Avestan vaēsmәnda

_...1_­

MICHIEL DE

V

AAN

Avestan

vaesmenda

1.

In Proto-Indo-European, a separate inflexional ending for the ablative singular existed only in the pronouns and in the a-stem nouns. According to BEEKES 1995: 192, the a-stem ablative singular ending *-od may derive from a contraction of *-a-ed, i.e. of the thematic vowel plus an ablative singular ending *-ed; this would match his reconstruction (p. 209) of the ablative of the pronouns as *hJmed 'from me', *tued 'from you', *nsmed 'from us' and *usmed 'from you', in which *-ed is the ablative ending. In the other noun classes, the function of the ablative singular was usually performed by the genitive singular.

In Indo-Iranian, this distribution was preserved more or less unchanged, since Sanskrit displays a separate abl.sg. ending -tit in the a-stems (e.g. devdt),

and in the pronominal forms mat, ivai, asmat, yusmai, etc., but still has no difference between the gen.sg. and the abl.sg. forms in the other noun classes. In Avestan, this situation has been preserved in Old Avestan (REICHELT 1909: 245, BEEKES 1988: 110, 137ff.), but in Young Avestan the ending *-(a)t has spread beyond its original domain. Every noun class in YAv. distinguishes bet­ ween the gen.sg. form and the abl.sg. form, and the way in which the abl.sg. ending is formed can easily be described: the abl.sg. is built on the gen.sg. form, by means of a replacement of final *-s by *-t. All of the noun classes show this process: a-stems gen.sg. -auti

<

*-ajas, abl.sg. -aUa!

<

*-ajat; {-stems gen.sg. -uti

<

*-iiis, abl.sg. -iidt

<

*-iat; u-stems gen.sg. tanuuo

<

~

-

~

"tanuas, abl.sg. tanuuai

<

"tanuat; z-stems gen.sg. -ois

<

"-ais, abl.sg. -oil

<

*-ait; u-stems gen.sg. -aos

<

*-aus, abl.sg. -ao!

<

*-aut; root nouns and stems in -t, -d, -nt, -ar, -ah, -uudhl-us, -is,

-uS

gen.sg. -0

<

*-as, abl.sg. -a!

<

*-at; ar-stems e.g. gen.sg. ndrJs

<

*nrs, abl.sg. ndrJ!

<

"nrt, etc. (cf. HOFFMANN-FoRSSMAN 1996: 117-157).

In the n-stems, the same procedure applies. The two variants of the gen.sg. ending *-n-as and *-an-s correspond to the two attested variants of the abl.sg. ending *-n-at and *-an-t, for example gen.sg. yuno and abI.sg. yuna!

Munchener Studien zur Sprachwissenschaft, 61, 2001, S. 185 - 192

i

!

111._

MICHIEL DE VAAN

186

<

*-n-as, *-n-at

on the one hand, gen.sg. ddmqn and abl.sg. barssmsn

<

*-an-s, *-an-t

on the other. In fact, the latter form barssmsn

<

*barsmant

(to the stem barssman- 'bundle of straw used in the ritual'

<

*bar/man-)

is the

only example known so far of the abl.sg. ending

*-ant

in YAv. The develop­

ment of

*-ant

to

-an

is regular in YAv. and can also be found in the secondary 3pl. active verbal ending

-sn

<

*-ant.

The loss of

*-t

in

*-nt

was most likely posterior to the Indo-Iranian period, since the formation of the ab1.sg. "bars­

mant

(>

barssmsni

was a specific Old Iranian development.

Below, I shall argue that a second

n-stem

abl.sg. form

*-ant

is attested in

Yt 10.86 vaesmsnda, which has retained final

*-t

because of the close connec­

tion with the following postposition. This interpretation of vaesmsnda involves

a renewed analysis of the text of Yt 10.86.

2.

Yt 10.86 forms part of the 220d karde of Yast 10 (Yt 10.83-87), which

mentions various persons who invoke the help of Mithra, viz. a dainhupaiti-,

a zantupaiti-, a vtspaiti-, a nmdno.paiti- and a

driyu-.

All of these are used in

the nominative singular in the construction

miirrsm ... yim daitjhupaitis

baoa

usuinazasui rbaiieiti auuainhe 'Mithra, whom the D. calls for help with out­ stretched hands'. The passage relevant to our question starts with the last words

of Yt 10.85 (in the translation of GERSHEVITCH 1959: 115; in 10.86a I read

azsmna with KELLENS 1984: 47):

Yt 10.85

e

ya!

gaoscit.

Yt 10.86

a

yii

varna azsmna

b

biioa

ustanazasto zbaiieui

auuainne

gauuaiirim paitismarsmna

d

kaoa no

arsa gauuaicim e

apaiiat paskdt vazsmno

f

midrd yo vouru. gaoiiaoitis

'as (that) of the cow, too, (does),

who, being led away captive, calls at times for assistance with

outstretched

hands, longing for the herd: "When will grass-land magnate Mithra, the

hero, driving from behind,

make us reach the herd,

Avestan

vaesmonda

187

g

kaoa no fraouruuaesaiiaiti

h

asahe paiti pa1Jt(Jm

drujo vaesmsnda azsmnam

when will he divert to the path of Truth us

who are (lit. is) being driven to the estate of

Falsehood? "

GERSHEVITCH takes

gaoscit

at face value as a gen.sg. form of gdu- 'cow',

which could have been used attributively to a preceding noun 'voice' in Yt

10.85. Yet I follow LOMMEL 1927: 12, who regards

ya! gaoscit.

as the first

words of the following sentence, with

gaos"

spelled instead of the original

nominative

*giiuS°

(cf. KELLENS 1974: 183). LoMMEL translates "(85e) wenn

auch die Kuh, (86ab) die gefangen fortgefiihrt wird, wahrlich mit erhobenen

I . Handen zu Hilfe ruft", The nom.sg.

giiuS

has been used in a parallel con­

struction to that of the preceding nom.sg. forms

dainhupaitis

etc. in

yim

daujhupaitis

baoa

ustdnazasto zbaiieiti auuainhe

etc., which may imply that § 85e

ya! *giiuSci!

represents

*yim giiuSci!,

with a replacement of original

*yim

by

ya!

on the model of § 85d

ya!ci!.

As LOMMEL 1927: 13 indicates, this

proposal had already been made by WINDISCHMANN 1857: 86.

3.

The translation of 86g-i

kaoa no fraouruuaesaiuiiti asahe paiti pantam drujo

vaesmsnda azsmnam must also be revised. WOLFF 1910: 211, following BAR­

THOLOMAE 1904: 22311534, translates "Wann wird er uns, die zur Wohnung

der Drug hinweggetriebenen (Rinder), zum Weg des A~a zuriickfiihren?"

BARTHOLOMAE 1904: 223 adds the remark "Aprn." (acc.p1.m.) after

azsmnam.

Thus, to BARTHOLOMAE's mind,

azsmnam

really represents original

"azsmna.

the expected spelling of an acc.pl.m. form

<

"azsmnans which would agree in

gender, case and number with

no.

GERSHEVITCH 1959: 233 objects that this

would imply that the cow was referring to her own sex with the wrong gender,

because a cow

(giiusj

is feminine but

rassmna

is masculine. This objection does

not hold because the personal pronoun of the first person is indifferent to gender, the YAv. form for the enclitic ace. being

no

<

*nas,

cf. Skt. nah,

GERSHEVITCH's own translation of 86i-g as "when will he divert to the path of Truth us who are (lit. is) being driven to the estate of Falsehood", which

takes azsmnam as the acc.sg.f. form referring to an unexpressed object gam,

must be rejected because it assumes a plural form

no

to agree with a singular

188

_ t. . .

MICHIEL DE VAAN

form azsmnam.

We must return to BARTHOLOMAE'S suggestion that

azsmnam

is not an acc.sg.f. form but an acc.p1.m. Original "azsmna would be the expected result of "azsmnans, since the ending *-ans regularly) yields YAv.

-a

after a prece­ ding nasal -n- (cf. HOFFMANN 1970: 189 = 1975: 276). In the pronunciation of the scribes who wrote down our Avestan manuscripts, final

-a

could easily be confused with

-an

and

-am,

which is why we frequently find the latter variants for expected -a, e.g. in V 3.18 pairi.daezan for "pairi.daeza, or in Yt 10.120 vtsP;j mazdaiiasnam yam asaonam for *vtsP;j mazdaiiasna ya asaona (SCHINDLER 1982: 204).

The form

vaesmsnda

was analyzed as

"vaesmsm

+

da by BARTHOLOMAE 1904: 1328 and also by GERSHEVITCH 1959: 233, i.e. as the acc.sg. of a noun

vaesma-

m. 'abode' with a postposition da. This analysis meets with three decisive objections. In the first place, vaesmsnda must be connected with Sanskrit vesman- 'house, abode', which is a neuter man-stem. The acc.sg. of IIr. "vaicman- would be YAv.

*

vaesma , not "vaesman. BARTHOLOMAE seeks to save his interpretation by referring to Skt. "dharma- neben °man-, homa­ neben "man- usw.", but whereas a certain degree of interchange between man­ and ma-stems must be acknowledged for Sanskrit (cf. DEBRUNNER 1954: 767), there is no evidence in Avestan for a switch of man-stems to a thematic in­ flexion in -ma-. As we will try to show, there is no need to derive Av. vaesmsnda from another stem than IIr. "vaic-man-, In the second place, the assumption of a postposition da, which BARTHOLOMAE 1904: 662 connects with Greek

-ot:,

is completely gratuitous, because it is unknown elsewhere in Ave­ stan, and it lacks a Sanskrit cognate. One might refer to the PIE suffix *-dhe, as reflected e.g. in Skt. i-ha, OAv. i-da, YAv. i-oa 'here', Old Church Slavo­ nic ks-de 'where?', etc., but this indicates locality, not direction.

The third objection against the interpretation of vaesmsnda as an accusati­ ve concerns the syntax of the passage-, Previous scholarship has analyzed 86gh as 'When will he make us (re)turn towards the path of truth?' vel sim., assu­ ming asahe paiti pantam to indicate the direction of the verb fraouruuaesaiiditi.

1 As appears from the enumeration of acc.pl. forms in *-nans in SCHINDLER 1982: 204.

The only exception A 1.11 daeuuaitasns can easily be explained from influence by the pre­

ceding ace. pI. vispJ in the text of A l.ll.

2 lowe the following observations to Alexander Lubotsky.

~

Avestan

vaesmsnda

189

In that case, the lines 86g-i would express the lament 'Help us return to the straight and narrow, for we are now being lead astray!', which seems a natural continuation in the light of the cries of help that the 22nd _{karde of Yt 10 pre­}

sents.

However, the preposition paid governing the accusative is not only used to indicate a direction, but also for locality (BARTHOLOMAE 1904: 822, REI­ CHELT 1909: 271), e.g. in Yt 19.26 xsaiiata paiti btimim 'he ruled on the earth', and with panui- in Yt 13.54

ttl

fratacinti mazdaodtsm paid pantam 'they flow forth along the Mazda-created path'. The expression asahe paid pantam of Yt 10.86 is attested once more, in a legal context in V 4.43, where it means 'according to the path of Truth'. This is a strong indication that we may have to translate Yt 10.86

asahe

paid

pantam

in a similar way as 'along the path of Truth' .

Furthermore, it seems to me that azsmna may govern not only drujo vaesmsnda, but also the preceding asahe paid pantam. Logically, it is not the act of 'turning someone around' ifraouruuaesaiia-) which can take place 'a­ long' a path, but the act of 'driving' someone. This is confirmed by Yt 10.38, where we find the middle of

aza-

governing the ace.

pantam

without a preposi­ tion, in the meaning 'to be driven along': giiul varaiirim pantam azaite 'the cow is driven along the road of captivity'. For Yt 10.86, this implies that asahe

paiti pantam ... azsmna means 'driven along the path of Truth'. Under this analysis, it becomes impossible for drujo vaesmsnda to mean 'towards the abode of falsehood', and we can interpret vaesmsnda as an ablative form: 'out of the abode of falsehood'. These considerations lead me to the following

translation of Yt 10.86: _I

_I

Yt 10.85

ya! xgiiusci! 'when even the cow

Yt 10.86

yii varsta azsmna who, being led away captive,

biioa ustanazasto zbaiieiti calls for help with outstretched hands, auuainhe

gauuaiirim

paitismarsmna

longing for the cowshed:

kaoa no arsa gauuaiiiim "When will the bull, driving from apaiiai

paskai

vazsmno behind,

190

...

MICHIEL DE VAAN

reach the cowshed?"

kaoa

n6

fraouruuaesaiiaiti

"When will he make us turn around,

asahe paiti pantam (so that we will be) driven along the path of truth,

druj6 "vaesmsnda "azsmna

out of the abode of falsehood?"

4.

We may thus assume that vaesmsnda represents the abl.sg. *vaismant of the

stem

vaesman-,

to which the postposition *a 'out of' was affixed; subsequent­

Iy, the noun and the postposition were univerbated and as a result, final *-t of

*vaismant was retained and is reflected by -d-. We can directly compare other YAv. instances of a noun in the abl.sg. which came to be univerbated with the postposition *a, e.g. xVafnaoa 'from sleep'

<

*xvafna! a or

zraiianhaoa

'out of the lake'

<

"zrailasjhat.

a

(cf. REICHELT 1909: 269). It is important to note that the univerbation in these cases took place after *-t had yielded Avestan

-!

in word-final position, since the result of univerbation into *xvafntita and

*irajahata at an earlier stage would have been txvafnata and tzraiia1]hata; we

must assume that the unreleased Avestan *-t in e.g. *xvafnat a was a voiced

consonant which partook in the YAv. sound change of interva"'calic *-d-to _0_3 .

In the case of

vaesmsnda

<

*vaismant a, we also need to assume as a

first step the development to

rvaesmani

a.

Subsequent univerbation yielded

"vaesmandd, whence with shortening of the final vowel and the development of *-and-to *-a'}d-we get vaesmsnda. The irregular spelling with -nda instead of -nda is a peculiarity of the manuscript Fl, compare Yt 10.85 pantam, Yt

10.86 jrascindaiieiti and zantumca, which can be found in the facsimile of F 1 (JAMASpAsA 1991).

In order to explain

barssmsn

beside

vaesmsnda,

we need the following

relative chronology of changes: 1. the replacement of the gen.sg. ending *-ans

by *-ant to form the abl.sg. of n-stems; 2. final *-ant

>

*-an, but in sandhi

3 Compare the similar conclusion dra wn for OAv. by PIRART 1988: 44f., who discusses

the OAv. reflexes of the lIr. particle sequence *Gd u. Whereas JJiida can be the direct result

of an early univerbation into *adu, the form a!ta presupposes the retention of two separate

words until the stage *a! u. The latter stage also underlies the reflex ada, viz. through the stages *a! u ~ *a! a > "ada. The form ada thus shows the same intervocalic -d-

<

*-!­

which turned into -0- by regular sound law in YAv.

191

Avestan

vaesmsnda

with a following postposition "a, *-ant is preserved; 3. *-ant a

>

*-ar:z! a; 4. *-ant a

>

"<andd, *-an

>

-sn, "<an-

>

*-ar:z-.

5. As we have seen in section 1, the replacement of the gen.sg. ending by the

abl.sg. *-(a)t postdates Proto-Iranian. This implies that the final consonants

which were replaced by abl.sg. *-t in YAv. actually were the I1r. allophones

of *s, viz. *-5 (after *i, *u and *r) and *-h (after *tl and *n). The replacement must have been prior to the loss of final *-h, since otherwise the model for the

attested abl.sg. endings in -(a)! would have disappeared. For instance, the

ending *-ah of consonant stems became *-aand then -6, but the abl.sg. -a! must

be based on the stage *-ah. Similarly, the a-stem gen.sg. ending -aUd can

hardly have been the basis for the abl.

-aiidt;

rather, we must date the replace­ ment to the gen.sg. stage *-aj,ah. In the n-stems, the gen.sg. -al-s (depending

on the preceding consonant) must still have been *-anh when the formation of

the abl.sg. -ant took place.

Regardless of one's interpretation of the differences between GAv. and

YAv. as being due only to a chronological difference between two stages of the

same language, or to (additional) dialectal differences (cf. HOFFMANN-FoRSS­ MAN 1996: 32), the preceding observations show that the phonological system of GAv. and YAv. must have been much closer to that of Proto-Iranian than appears from the written form of the texts. The survival in YAv. of the inheri­ ted gen.sg. ending in *-s, which can be found especially in abl.sg. forms not governed by a pre- or postposition (cf. REICHELT 1909: 245, 254), beside the

productive abl.sg. in *-t, seems to suggest that YAv. was indeed a more recent

variety of one and the same Avestan language. This view requires the un­ changed preservation of the phonological" sequences *-ah, *-anh, etc. at the GAv. stage, as has been argued by BEEIrnS 1988: 82f. and passim. If one pre­ fers the view that GAv. and YAv. split off after a common Proto-Avestan stage, the new abJ.sg. ending ofYAv. must have arisen after that PAv. period,

but before final *-h was dropped. The productivity of -(a)! in YAv. suggests

that this could not have happened long before the composition of the YAv. texts.

4 Of course, allophonically these may have been [-Jh], [-Jnh], etc.

192

_

....

MICHIEL DE VAAN

•

~1 References BARTHOLOMAE, Ch.

1904: Altiranisches Worterbuch, Strassburg. BEEKES, R.S.P.

1988: A grammar of Gatha-Avestan, Leiden.

1995: Comparative Indo-European linguistics, Amsterdam - Philadelphia. DEBRUNNER, A.

1954: Altindische Grammatik II,2: Die Nominalsuffixe, Gottingen. GERSHEVITCH, I.

1959: The Avestan hymn to Mithra, Cambridge. HOFFMANN, K.

1970: Zur avestischen Textkritik: Der Akk.Pl.mask. der a-Stiimme, in: Henning memorial volume, ed. M. Boyce and I. Gershevitch, London, p. 187-200.

1975: Aufsatze zur Indoiranistik, ed. J. Narten, Band 1, Wiesbaden. HOFFMANN, K. and B. FORSSMAN

1996: Avestische Laut-und Flexionslehre, Innsbruck. JAMASPAsA, K.

1991: The Avesta Codex Fl (Niyayisns and Yasts). Facsimile edition with an intro­ duction, Wiesbaden 1991.

KELLENs,1.

1974: Les noms-racines de l'Avesta, Wiesbaden. 1984: Le verbe avestique, Wiesbaden.

LOMMEL, H.

1927: Untersuchungen iiber die Metrik des jungeren Awesta, Zeitschrift ftir Indologie und Iranistik 5, 1-92.

PiRART, E.

1988: Phonetique et graphie, in J. KELLENS and E. PIRART

1988-91: Les textes vieil-avestiques. Volume I, Wiesbaden, p. 42-88. REICHELT, H.

1909: Awestisches Elementarbuch, Heidelberg.

SCHINDLER, J.

i

1982: Zum Nom.Sing.m. der nt-Partizipien im Jungavestischen, in: NEU, E. (ed.) Investigationes philologicae et comparativae: Gedenkschrift fur Heinz Kronasser, Wiesbaden, p. 186-209.

WINDISCHMANN, F.

1857: Mithra. Ein Beitrag zur Mythengeschichte des Orients, Leipzig (= Abhandlun­ gen fiir die Kunde des Morgenlandes Band I, Lieferung 1). Reprint Nendeln (Liechten­ stein) 1966.

WOLFF, F.

1910: Avesta, die heiligen Bucher der Parsen, iibersetzt auf der Grundlage von Chr. Bartholomae's altiranischem Worterbuch, Strassburg.