The origin of Indic and Iranian feminines in -ani-(d).

• Author: Rau, Jeremy

1.0 Vedic has a feminine-forming suffix -ani(d) with devi-inflection (1) that is derived nearly exclusively from thematic stems and that makes derivatives in the following two semantic groups:

1. 'wife of x', with x being the name of a divine or (semi-) mortal figure:

   indrai- (RV 1.22.12a, 2.32.8c, 5.46.8b, 11a, 12a): indra-

   varunani-(RV 1.22.12b, 5.46.8c): varuna-

   usinarani- (RV 10.59.10b)" usinara-(Br.) (2)

   Purukutsani- (RV 4.42.9a): Purukutsa-

   mudgalani- (RV 10.102.2c, 6d): mudgala-

2. 'the divine woman/female genius of x', with x being an appellative:

   arayani- 'female genius of the forest' (RV 10.146): aranya-'forest'

   urjani- 'female genius of urj- (who attends the Asvins)' (RV 1.119.2d): urj-/urja- 'energy' (3)

   The type remains productive in later Vedic and in the epic and classical language, where it largely preserves the same derivational behavior and semantic functions it has in the Rigveda. For a survey of the material in the later language and the details of its development, see Debrunner 1954: 279f. and Narten 1986: 213f.

   The type also appears in two residual cases in ancient Iranian. The first is the famous Old Avestan form ahurani-(d), which is used as an epitnet of the waters at Yasna Haptanhaiti 38.3. (4) This stem has recently been treated by Narten (1986: 212ff), who has convincingly shown that it should be aligned with its Vedic correspondents and translated 'Herrinnen' or, more originally, 'wives of the ahura'. This analysis is supported with reference to the other Iranian reflex of this type, Pahlavi m'sy'nyd < Av. *masiianl-(d) and Turfanian Middle Persian mwrdy'ng < Old Persian *martiyanl-(d), the feminine corresponding to Avestan masiia- 'mortal, man' and the female member of an early Iranian mythical pair *martiia-and *martiiani- of the sort that is familiar from Vedic manu-: manavi. (5)

   2.0 On the basis of these forms it is possible to reconstruct a Proto-Indo-Iranian suffix of the shape (6) *-aniH-(d) or *-aHniH-(d) that made feminine derivatives meaning 'the wife/woman of x' to masculine thematic nouns. More complicated however is the question of the origins and derivational history of the type. Although it has been the subject of numerous discussions in the last two hundred years, today there are still two competing approaches to its origins, and no clear consensus as to which is likely to be correct. (7) The purpose of this short contribution is to decide which of these approaches best explains the type.

   2.1 The first approach to be discussed--which was advanced by Theodor Benfey (1854: 455ff.), Ernst and Manu Leumann (1893: 294ff.; 1952: 14)--takes its start from the athematic-stem counterpart of the -ani- (d) formations. This type, which is limited to a handful of examples in the Rigveda and later Vedic texts, is formed by lengthening the suffix vowel of the derivational basis and adding the devi-suffix -i-(d). Like -ani-(d) it mostly makes derivatives meaning 'the wife of x', where x is the name of a divine or (semi-) mortal figure: agnayi RV (1.22.12c, 5.46.8b): agni-, jahnavi RV (1.116.19c): jahnu-, (8) manavi-SB (manavi p.4.1.38), manayi- (9) MS: mdnu-, (+) putdkratayl- (10) RV (8.56.4b): putdkratu- RV (8.68.17c). vasavi- RV (10.73.4c): *vasu-, (11)' vrsdkapayl- (12) RV (10.86.13a): vrsakapi-.(13) Benfey and his followers argue that this "athematic" type is the older of the two formations, and that the apparent n-stem suffix we have in the "thematic" -ani-type above has been segmented out of athematic n-stem pairs like brahmani-: brahman- and secondarily extended to thematic stems. (14)

   This approach has two fatal weaknesses. The first is that contrary to what we would expect in a morphological category with a unitary origin, the accentual patterns of the two types fail to match up. In the "athematic" type paroxytonesis is the rule apart from two exceptions: + putakratayi-, where the stem replicates the accent of its derivational basis, and manavi- (SB), which is essentially a varia lectio with the same accentuation that is common in the "thematic" type. The "thematic" type, on the other hand, seems to have originally been oxytone, the exceptions being two instances, usinarani- and purukutsani-, where the derivative imitates the accent of its basis, and one further case, mudgalani-, (15) where the accent is paroxytone just as in the "athematics."

   A second and more significant weakness involves the supposed evidence for the "athematic" agnayi-type to n-stems. The form that is usually taken to support this analysis, brahmani 'wife of a Brahmin', is attested only in the epic/classical period, and is not likely to be older than this, since -ani-(d) feminines meaning 'wife of x' made to appellatives are not attested before this period. Since all the feminines of the Vedic period apart from personifications like aranyani- and urjani- are made to proper names, to motivate this analysis it would be necessary to assume (1) that to an n-stem compound name like Ved. susaman- speakers of Proto-Indo-Iranian made a feminine *susamani- 'the wife of S[degrees]' according to the regular "athematic" pattern, (16) and (2) that because compounds with an n-stem as second member could optionally replace the n-stem suffix with the thematic vowel--cf., e.g., RV visvakarma- beside visvakarman- (17)--they got the idea that a-stems could also make feminines in [TEXT NOT REPRODUCIBLE IN ASCII] and then generalized this pattern by proportional analogy: *susama- ([congruent to] susaman-): * susamani-:: indra-: x, x [right arrow] indrani-. But beyond the fact that this analogy is somewhat artificial, there are no n-stem proper name pairs attested in the languages that support this analysis, nor do proper names with n-stem second members seem to have regularly replaced this suffix with the thematic vowel. Of all the Vedic, Avestan, and Old Persian names given in Mayrhofer 2003 and 1979, there are at least twenty compounds based on n-stems, and all have n-stem inflection, there being no certain cases of -a- replacing -an-. (18)

   2.2 A second direction has been highlighted by Meid (1956: 277f.), who aligns the -ani- (d)feminies with the Greek suffix -[TEXT NOT REPRODUCIBLE IN ASCII], which makes genitival derivates (19) to thematic stems. In particular, Meid compares the kinship term [TEXT NOT REPRODUCIBLE IN ASCII] (Hom.+) 'grandson': [TEXT NOT REPRODUCIBLE IN ASCII] 'son' and feminine derivatives to masculine proper names like [TEXT NOT REPRODUCIBLE IN ASCII] (Hom. +) 'daughter of Akrisios': [TEXT NOT REPRODUCIBLE IN ASCII] and [TEXT NOT REPRODUCIBLE IN ASCII] (Hom.+) 'wife of Zeus': [TEXT NOT REPRODUCIBLE IN ASCII] (20) These are closely related to nouns like [TEXT NOT REPRODUCIBLE IN ASCII] (Hom.+) 'bird of prey': [TEXT NOT REPRODUCIBLE IN ASCII] (Att.) 'egg' (21) or [TEXT NOT REPRODUCIBLE IN ASCII] 'wind-flower': [TEXT NOT REPRODUCIBLE IN ASCII] 'wind', and have their origins in a specialization of what in today's terminology can be termed the thematic version of the "Hoffmann suffix," viz. *-[h.sub.1]-n-o-. (22) This suggestion seems promising, but requires elaboration and further motivation. (23)

   What Meid and others have crucially failed to appreciate in making this identification is that the patronymic use of Greek -[[TEXT NOT REPRODUCIBLE IN ASCII] has a very close correspondent in lndic and Iranian in the suffix -ana-. (24) In Avestan this suffix is well attested for the formation of patronymics and pro-patronymics, and like its Greek correspondent is mostly limited to thematic stems: (25)

   Old Avestan friiana- (Y.46. 12, epithet of tura-) (26) 'descendant of F[degrees]': friia- (Yt. 13.110, 119) haecat.aspana- (Y.53.3, epithet of pourucista-) 'descendant of H[degrees]': haecat.aspa-(27) (Y.46.15) Young Avestan (28) fraiiazentana- (Yt. 13.113, epithet of frenah- and jaro.vohu-): fraiiazenta- (Yt.13.113) (29) gaiia[delta]astiiana- (Yt. 13.114, epithet of asa.siiaona-): gaiiaasti- (Yt. 13.112, 140) gaoraiiana- (Yt.13.118, epithet of +yoista-): gaoiri- (Yt.13. 118) jamaspana- (Yt.13.104, epithet of hanhaurus-): (de)jdmdspa- (OAv.+) neremiiazdana- (Yt.13.110, apparently patronumic as...