Abbildungen der Seite
PDF
EPUB

'battle', Greek ruάw 'I honor': Tuń 'honor', Latin ceno 'dine': cena 'dinner'). The only other IE type that could come into consideration at all would be the verbs in ejo (e.g. Skt. amitrayati 'is hostile': amitras 'enemy', Gk. ȧpiμéw 'count': ȧpuós 'number', Lat. claudeo 'limp': claudus 'lame'); but IE e regularly appears in Hittite as e (often written i), and a change of e to a before i is disproved by the nominative plural of the i-stems already cited (PIE eies becomes Hittite es).

The identification of the Hittite denominatives in a(i) with IE denominatives in ajo is supported by several etymologies.

=

Hittite aššuš 'good' has been plausibly identified with Gk. éús 'good'.4 The difference in the initial vowel may indicate a variation in ablaut; but there are still a number of unexplained instances of initial a for e (e.g. anta(n) 'within'; cf. Sturtevant, LANGUAGE 3. 168). Now, if aššušéús, the derivative verb should appear in Gk. as eifáw < *kofāw and the original digamma of éáw is actually recorded by the lexicographers (Hesychius and the Etymologicum Magnum report Syracusan ἔβασον ἔασον, Baσov čarov, and Hesychius has eva. . . ëa). The semantic development from 'be good to, favor' to 'permit' is particularly easy. The initial vowel of éáw is difficult, but not inexplicable. Phonetic laws would yield *ús, *eifáw, and *Faov; but a system so complicated and so unparalleled could scarcely survive. In particular, the alternation of e (i.e. e) with n possibly occurred in no other verb, except eiuí: v; and so one might expect the augmentless *elfaov to be preferred to *Faov. At the same time the adjective *ús tended to change the present to *fáw, and both processes were favored by verbs which had lost initial σ or before e, e.g. exw:elxov. Precisely this verb seems to have been particularly influential, since it accounts for the loss of rough breathing in eύς, έω, and είων.

[ocr errors]

The identity of aruwa (i)- 'bow, make obeisance, worship' with Gk. ápáοual 'pray' seems quite evident, since Attic ȧpá and Ionic ȧph 'prayer' are shown by Arcadian κáтapfos to have lost digamma.5

Götze (Madd. 84 and fn. 17) suggests that Hittite mita (i)- may mean 'befestigen', and he points out that it must at any rate be similar in meaning to tarma(i)-, for which he suggests the same translation. Since

4 See Friedrich, Indogermanische Forschungen 41. 371 (1923).

'Persson's (Studien zur Lehre von der Wurzelerweiterung und Wurzelvariation 243) connection of apá with Lat. ōro 'plead' and Skt. āryati 'he praises' is semantically improbable, and the Lat. word can scarcely be separated from ōs 'mouth' (see Buck, Grammar of Oscan and Umbrian 21). Our Hittite etymology is not affected in any case.

the latter word in the Yale tablet denotes something that is done to a road with pegs, I translate it 'mark off, delimit' and I connect it with Lat. terminus, Gk. Tépμa 'boundary'. If so we must certainly identify mita(i)- (i.e. meta(i)-) with Lat. metari 'measure, lay out (a camp)'. Note that the corresponding noun, meta 'goal-post', would be translated into Greek as тépμа. Probably tarma(i)- may also be identified with Lat. terminare, for there is nothing to interfere with the assumption that Hittite simplified the sound-group rmn to rm.

I have recently connected Hittite happara(i)- 'sell' with Lat. parare 'buy', on the assumption that the Hittite word contains the prefix ha = IE *o. In this case Latin does not possess a noun from which the verb could be derived; but, as noted above, Hittite has a noun happar 'business transaction'.

Particularly striking is the correspondence between such Hittite denominatives as aššuwa(i)- 'be good' from aššuš 'good' and the Latin type seen in aestuare from aestus 'billowy motion', fluctuare from fluctus 'wave', arcuare from arcus 'bow', deartuare beside artus 'limb', sinuare from sinus 'curve', manuari 'steal' from manus 'hand', tumultuari from tumultus 'uproar'. To this list we must add gravare beside gravis 'heavy', levare beside levis 'light', and tenuare beside tenuis 'thin'; for these adjectives were originally u-stems.

The agreement of Hittite with Latin in the formation of denominatives from u-stems is the more significant, since they also agree in treating the i-stems quite otherwise. Latin employs the suffix io to form finio from finis 'boundary', vestio from vestis 'clothing', tussio from tussis 'cough', sitio from sitis 'thirst'; and just so we find in Hittite (Götze, Madd. 97 f.) armizziya- from GIšarmizzi- (meaning uncertain), išhamiya- 'sing' from išhamais 'song', tuzziya- 'encamp' from tuzziš 'camp', hapatiya- 'obey' from hapatiš 'servant', hulaliya- ‘entwine' from GIšhulali 'distaff', lazziya- 'grow well, recover' from lazzaiš 'orderly condition', šankunniya- 'be a priest' from šankunniš 'priest', zahhiya'fight' from zahhaiš 'battle'. When Latin forms verbs of the first conjugation from i-stem nouns, the stem-vowel disappears, thus: piscor from piscis 'fish', testor from testis 'witness', illustro from illustris 'bright', infelico from infelix 'unlucky'. With such formations corre

• See Sturtevant, Transactions of the American Philological Association 58. 20 (1927).

7 LANGUAGE 4. 164 (1928).

See Stolz, Historische Grammatik d. lat. Sprache 1.592.

spond Hittite arša (i) from aršiš, gangala (i)- from gangatis, and warra(i)from warris.

In Greek also we have a few verbs of the type of Hittite aššuwa(i)from aššuš. We have already noticed the Greek etyma of this particular pair, namely éáw and èús. Other examples are Teλekάw, whose original digamma is guaranteed by πελέκκησε (Odyssey 5. 244), from πελεκύς 'axe', ixováw 'angle' from lx0ús 'fish', and yyváw 'give security' (originally 'give into one's hand') beside Avestan gav(a)- 'hand'.9 We must include also ἀράομαι from *ἀρράομαι, on account of its identity with Hittite aruwa (i)- (see above p. 10). Greek, however, unlike Latin and Hittite, exhibits similar formations from i-stems, e.g. dŋpiáoμai from δήρις fight', μητιάομαι from μῆτις ‘wisdom', ὀκριάομαι from Ŏкp is 'roughness'. The lack of such denominatives in the related languages makes it probable that Greek is here the innovator. No doubt the source of the type is to be found in the more numerous derivatives of stems in a and to, such as oxiάw from oxía 'shadow', ἀνιάομαι from ἀνία 'sorrow', ἀντιάω from ἀντίος 'opposite', and αἰτιάομαι from airios 'blameworthy'; for Latin shows formations analogous to these, such as insidior from insidiae 'ambush', glorior from gloria 'fame', vario from varius 'diverse', and consilior from consilium 'counsel'.

We have, then, in Hittite, the familiar verbs in ajo, and many of them are as clearly denominative in origin as in any IE language. But, oddly enough, no type of Hittite noun can plausibly be connected with the IE ā-stems. It is conceivable that this declension was merged with the o-stems in the Hittite a-declension; but such an assumption will not help us out of our present difficulty, since only one or two Hittite a-stems have so far turned up beside verbs in a(i). As far as Hittite is concerned the suffix a(i) from original **āio is to be regarded, not as a fusion of two suffixes, but as a single formative element, by which verbs are derived from u-stems and consonant stems, and, after loss of the stem vowel, from i-stems and a-stems as well.

Now, the ajo suffix often behaves in the same way in the IE languages. We have already cited several examples, such as Lat. sinuare and testari, and Gk. ix@váw. Skt. priyāyate 'be friendly with' from priyas 'friend' is a denominative with suffixal aio that must date from proethnic times on account of the cognates, Gothic frijōn and Church Slavonic prijają 'treat kindly'. Similarly Lat. novare and Old High German niuwōn 'renew' imply IE *neyājo- from *neyos 'new'. In Sanskrit

Cf. Boisacq, Dictionaire Etymologique de la Langue Grecque, s. v. ¿yyún.

denominatives in aya are freely made from a-stems, but scarcely at all from consonant stems. Greek has many denominatives in aw from o-stems, as ȧτáw from åтuos, and a few from consonant stems, as στixáομαι from orixes 'ranks'. In Latin both these types are common, e.g. donare from donum 'gift', nominare from nomen 'name'. Similarly Old Irish shows marbaim from the o-stem marb 'death' and athrīgaim 'dethrone' beside the consonant stem of ri 'king'. In Germanic we have many aio-denominatives from o-stems (Goth. frijōn, OHG niuwōn) and particularly from stems in s and n, e.g. Goth. hatizōn from hatis 'hate' and fraujinon 'be master' from frauja 'master'. Typical Balto-Slavic denominatives are Lithuanian kilnóju from kilnas 'high' and Ch. Sl. prijają. Ch. Sl. znamenają comes from the n-stem znamę 'symbol'. Peculiarly significant is the fact that in the oldest documents of IE speech there are numerous ajo-verbs for which no nouns of any form can be cited as primitives. Whitney (Sanskrit Grammar §1066) says:

"A number of denominative stems occur in the Veda for which no corresponding noun stems are found, although for all or nearly all of them related words appear.... A Vedic group of stems in āya. have allied themselves to present systems of the na-class, and are found alongside the forms of that class: thus, grbhāyati beside grbhṇāti. Of such the Rig-Veda has grbhāya-, mathāya-, pruṣāya-, muṣāya-, śrathāya-, skabhāya-, stabhāya-. A few others have no nā-class companions: thus, damāya-, śamāya-, tudāya- (Atharva-Veda); and panāya-, naśāya-, vṛṣāya-, (vṛṣ 'rain'), vasāya- (vas 'clothe'), and perhaps aśāya(as 'attain')".

From Homer we may cite as similar verbs without known nominal source: ἀντάω, ἀσχαλάω, ἀπολιχμάομαι, διφάω, εἰλυφάω, εἰρωτάω, ἰάομαι, κοιμάω, κυβερνάω, κυκάω, λωφάω, μυκάομαι, νωμάω, οἰμάω, σκιρτάω, στρωφάω, τηλεθάω, τρωπάω, τρωχάω, ψηλαφάω.

Brugmann1o mentions as examples of old ajo-verbs of unknown source Lat. hio, Lith. žióju 'yawn'; Gk. λáw 'bark', Lith. ulóju 'shout, call', Lat. ululo 'yell'; Lat. iuvo 'help'; Lat. mico 'vibrate'; OIr. scaraim 'separate'; Goth. mitōn, OHG mezzōm 'measure'; Lith. lindoju 'stecke worin'; Ch. Sl. razvrizają 'open'.

The IE evidence is enough to show (1) that the suffix ajo is of extreme antiquity, and (2) that it was not always associated with nouns in a at the close of the IE period. The Hittite evidence indicates that the suffix was originally independent of nouns in a, and that it was probably

1o Grundriss der Vergleichenden Grammatik der Indogermanischen Sprachen1 2. 1109.

more ancient than they. I conclude that ajo is to be considered as a unit; it was not, as has usually been supposed, a conglomerate of stemfinal ǎ and suffix jo.

If the ajo-verbs antedate the corresponding a-stem nouns, it follows that our conception of the relationship between the two categories must be reversed; these a-stem nouns are pendants of the verbs in ajo. It has long been recognized that certain ā-stem nouns are 'backformations', e.g. Lat. pugna 'fight' from pugnare, which is a denominative from pugnus 'fist'; Gk. κarápa 'curse' from Kaтарáομаι, a compound of ȧpáoμa 'pray'. But the a-stem nouns as a class cannot have originated in precisely this way; pugna from pugnare was formed on the model of such pairs as cena: cenare, flamma: flammare, multa:multare, and without some sort of models the process of inverse derivation is scarcely possible. The original impulse must therefore have been supplied by denominatives from o-stems and i-stems with the suffix jo; e.g. *senejo- 'grow old': *sene/o- 'old' = *aryājo- 'pray': *aryā'prayer' or *mēti-jo- 'plan, measure': *mēti- 'a plan, a measure' *aryājo-: *aryā-.

=

It is doubtful, however, whether inverse derivation of this sort would occur if it resulted in words of unfamiliar type. I am therefore more than ever inclined to agree with Hirt" in deriving a number of ā-stem nouns directly from dissyllabic bases; e.g. Bla 'strength', åyn ‘astonishment', dún 'misery', ovn 'growth'. The existence of such words rendered plausible the products of inverse derivation from verbs in ajo, and thus secured their adoption into the language.

This theory of the origin of the IE a-stem nouns favors the current belief that the European languages preserve the original inflection of the a-stems better than the Indo-Iranian languages. The opposite view. however, which has been supported by Collitz, 12 is not excluded. If we assume IE denominatives with suffix o, these would furnish a model by which the ajo-verbs may have given rise to nouns in ai, which would have fallen into line with the diphthongal stems like *dieus and *rēis. For details of the later development, reference to Collitz will suffice.

11 Indogermanische Grammatik 2. 204 f.

12 Bezzenberger's Beiträge zur Kunde der Indogermanischen Sprachen 29. 81-114 (1905).

« ZurückWeiter »