You are on page 1of 25

Linguistic Society of America

The Personal Endings of the Present and Imperfect Active and Middle Author(s): Louis H. Gray Reviewed work(s): Source: Language, Vol. 6, No. 3 (Sep., 1930), pp. 229-252 Published by: Linguistic Society of America Stable URL: http://www.jstor.org/stable/409219 . Accessed: 28/01/2012 01:33
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Linguistic Society of America is collaborating with JSTOR to digitize, preserve and extend access to Language.

http://www.jstor.org

THE PERSONAL ENDINGS OF THE PRESENT AND IMPERFECT ACTIVE AND MIDDLE Louis H.
GRAY

COLUMBIA UNIVERSITY [See the first two paragraphs] INTRODUCTORY

In 1927 I published in this Journal (3. 71-86) a study on 'The Inflection of the Present Indicative Active in Indo-European.' Since that time, further investigation, particularly along the lines of vocalic alternation (apophony, 'Ablaut'), with its underlying factor of accent, has led me to conclusions which seem to me so much deeper in foundations and, in a measure, so much broader in results that I regard that older discussion as practically supplanted by the one which I now present. This investigation forms part of a series of three, the other two being 'Sur l'inflexion des pr6tendus th6mes en -i' and 'Vocalic Alternation in the Disyllabic Base in Indo-European', to appear almost simultaneously with this in the Bulletin de la Societe de Linguistique de Paris and the American Journal of Philology respectively. To the 'Disyllabic Base' I must refer for the theory of alternation here assumed, and there one will see that the difference between athematic and thematic bases still appears to me to be due solely to apophonic (i.e. accentual) factors, as I suggested in LANGUAGE 3. 85-6. Concerning the origin of the personal it seems premature (if not impossible) to express any reasoned endings it at least be obvious that I do not share the view that but will opinion; composition (or agglutination) and nominal formations have played an important part in verbal inflexion.'
FIRST PERSON SINGULAR

The normal athematic type of the present active is represented by Ved. ddmi, GAv. staomi, OPers. amiy, Turf. ('Tokharian A') gem, Kanes.

1See, for the most recent exposition of this theory, H. Hirt, Indogermanische Grammatik4. ?? 45 sqq., Heidelberg, 1928. On the accent of the verb in general see R. Loewe, Der freie Akzent des Indogermanischen 1-20, Berlin, 1929; cf. A. Meillet, 'Remarques sur les desinences verbales de l'indo-europ~en', BSLP 23.64-75 (1927). 229

230

LOUIS H. GRAY

('Hittite') epmi (but also arhi), Arm. em, Gk. el.u, Lat. sum, OIr. am(m), Goth. im, Lith. esmi, OCSlav. jesmi; and the thematic by Kuch. Lat. fer3, OIr. ('Tokharian B') laka-u, Kanes. tittanumi, Gk. qpepw, OCSlav. bera, < *bert-m, -biur, Goth. baira, OHGerm. biru, Lith. vedza, etc. In the thematic type Indo-Iran. shows -ami, as Ved. bhdrdmi, YAv. bardmi, OPers. ddraydmiy. This -d- is an obvious survival of an earlier *bhdrd < *bher5 (Gk. cpcpw, etc.), still preserved in ten different subjunctives in Ved., in nine indicatives (with nine subjunctives) in GAv., and in two indicatives in YAv., the latter having also the more primitive form zbayemi.2 The type in -dmi is, accordingly, a 'double first'. The corresponding ending for the present middle is *-ei, *'-oi for athematic verbs, and *'-ai (ai : oi) for thematic, as Ved. uv6, vdste, bhdre, GAv. mruye, aroi, yese, Kuch. aikema-r,3 OIcel. heite, OLith. duomie-t (?). It would appear, from a comparison of these forms, that Gk. ?pipolaL (like the OLith. form) received its -m- from the first person plural, so that the original form would have been *op.paL,its historic form being due to the analogy of spiE~raL,etc.4 On the other hand, the termination of the perfect active type of Ved. -i , OLat. fecei, Lat. tetudi, OCSlav. vde' is more probably derived from *-ei, *'-oi than from *-ai, the current view notwithstanding. The imperfect active shows -m, as Ved. adoham, dbravam, dgaccham, GAv. 5isiam,darasam, OPers. ajanam, abaram, Kuch. takoym, yamim, Kanes. hinkun < *-5m,5Gk. CZ0y, kpov, Lat. -bam, OIr. -ba, -ber < *-dm, OCSlav. mogil. The athematic type of Ved. adoham, GAv. 65isam, etc., seems to have borrowed its -am from the corresponding thematic form, and its vocalism from its own present singular, its strictly
2 B. DelbrUick,Altindisches Verbum26, Halle, 1874;C. Bartholomae, in KZ 29. 272-3 (1888), and GirPh. 1 a ? 333, Strasbourg, 1901; K. Brugmann, Grundrissder derindogermanischenSprachen 2. 3. ?503,2, Strasbourg, Grammatik vergleichenden 1916. ith Kanes. yahha-ri, a'haha-ri cf. the perf. ending -a, as Ved. vida, Gk. W3 o13a. For the Kanes. verb in general see L. Delaporte, Al1ments de la grammaire hittite, Paris, 1929, and 'Etude sur le verbe hittite aux modes personnels', in JA 212. 267-321 (1928); F. Hrozn?, 'Das hethitische Mediopassivum', in Actes du premier congres international des linguistes 155-64, Leyden, 1928; J. Friedrich, in ZdmG76. 167-73 (1922); E. H. Sturtevant, in LANGUAGE 3. 165-9 (1928); for the Kuchaean, S. L6vi and A. Meillet, in MSLP 18. 2-33 (1914).

-o- to analogy with secondary terminations in -o. r'iKrroL, 5 For the phonology cf. Kanes. acc. sg. antuhg-an < *-om, acc. pl. antuhg-us < *-ons.

4 Such Arcad. forms as rwtavilo-raro&, e'oEoL,36XArot,

etc., may owe their

ENDINGS OF PRESENT AND IMPERFECT

231

etc. The terminaschematic form having been *dduha < * -dhught, tion of Arm. beri is doubtful.' The imperfect middle has *-i for athematic and *-d-a (a : o; probably on the analogy of the present middle) for thematic verbs, as Ved. dvrn.i, dhuve, GAv. -aoft, YAv. aguze, OPers. patiydvahyaiy, Kuch. takoymd, (Dor. wlawoyma-r. The Gk. middle form cip , pp9 Y < *-a 'pa4av,Argol. E6Idacv) has given rise to much discussion, for one would etc., or *'p~Epa on the expect either *,pipepoon the analogy of ApkpEro, As of Ved. bhdre (injunct.). Meillet-Vendryes and Hirt analogy is probably for *'p'pav have already pointed out, however,' opepbqmp is be that it to be derived from a and it suggested may (<*cpdpah), innovation > > very early *6bherom, *,'ppa-.8 *e-bheroma The earliest inferable terminations of this person would seem to have been: Athematic Thematic Active present -mi -5 -m " imperfect -o-m Middle present -ai -ei, '-oi " imperfect -i -0-9 Tabulating these in the alternation-scheme of a termination-base *-ai, *-5i, the results are: N a *-i, *-5i (*bhergi,*bherai) b *-j, *-5 Ved. stivd, Gk. Appo, etc.9 S a *-ei, *-oi Ved. yuje, i'?e, GAv. drbi, YAv. mruye, OLat. fecei, OCSlav. vde' b *-e, *-o Ved. veda, jagrdbha Ved. bhdre, GAv. yesi, Gk. ppopac < *4pp(oA)at, R a *-i OIcel. heite, OLith. duomie-t (?) Ved. dhuve < *-0-a,Gk. 'pe~p(b6).Lv < *p(~ppo0p b *-a
A. Meillet, Esquisse d'une grammaire compardede l'arminien classique ? 95, Vienna, 1903. A. Meillet and J. Vendryes, Trait de grammairecomparee des langues classiques 1912, Grammatik4, ? 64, 7; for Kanes. eshat(i) see Hroznr 162. 8 Either < *-ome which > Aa internally and ag finally, or < *-Aa > -aA through
? 473; H. Hirt, Handbuch der griechischen Laut- und Formenlehre2 ? 404, Heidelberg,
6

i etc. (for examples see Hirt, Formenanalogy with such aorists as kx&prY, ja7Y,

lehre ? 419). 1 The type of pipw,etc., might also be explained as < *bhdro-o, in which case the grade would be Sb. This encounters grave difficulties, however, in the type of fipeL, etc., unless it be assumed that the ultra-reduced vowel was contracted with a following short vowel, but vanished before a diphthong.

232

LOUIS H. GRAY

1 a *-,i, *-oi (*bhdrei,*bheroi) Ved. dgaccham, Gk. i epov,etc., < b* * *&-gipso-m, *6-bhero-m, etc. Z a *-i (OPers. adarsiy, etc.) b *-0 (*bhrO).
SECOND PERSON SINGULAR

The normal athematic type of the present active is represented by Ved. dtsi, YAv. pahi, Turf. ?es, Kuch. nesdh < *nes-tu (?),1o Kanes. epsi, Arm. es < *es-si, Hornm. al, OLat. ess, OIr. a-t, Goth. is; and the thematic by Ved. bhdvasi, YAv. barahi, Kanes. iyasi, Lat. legis < *lege-si, OIr. beni <*bhina-si, Goth. bairis, OHGerm. biris. Beside this stand forms in *-ei: OIr. -bir, OPruss. sdtuinei, Lith. suki, sukie-s, Lett. metti, mettie-s; and in *-eis: Ossete fdvis, Talif bis, Wai-Ali vias', Gk. pELts.The form in *-ei seems to be the grade Sa of a termination-base *-i, *-5i, whose grade Sb has already been seen in the thematic first singular active. While the grade *-ei may indeed underlie the Iran., OIr., and Balt. forms, Gk. p'ptL-s can arise only from *-ei-s. This -s in the Iran. and Gk. seems best explicable as the grade Zb of the termination-base *-sji, *-sai, added to the base *- i *-5i on the analogy of the characteristic consonant s of the athematic inflexion to distinguish the second from the third singular (Ossete fdvi, Talig bi, Gk. ?pEt < *?ppeL-.r). Originally thematic, *-ei was transferred in and perhaps Lat. Gk. to athematic verbs (cf. Ved. bhdrti, Gk. qPEpre, fers," etc.), just as the athematic ending *-si was transferred to thematic verbs not only in Indo-Iran., but also in Italo-Celt. and Teut. The original form would appear to have been *bhreis. Besides *-ei one also finds *sei in OPruss. waise(i), segg~sei, druwjse, asse(i), etc., OLith. duosi, desie-s, and OCSlav. beresi, the latter best explained as < *berifi (with e instead of i on the analogy of the second plural berete = Gk. popere ?) < *ber i-sei, i.e. a 'double second'.'2 On the analogy of *-ei, *'-oi in the first singular perfect Ved. ij6, etc., it may be suggested that one has here a survival of an old perfect middle
L6vi-Meillet, in MSLP 18. 8 (1914). " Lat. fers, etc., may, however, be syncopated from *fer-es,etc. (Stolz-Schmalz, see P. PersLateinische Grammatik6 ? 219 e, Munich, 1928); for *bherbeside *bhere 693, Upsala, 1912. son, Beitrdge zur indogermanischenWortforschung 12 To the references in LANGUAGE 3. 7528 add A. Meillet, Le Slave commun29, Paris, 1924;J. Endzelin, Lettische Grammatik? 596,Heidelberg, 1923;N. Van Wijk, Altpreussische Studien 13-23, 59-60, 127, The Hague, 1918, and 'Zu den altpreussischen Personalendungen', in IF 47. 148-60 (1929).
10

ENDINGS OF PRESENT AND IMPERFECT

233

ending *-sei found in Ved. cakrfe; and this hypothesis seems borne out by the accentuation (cf. Ukrain. dasi, Stokav. jesi), which obviously was *bereAi. (Kanes. also shows -ti, -4ti in this person, as apti, pelti; and in the present has egtat beside epsi.) The corresponding present middle ending is *-sei, *'-soi for athematic verbs, and *-sai < *-sai (ai : oi) for thematic, as Ved. vitse, i'kge, bhdrase, YAv. parasahe, Gk. "ata, (ppeat, opEp < *"?pEpoaL,OPruss. assai; and for the imperfect active, *-s, e.g. Ved. d's, dyds, dbharas, GAv. da, mraos, daray3, YAv. vaodas-ca,OPers. gaudayah(injunct.), Kanes. dag (but also Gk. iya-t, ep-ta, memis-ta, pes-ta, arnut-ti), Arm. berer < *&-bheres-r(e),LS Lat. OIr. Corn. nese < OCSlav. -ba, 7s, epO~Ep, fuas, -bas, cares, *neses (cf. OPruss. opt.-impv. wedais, wedeys, weddeis'4). Goth. bairaza stands alone in having the ending *-s5.15 The imperfect middle has *-so, as GAv. -aoyz~a <*6-euguh-so, YAv. -zaya~ha, Gk. kio, 'p~po,EVppov<*V4peao, Lat. sequere. Side by side with this is *-thes in Ved. ddhatthds, dbharathas, Kuch. dhatmasdha-r (?), Gk. a660ls,OIr. cluinte (impv.), suidigthe-r (pres.).'1 It may further be suggested that we again have a 'double second' as in the cases of Gk. pipetsand OCSlav. bereli, *-thes being really a conglomerate of *-the + *-s, the latter parallel with Gk. qpcpe-s, etc., and the former in apophonic relation with *-tha of the second singular perfect active (Ved. vettha, cakdrtha, GAv. dada0d, Gk. olO0a, Goth. farft). The vocalism of *-this implies an original accent, so that its connexion with the imperfect must be late. It is obviously connected with the *-te-s(i) of the second plural present active, but one may suspect that the original ending was *-so, *-o-so.17 The earliest inferable terminations of this person would seem to have been: Athematic Thematic Active present -si -ei-s -s imperfect -,-s Middle present -sei, '-soi -,-sai -so " imperfect -thW'-s, -o-so,-ethe'-s
Meillet, Esquisse ?? 88, 94. type of Lith. sakar, Lett. saki see Endzelin ? 596 c. 16 A. Meillet, in BSLP 23. 68 (1922), and MSLP 23. 219 (1929), and similarly for the other persons. 18OIr. berthaseems to show *-thds (H. Pedersen, VergleichendeGrammatik der keltischenSprachen 2. ? 605, 2, G6ttingen, 1913). Kanes. has estat. 17Brugmann, 2. 3 ?577, 2, suggested that *-so belonged originally to the thematic type, and *-this to the athematic.
13

14For the

234

LOUIS H. GRAY

Tabulating these in the alternation-scheme of a termination-base *-si, *-soi, the results are: N a *-sai, *-s.i, (*bher(e,)si,*bher(o)sai) b *-sa, *-sa, Goth. bairaza S a *-sei, *-soi, Ved. krg', (cakr?e), OPruss. waisei, OCSlav. beresi, Ukrain. dasi GAv. -ao-yza,YAv. -zayagha, Gk. aco, Lat. sequere, b *-se, *-so etc. Ved. bhdras?,Gk. ~aat, OPruss. assai, etc. R a *-sai b *sa (*bherese) 1 a *-sei, *-soi (*bher(e)si, *bhrr(o)soi) b *-se, *-so *bhdr(o)so) (*bher(e)se, Z a *-si Ved. dtsi, bhdvasi, Hom. baai,Goth. bairis, etc. Ved. d'gas, dbharas, Gk. pcp-s, EpEPES, b *-s etc.
THIRD PERSON SINGULAR

This person closely resembles the second singular, except that its characteristic consonant is t instead of s. The normal athematic type of the present active is represented by Ved. pd'ti, bhdrti, GAv. vasty, OPers. aitiy (YAv. asiti), Dor. &OiTrt,Lat. est, OIr. is, Goth. is, OPruss. ast, Lith. est(i), ORuss. jest?; and the thematic by Ved. bhdrati, GAv. baraitT, OPers. tarsatiy, Arm. berg <*beriy Goth. bairi], OHGerm. birit, ORuss. <*bhereti, Lat. agit, OIr. berid,"8 Besides this are forms in <*qpE -7 <*pEtpi-7, *-ei: Gk. pE~p beretf.X9 Ossete fdvi, Tilif bT,and probably OIr. -beir. In Kuch. the simple verb-base serves without ending, as yapi (in swdsam etc., -am? the case with Aeol. laitam., appears to be a suffixed particle),20 as is also W p,.21 elKPVU,'5u4pk, 16a, (Kanes. here shows *-tvi and -i, as T7a 7, iyazi,22ddi, ari; and Turf. has ses.) The corresponding present middle ending is *-tei, *'-toi for athematic verbs, and *-tai <*-tdi (di : oi) for thematic, as Ved. bratet, vdste, GAv. vasti, haatJa, YAv. mr'ate, baraite, OPers. gaubataiy, Kuch. mrauskdte
18 R. Thurneysen, Handbuchdes Alt-Irischen ? 559, Heidelberg, 1909;less probably Pedersen 2. ? 603. 19 OCSlav. sometimes shows -tU beside the far more usual -ti~, e.g. est), moietl (0. Wiedemann, Beitrdgezur altbulgarischenConjugation 11-13, Petrograd, 1886).

original i.

0o L6vi-Meillet 6, 8. 21F. Bechtel, Die griechischenDialekte 1. 97, Berlin, 1921. 22Sturtevant, in LANGUAGE 4. 228-9, 230-1 (1928), holds that t > z before

For Kanes. z < t + J see Delaporte, Rl4ments ? 69.

ENDINGS OF PRESENT AND IMPERFECT

235

(?) (also maskit-r, dhatmast-r),23 Phryg. aoepEraG,24 Gk. arat, iperat, Anglo-Sax. hitte. Besides this, one also finds *-ei *'oi, *-ai in Ved. duhi, dye, stdve, YAv. mruye, which clearly stand in apophonic relation with the *-e of the third singular perfect active (Ved. vida, jajd'na, GAv. va&da,nnasda, Gk. otCe,yc'ov~, OLat. fhefhake-d, Osc. dede-d,25Gaul. 6E8,OIr. -cechuin <*kekane, Goth. bauf (perfect middle: Ved. jajn6, GAv. dddrE,YAv. vaode).Goth. bairada, however, sho'ws*-t4. The imperfect active has *-t:26 Ved. dpdt, dcarat, GAv. mraot, ddrayat, OPers. agarbdyah,Kanes. iyat, dds' <*d-s-t (-s-)aorist, Arm. ed, eber, Phryg. acfeper (from the context almost certainly injunct.),27 Gk. Lat. -bat, OIr. -beir, -cria (conjunct presents), Lith. mita, gpoy, epE, Lett. met (presents),28 OCSlav. moze. In Kuch. the termination is zero (wefia, dra, lacd, gem).29 In the imperfect middle the ending is *-to, as Ved. dsiita, dmandata, GAv. yfxtd, giisatd, OPers. akutd, agaubata, Kanes. arta-ti, Phryg. ac/tpero (reading not absolutely certain), Venet. OIr. -bered. zoto, Gk. oTro, sPcPETro, The earliest inferable endings of this person would seem to have been: Athematic Thematic Active present -ti -ei-t -t " imperfect -e-t Middle present -,-tai -tel, '-toi -to " imperfect -,-to Tabulating these in the alternation-scheme of a termination-base *-ti, *-tai, the results are: N a *-tei, *-tai (bher(c)tai, bhegr(o)ti) b *-t3, *-to Goth. bairada S a *-tei, *-toi Ved. briitd,GAv. vasts Ved. dsi7ta, GAv. yiaxtd, Gk. 3aTro, etc. b *-te, *-to 6, 12, 13. With Kanes. yatta-ri (also epta, memilta, ela[t(ari)]) Lat. Umb. fertur, ferar, etc. (Hrozn4 160). of. 24 The suggestion of J. Fraser (Transactions of theCambridge Philological Society 6. 2. 124[1913]),that 'it is almost certain that the [Phrygian] middle endings are borrowed from Greek', is perhaps not quite so sure. 25Cf., however, Stolz-Schmalz ? 246, 3. 28 A. Meillet, 'Sur les desinences secondaires de 3e personne du singulier', in MSLP 23. 215-21 (1929).
27 But see Fraser 74, 81, 123. Whether Lye. prnnawat? 'EnoL7jraTo', prAnawate 'pyd4cravTro' belong here is uncertain in view of the obscure affinities of the lan23 L6vi-Meillet

guage; and Lyd. ftnshifid (sing.), vq(?)bapint (twice with a single subject and twice with two subjects) are likewise dubious (Lyd. is probably not IE; cf. Gray, JRAS 1930. 625-7). 28 Endzelin ? 597.
29 Lvi-Meillet

3.

236

LOUISH. GRAY

Ved. bhdrati, GAv. haca'tt, Gk. pepEr7at, etc. R a *-tai b *-ta (bhdr(e)ta) 4 a *-tei, *-toi (*bh`r(e)tji,*bher(o)toi) b *-te, *-to (*bhdr(e)-te, *bher(o)to) Z a *-ti ORuss. jestf, Ved. bhdrti, GAv. vasti, Dor. iL&3T, etc. 'Ved. dpdt, GAv. mraot, Gk. 4, epe, etc. b *-t
FIRST PERSON PLURAL

The characteristic ending of this person seems to be *-me'-, *'-m6-, to which *-s(i) or *-n(i) is usually added, thus forming a conglomerate termination. The normal athematic type of the present active *-mi'-s(i), *'-m3-s(i) is represented by Ved. imdsi, stumdsi, dvifmds, braimds,GAv. imahi, YAv. garesmahi, OPers. amahy, Turf. ymas, Cret. O0AEs(subj.), Lat. sumus, OIr. ammi <*esmisi, the ending being *-mis(i), with *-me in Czech neseme; apparently *-mj's in OHGerm. beramis, tuomis; and *-mq' in OPruss. asmai, druwjmai, Lith. esme, si2kame,sukamie-s, OLett. esmi, Lett. essam, sitam, sitamig-s.30 The thematic type *'-mos(i) is seen in Ved. drcdmasi, hdramas, ydjamas(i), GAv. sanghdmah;,YAv. baramahi, OPers. Oahydmahy(with -d- instead of -a- through analogy with the first person singular, the older forms appearing in GAv. vajdayamahi, YAv. zbayamahi), Turf. kalkamas, Dor. pipojes <*"pipoos through analogy with the athematic type, Lat. ferimus (by the same analogy), OIr. bermi <*bheromosi; *'-mo appears in OIr. beram, Goth. sijum, qifam,3~ OCSlav. nesem'i, Stokav. jesmo, imdmo (Ukrain. im6), plet6mo <*-m6 (analogically changed from the thematic type); and *'-mo in Goth. qifaima (opt.). Another conglomerate, *-me'-n(i), *'-m6-n(i), seems to occur in Kanes. wahnumeni, aummeni, tiyaweni, lekkueni, nemaweni, paiweni (also This Gk. koihv, qEpo'Erv <*-iov. paiwani), harnikweni,dasgaweni,32 in the relation with in *-na to stand apophonic <*-n, *-n(i) appears second plural active present, imperfect, and imperative (e.g. Ved. pdthdna, ydtdna, a'yatana), and with *-ni <*-na in the first singular subj. active (e.g. Ved. brdvdni).
For OLett. esma, dialectic Lett. eima, see Endzelin ? 598. See E. Kiekers, Handbuchder vergleichendengotischen Grammatik 191, Munich, 1928. 32For ICanes. w:m see Delaporte, ?? 54, 56, 122 b, 213; Sturtevant, m after u or 9', in AJPh 50. 360-9 (1929). 'A Pre-Indo-European Change of u to ,lments
30

~'

ENDINGS OF PRESENT AND IMPERFECT

237

The corresponding present middle ending was *-medhai, *'-modhaifor athematic verbs and *'-o-modhai (ai : oi) for thematic, though in the historic period only the former occurs in Gk., as Ved. hZamdhe, &'smahe, ydjamahe (with -d- instead of -a- through analogy with the first singular active), GAv. aog'mada-6a, yazamaid8, YAv. staomaide, yazamaide; with *-medha in Gk. 77jEOa, SeEpb6o/Oa. These terminations seem best explicable as the grades SbRa and SbRb of a conglomerate *-m6i-dhei, *-mni-dhbiwhose second component may be seen in the second singular active of imperatives of the type of Ved. viddhi, pdhi (accent!),33 GAv. idi, YAv. ibi, OPers. idiy, Gk. IN. Kanes. shows esuwaita(ti), arwaita(t),
possibly comparable with Hom. (ppb6eLo'a.34

The imperfect active has *-me(s), *-mo(s), and *-men: (a) Ved. dpdma, Lith. bzivome; (b) Ved. GAv. d&m&,YAv. haxma, Dor. `ues, WKO1E~, the more original form YAv. bav&ma GAv. (also taurvayama, dbhardma, ta"rvayama), OPers. ataraydmah, Lat. -bamus, fudmus, OIr. -beram, OCSlav. mogomil; (c) Kanes. arnummen, elwen, Gk. $iLEv,~fppop1E. In the imperfect middle the endings are *-medhi, *-modha(a : o), as Ved. Skt. abhardmahi, GAv. dvaidi, Gk. /.EOa, C'epbo'ea (Kanes. dvrn~imahi, pariyawastati). The earliest inferable endings of this person would seem to have been: Thematic Athematic Active present -me'-(s) (i), -me'-n(i), '-o-m6-(s) (i), '-0-m6-n(i) '-m6-(s) (i), '-m6-n(i) -me-(s), -me-n " imperfect -o-mo-(s),-o-mo-n (-mo-(s), -mo-n?) Middle present -medhai, '-modhai '-o-modhai " imperfect -medhi, (-modhi?) -o-modha Tabulating these in the alternation-scheme of a termination-base *-myi, *-mai, the results are: N a *-m6i, *-mai (*bher(,)mai,*bher(o)mai) b *-m , *-m3 Goth. qipaima (opt.), OHGerm. beramis, Lith. esme, etc.
33For the repartition of Ved. -dhi and -hi see J. Bloch, in MSLP 23. 175-8 (1929). The impv. type of OLith. duodi, OCSlav. da'di scarcely belongs here (Meillet,

Slave? 360).

SHrozn:V161; Hornm. -AeaOais probably by analogy with -0Oe,etc. (K. Brugmann-A. Thumb, GriechischeGrammatik4 ? 423, Munich, 1913).

238

LOUISH. GRAY

S a *-mei, *-moi (*bher(e)rmei, *bher(o)moi) b *-me, *-mo Ved. imdsi, dvigmds, ydiamas(i), Cret. Ocles, OIr. bermi, beram,OCSlav. nesemil, Czech. neseme; Kanes. wahnumdni, Gk. okiivY, plpo1vE; Ved. himdhe, ydjamahe; Gk. Q'.eOa, p~pb~pOa;Ved. dpama, Dor. jies, VKO/es,Att. iLEV, z pojEv;Ved. dbhar&ma, Skt. abhar&mahi, Gk. 'iaIe'a, qEpb6pEOa R a *-mai dvr.?imahi, (*bhdr(e)mai) b *-ma (*bher(,)ma) I a *-mei, *-moi (*bhdr(e)mei, *bh6r(o)moi) b *-m,, *-mo (*bhdr(e)me, *b6hrr(o)mo) Z a *-mi (*bhdr(e)mi,*bher(o)mi) b *-m (*bhdr(e)m,*bher(o)m)
SECOND PERSON PLURAL

In the historic period the characteristic endings of this person seem to be *-te (often with conglomerate *-s(i) or *-n(i), as in the first person plural) for the active, and *-dhue(i), *-dhue-mfor the middle. The normal athematic type of the present active is represented by Kuch. trigcer <*-te-r,35 Gk. eorT, Lith. este, OCSlav. jeste; and the thematic by Kuch. yatmacer, Gk. (pper, Goth. bairib, Lith. s'ikate, Lett. sitat," OCSlav. berete. Indo-Iran., however, shows *-the, probably through analogy with the second singular imperfect middle (Ved. ddhatthds,etc.) and the same person perfect active (Ved. vbttha,etc.), as Ved. hathd, bhdrata, GAv. utad, is"aOa, YAv. dasta; and this may be extended in the present by *-na <*-n~, as in Ved. sthdna, ydthdna.37 The conglomerate *-te-si perhaps occurs in OIr. berthe if <*bertsi <*bhdretesi;3a*-te-s may be found in Lat. agitis <*agetes;" and *-the-s in Hindi, etc., calo <*carathas;4P while *-te-ni is present in Kanes.
36 L6vi-Meillet 10; for c (i.e. 6) <t before prepalatal vowels and y see Meillet, in JA 10. 17 (1911). 450. 16For further details see Endzelin ? 599; Van Wijk, Studien 61-2.
'7

development, not as an ancient survival (A. Thumb, Handbook of the Modern Greek Vernacular ? 214, 4, Edinburgh, 1912).
38

In Naxian Mod. Gk. one finds 9Oreve, K'obreve, but probably as a modern

31For various other theories see Stolz-Schmalz 304-5. Here, too, probably belongs the type of Arm. sirik' (H. Pedersen, in KZ 38 [1905].213). 40 For another view see J. Bloch, in Kuiatsava 481-3, Benares, Smdrakasarihgraha 1928.

Gray, in LANGUAGE 3. 80.

ENDINGS OF PRESENT AND IMPERFECT

239

peiteni, iyateni. In the perfect active, simple *-e is found, as Ved. vidd, cakrd, babhftvd. In the imperfect active, the ending is *-te, as Ved. d'ita, dbhavata,GAv. Lat. datd, va nata, YAv. pata, fasata, OPers. itd, Gk. E3-r7e, E~ppere, agite (impv.-injunct.), OIr. -berid, Lith. si~kote,OCSlav. molete; *-tana <*-te-ne in Ved. d'itana, dyatana; and *-ten <*-te-n in Kanes. iyaten, kappuwatten. The corresponding presentmiddle ending is *-dhyei for athematic verbs, and *-dhyui (ai:oi) for thematic, as Ved. augdhvy, bhdradhve, GAy. daduy ,4'YAv. caraOwe, besides, apparently, *-dhuain Hom. mer"ng'duy?, with with XEXoLbare, <*-6a -OE by analogy etc.42 A con&yp7-yopO~ glomerate *-s-dhua seems best to explain the type of Gk.;Ts, pppaEOwith -Be <*-Oa by analogy with ppETcr,etc. In the imperfect middle the endings are the conglomerates *-dhue-m for athematic verbs, and *-dhu,-m (a:o) for thematic, as Ved. drddhvam, djugadhvam, GAv. didum, syodfim, YAv. zambayaswam,Kanes. eitumm-at, pahhaldum-at (impvs.) (this -dum- may also be regarded as the grade ZZb of *-dheuem-; the Kanes. present also has yadduma, sarkaliyatumari). Gk. here shows the conglomerate *-s-dhua as in the thematic present, e.g. r0Go0E, etc., with -GE<*-Oa as above. 8i5LooGe, EapEpEcTe <*o~tooa, The earliest inferable endings of this person, from the historic evidence, would seem to have been: Athematic Thematic Active present -te-s(i), -te-n(i) -e-te-s(i), -,-te-n(i) -te-n(e) -,-te-n(,) " imperfect Middle present -(s)-dhyei -o-(s)-dhyX(i) (-(s)-dhyoi?) " imperfect -(s)-dhue-m -_-(s)-dhu,-m One must observe, however, that this person alone shows a distinction in consonant-characteristic for the active and middle-*-t- and *-dhu-and that the former is identical with the characteristic consonant of the third person singular, only the vocalism being unlike. Moreover, all the other persons show a gradation in vowels, whereas in the active of the second plural the grade *-te is constant throughout. It may be suggested, in the light of these facts, that the original characteristic
Cf. Bartholomae ? 268, 19. For other views see Brugmann-Thumb ? 417 note; J. Wackernagel, in Nachrichten der k6niglichen Gesellschaftder Wissenschaftenzu G6ttingen,philolog.-hist. Klasse, 1914. 98.
41 42

240

LOUISH. GRAY

here was *-dhu- for both voices, so that the earliest active conjugationforms were:
Athematic Thematic

Active present

"

imperfect

-dhy'-(s)(i), -dhye'-n(i), '-dhy'-(s)(i), '-dhyd- '-o-dhy6-(s)(i), '-o-dhu6-n(i) n(i) (Ved. *studhvds(i), Gk. (Ved. *bhdradhvas(i),Gk. etc.) *Ef0o,etc.) *"ppeO <*cppoOo, -dhue-s, -dhue-n -o-dhyo-s, -o-dhyo-n (-dhyo-s, -dhyo-n?) (Ved. *dstudhvas, Gk. (Ved. *dbharadhvas, Gk.
*tOE, etc.) *yppeOe < *6pepaOo, etc.)

Reducing the *-dhy-type (with the exception, of course, of the hypothetically restored active forms) to the alternation-scheme of a termination-base *-dhyei, *-dhyui (Z grades of *dheyui,*dheyui), the results are:43 N a *-dhu6i, *-dhuyi (*bherdhyui,*bhdrdhuyi) b *-dhy?, *-dhy3 (*bherdhVu', *bherdhyu) S a *-dhuei, *-dhuoi Ved. a:gdhvo, etc. b *-dhye, *-dhyo Ved. drddhvam Ved. bhdradhve,etc. R a *-dhuai b *-dhya Gk. acre, (-)TrlO'wee, (')peipeOE,Hom. [yprfyopOe <*-Oa(?) R a *-dhui, *-dhuoi (*bhdr(e)dhui, *bher(o)dhuoi) b *-dhye, *-dhyo *bher(o)dhuo) (*bher(e)dhue, Z a *-dhui (*bher(e)dhui,*bher(o)dhui) b *-dhu (*bh r(e)dhu, *bher(o)dhu)
THIRD PERSON PLURAL

This person is formed very much like the corresponding singular, except that the various grades of the termination-base *-tei, *-tei are preceded by -n- (apparently the grade Zb of the termination-element *-nai, *-nai already found in the first singular active [Ved. subjunct. brdvdni],first plural active [Kanes. wahnumeni, Gk. ?ppoyeJ], and second plural active both indicative present [Ved. sthdna, Kanes. iyateni] and as well as imperaimperfect [Ved. ditana, Kanes. iyaten, cf. Gk. ApfpoMEv], tive [Ved. yeatdna]). The normal type of the present active is represented by *-nti for athematic verbs and by *-o-nti (i : oi) for thematic, as Ved. didati,
3 All the historic forms in *-te fall in the grade Sb of the termination-base discussed under the third person singular. *-t~i, *-tWi,

ENDINGS OF PRESENT AND IMPERFECT

241

juihvati (< *'-Oti), badhndnti (< *-dnti [!]), GAv. dada'ti, henti, YAv. hanti, OPers. ha"tiy, Turf. sencd, Arm. en, Delph. vrit, Att. elo-i, Lat. sunt, OIr. it, OWel. hint, Goth. sind, OCSlav. dadotii (< *-ntl) for athematic verbs; and Ved. bhdranti,GAv. maranti, YAv. baranti,OPers. Att. ixovrLt, OLat. bara"tiy, Turf. kalkencd, Arm. beren, Lacon. Exovt, tremonti (?), Lat. ferunt, Umb. furfant (?), Gaul. cartaont (?), OIr. birit, Goth. bindand, OHGerm. bindant < *bindandi, Anglo-Sax. bindab OLith. gina,46 <*binddnbi (cf. Ved. badhndnti, above), ORuss. beruti,44 for thematic. In the imperfect active the ending *-nt for athematic verbs is represented by Ved. ddihan, GAv. jigar'zat, g"man, YAv. usan, OPers. aha", Lacon. fiov (aorist),46 OCSlav. dasa, (< *sFt; -s-aorist); and -o-nt for thematic by Ved. dbharan, GAv. mar~ndan, YAv. baran, OPers. Lat. -bant, OIr. -berat, OCSlav. pletan. abara",Gk. ypepov, The corresponding athematic present middle ending *-ntei, *'-ntoi is represented by Ved. vdsate, pundte, punate, GAv. s'avait,Yav. aojaite, Gk. 7r10vrat,Goth. haitanda (< *-nt5); and the thematic *-,-ntai (ai:oi) YAv. barante,Kuch. by Ved. bhdrante,GAv. visantW, (also .melem In the Gk. for the athematic yanem. ne),47 p~povrat. imperfect middle one finds Kanes. deantat, Gk. type *-nto: Ved. dmrjata, Gay. varatd, YAv. "r"uanta, er&levro three with the thematic (these by analogy type, cf. Ion. pluperf. for the and thematic perf. dnroME&xarat) ; ordhXaro, *-o-nto:Ved. dbharanta, GAv. dvdranta,YAv. yazanta, OPers. abara"td,Kanes. kls'anta(ti), Gk.
fpeporo.

Besides these forms, there are perhaps some made by a conglomerate of the termination-bases *-ntWi,*-ntai and *-sai, *-sai in the grades *-ntsi and *-nts (ZZZa and ZZZb respectively) in the present and imperfect active, the former in Apabhram. Prak. vattahi,, Kanes. appanzi, danzi,"1and the latter in Osc. fufans (impf.), Paelig. coisatens, Marruc. amatens, Volsc. sistiatiens, (perfects), Umb. dirsans (subjunct.), etc."
44OCSlav. rarely shows -tTbeside the much more usual -tui (e.g. so,tt, mogo,tj; Wiedemann 23-4). Cf. also such dialectic Lith. phrases as nira kas dara, 'one has nothing to do' S5 (lit., 'there is nothing that they do); Endzelin, in IF 33. 124-5 (1914). instead of *ddihat, etc., are 46 The forms ddihan, g'man, usan, dhan, and Mbov influenced by analogy with the corresponding thematic types. 41 L6vi-Meillet 6-7. 48 The Kanes. forms may, however, be explained as <-nti (cf. under the third singular). " Cf. Gray, in LANGUAGE 3. 82-4; H. Ehrlich (IF 11 [1900],299-307) derives the type from *-ens < *-enes (nom. pl. of an -en-stem) + (sent). The types of Skt.

242

LOUIS H. GRAY

The earliest inferable endings of this person would seem to have been:
Athematic Thematic

Active present -nti -o-nti -nt -o-nt " imperfect Middle present -ntei, '-ntoi -o-ntai -nto imperfect " -o-nto For their tabulation in alternation-scheme cf. the third singular.
FIRST PERSON DUAL50

The normal type of the present active is represented by the endings i.e. by conglomerates of the termination*-.y'-s(i), *'-Xy-s(i), *'-o-uo-s(i), + *-sai, *-sai. Only GAv. shows *-ui-si, *'-uo-si, as bases *-Xyi, *-upi have *-uy-s, *'-yo-s, as Ved. cdravas (with -d- instead all the rest usvah!; the first singular), Argive A&ycyts(i.e. *&ywFts). with -aof by analogy Goth. ga-leibas (< *-o-uz < *-o-Uiz < *-o-Vaz52), &t'ywjEv (Hesychius),51 Lith. Jsva, sizkava(reflex. s'ukavo-s),53 OCSlav. jesve, vedeve' (with -e either or first dual the < I-E *-.us),54 OSCr. sva, pronoun v' by analogy with OPol. jesua. In the imperfect active the endings are *-ue, *-u6, as Ved. hdnava, cdrava (subjunctives), GAv. ahvd, YAv. Jvdva, Goth. siju, magu all < *'-?o; Goth. (< *-uy, *'-uuo), Run. varitu < *writuwe,55 *-ud, or sitaiwa (opt.) < *'-Xu *-y'.
abharann iha, abhara,s tatra (the second not Ved.), Mod. Gk. 58vovvmust be rejected (see respectively Bartholomae, in KZ 29. 514-8; A. Macdonell, Vedic Grammar, ?77, Strasbourg, 1910; J. Wackernagel, Altindische Grammatik 1. ??279-80, Gittingen, 1896; A. Thumb, Handbuchdes Sanskrit ??178, 180, Heidelberg, 1905; H. Oldenberg, Hymnen des Rigveda 1. 424-34, Berlin, 1888; and A. Thumb, GriechischeSprache im Zeitalter des Hellenismus 197-8, Strasbourg, 1901; der griechischenSprache247, Leipzig, zur Geschichte K. Dieterich, Untersuchungen der byzantinischenChroniken209, Gjttingen, 1913. 1898;S. B. Psaltes, Grammatik 60See A. Cuny, Le Nombreduel en grec 56-64, 84, 108-17, 179-86, 286-99, 371-2, 390-1, 419-20, 455, 462-3, 470, 474-5, 482, 496, Paris, 1906.
R6ib.
52

58, 474. Kiekers 191.

53Memel dirbau, sec. dual dIrbatau, is apparently by analogy with standard Lith. first sing. impf. dirbau. 5 For the Slav. dual in general see Meillet, Slave ??351-3;F. Miklosich, Vergleichende Grammatikder slavischen Sprachen 32. 64-7, Vienna, 1876; W. Vondrik, Vergleichendeslavische Grammatik22. 115-6, G6ttingen, 1928. 15E. Noreen, Geschichteder nordischen Sprachena?251 d, Strasbourg, 1913; A. der urnordischenRuneninschriften65, 68, 93, Heidelberg, J6hannesson, Grammatik 1923, however, regards it as first sing. pres. act. OSax. wita 'allons' may not be a verb-form (F. Holthausen, AltsdchsischesElementarbuch ??328,1, 395, 4083,Heidel1899). berg,

ENDINGS OF PRESENT AND IMPERFECT

243

In the present middle the endings are *-u~dhai, *'-uodhai, *'-u-yodhai (9i : oi), as Ved. dddvahe,sdcavahe; and in the imperfect middle *-uedhi, *-yodhi, *-o-yodha, as Ved. gdnvahi (root-aorist), sacevahi (opt., A-V), GAv. dvaidi < *duuadhi.56 The earliest inferable endings of this person would seem to have been:
Athematic Thematic '--ut6-(s) (i) -o-Ui-(s)

Active present -y'-(s) (i), '-Y6-(s) (i) -y-(s) " imperfect Middle present -iydhai, '-uodhai imperfect -yedhi, -yodhi For their tabulation in alternation-scheme cf. the substitution of V for m.
SECOND PERSON DUAL

'-o-yodhai -o-uodha first plural, with the

The normal ending in the present active seems to have been *'-to-s, *-.-to-s, (for which Ved.-Iran. affords no evidence here-has, as in the corresponding plural, -th-), as Ved. hathds (< *-tes), bhdrathas, Goth. Lith. ista, sz~kata(reflex. szikato-s), OCSlav. jesta, vedeta;and in bi-gats,57 the imperfect, *-to-m, -o-to-m, as Ved. a'stam, dbhavatam, Gk. Jorov, < *ij poTov.68 In the present middle one finds *'-toi, *-0-tai wpipETo (ai : oi; Ved. -th- instead of -t-), as Ved. d'sathe, vdrethe,YAv. arbiOe (second dual used as third); and in the imperfect,*-t3-m, *-o-t3-m,as Ved. mimatham (impv.), drabhethdm.59 On the evidence of Balto-Slav., which alone shows a real distinction
zu Homer ings, instead of -/EO0,, see J. Wackernagel, Sprachliche Untersuchungen 55, Gittingen, 1916).-eoov--if it really existed (it is recorded only as a variant for -peOain Iliad 23. 485; Soph. Elec. 950; Phil. 1079)-was formed from -JAeOa on the analogy of the second dual -aooz (Brugmann-Thumb ? 426). Grammarians record Aeol. etc. qp6eOeiV, 65For the possible development from P-Teut. *-egis < *-etes (more probably der *-e8as <*-etos) see Kiekers 191; R. Bethge, in F. Dieter, Laut- und Formenlehre altgermanischen Dialekte ?210, e, Leipzig, 1900; F. Kluge, Urgermanisch' ?198, Strasbourg, 1913. 58The present type of Gk. must be analogical with the impf., since the iyateni, Ved. stdna).
T7rT true present would probably be *a7rovL or *garvo (cf. present second plur. Kanes.

are plurals used as duals (for these readXXeXElpeOa 66 Gk. repL&~6Oea, 6ppuOceOa,

59 Gk. (k)Xp?peOov seems to be a contamination

of (k)>pipeaoe and (i)pSperou

(Brugmann-Thumb ?426).

244

LOUIS H. GRAY

between the second and third dual, the earliest inferable endings of the second person would seem to have been: Athematic Thematic '-to-s Active present -o-to-s -to-m -o-to-m " imperfect Middle present '-toi -o-tai imperfect " -t5-m -o-t-m For their tabulation in alternation-scheme cf. the third singular.
THIRD PERSON DUAL

The normal ending in the present active seems to have been *-t9-s, as Ved. hatds, bhdvatas (< *'-to-s), YAv. mrftW, OCSlav. jeste, jesta, and in the imperfect, *-tam, as Ved. a'stdm, dbhavatam, vedete, vedeta;60 El. XEOLTrY, YAv. ditam, Jasatdm, OPers. ajivat(?)am,61Boeot. bVEO~iav,

Att.ITarfl,i(pEp~lV.62

In the present middle one finds *-tai < *-tei and *-.-tai, as Ved. st`vdte, myasaitm,YAv. caraite; and in the imbruva'te, vddete, GAv. var'nvaitM, as Ved. perfect, *-tam, duhdatm, (impv.), drejetdm, GAv. asrvdtam, is apparently a contaminaAtt. eppoefrlv apar'sastam (Dor. KTrLOaoeOav, tion of e'opeo~aand peper'av). The earliest inferable endings of this person would seem to have been:
Athematic Thematic

Active present -t -s -,-t0-s -tam " imperfect -,-tdm -tei Middle present -,-tai -tam " imperfect -,-tam Summarising the results obtained from study of each person separately, one may picture the inflexion of the present and imperfect active and middle of the athematic base *ed- 'eat' and of the thematic *bhere-'bear' somewhat as follows:
60For the distribution see Wiedemann 25-7. 61 Foundation-Charter 14 (V. Scheil, Inscriptions des Achiminides 4 Suse 16, 25, Paris, 1929). 62 One may have here an instance of the alternation a : o (Hirt, Grammatik2.
?191, Heidelberg, 1921). Such rare Gk. forms as sec. dual impf. eE77I), EtXTr1Y, third dual impf. eir1bxerov, UbtCKETOV, are probably due to confusion, especially as

the dual was already moribund in Gk. (cf. Brugmann-Thumb ?419).

ENDINGS OF PRESENT AND IMPERFECT

245

Act. pres. 1 s. gdmi bhra5 2 " idsi bhereis 3 i " dti bhereit 1 p. edme'si, edme'ni bherom6ni bhdrombsi, idmoni ?) (ddmbsi, 2 " eddhug'si,eddhyM'ni bherodhudsi, bhdrodhMuni ?) (eddhu6si, iddhyodni 3 " bhdronti dnti 1 d. edu/'si,yduosi bheroyosi 2 " idtos bhdrotos 3 " ,dtes bher,tes <bh,r,tis Act. Impf. 1 s. ?'dip <e-6ddp bhront <e-bhdrom 2 " e'ds <e-eds ebh,res <e-bhdres 31 ebh.ert <e-bhgret " 'dt <e-edt 1 p. ?'dmes, ?'dmen <e-edmis, ybhromon ebhromos, e-edmnen (e-edmos, <e-bheromos,e-bhgromon e-edmon?) 2 " 'ddhyes, 'ddhyen y bhrodhmos, bhrodhyon < e-bhe'rodhuos, <e-eddhyes, e-.ddhyin e-bherodhuon (e-eddhuos, e-eddhyon ?) 3 " ?'dnt <e-ednt 8bhront <e-bheront 1 d. Z'dXus<e-edyu's, e-ldy6s bheroius <e-bheroy6s 2 " ?'dtom <e-edtom ebh.rotom<e-bherotom 31 " 'dtam <e-edtdm (e-,dtd'm?) ebh,r.tdm <e-bhdr,tdm (ebhretd'm?) bherai (ai : oi) Mid. pres. 1 s. ,dei, doi 2 " ,dsei, idsoi bherosai(ai : oi) 3 " edtei, idtoi bherotai(ai : oi) 1 p. edmedhai,idmodhai bheromodhai (ai : oi) 2 " eddhuei (eddhuoi?) : oi) (ai bherodhiui bher.ntai (ai : oi) 3 " edntei, dntoi 1 d. eduedhai, dduodhai (ai :oi) bherouodhai (ai : oi) 2 " idtoi bherotai(ai : oi) 3 " edtei bhdretai(ai : ei) Mid. impf. 1 s. ?di <e- di <e-bhe'roa(a : o) ebheroa 2 " ?'dso <e-&dso <e-bheroso dbheroso <e-bh rethes ?'dthes< e-Adth's dbherethes <e-bhereth's 3 " ?'dto <e-edto ebh,roto <e-bheroto

246 1 p. 2" 3 " 1 d. 1 2 " 3"

LOUISH. GRAY 'dmedhi <e-edmidhi (e-edmodhi?) i'ddhAe-m <e-eddhy,-m (e-eddhuo-m?) ?'dnto <e-drdto ?'duedhi <e-edyudhi, e-edyodhi 'dt'm <e-edtlm 'dt&m <e-edt&m (e-,dtd'm?) ebheromodhe <e-bheromodha (a : o) bheredh,-m<e-bherdhy,-m (e : o) bhronto <e-bh'ronto bhroyodhy <e-bh royodha (a : o) <e-bh rot m bherotam bheret&m <e-bh retrm (e-bhrrt 'm?)

Besides the simplest characteristics of the personal endings inferred in the foregoing pages, there are two which do not fall wholly within this category: *nji, *n7i in 4%b(*na) in the first sing. subjunct. act. (Ved. bradvdi), I~b (*n,) in the second plur. pres. and impf. act. (Ved. vddathana, d'itana) and impv. (Ved. yatdna), Za (*ni) in the first and second plur. pres. act. (Kanes. wahnum&ni, iyateni), and Zb (*n) in the second plur. impf. act. (Kanes. iyaten; cf. also Kuch. third plur. pres. act. ne); and *dhsi, *dhai in Ra and Rb (*dhai, *dha) in the first melem, and Za (*dhi) in the plur. pres. mid. (Ved. hamdhe, Gk. qp~p6pEOa), second sing. impv. act. (Ved. viddhi, pihi). The force of these two bases is not fully apparent, but their distribution in verb-terminations suggests that *nai, *ndi was an intensifying (and usually pluralizing) particle, while *dhei, *dh3i was another characteristic of the second sing., perhaps with a middle force as distinguished from the active s. One would have, then, the following distribution of characteristic terminations: Dual Plural Singular 1 m, vowel (diphthong) m V t 2 s, th, dh (?), diphthong t, dhA t nt 3 t, diphthong (With the t of the second and third dual cf. the t of the second plur. and the t of the third sing.) The various conglomerates considered above would seem, in the light of these characteristics, to have the following distribution and raisons d'gtre: 2 sing.: *-ti-sei, *-ei-s (Na[Sa]Sa, SaZb; OPruss. saggesei, OCSlav. bereli, Gk. (pipet), 'double second sing.'; *-ths-s (NbZb; Ved. ddhatthas), 'double second sing.';

ENDINGS OF PRESENT AND IMPERFECT

247

3 sing.: *-ei-t (SaZb; Gk. op~p <*4ppEtT), 'double third sing.'; 1 plur.: *-m-ni, *-me-ni, *-me-n (NbZa, SbZa, SbZb; Kanes. wahnum&ni,aummeni, Gk. kowiv, /ufE),'we indeed' (' we ourselves'?); *-me-s, *-me-si, *-me-s (NbZb, SbZa SbZb; OHGerm. berames, Ved. ydjamas(i), Dor. ?pfpoesg, 'ET), 'we and thou (active)'; *-me-dhai, *-me-dha,*-me-dhi (SbRa, SbRb, SbZa; Ved. Gk. eu/0a, Ved. dvrimahi), 'we and thou (middle)'; hfZmdhe, 2 plur.: *-te-n., *-te-ni, *-te-n (SbPlb, SbZa, SbZb; Ved. d'itana, Kanes.

iyateni, iyaten), 'ye indeed' ('ye yourselves'?); agitis <*agetes[?]), 'ye and thou especially';

*-te-si, *-te-s (SbZa, SbZb; OIr. berthe <*bhWretesi[?], Lat. 'thou and ye especially'; *-s-dh~a (ZblFb; Gk. ii4E, TrOlerOEa), Ved. drtdhvam, djugadh*-dhye-m, *-dhq,-m (SbZb, I4bZb;

vam), 'ye and we';


3 plur.: *-ntei, *-nto, *-nti, *-nt (ZbSa, ZbSb, ZbZa, ZbZb; Ved. bhdrante,dbharanta, bhdranti, Lat. fer?-bant), 'he intensified' (i.e. pluralised) = 'they';

Kanes. appanzi[?],Paelig. *-ntsi,*-nts(?) (ZbZbZa, ZbZbZb; two, of whom thou (active) art one';
coisatens), 'he intensified (i.e. pluralised) and thou' = 'they'; 1 du.: *-ye-si, *ye-s (SbZa, SbZb; GAv. usvahi, Ved. cdravas), 'we *-ye-dhai, *-*e-dha, *-ye-dhi (SbRa, SbRb, SbZa; Ved. ddd-

vahe,sacevahi,gdnvahi),'we two, of whom thou (middle)art one'; 2 du.: *-to-s (SbZb; Ved. hathds),'ye two, of whom thou (active) art one';
*-td-m, *-to-m (NbZb, SbZb; Ved. drabhetham, dbhavatam),

'ye two, of whom I am one'; 3 du.: *-te-a(SbZb;Ved. hatds), 'they two, of whom thou (active) art one';
Ved. dr jetdm, *-ta-m, *-to-m (NbZb, SbZb; Boeot. Av~E~7tw, GAv. asrvdtam,YAv. jasatam, 'they two, of whom I am one'.6

The simple forms without conglomerates,i.e. those expressingdual or plural alone with neither inclusive nor exclusive connotations,are seen in the followingtypes: 1 plur.: OPruss.asmai, Lith. Jsme;Czech
6* One may suspect that this m appears also in the impv.: 2 sing. act. Gk. (aor.) 3 du. act. Ved. XV^ov, Dor. etbov; 2 du. act. Ved. pdtdm, jdyatam, Gk. Irov, ppe~roz,; pata'm, jdyatdm; Gk. rcw, peprTco; 3 plur. act. Gk. ea7rv, Pep6vrwo, Lesb. Treixovrov;

3 sing. mid. Ved. duhd'm, GAv. u6a,; Ved. vdrdhatdm, YAv. vardzyata,, OPers. 3 du. mid. Ved. varnavatdm; 2 du. mid. Ved. vdrdhethdm, Gk. IoGo, pepeaOov; jihatdm, Gk. iiawv, ?ppaeOo'; 3 plur. mid. Ved. pdvantdm, GAv. x'ntam, Gk. iG0OPv, Lesb. pOepe,v, brtXeeov.

248

LOUIS H. GRAY

neseme; OIr. beram; Goth. qi]aima; Ved. dpdma, dbhardma;2 plur.: Gk. d'sethe; 3 du.: OCSlav. jeste; Ved. siTvdte.

1 du.: Lith. hsva,Ved. hdndva; Goth. sitaiwa; 2 du.: Lith. Ista; Ved.

The exclusiveforms seem to be: 1 plur.: Gk. pckpo;iev; 2 plur.: Ved.

d'itana; and the inclusive, 1 plur.: Ved. ydjamas(i), hfimdhe; 2 plur.: OIr. berthe (?) Gk. 0ia00,Ved. drddhvam; 1 du.: GAv. usvahi, Ved. dddvahe;2 du.: Ved. hathdis, dbhavatam;3 du.: Ved. hatds, drabhethdm, Boeot. Ave0rTav, GAv. asrvatam. sian, Tibeto-Burman, MunddT, Dravidian, North-Caucasian, etc.;""but

The presenceof exclusive and inclusive forms in the first plur. (and and in verbsis frequent in American,Melanedu.) of personalpronouns

its existencein Indo-European seemsnot to have beensuggestedhitherto. Finally, the personalending characterized by r, to which much study has been devoted,"5 calls for some attention. The distributionof forms is as follows: Vedic: second and third du. perf. ind. act.; third plur. impf., aor., perf., pluperf.ind. act., and pres., perf. opt. act.; third plur. pres. and perf. ind. mid.; Iran.: third du. ind. perf. act.; third plur. aor., perf. ind. act.; third plur. pres. and aor. opt. act.; third plur. pres. and perf. ind. mid., and impv. mid.; Kanes: first and third sing. and second and third plur. pres. ind.;
64 F. Boas, Handbookof American Indian Languages 1. 527, 529, 580-1, 817, 820, 851, 908, 914, etc.; 2. 321, 351, 395, etc., 467-8, 470, 473-4, 575-6, 578, Washington, 1911-22; R. H. Codrington, The Melanesian Languages 111-4, 118-9, 122-3, 171-2, 413, 425, 453, 496, 502, 519, 530, 543, 550, Oxford, 1885; S. H. Ray, Comparative Study of the Melanesian Island Languages 77, 88-9, 98, 115, 148-9, 157, 159-60, 179, 181, 188, etc., Cambridge, 1926; Linguistic Survey of India 3. 1. 36, 289-92, 319, 329-33, 376, 386-7, 433, 435-6, 462, 472; 3. 2. 71, 266, 385; 4. 42, 170, 193-4, 196, 198-9, 293, 312, 356, 414, 418, 452, 460, 462-3, 481, 547, 588, Calcutta, 1909, 1903, 1906;A. Dirr, Einfiihrung in das Studium der kaukasischen Sprachen 166, 188, 209, 270, 281, Leipzig, 1928; in general, W. Schmidt, Sprachfamilien und Sprachenkreise der Erde 327-34, 505-6, 518, 520, Heidelberg, 1926. 65All previous studies are fully discussed by A. W. M. Od6, De uitgangen met R van het deponens en het passivum in de indoeuropeeschetalen, Haarlem, 1924; see also A. Meillet, 'Sur les d6sinences en -r', in BSLP 24. 189-94 (1924), and Esquisse d'une histoire de la langue latine 25-7, Paris, 1928;Edith F. Claflin, 'The Hypothesis of the Italo-Celtic Impersonal Passive in -r', in LANGUAGE 5. 232-50 (1929); Hirt, Grammatik 4. ?71. Both personal and impersonal endings in -r are found in Checheno-Lesghian (Prince N. Troubetzkoy, in BSLP 29 [1929].160, 164-5).

Kuch.: throughout, so far as forms available go;

ENDINGS OF PRESENT AND IMPERFECT

249

third plur. impf. ind.; first sing. voluntative;third sing. and plur.impv. (all mid.); Phryg.: third sing. impf. (in view of the paucity of material,no conclusion as to distributionis warranted); Osco-Umb.:thirdsing. and plur. pres.ind. subj.; third sing. fut. perf. ind., impf. and perf. subj.; fut. impv.; Lat.: throughout in non-periphrasticmid. formations, except in second plur.; third plur. perf. ind. act.; OIr.: all mid. formsof pres. and pret. ind. and subj.; and simplefut. (except,as in Lat., in the secondplur.). Of all the grades,Zb (r) is commonest:Ved.ihdthur,a6 ihdtur (YAv.
vaodatar"), djuhavur, dbhaigur (GAv. dare, YAv. alkar"), cakrt'r (GAy. aiharj), dcucyavur, syir (YAv. hydr"),vavrjyz'r;Kuch. jamantr, Turf. kljamtr, Kanes. es'er,ddir, Phryg. a&daxerop (sing.),67 Arm. berer (impf. ind. act.), Lat. feruntur, Umb. terkantur, OIr. labritir, bertir).

The grade Za (ri) is found in the Kanes. third plur. perf. ind. act. the grade Sa (rei, 'roi) in the third plur. pres. and perf. ind. (es'antari);
mid. (Ved. duhre,Ure, YAv. saire; Ved. cikitre, YAv. caxrare); the grade Sb (rd, 'ro) in the third plur. perf. ind. act. (Kuch. weiadre,Lat. fecdre, OLat. dederi,and perhaps in OLat. ded(e)ro),68and once in the third plur. impf. mid. (Ved. aduhra [Maitr. Saihh.]).

in which r Besides the simpler, one finds a numberof conglomerates is the prior member; where, in other words, it seems to be an infix. These conglomerate terminations,which appearonly in the third plur., are as follows: in the pres. impv. mid.: Ved. duhratdm *-r-dtam (AV); in the pres.ind. mid.: Ved. duhrate, serate(AV., Tait. Sarih.); *-r-ntei
*-r-nto in the pres. opt. mid.: Ved. bharerata; *-r-ontoin the pluperf. ind. mid.: Ved. dvavrtranta; *-r-ont in the perf. ind. act.: OLat. dederont,coiraveront,Lat. fecdrunt, institerunt (Ital. f cIro, faciron[o]);9 and the impf. and aor. ind. mid., and pres. opt. mid.: Ved. d.eran, dyujran, dadiran;
68 See, most recently, C. Bartholomae, Zur Kenntnis der mitteliranischen Mundarten,6. 75, Heidelberg, 1925. 67Cf. W. M. Calder, 'The Medial Verbal -r Termination in Phrygian', in Jour-

nal of theManchester Society10. 25-33(1923). EgyptianandOriental


68

Laut-undFormenF. Sommer, Handbuch derlateinischen guage 531-2,Oxford,1894;


lehre2 277, Heidelberg, 1914. Kanes. shows an imperatival -u in the first sing. voluntative mid. (arhaharu;cf. Ved. mdndatu,mdndantu, etc.). 89 Cf. Lindsay and Stolz-Schmalz locc. citt.; Sommer 578-80.

Against this view, Stolz-Schmalz 305, 338; W. M. Lindsay, The Latin Lan-

250

LOUIS H. GRAY

*-r-amin the aor. and pluperf. ind. mid.: Ved. ddtbram,YAv. vaoziram; *-r-dm in the impv. mid.: Ved. duhrdm (AV.); *-r-s in the perf. ind. and aor. opt. act.: GAv. ikaitar'), YAv. sa*-ra-rei (reduplication) in the perf. ind. mid.: Ved. jagrbhrir6 (cf. also the OIr. reduplicated forms of the pret. ind., -ldmratur, -midratar,
-ferartar).70

Of all these types, the most important seems to be that in which r is added, not to another personal ending (or infixed before it), but directly to the verbal base. This is found at the two extremes of the Indo-European area: Indo-Iranian and Kanesian, and Italo-Celtic. Here belong the types of Ved. third plur. ind. mid. duhr6, ire, sunvirg, and perf. ind. act. cakricr, jagmuir, vid'ir,7"YAv. third plur. pres. ind. mid. saire, third plur. perf. ind. act. GAv. adxnar5,YAv. ddaar', and mid. mrava~re, aor. mid. GAv. dare, YAv. alkar';72 Kanes. third plur. perf. and caxrare, act. ind. eger, ddir; Osc. third sing. pres. ind. and perf. subj. loutir, sakrafir; Umb. third sing. pres. and perf. subj. ferar, ier;73 OIr. third sing. pres. ind. pass. berir, pret.-pres. depon. -fitir (Wel. gwyr, Corn. gor, Bret. goar, the phonological equivalent of the Ved. third plur. perf. ind. act. viduir),pret. ind. depon. -comnocuir, -ddmair, -genair, -ldmair, -minair, -midair, -siasair, and third plur. pres. subj. depon. -clorat.74 Taking the evidence as a whole, the writer is at present strongly inclined to feel that the -r-endings were originally neither active nor passive, and neither singular, dual, nor plural, but that they were primarily merely impersonal terminations of the third person attached directly to the verbal stem. All differentiation into singular and plural (Ved. viditr : OIr. -fitir) arose from the equivalence of 'one says' and 'they say' (also 'it is said'); and the extension to other persons than the third, as well as the use of r to indicate the real passive, seems to be of secondary development. The original distinction between the impersonal and the deponent (e.g. Umb. ferar : Lat. fertur; OIr. berir, -berar : labrithir, -labrathar)would appear to have lain in the fact that in the former r was added directly to the verb-stem, in the latter to the inflected active. In Ved. the passive is used, very rarely, in an impersonal sense, as
70Pedersen 2. 386; cf. Thurneysen ?699. 71 Full collection of material in Macdonell ??451,470, 485 (pp. 337, 347, 358). 72 Bartholomae, GirPh 1. ?121. of Oscan and Umbrian2??238,2, 239, Boston, 1928. 73C. D. Buck, Grammar 74Pedersen 2. 319, 385, 386, 388, 405-6; cf. A. Walde, Ueber dlteste sprachliche Beziehungen zwischen Kelten und Italikern 16 sqq., Innsbruck, 1917.

ENDINGS OF PRESENT AND IMPERFECT

251

prd jVdyate 'one finds' (Maitr. Safihh.), sdm amyate 'one binds oneself' (Taitt. Sarhh.), though it becomes extremely common in Skt.; and the 'one runs same construction is occasionally found in Gk., as KLvwvvErerat made had 'one preparation' risk', 7rapeaKEibaoro (Thucyd.).76 The most archaic system seems to have been preserved only in ItaloCelt., for the type of Ved. vidfir = OIr. -fitir 'on sait' had already come to be felt as a plur. in the prehistoric period. Thus one finds in Osc. Itiviass messimass... sakrafir 'Iovias medioximas... sacrato'; in Umb. pone esonomeferar 'cum in sacrificium feratur', nosue ier ehe esu poplu 'nisi itum sit ex hoc populo';76 in OLat. errat animus...vitam vivitur 'the mind wanders... one leads a life', itur in antiquam silvam 'one goes', amatur atque egetur acriter 'there's love and bitter want' parallel with neque ament nec faciant bene (Plautus, Pseud. 272-3); and in Late Lat., asfaciatur, si tibi videtur, et triclinia 'let one make' (Petron. 71).77 This seems to recur in Vulg. Lat. of the Merovingian period, as reliquas reddebucionesexigebatur 'one demanded the other taxes', and the rule of the Doctrinale of Alexander de Villa Dei (1199), Mathaeum legitur, psalmos erat ante legendum'one reads.., .one should have read'.78 It is also noteworthy that this construction appears to survive in Ital. and (rarely) in OSpan., as OItal. trouaualise stormi e batallie 'one found storms and battles', Mod. Ital. si visito insieme le collezioni 'one visited the collections together', OSpan. se lo cerc6 de sdlidas murallas 'one surrounded it with solid walls.'79 Turning to OIr., such passages may be cited as niimirchdi nachggin (read, nimirchdi nach n-cein) libsi 'one will not delay me long with you (Wb. 7a, 11; Stokes-Strachan, 'I shall not be delayed'), arnach naurchoissed som 'that one might not hinder him' (Wb. 8a, 4; Stokes-Strachan, 'that he might not be hindered'); and here belong forms like tiagar hudin 'ito de nobis, let one of us go' (Ml. 16c, 5), rigthir cuccuib 'ibitur ad vos, one will go to you' (Wb. 9a, 23), Mid. Ir. docuiasiarsin...ocus tuccad Etain immach 'itum est post eos. . .et E. foras educta est, one went after them,' etc. (Tochmarc Etdine 20). From such constructions are 76 Delbriick, VergleichendeSyntax der indogermanischenSprachen3. ?16, Strasbourg, 1900; J. S. Speyer, Vedischeund Sanskrit-Syntax ?245,do., 1896.
76Buck ?239.

7 Lindsay 520-1, 523; Stolz-Schmalz 623. are given by Jeanne Vieillard, Le Latin des dipl6mes royaux et chartes privees de l' poque mirovingienne 220-2, Paris, 1927. Against the view advanced in the text, see E. L6fstedt, Philologischer Kommentarzur Peregrinatio Aetheriae 291-3, 319-20, Upsala, 1911; Stolz-Schmalz 596. 7' W. Meyer-Liibke, Grammairedes langues romanes 3. ?94, Paris, 1900.
78 Examples

252

LOUIS H. GRAY

formed all OIr. persons of the passive except the third, e.g. 1 sing. ni-m-tharberar 'I am not brought' (lit., 'one brings me not'), 2 sing. arnach-it-r-indarpither 'lest thou be cast away', 1 plur. ro-n-mess ni 'aestimati sumus', 2 plur. ni-b-iccfither'ye will not be saved'.80 One of the main sources of the entire passive voice, which, as is well known, was not Indo-European, must, then, have been the impersonal third, with analogical extensions to the other two persons and to all three numbers.8'
so Pedersen 2. ??589,3; 624, 2. s1 Even the Indo-Iran. third sing. aorist 'passive' in -i (Ved dkeri, GAv. srdvi, OPers. addriy) was probably originally middle in force (Delbriick, Verbum 181-2; Brugmann, 2. 3. ? 622, 2).

You might also like