle of place do signifies where; if transferred to express the idea of time, it would naturally denote when or then, which may be employed with equal propriety in a future or a past acceptation. Thus the Irish ro = moreover, is used with preterites, and its counterpart, the Welsh rhy, both with the preterite and future.

The languages of Central Asia also present a few analogies. The ancient Armenian prefixes e to the preterite, exactly like the Greek : but this formation is confined to the third person of one particular conjugation. The Kurdish also employs several particles in the formation of past tenses; as che kiria, fecit, from the root ken, make, &c. We have no means of ascertaining whether those particles have a distinct meaning, or what it is. In Persian the particle be or bu, prefixed to the present, converts it to a future: pursem, I ask, bupursem, I shall or will ask. The Georgian also employs a variety of preformative particles in conjugation, but the precise analysis of them has not hitherto been made known.

In the Coptic language the system of verbal preformatives is more fully developed than in any of the Indo-European. Every tense has its appropriate particle, apparently meant to express the particular modification of time supposed to be included in the entire phrase. Thus e is the sign of the present tense; na of the imperfect; a of the preterite definite; sha of the preterite indefinite; ne a and ne sha of the pluperfect; e-na the future; ta or ta-re the future indefinite, &c. It is true that several grammarians regard those prefixes as auxiliary verbs; but there are reasons, which we cannot here discuss at length, for believing that most of them are of pro nominal origin. The particle ent or et, used in one of the preterite formations, is confessedly identical with the relative pronoun in form; and Benfey admits that they are, in all probability, of common origin.

In the languages of Southern India the system of verbal formatives expressing the time of the action is carried to a great extent. Thus Anderson observes (Rudiments of Tamul Grammar, page 44), "The Tamul grammarians resolve most of the derivative forms of a verb into three parts; viz. paghudi the root, vighudi the form of termination [distinguishing the persons], and ideimilei the intermediate argument, which is generally employed as the formative of the different tenses." MacKerrell also remarks (Grammar of the Carnataca Language, p. 85), "Verbs in the Carnātaca language, whether possessing an intransitive or a transitive meaning, are conjugated by adding to their roots, in three of the tenses (viz. present, past and future), certain affixes expressive of time; and to these the affixes denoting persons being attached, the inflexion is complete." The particles thus employed are pretty numerous, and the rules for the application of them are rather intricate; but it is obvious that they are all regulated by the same general principle, of specifying the time of the action more precisely than could be done by merely using the verbal root with its personal affixes.

The Polynesian languages, especially those of the eastern division, furnish copious and valuable materials for the illustration of the point in question. The whole conjugation of the verb-as far as distinction of tense is concerned-depends on the employment of certain particles, which, allowing for the difference of dialect and pronunciation, are nearly common to the great body of the east insular Polynesians, properly so called. A minute account of them