CC BY
18
Acta Linguistica Petropolitana. 2022. Vol. 18.1. P. 147-164 DOI 10.30842/alp23065737181147164
Numerative and numeral inflection in Ossetic
Ronald I. Kim
Adam Mickiewicz University in Poznan (Poznan, Poland); rkim@amu.edu.pl
Abstract. Ossetic, like many Middle and Modern Iranian languages, has a special form for nouns in the nominative when modified by a numeral. This numerative form takes the ending -i in the Digor dialect and -y in the Iron dialect, e.g. Digor avd furt-i, Iron avdfyrt-y 'seven sons'. The oblique cases are inflected normally in Iron, e.g. genitive avd fyrt-y, dative avd fyrt-an, but Digor adds -e(m)- to nouns before the case endings, as well as to numerals when used alone, e.g. genitive avd furt-e-j 'of the seven sons', avd-e-j 'of the seven', dative avdfurt-em-an 'to the seven sons', avd-em-xn 'to the seven'.
The origin of the numerative markers remains a controversial question of Ossetic historical grammar. Although the numerative in Sogdian and Khwarezmian goes back to the Proto-Iranian dual, a change with typological parallels in eastern South Slavic, the Ossetic numerative in -i/-y cannot reflect any of the inherited dual markers. It is proposed that this ending rather continues the Proto-Iranian nominative plural in *-ah of a-stems and animate a-stems. Although it was replaced by *-ahah and collective *-a in most Iranian languages, and indeed already in Avestan, *-ah could have survived in pre-Ossetic. There it regularly developed to -i/-y after numerals, but was ousted in all other contexts by an innovative (originally collective) formation in *-ta- > -te, a situation paralleled in Middle Welsh. It is further hypothesized that the suffix -e(m)-of the Digor oblique cases was generalized to numerals from adjectives of quantity, or it could have originated in Proto-Ossetic *duwe 'two' < Proto-Iranian *duwai. From numerals, it then spread to nouns governed by numerals.
Keywords: Ossetic, Sogdian, Iranian, Slavic, numeral systems, numerative, plural, dual, pronominal inflection, "laws of finals".
Acknowledgment. The research for this article has been supported by grant no. 2019/35/B/HS2/01273: "Ossetic historical grammar and the dialectology of early Iranian" from the Polish National Science Centre (NCN). I am also grateful to Nicholas Sims-Williams for his many useful comments and suggestions.
© Ronald I. Kim, 2022
Нумератив и склонение числительного в осетинском языке
Р. Ким
Университет им. Адама Мицкевича в Познани (Познань, Польша); rkim@amu.edu.pl
Аннотация. В осетинском, как и во многих ново- и среднеиранских языках, имеется особая форма номинатива существительных, входящих в именную группу с числительными. Эта форма, нумератив, принимает в дигорском окончание а в иронском -у, например, дигор. avd/угЫ, ирон. avd/и^-у 'семь сыновей'. В косвенных падежах в иронском употребляются обычные падежные формы, а в дигорском падежным окончаниям таких существительных, как и числительных в самостоятельном употреблении, предшествует суффикс -е(т)-.
Происхождение показателей нумератива остается спорным вопросом исторической грамматики осетинского языка. В согдийском и хорезмийском нумера-тив восходит к праиранским показателям двойственного числа, что находит типологическую параллель в южнославянских языках. Осетинский нумератив -И-у, напротив, не может быть возведен ни к одному из унаследованных показателей двойственного числа.
В статье приводятся аргументы в пользу происхождения осетинского ну-мератива от праиранского окончания номинатива множественного числа * основ на -а- и основ на -а- одушевленных имен существительных. Хотя окончание во многих иранских языках было вытеснено окончаниями собирательного числа *-а и *-аhah, оно могло сохраниться в праосетинском. Предполагается, что именно было вытеснено во всех остальных контекстах новым показателем множественного числа *-Ш- > -Ш, развившимся на базе собирательных форм.
Высказывается гипотеза о том, что суффикс -е(т)- в формах косвенных падежей дигорского диалекта сформировался в числительных под влиянием прилагательных, выражающих количество, либо под влиянием праосетинского *duwe 'два' < пра-иран. *duwai. От числительных этот показатель мог дальше распространиться на синтаксически зависимые существительные.
Ключевые слова: осетинский, согдийский, иранские языки, славянские языки, числительные, нумератив, множественное число, двойственное число, прономинальное склонение, закон конечного слога.
1. Introduction
It is a pleasure to contribute to this volume in honor of Professor Kazansky, an esteemed scholar and colleague who has done so much to make St. Petersburg an internationally renowned center of Indo-European linguistics. The following study reflects my longstanding enthusiasm for one of the many fascinating languages spoken in his native country, as well as the historical evolution of his mother tongue and its closest relatives.
Like many Middle and Modern Iranian languages, Ossetic has a special form for nouns when modified by a numeral or certain quantifiers. 1 In the nominative, nouns take the ending -i (Digor), -y (Iron), synchron-ically identical to the genitive and inessive in -i/-y, e.g. Digor avd furt-i, Iron avdfurt-y 'seven sons'. In the other cases, the modified noun has the usual case markers in Iron, but Digor shows a more complex pattern: the genitive, allative, and superessive endings are preceded by -e-, the dative, ablative, and inessive endings by -em-. When used alone, numerals inflect as ordinary nouns in Iron, e.g. avd 'seven', gen. avd-y, dat. avd-an, etc. In Digor, however, they show the same exceptional inflection as nouns modified by numerals, e.g. gen. avd-e-j, dat. avd-em-an; so also do adjectives of quantification such as ijnna 'other (of two)', be(w)ra 'much, many', agas 'all'. 2 The following table illustrates these facts with the inflection of Digor avd 'seven' and avdfurti 'seven sons', together with the
1 Where two Ossetic forms are given separated by a slash, the first is in the Digor dialect, the second in Iron. I retain the usual scholarly transcription system for Ossetic, except that w is distinguished from u (both represented by <y> in Cyrillic orthography). The more phonetically based transcriptions of Belyaev and Erschler reflect the shifts in Iron consonant articulation since the 19th century, e.g. retraction of s [s], z [z] > [§], [3] and deaffrication of c [te], j [A] > [s], [z] except in geminates ([Thordarson 1989a: 463, 1989b]).
2 See [V. F. Miller 1881-1887, 2: 159; W. Miller 1903: 49 §55; Abaev 1949: 397398, 398-399; Isaev 1966: 50-52, 1987: 592; Christol 2003: 30-32; Belyaev 2017: §§1-2; Erschler 2019: 868, 870 (Table 6)]. An interesting point of divergence between the two dialects is that the numerative occurs with pluralia tantum in Digor but is
pronominal inflection of ci 'which' (oblique stem ca-) and ijnna 'other (of two)' for comparison, followed by the corresponding forms in Iron.
Table 1. Inflection of 'seven', 'seven sons', 'which', and 'other (of two)' in Digor and Iron Ossetic
Digor
nominative avd avd furt-i ci ijnnx
genitive avd-e-j avd furt-e-j Cte-j ijnn-e-j
dative avd-em-xn avd furt-em-xn cx-m-xn ijnn-em-xn
allative avd-e-mx avd furt-e-mx cx-mx ijnn-e-mx
ablative avd-em-xj avd furt-em-xj cx-m-xj ijnn-em-xj
inessive avd-em-i avd furt-em-i cx-m-i ijnn-em-i
superessive avd-e-bxl avd furt-e-bxl cx-bxl ijnn-e-bxl
equative 3 (avd-ej-aw avd furt-ej-aw) cxj-aw ijnn-ej-aw
Iron
nominative avd avd fyrt-y cy innx
genitive avd-y avd fyrt-y cx-j innxj-y
dative avd-xn avd fyrt-xn cx-m-xn innx-m-xn
allative avd-mx avd fyrt-mx cx-mx innx-mx
ablative avd-xj avd fyrt-xj cx-m-xj innx-m-xj
inessive avd-y avd fyrt-y cx-m innxj-y
superessive avd-yl avd fyrt-yl cx-wyl innx-wyl
equative avd-aw avd fyrt-aw cxj-aw innxj-aw
comitative avd-imx avd fyrt-imx cx-imx, cemx innx-imx
blocked in Iron, e.g. D. xrtm bijnont-i 'three families', dat. xrtx bijnont-em-xn vs. I. xrtx binontx 'id.', dat. xrtx binont-xn (Belyaev 2017:§3.1).
3 No equative forms are given for Digor by Isaev [1966: 50-51; 1987: 592] or Chris-tol [2003: 31], though it is not clear if this is because they are "pragmatically improbable" [Erschler 2019: 870, fn. 8]. They do occur in Iron, e.g. Dywx gorxt, —Xiro-simx, Nagasaki... dywa cwstaw fxsoqqyr kodtat zxxx! 'Two cities, Hiroshima and Nagasaki... like two eyes, you blinded the earth!' (E. Tedety, Byndary kadxg [2005];
Erschler [2019: 868] refers to the elements preceding the oblique case endings in Digor as the "dedicated numeral declension suffixes" -e-and -em-, e.g. D. duw-e-j 'two-NUM-OBL', duw-em-aj 'two-NUM-ABL'. 4 Since however the same distribution of -m- is found in both dialects in demonstrative and interrogative pronouns, the numeral inflection may synchronically be considered a subset of the pronominal inflection. 5 The numeral suffix is homophonous with the associative plural marker D. -e-, I. -i-, e.g. D. Worazmag-e-ta 'Worazmag and his companions', I. Aslan-i-ta 'Aslan and the others', 'Aslan and the gang' [Bailey 1956: 125-126; Erschler op. cit.], although whether this identity is original or has arisen secondarily remains unclear.
The origin of this inflectional pattern, for which Digor undoubtedly preserves the more archaic situation, remains an unresolved mystery of Ossetic historical grammar. Most grammatical studies simply equate the numerative in -i/-y with the homophonous genitive ending, e.g. [Bagaev 1965: 214] ("beginning with the numeral dywa 'two', in the nominative and accusative the substantive in combination with the numeral occurs in a form similar to the genitive, i.e. with the ending -y"; tr. mine), [Kim 2003: 49, fn. 18] ("the ordinary genitive"), [Thordarson 2009: 132] ("the numerative use of the genitive after cardinal numerals higher than 'one'"). As observed by Erschler [2019: 870, fn. 7], "[i]n numeral
cited from the Ossetic National Corpus) or, from the Ossetic Bible translation by the Jehovah's Witnesses, Wycy artajaw wyman dar ja kad ajqyst 'He became as famous as those three' (1 Chronicles 11: 20, 2 Samuel 23: 18; https://www.jw.org/os/ 6H6jiHOTeKffi/6H6jiH/bi12; accessed on: 20.08.2020).
4 Erschler glosses the case ending in D. duw-e-j as "oblique", as the genitive and inessive are identical for lexical nouns; but since the two cases are distinguished in Digor numeral inflection (and more generally in pronominal inflection), I have retained the genitive label here and below.
5 This -m- must ultimately be connected with the PIr. pronominal case forms in *-hm-, i.e. dat. sg. *-hmai, abl. *-hmat, loc. *-hmi (cf. OAv. rel. pron. yahmai,yahmi, YAv. yahmat; [Kim 2003: 45, fn. 9; Cheung 2008: 99, 102]), but the details remain to be clarified. Belyaev [2010: 298-299, 302] derives dat. -m-an from a construction acc. + *ana/u, but it is unlikely that PIr. word-final *-m in the acc. sg. would have survived long enough for the creation of new cases from postpositional phrases.
phrases in the nominative, the oblique marking surfaces on the nouns: vrtv qaz-3 three goose-OBL 'three geese'. In other cases, the oblique is replaced by the respective case marker: vrtv qaz-vn three goose-Dar." The morphosyntax of quantified noun phrases in Ossetic is thus comparable to that of Russian, with heterogeneous case distribution in the nominative and homogeneous case distribution in the oblique cases in the terminology of Babby [1987: 100-101] (see below).
However, while valid on a descriptive level, the identification of the numerative with the genitive offers no insight into its actual origin. Furthermore, Belyaev [2014: 46-47; 2017: §3.2; 2021: 259, 263-264] adduces evidence from coordinate phrases that the numerative is syn-chronically distinct from the genitive: in both Iron and Digor, if two or more nouns modified by a numeral are coordinated and the case affix on the nonfinal conjunct(s) is suspended, the numerative ending appears, e.g. D. [arta Yaw-i ama duwa gorat]-e-ma, I. [arta qaw-y ama dywa gorat]-ma 'to three villages and two towns'. Since nouns can only surface in the unmarked form in instances of suspended case affixation (e.g. I. [Za-wyr ama Alan]-an 'to Zawyr and Alan'), the ending of D. arta Yaw-i, I. arta qaw-y cannot be the genitive marker -H-y but, like singular Yaw/ qaw and plural Yawta/qawta, must be unmarked for morphological case.
One point that must be made clear at the outset is that the numerative ending -i/-y need not have the same source as the inflected (oblique) case forms of the numerals themselves. A quick glance at the situation in Slavic will make this clear. In Proto-Slavic and Old Church Slavonic, the numerals inflected as substantives of various inflectional classes, but in many modern Slavic languages they have been influenced by the inflection of pronouns and definite adjectives: 'two', 'three', 'four' in Russian gen./loc. dvux, trex, cetyrex, Polish gen./loc. dwóch, trzech, czterech; optionally for higher numerals in Ukrainian, e.g. gen./loc. p'jat'óx 'five', desjat'óx 'ten' (beside inherited p'jaty, desjaty). In contrast, nouns in the nominative (and, for inanimate nouns, accusative) take the genitive singular in Russian and BCSM when modified by the numerals 'two', 'three', and 'four', e.g. Ru. dva/tri/cetyre stolá 'two/three/four tables' (to stol 'table', cf. nom. pl. stoly). This pattern originates in the Proto-Slavic dual, which in the largest inflectional class of masculine o-stems fell together with the genitive singular;
this led speakers to abstract a rule of heterogeneous case agreement, which was then extended to 'three' and 'four'. 6 In eastern South Slavic, the originally dual ending -a has been extended to all higher numerals and quantifiers for masculine nouns, yielding a true numerative in e.g. Bulgarian tri/ deset/mnogo bileta 'three/ten/many tickets' (to bilet 'ticket', cf. pl. bileti).
2. The numerative: an old dual?
The origin of the Ossetic numerative has most recently been discussed by Sims-Williams [2020]. In Sogdian, as in eastern South Slavic, nouns immediately preceded by any numeral take a form that goes back to the Old Iranian dual, e.g. m. a-stem 'Sw'kp' 'two fishes', xii fiy' 'twelve gods' with -' [-a] < PIr. *-a; f. a-stem xii Srxwsy 'twelve she-devils', n. a-stem ii fiynyy 'two temples' with -y(h) [-e] < PIr. *-ai; aka-stem dw' z't' 'two sons', ds'nyzb'n' 'ten passions', xwsrts wyc'w' 'sixteen martyrs' with -' [-a] < PIr. *-aka [Sims-Williams 1979: 339-342, 2020: 955-958]. 7
6 The oblique cases show a homogeneous case distribution with the usual plural endings, e.g. Ru. gen. trex stolov three-GEN table-PL-GEN, instr. tremja stolami three-inst table-PL-iNST. No position is taken here on the theoretical analysis of the numeral agreement patterns of Russian and other Slavic languages, which have been extensively investigated in the syntactic literature (in addition to [Babby 1987], see [Franks 1994; Rappaport 2002; Pesetsky 2013]). Since a few nouns in Russian maintain a distinction of stress between the gen. sg. and the count form (e.g. dva/tri/cetyre casa 'two/three/four hours' to cas 'hour', cf. gen. sg. casa), it could be argued that the latter is in fact synchronically a minor numeral category of "paucal", homopho-nous with the gen. sg. for the great majority of nouns ([Rakhlin 2003; Bailyn, Nevins 2008: 263-268]; otherwise [Pesetsky 2013: 21-34, 89-91]). Such an analysis seems warranted for BCSM, where attributive adjectives and (optionally) predicates also exhibit the ending -a with masculine nouns, e.g. tri b(ij)el-a stol-a su bil-a 'three white tables were' [Naylor 1972: 4-5; Belie 2003, 2008].
7 The m. a-stem numerative could alternatively continue the Old Iranian nom. pl. *-a (OAv. -a, Khot. -a), which survives in the early Sogdian fiy' 'gods; lords'; but the forms of the other inflectional classes can only go back to the dual.
A similar system seems to have existed in Khwarezmian [Durkin-Meis-terernst 2009: 343; Sims-Williams 2020: 959-960], and several Modern Iranian languages also have numerative forms that may be of dual origin, including Pashto and Parachi [Sims-Williams 2020: 960-963]. Following an earlier suggestion [Sims-Williams 1979: 342] (cf. [Tremblay 1997: 158-159, fn. 5; Kim 2003: 49, fn. 18]), he therefore proposes that the Ossetic numerative ending -H-y, like the Sogdian numeratives in -y(h) [-e] cited above, goes back to PIr. a-stem and n. a-stem dual *-ai (YAv. -e, cf. Ved. -e; [Sims-Williams 2020: 963-968]).
The problem with this attractive idea is that word-final *-ai is reflected in all clear cases as Oss. -x, as in the enclitic pronouns mx, dx < PIr. *mai, *tai [Cheung 2002: 63-66, 2008: 97; Kim 2003: 46, fn. 10]. Sims-Williams raises the possibility that PIr. *-ai yielded Oss. -x in monosyllables but -i/-y in polysyllables, but this too is contradicted by the evidence of the following:
xndx/xddx 'outside' < POss. *xnde < PIr. loc. *antai (cf. Ved. anta- 'end, limit');
dxlx 'below, beneath, under' < POss. *dxle < PIr. loc. *adarai (cf. YAv. aSara- 'lower, western');
falx 'beyond, on the other side' < POss. *fale ^ PIr. loc. *parai, loc. of *para- (YAv. para-, Khot.para- 'border', NP bar 'bank'); 8
and
duwx/dywx 'two' < POss. *duwe < PIr. f./n. *duwai (YAv. duiie, (neut.) duuae-ca; cf. Ved. duve) < PIE f. *d(u)weh2ihi, n. *d(u)woihi (see [Cheung 2002: 65; Kim 2020: 261 with refs.]).
These forms suggest that PIr. word-final *-ai underwent regular monophthongization to POss. *-e as in other positions, and that this
8 Cf. fal-,fate 'beyond, on the other side',fale (<fateja) 'beyond', likewise with analogical -l- from faldar 'farther' (dissimilated from *far-dar), fallag 'located on that side' (< pre-POss. *par-yaka-); the older form far- survives only in far-ast 'nine' < 'beyond eight'. See [Kim 2020: 260 with refs.].
gave -a in both Digor and Iron [Kim 2007: 53-541; Cheung 2008: 101]. 9 Such a development is supported by the adjectives ijnna/inna 'other' and be(w)ra/bira 'many, much; very', where Digor preserves the stem-final *-e- in the plurals ijnne-ta, bere-ta. As seen by Cheung [2002: 64, 65-66], the innovative plural marker *-ta must have been added to the inherited pronominal m. nom. pl. *-ai, then the unextended form was reanalyzed as singular. Hence D. ijnna, pl. ijnne-ta 'other' < POss. *ijnne < PIr. *anyai (YAv. aniie, OP aniyai; cf. VeD. anye) and D. be(w)ra, pl. bere-ta 'much, many' < POss. *bewre < *baiwarai ^ PIr. *baiwar/n-'ten thousand' (YAv. baeuuars, pl. baeuuqn, baeuuani).
As for the other common dual ending, PIr. m. a-stem dual *-a, it is universally agreed that word-final *-a became POss. *-a > D. -a, I. -0. 10 Since neither PIr. *-a nor *-ai would have led to the desired outcome, I conclude that the Ossetic numerative ending -i/-y cannot be of dual origin.
3. The numerative as inherited plural
If the dual is ruled out, the most likely remaining source for the Ossetic numerative is an inherited plural formation that has been ousted in all other contexts by the innovative (originally collective) formation in *-ta- > -tœ. Such a situation has parallels in Indo-European,
9 The inessive case ending -i/-y thus cannot reflect PIr. a-stem loc. sg. *-ai (pace [Christol 1990: 32, 2003: 37, 38]). Although some details remain to be clarified, I continue to believe that it goes back to PIr. case endings of the shape *-aya(h), which developed via syncope and other regular changes to POss. *-i (Kim 2003); also possible on formal grounds is the PIr. relational adjective suffix *-iya- ([Cheung 2008: 94; Belyaev 2010: 300]; see already [Miller 1903: 46]).
Cheung [2008: 101] suggests that post-PIr. *-ai > POss. *-e is reflected as -e/-i in tonic monosyllables, as in the interrogative-relative pronoun D. ka (gen. ke), I. ci 'who'. This is phonetically plausible, but much about the inflection of this pronoun remains obscure [op. cit.: 101-103].
10 Or -a in both dialects in the special case of pl. *-ta > POss. *-ta > D., I. -ta. On the phonological "semi-independence" of the plural suffix, see [Bailey 1945: 25].
e.g. in Middle Welsh numeratives such as tri broder 'three brothers',ped-war broder 'four brothers' (to brawd 'brother') or tri meib 'three sons', pedwar meib 'four sons' (to mab 'son'), where broder and meib preserve the Proto-Celtic plurals *brater-es, *makw(kw)-ï, replaced elsewhere by the innovative formations brodyr and meibion. 11
Of the possible starting points in Proto-Iranian, athematic animate nom. pl. *-ah may immediately be left aside, since it should have been lost by apocope (pace [Bielmeier 1982: 59, 66; Thordarson 1989a: 468, 2009: 132-133]); and pronominal m. nom. pl. *-ai cannot have given -i/-y for the reasons put forth in §2 (pace [Christol 1990: 35, 2003: 37; Tremblay 1997: 159, fn. 5]). By process of elimination, we are left with the nom. pl. *-ah of a-stem and animate a-stem nouns. 12 This seems promising, as generalized nom. pl. *-ah has left reflexes elsewhere in Iranian, e.g. the end-ingless form of nouns after numerals in Modern Persian or the Balochi nom. pl. in -0 [Tedesco 1921:64-65]. But would this ending have given Oss. -i/-y by the regular laws of finals, or Auslautgesetze?
PIr. *-ah appears to give Oss. -0 in reflexes of adjectives in *-mant-, *-want- with nom. sg. *-mah, *-wah, e.g. fusun/fysym 'host, person who practices hospitality toward guests', rœdaw 'generous' < PIr. *fsu-mah, *fra-da-wah. Also pointing in this direction is the equative case ending D., I. -aw (e.g. furtaw/fyrtaw 'like a son'), if it too continues adjectives
11 The respective forms after 'two' go back to the inherited dual, which has fallen together with either the singular or the plural: cf. dau froder 'two brothers' < PCelt. du. *brater-e vs. dau fab 'two sons' < PCelt. du. *makw(kw)-o. Cases such as the latter, as well as those where both dual and plural fell together with the singular by sound change (e.g. dyn 'man' < PCelt. sg. *doniyos, du. *doniyo, pl. *doniyi), gave rise to the selection rule of Modern Welsh by which nouns appear in the singular after all numerals. For details and discussion, see [Nurmio, Willis 2016: 309-312; Nurmio 2019: 214-241, esp. 220-223].
12 The latter have largely replaced this ending with originally collective *-a (> OAv. masiia, YAv. masiia 'people', Khot. -a) and *-ahah (> OAv. masiiayho 'id.', OP bagaha 'gods'; cf. Ved. -asah), but *-ah could well have survived in the dialects ancestral to Ossetic. OP -a and early Sogd. fty' [Pay-a] 'gods' can go back to either PIr. *-a or *-ah, but the Av. and Khot. endings can only continue *-a.
in *-want-. 13 However, Cheung [2002: 62] rightly points out that these forms could have been remodeled to *-mah, *-wah after the preponderant masculine a-stems, so they are not probative examples of a phonetic development to Oss. -0 (pace [Tremblay 1997: 159, fn. 5]). Cheung [2002: 61, 185-186] adduces fanda/fand 'plan, intention' < PIr. *pantah 'way, road' < PIE *pent-oh2-s (YAv. panta, Khot. pande; cf. Ved. panthah) as evidence for the regular treatment of word-final *-ah, but this noun could also reflect a remade a-stem *panta (cf. NPpand 'advice, opinion'; [Abaev 1958-1995, 1: 445], which in turn served as the basis for the extended *panta-ka- > Oss. fandag 'way, road'.
There are thus to my knowledge no clear counterexamples to a development PIr. *-ah > POss. *-i. The phonetic evolution cannot be determined with certainty: possibly *-ah first became *-e, as in Khotanese a-stem nom./acc. pl. -e < PIr. *-ah and pande 'path', ysare 'old age', urmaysde 'sun' < PIr. *pantah, *jarah, *ahura mazdah [Emmerick 1968: 278, 309, 347-348], 14 then was raised to *-i, merging with the genitive and inessive cases. The same ending may also have given rise to numera-tives in other Iranian languages: for instance, the Pashto numerative in -a (e.g. dwassla '40', dre zsra '3000',pinjsplara 'five fathers') could reflect m. nom./acc. du. *-a [Sims-Williams 1979: 341; 2020: 961] or n. nom./ acc. pl. *-a [Tremblay 1997: 171, fn. 31], but an origin in m. nom. pl. *-ah is also possible [Morgenstierne 1942: 94].
13 Since the original value of this suffix was 'provided with X, abounding in X' (cf. Ved. -va(n)t-, Gr. -(p)svx-), Cheung [2008: 95] argues that the equative was abstracted from *ma-want- 'like me', *0wa-want- 'like you' (OAv. mauuant-, Qfiauuant-, cf. Ved. ma-vant-, tva-vant-). Pace Cheung, the language suffix -aw (e.g. Ironaw 'in Ossetic'; cf. Sogd. - 'w, Khot. -au) is surely of the same origin.
14 The nom. pl. -e found with several masculine a-stems (beside the majority ending -a < PIr. *-a) may likewise continue PIr. *-ah [Emmerick 1968: 265], but of the examples listed there, ggare 'mountains' is an old i-stem and so may in fact continue PIr. *garayah, while old neuters such as ggune 'hairs', date 'wild beasts' could have taken over the ending from the a-stems. (I thank Nicholas Sims-Williams for helpful discussion of the Khotanese facts.)
4. The oblique cases
As for the oblique case forms, it is commonly assumed that pronominal inflection was generalized to numerals from adjectives of quantity, e.g. ijnna/inna 'other', agas 'all, whole', 15 then spread to nouns governed by numerals. The stem vowel *-e- (> D. -e-) allows for multiple sources. It could go back to PIr. pronominal m. nom. pl. *-ai reinterpreted as a stem suffix, comparable to what happened in D. ijnna 'other', be(w)ra 'many, much; very' (pl. inne-ta, bere-ta) < POss. *ijnne, *bewre < (post-)PIr. *anyai, *baiwarai 'ten thousands' ([Cheung 2002: 64, 6566]; see above, §2). Another possible source is the PIr. pronominal endings gen. pl. *-ai-sam, dat. *-ai-byah, loc. *-ai-su (cf. OAv. rel. pron. yaesqm, yaebiio, yaesü [Kim 2003: 45, fn. 9]). 16
However, the oblique *-e- of numerals could have an entirely different origin, namely POss. *duwe 'two' (> duwa/dywa), which continues PIr. f./n. *duwai rather than m. *duwa. 17 From the case forms of 'two', e.g. gen. *duw-e-j, dat. *duw-em-an (> D. duwej, duweman), the stem vowel *-e- spread to the higher numerals, producing e.g. gen. *avd-e-j, dat. *avd-em-an 'seven' (> D. avdej, avdeman). The extension of a feature from 'two' to all higher cardinal numerals might seem surprising, but an approximate parallel may be adduced, once again from Slavic: the instrumental dual ending -ma has spread in several modern Slavic languages from 'two' to 'three' and 'four' (e.g. Ru. dvumjá, tremjá, cetyr 'mjá, Pol. m. dwoma/f. dwiema, trzema, czterema), and in Polish to all numerals up to 'nine hundred' (e.g. piqcioma 'five', dziesiqcioma 'ten', trzydzie-stoma 'thirty', stoma 'hundred', trzystoma 'three hundred'). Even more
15 Cf. [Miller 1903: 49, §55 ("durch den Einfluss der Pronominaldeclination"); Isaev 1987: 592; Edel'man 1990: 199; Kim 2003: 49, fn. 18; Thordarson 2009: 185]. For a parallel from Slavic, see §1 above on the (optional) pronominal inflection of the numerals 'five' and over in Ukrainian.
16 Pace [Belyaev 2010: 291, 294], I see no grounds for assuming a postposed demonstrative pronoun *ai-.
17 See [Cheung 2002: 63; Kim 2003: 62, fn. 48; 2007: 50; 2020: 261].
Table 2. Evolution of 'other', 'two', 'seven', 'son(s)', and 'seven sons' from Proto-Iranian to Ossetic
Proto-Iranian POss. Digor
*anyah ^ *anyai ^ *ijnne ^ obl. *ijnne(m)-*anyai ^ *anyai-ta ^ *ijnne-tœ *ijnne *ijnne(m)- *ijnne-tœ ijnnœ ijnne(m)- ijnnetœ
*duwai > *duwe ^ obl. *duwe(m)- *duwe *duwe(m)- duwœ duwe(m)-
*hafta ^ *avd ^ obl. *avde(m)- *avd *avde(m)- avd avde(m)-
*puQrah > *puQri > *puQräh ^ *puQri-tä > *hafta puOräh > *avd furte ^ obl. *avdfurte(m)- *furt *furt-tœ *avd furti *avd furte(m)- furt furttœ avd furti avd furte(m)-
strikingly, the genitive/locative dual ending -u of Pol. dwu 'two' has also spread to the numerals 'five' to 'ten' (e.g. piqciu 'five', dziesiqciu 'ten'), then to higher numerals (e.g. trzydziestu 'thirty', stu 'hundred', trzystu 'three hundred'). 18 Once the oblique numeral stem in -e- was established, with the allomorph -em- influenced by pronominal inflection,
18 The Russian development is actually more complex, since the instrumentals of 'two' (ORu. duvema) and 'three'/'four' (ORu. trlmi, cetyrlmi) have mutually influenced each other. The Polish numerals in -u have become the general form for all oblique cases, instrumental (alongside -ma) and dative as well as genitive and locative, and also the nominative for masculine personal ("virile") nouns. For the chronology and details, see [Bulakhovskij 1958: 197-199] for Russian and [Klemensiewicz et al. 1965: 338-353, 354-356] for Polish. On the role of the dual in bringing about replacement of the acc. pl. with the gen. pl. in these languages, and the shift of -u and -ma from dual to virile markers, see [Janda 1996: 175-202, 1998, 2000].
it would have spread to nouns governed by numerals as under the preceding analysis.
Thus even if a dual source is ruled out for numerative -i/-y (§2), a trace of dual inflection could survive in the inflection of the numerals themselves, as suggested by Sims-Williams [2020: 966]. The inflectional evolution of these categories therefore passed through the approximate stages in Table 2, illustrated with the reflexes of PIr. *anya- 'other', *duwai 'two', *hafta 'seven', and *pu6ra- 'son'.
5. Summary
Despite parallels in Sogdian and other Iranian languages, the Ossetic numerative in -i/-y is unlikely on formal grounds to go back to a dual ending. It could instead continue PIr. a-stem and m. a-stem nom. pl. *-ah, which in all other contexts was replaced by the new plural formation in *-ta-. The pronominal inflection of the numerals, with stem-final element *-e(m)- in the oblique cases, could have been taken over from adjectives of quantity; alternatively, it could have begun in POss. *duwe 'two' and spread from there to the higher numerals. Finally, the inflectional pattern was extended to nouns governed by numerals. The Proto-Ossetic state of affairs is preserved in Digor, while Iron has generalized ordinary noun inflection for numerals and numeral-governed nouns.
Abbreviations
abl. — ablative; dat. — dative; du. — dual; f. — feminine; gen. — genitive; in-str. — instrumental; loc. — locative; m. — masculine; n. — neuter; nom. — nominative; obl. — oblique; pl. — plural; rel. pron. — relative pronoun; sg. — singular.
Av. —Avestan; BCSM—Bosnian, Croatian, Serbian, Montenegrin; D. — Digor; I. — Iron; Khot. — Khotanese; NP—New Persian; OAv. — Older Avestan; OP—Old Persian; ORu. — Old Russian; Oss. — Ossetic; PCelt. — Proto-Celtic; PIE—Proto-Indo-European; PIr. — Proto-Iranian; Pol. — Polish; POss. — Proto-Ossetic; Ru. — Russian; Sogd. — Sogdian; Ved. —Vedic.; YAv. — Younger Avestan.
References
Abaev 1949—V. I. Abaev. Osetinskijjazyki fol'klor [Ossetic language and folklore]. Vol. 1. Moscow; Leningrad: USSR Academy of Sciences Press, 1949.
Abaev 1958-1995 — V. I. Abaev. Istoriko-etimologicheskij slovar osetinskogo jazyka [Historical-Etymological Dictionary of the Ossetic language]. 5 vols. Moscow: Russian Academy of Sciences, Institute of Linguistics, 1958-1995.
Babby 1987—L. H. Babby. Case, prequantifiers, and discontinuous agreement in Russian. Natural Language and Linguistic Theory. 1987. Vol. 5. Iss. 1. P. 91-138.
Bailey 1945 — H. W. Bailey. Asica. Transactions of the Philological Society. 1945. Vol. 44. Iss. 1. P. 1-38.
Bailey 1956—H. W. Bailey. Armeno-Indoiranica. Transactions of the Philological Society. 1956. Vol. 55. Iss. 1. P. 88-126.
Bailyn, Nevins 2008 — J. F. Bailyn, A. Nevins. Russian genitive plurals are impostors. A. Bachrach, A. Nevins (eds.). Inflectional Identity. Oxford: Oxford University Press, 2008. P. 237-270.
Belie 2003 — B. Belie. Grammatical number in Serbian: singular—paucal—plural. Ohio State University Working Papers in Slavic Studies. 2003. Vol. 2. P. 1-22.
Belie 2008—B. Belie. Minor paucal in Serbian. G. Zybatow, L. Szucsich, U. Junghanns, R. Meyer (eds.). Formal Description of Slavic Languages. The Fifth Conference, Leipzig 2003. Frankfurt am Main: Peter Lang, 2008. P. 258-269.
Belyaev 2010 — O. Belyaev. Evolution of case in Ossetic. Iran and the Caucasus. 2010. Vol. 14. P. 287-322.
Belyaev 2014 — O. Belyaev. Osetinskij kak jazyk s dvukhpadezhnoj sistemoj: grup-povaja fleksija i drugie paradoksy padezhnogo markirovanija [Ossetic as a language with a two-case system: group inflection and other paradoxes of case marking]. Voprosy jazykoznanija. 2014. Vol. 6. P. 31-65.
Belyaev 2017 — O. Belyaev. The numerative in Iranian. Paper read at the 7th International Conference on Iranian Linguistics, Moscow, 28-30 August 2017.
Belyaev 2021 — O. Belyaev. Paradigm structure influences syntactic behaviour: Ossetic case inflection. I W. Arka, A. Asudeh, T. H. King (eds.). Modular Design of Grammar: Linguistics on the Edge. Oxford: Oxford University Press. 2021. P. 251-281.
Bielmeier 1982—R. Bielmeier. Zur Entwicklung der ossetischen Deklination. Indogermanische Forschungen. 1982. Vol. 87. P. 58-69.
Bulakhovskij 1958 — L. A. Bulakhovskij. Istoricheskij kommentarij k russkomu lite-raturnomu jazyku [Historical commentary on the Russian literary language]. 5th ed. Kiev: Radjanska Shkola, 1958.
Cheung 2002 — J. Cheung. Studies in the Historical Development of the Ossetic Vo-calism. Wiesbaden: Reichert, 2002.
Cheung 2008—J. Cheung. The Ossetic case system revisited. A. Lubotsky, J. Schae-ken, J. Wiedenhof, R. Derksen, S. Siebinga (eds.). Evidence and Counter-evidence: Essays in Honour of Frederik Kortlandt. Vol. 1: Balto-Slavic and Indo-European Linguistics. Amsterdam; New York: Rodopi, 2008. P. 87-105.
Christol 1990 —A. Christol. Introduction à l'ossète: éléments de grammaire comparée. LALIES: Actes des sessions de linguistique et de littérature. 1990. Vol. 8. P. 7-50.
Christol 2003 — A. Christol. Les deux pluriels de l'ossète. B. Hourcade (ed.). Iran: Questions et connaissances. Actes du IVe Congrès Européen des Études Iraniennes, organisé par la Societas Iranologica Europœa, Paris, 6-10 Septembre 1999. Vol. III: Cultures et sociétés contemporaines. Leuven: Peeters, 2003. P. 29-44.
Durkin-Meisterernst 2009 — D. Durkin-Meisterernst. Khwarezmian. G. Windfuhr (ed.). The Iranian Languages. London; New York: Routledge, 2009. P. 336376.
Èdel'man 1990 — D. I. Èdel'man. Sravnitelnaja grammatika vostochnoiranskikh ja-zykov. Morfologija, elementy sintaksisa [Comparative grammar of Eastern Iranian languages. Morphology, elements of syntax]. Moscow: Nauka, 1990.
Emmerick 1968 — R. E. Emmerick. Saka Grammatical Studies. London: Oxford University Press, 1968.
Erschler 2019—D. Erschler. Ossetic. G. Haig, G. Khan (eds.). The Languages and Linguistics of Western Asia: An Areal Perspective. Berlin; Boston: Mouton de Gruyter, 2019. P. 861-891.
Franks 1994 — S. Franks. Parametric properties of numeral phrases in Slavic. Natural Language and Linguistic Theory. 1994. Vol. 12. Iss. 4. P. 597-674.
Isaev 1966 — M. I. Isaev. Digorskij dialekt osetinskogo jazyka. Fonetika. Morfologija [The Digor dialect of Ossetic. Phonetics. Morphology]. Moscow: Nau-ka, 1966.
Isaev 1987—M. I. Isaev. Osetinskij jazyk [Ossetic]. V. S. Rastorgujeva (ed.). Os-novy iranskogo jazykoznanija. Novoiranskie jazyki: Vostochnaja gruppa. Moscow: Nauka, 1987. P. 537-643.
Janda 1996 — L. A. Janda. Back from the Brink: A Study of How Relic Forms in Languages Serve as Source Material for Analogical Extension. Munich; Newcastle: Lincom Europa, 1996.
Janda 1998 — L. A. Janda. Linguistic innovation from defunct morphology: old dual endings in Polish and Russian. R. A. Maguire, A. Timberlake (eds.). American Contributions to the Twelfth International Congress of Slavists. Blooming-ton, Ind.: Slavica, 1998. P. 431-442.
Janda 2000—L. A. Janda. From number to gender, from dual to virile: bridging cognitive categories. E. Contini-Morava, Y. Tobin (eds.). Between Grammar and Lexicon. Amsterdam; Philadelphia: Benjamins, 2000. P. 73-86.
Kim 2003 — R. I. Kim. On the historical phonology of Ossetic: the origin of the oblique case suffix. Journal of the American Oriental Society. 2003. Vol. 123. Iss. 1. P. 43-71.
Kim 2007—R. I. Kim. Two problems of Ossetic nominal morphology. Indogermanische Forschungen. 2007. Vol. 112. P. 47-68.
Kim 2020 — R. I. Kim. The numerals 'one' to 'ten' in Ossetic. H. Bichlmeier, O. Sefcik, R. Sukac (eds.), Etymologus: Festschrift for Vaclav Blazek. Hamburg: Baar, 2020. P. 257-265.
Klemensiewicz et al. 1965 — Z. Klemensiewicz, T. Lehr-Splawinski, S. Urbanczyk. Gramatyka historyczna jgzyka polskiego. Warszawa: Panstwowe Wydawnictwo Naukowe, 1965.
Miller 1881-1887—V. F. Miller. Osetinskie etjudy [Ossetic studies]. Moscow: Im-peratorskij Moskovskij Universitet: 1881-1887.
Miller 1903 — W. Miller. Die Sprache der Osseten. G. von Wilhelm, E. Kuhn (eds.). Grundriß der iranischen Philologie. Anhang zum ersten Band. Straßburg: Trübner, 1903.
Morgenstierne 1942 — G. Morgenstierne. Archaisms and innovations in Pashto morphology. Norsk Tidsskrift for Sprogvidenskap. 1942. Vol. 12. P. 88-114.
Naylor 1972 — K. E. Naylor. On some developments of the dual in Slavic. International Journal of Slavic Linguistics and Poetics. Vol. 15. P. 1-8.
Nurmio 2019 — S. Nurmio. Grammatical Number in Welsh: Diachrony and Typology. Malden, MA: Wiley-Blackwell, 2019.
Nurmio, Willis 2016 — S. Nurmio, D. Willis. The rise and fall of a minor category: the case of the Welsh numerative. Journal of Historical Linguistics. 2016. Vol. 6. Iss. 2. P. 297-339.
Ossetic National Corpus. Available at: http://corpus.ossetic-studies.org (accessed 20.08.2020).
Pesetsky 2013 — D. Pesetsky. Russian Case Morphology and the Syntactic Categories. Cambridge, MA; London: The MIT Press, 2013.
Rakhlin 2003 — N. Rakhlin. Genitive of quantification in Russian: what morphology can tell us about syntax. M. van Koppen, M. de Vos (eds.). Proceedings of CONSOLE XI. Leiden: Student Organization of Linguistics in Europe, 2003. Available at: http://www.hum2.leidenuniv.nl/pdf/lucl/sole/console11/console11-rakh-lin.pdf (accessed 20.08.2020).
Rappaport 2002 — G. C. Rappaport. Numeral phrases in Russian: a minimalist approach. Journal of Slavic Linguistics. 2002. Vol. 10. Iss. 1-2. P. 329-342.
Sims-Williams 1979 — N. Sims-Williams. On the plural and dual in Sogdian. Bulletin of the School of Oriental and African Studies. 1979. Vol. 42. Iss. 2. P. 337-346.
Sims-Williams 2020 — N. Sims-Williams. Dual and numerative in Middle and New Iranian. S. Badalkhan, G. P. Basello, M. de Chiara (eds.). Iranian Studies in Honour of Adriano V. Rossi. Napoli: UniorPress, 2020. P. 955-970.
Tedesco 1921 [1922] — P. Tedesco. Über den Nominativ Pluralis im Mittel- und Neuiranischen. Anzeiger der Akademie der Wissenschaften in Wien, Philosophischhistorische Klasse, 58. Jahrgang. 1921 [1922]. P. 58-70.
Thordarson 1989a—F. Thordarson. Ossetic. R. Schmitt (ed.). Compendium Lingu-arum Iranicarum. Wiesbaden: Reichert, 1989. P. 456-479.
Thordarson 1989b—F. Thordarson. Sibilanten und Affrikaten im Ossetischen. Georgica. 1989. Vol. 12. P. 14-17.
Thordarson 2009—F. Thordarson. Ossetic Grammatical Studies. Wien: Verlag der Österreichischen Akademie der Wissenschaften, 2009.
Tremblay 1997—X. Tremblay. Numeratifs et comprehensifs dans le Videvdat. Studia Iranica. 1997. Vol. 26. Iss. 2. P. 157-172.
