The prehistory of Ossetic verbal inflection (II): The copula

The prehistory of Ossetic verbal inflection (II): The copula Ronald I.Kim Adam Mickiewicz University in Poznań ABSTRACT The forms of the Ossetic copula, which diverge significantly between the two principal dialects Digor and Iron, have all been derived from Proto-Iranian sources, but many details of their development remain uncertain. In the singular, apronominal stem, probably *ta-, was grammaticalized as acopula and extended to 1sg d-æn and 2sg d-æ, the remaining portions of which go back to PIr. 1sg *ahmi, 2sg *ahi. Other pronouns were then grammaticalized in 3sg Iron uand Digor æj, whereas Digor je(s), Iron i(s) reflects PIr. 3sg *asti. In the plural, Digor 1pl an, 2pl atæ continue PIr. subjunctive *ah-aor present indicative forms remodeled on the stem *ah-; while Iron ystæm, ystut, ysty are in origin present forms of PIr. *staH- ‘stand’. KEYWORDS Iranian, Ossetic, copula, suppletion, grammaticalization 1. INTRODUCTION1 The first installment of this series (Kim 2022) examined the prehistory of the Ossetic person-number verbal endings in the present indicative and imperative, but due to space restrictions the copula could not be included. As seen in (1), the forms of the copula diverge significantly between the two principal dialects, Digor and Iron. (1) Digor Iron Digor Iron 1sg dæn dæn 1pl an (y)stæm 2 dæ dæ 2 ajtæ (y)stut 3 æj, je(s) u, i(s) 3 æncæ (y)sty In addition to brief discussions by Miller (1903: 74–76) and Thordarson (1989: 477), the prehistory of these paradigms was examined by Weber (1983) and Cheung (2002: 141–142), but aconsensus has not yet been reached on the interpretation of their 1 Iam indebted to Agnes Korn for her comments and suggestions, which have enormously improved the clarity of argumentation. Where two Ossetic forms are separated by aslash, the first is in the Digor dialect, the second in Iron. Abbreviations: B, C, MSo. = Buddhist, Christian, Manichean Sogdian; D = Digor; I= Iron; ModP = Modern Persian; O, YAv. = Older, Younger Avestan; (P)Oss. = (Proto-)Ossetic; PIE = Proto-Indo-European; PInIr. = Proto-Indo-Iranian; PIr. = Proto-Iranian; Ved. = Vedic; Yagh. = Yaghnobi. OPEN ACCESS 20 CHATREŠŠAR 2/2021 numerous peculiarites. It will be shown in the following pages that all forms may be plausibly derived from Proto-Iranian sources, and that the evolution of the copula illustrates phenomena of crosslinguistic interest far beyond Ossetic, including grammaticalization of pronominal stems and introduction of stem suppletion. 2. THE FIRST AND SECOND SINGULAR As already seen by Miller (Oss. St.II: 191), the elements -æn, -æ in 1sg dæn, 2sg dæ continue PIr. *ahmi and *ahi, respectively. This view is generally accepted,2 but the phonological development has never been investigated in detail. In the 1sg, final -n may be explained in the same three ways as in prs.ind.1sg -un/-yn, i.e. by (2) a. early apocope of *-ifollowed by *-m > *-n; b. regular apocope of *-ifollowed by shift of newly final *-m > *-n (with *-m then restored in nominal paradigms, e.g. *nam ‘name’ > non/nom); or c. analogical generalization from PIr. prs.subj.1sg *-āni > Oss. -on.3 In the 2sg, *ahi was contracted to *ai and monophthongized to POss. *e. As this should have remained in afully stressed monosyllable (cf. PIr. prohibitive particle *mā > POss. *ma > ma), the weakening to æ in both dialects must have been generalized from atonic position, as in (3a); cf. also the regular treatment of POss. *-e> D, I-æ in polysyllables (3b).4 (3) a. gen/acc clitic mæ ‘me’, dæ ‘you’ < PIr. gen/dat *mai, *tai b. ændæ/æddæ ‘outside’ < POss. *ænde < PIr. loc *antai ‘at the end’ (cf. Ved. ánta- ‘end, limit’) dælæ ‘below, beneath’ < POss. *dæle < PIr. loc *adarai (YAv. aδara- ‘lower, western’) duuæ/dyuuæ ‘two’ < POss. *duwe < PIr. f/n *duwai (YAv. duiie, duuaē-ca, Khotanese dvī; cf. Ved. duvé) The forms -æn, -æ are retained in future 1sg -ʒæn-æn/-ʒyn-æn, 2sg -ʒæn-æ/-ʒyn-æ, which goes back to aperiphrastic construction with PIr. *čanah- ‘desire’; 5 otherwise they have been expanded to dæn, dæ with an initial dof disputed origin. Older hy2 See Benveniste 1959: 75, Testen 1997: 715. 3 For details, see Kim 2022: 156–157. Cheung (2002: 141) and Korn (2011: 56fn.10) favor the last option, in which case the subj ending would surely have influenced the general prs. ind.1sg *-un as well. 4 On clitic mæ, dæ see Testen 1996: 369, Cheung 2008: 97. Weber (1983: 89) suggests adevelopment of PIr. *ahi > *ah > *æ, but the “frühe[r] Verlust” of *-iwould have to have been early indeed, before loss of intervocalic *h and contraction. For POss. *-e> -æ in polysyllables, see Cheung 2002: 65, 2008: 101, Kim 2007: 53–54, 2020: 261. 5 See Benveniste 1959: 75, Testen 1997: 715. OPEN ACCESS RONALD I.KIM 21 potheses compared this dwith preverbal elements or adverbs elsewhere in Iranian or assumed an analogical origin, but all are formally or semantically problematic.6 Weber (1983: 90) first suggested that the dof dæn, dæ goes back to pronominal *taor *aita- ‘this’. This explanation is attractive, since grammaticalization of demonstrative pronouns as copulas is acrosslinguistically common phenomenon with numerous parallels in Iranian, e.g. So. 3sg/pl (’)xw, Yagh. 3sg =x from the distal pronoun So. (’)xw, Yagh. ax < PIr. *hau-/*awa-.7 Korn opts for *aita-, comparing Pashto 3sg.m dəy, 3sg.f dā, 3pl dī and Wakhi prs tey, prt tu (Korn 2011: 56, 2020: 477; cf. Benveniste 1959: 74–75). The voicing of intervocalic *t is regular; the aphaeresis of *aiis not impossible in agrammaticalized copula, but one would then also expect aphaeresis in 3sg je(s)/i(s), which she derives from PIr. *aiša- (see §3). Ithus prefer to start from PIr. neuter *tat (Av. tat); the voicing of word-initial *tin weakly stressed or unstressed position has an exact parallel in the 2sg pronoun: PIr. nom *tuwam, acc *tawa > POss. *du, *dæw > du/dy, dæu.8 The pronominal stem was extended to the 1sg and 2sg to yield the attested forms (cf. Korn 2011: 56–57, 61). These developments are represented below in (4). (4) PIr. POss. 1sg *ahmi > *am > *an → *dæn 2 *ahi > *ai > *e → *de 3 *tat > *da > *dæ 3. THE THIRD SINGULAR Of the 3sg copula forms, je(s)/i(s) can easily be inherited: PIr. *asti > *asty > *aist > POss. *es(t) > D je(s), Ii(s), with regular i-umlaut as in fest(æg)/fistæg ‘pedestrian; rolled-up (sleeve)’ < PIr. *pasti- (OP pasti- ‘foot-soldier’, Ved. pattí- ‘pedestrian, footsoldier’).9 Several authors have connected these forms with the PIr. demonstrative pronoun *aiša- (Weber 1983: 87, Thordarson 2009: 6, Korn 2011: 56), which is formally acceptable, though the absence of aphaeresis contrasts with that assumed by Korn for 6 See Müller 1864: 528, Salemann 1876: 85, Miller Oss. St.II: 191 (comparing preverbal elements of the shape dVin Mazandarani and Kurdish), Salemann 1883: 143 (from *du d(u) æ, whence by analogy 1sg æz dæn), Miller Oss. St.III: 165 (comparing ΥAv. haδa ‘always’), 1903: 75 (connecting dwith æd- < *hadā ‘(together) with’; semantically improbable, cf. Benveniste 1959: 74, Weber 1983: 90), Tomaševskij 1925: 77–80 (geminate dental introduced into intransitive preterites, e.g. cæddæn, cæddæ ‘I, you went’, whence copula dæn, dæ; why only 1sg and 2sg?), Axvlediani 1948: 82 [1960: 159] (borrowing from Ingush da). 7 See in general Korn 2011, and for this example Benveniste 1959: 75, Weber 1970: 21–24, 1983: 88, Korn 2011: 55–63, Benkato 2020: 7–8, 10–11, 11–15, 20. 8 Asimilar change has taken place in Armenian: dow ‘you.sg’ < PIE *tuh2; d-a, -d < PIE *to-. Cf. also the voicing of initial [θ-] > [ð-] in English function words: the, this, that, there, etc. 9 See already Müller 1864: 528, Salemann 1876: 84, Miller Oss. St.II: 190, 1903: 75 (tentative); further Benveniste 1959: 75, Isaev 1987: 622, Christol 1990: 39 (but not from *asty-), Cheung 2002: 100, 141, 194. OPEN ACCESS 22 CHATREŠŠAR 2/2021 *da- < PIr. *aita- (see §2). If the Alanic phrase τ φάρν-ετζ in Tzetzes’s Theogony stands for /dæw farn-e(s)c/ [pron.2sg.gen glory-cop.3sg] ‘yours is the glory’ (Lubotsky 2015: 58), it would constitute strong support for aderivation from *asti. The reduction of the final consonant cluster is widely paralleled in IE languages, from Germanic (Old English is vs. Gothic, Old High German ist) to Romance (Spanish es, Italian è < Latin est) to Slavic (Czech je vs. Polish jest) to Iranian itself (ModP colloquial -ebeside literary ast). Alongside this form are two others that are unrelated to one another and strongly recall pronominal stems: Iu< PIr. *hauand/or *awa- (cf. Iu, D obl wo- ‘that’); and D æj, probably from PIr. *aya- (cf. D nom je ‘that’).10 Apronominal origin for these forms is supported by their usage, with je(s)/i(s) serving as existential verb and æj/uin identificational function, as in the Iron examples in (5).11 (5) a. mæ xælar sk’olæ-jy i(s) pron.1sg.gen friend school-loc cop.3sg ‘my friend is in/at school’ b. mæ xælar student u pron.1sg.gen friend student cop.3sg ‘my friend is astudent’ Most likely, Iuand D æj replaced POss. *dæ after the latter fell together with 2sg dæ < POss. *de. Since that merger took place separately in the two dialects (see §2, (4)), it is not surprising that they chose different stems. 4. THE DIGOR PLURAL FORMS The forms of the two dialects diverge in the plural, where only the D paradigm stands achance of continuing the old present of *ah- ‘be’ (Weber 1983: 84–85). D 3pl æncæ could be from PIr. *hanti, with the same replacement of *t with *θ under the influence of the 2pl as in prs.ind.3pl *-anti → *-anθi > *-anty > *-ænc → POss. *-unc(æ) > 10 See Miller 1903: 75, Benveniste 1959: 74–75, Weber 1970: 24fn.57, Isaev 1987: 622–623, Christol 1990: 39–40 (on Iu); Weber 1983: 86–87, Thordarson 2009: 5–6, Korn 2011: 56. Weber (1983: 87) and Korn (2011: 56, 57) separate jes/is from je/iand take the latter along with D æj from PIr. *aya-/*i-, which is uneconomical and does not address the formal gap between D je and æj. 11 This functional distribution speaks against acomparison of Iuwith the present of PIr. *bawa-, which in Ossetic has the meanings ‘become, grow’ or ‘be/appear/happen sometimes, once in awhile’ (Miller Оss. St.II: 192, 1903: 75, Isaev 1987: 622, Cheung 2002: 141; see Kim 2022: 164–5). It also further disfavors the idea of Benveniste (1959: 75–76; cf. Christol 1990: 40) that D æj < *hati was remodeled after the 3pl *hanti, which is highly improbable on typological grounds (no comparable reshaping of the 3sg on the basis of the 3pl can be adduced from any other IE language). OPEN ACCESS RONALD I.KIM 23 -uncæ/-ync and impv.3pl *-antu → *-anθu → POss. *-æntæ > -æntæ/-ænt.12 In contrast, 1pl an, 2pl ajtæ can hardly continue PIr. *hmah(i), *sta (OAv. mahī, stā; cf. Ved. smási, sthá), which would have been perceived as irregular very early on and thus obvious candidates for remodeling. Generalization of full-grade *ahfrom the singular to 1pl *ahmah(i), 2pl *ahta would be astraightforward step, as in OP 1pl amahi.13 However, examples such as PIr. cop.1sg *ahmi > *am → POss. *d-æn (§2) and PIr. *hazahra- > *azara- > *azra- > POss. *ærze ‘thousand’ show that there was no lengthening of the vowel in OIr. sequences *ahC. The preforms *ahmah(i), *ahta would thus have given *amah, *ata > *æm, *ætæ, not the required POss. *am, *atæ. The only realistic source for the aof D 1pl an, 2pl ajtæ is asequence *ahā- with loss of intervocalic *h and contraction to *ā, as in PIr. *upa-had- > *bād- > POss. *bad- > badun/badyn ‘sit’.14 This leads to two possibilities. The D forms could go back to subjunctives, namely PIr. *ahāma, *ahaθa (OAv. 1pl āŋhāmā; cf. Ved. ásāma, ásatha).15 Their generalization could be attributed to confusion following the loss of *h, leading speakers to prefer the subjunctive forms to the opaque reflexes of PIr. *hmah(i), *sta. Alternatively, A.Korn (p.c.) calls to my attention the Parthian copulas ahāz and ahād, which have been explained as impf.3sg *āhat (OP āha) > *āδ and subj.3sg *ahati > *ād attached to the present stem to give *ah-āδ, ah-ād.16 Asimilar recharacterization could have taken place in pre-Ossetic, whereby 1pl *ahmah(i), 2pl *ahta with analogical fullgrade root (see above) became *amah(i), *ata and were remade to *ah-amah(i), *ah-ata. In either case, the forms would have given *āmah, *āθa with contraction of *aha > *ā and generalization of the usual endings, 1pl *-mah and 2pl *-θa (see fn. 12), whence POss. *am, *atæ. The expected D reflexes of these are †on, †atæ. The secondary -jof 2pl ajtæ is also found in other categories, e.g. subj.2pl -ajtæ;17 and the aof the 1pl was apparently leveled from the 2pl, since sequences of aplus nasal are extremely rare in both Oss. dialects due to the late (post-Alanic) change of *aN > oN. 5. THE IRON PLURAL FORMS Turning to Iron, Miller, following aproposal of Salemann (1883: 143), understood I 1pl (y)stæm, 2pl (y)stut to be formed on the basis of 3sg *asti, citing as parallels ModP hast-īm, hast-īd and Polish jest-eśmy, jest-eście (Miller Оss. St.II: 191, 1903: 75–76; Isaev 12 See Kim 2022: 162–163; first observed for the imperative by Benveniste (1959: 76). Ossetic has generalized PIr. *θ > *t from prs.ind.2pl *-aθa > POss. *-tæ (→ -etæ/-ut) to all 2pl endings. 13 See Miller Оss. St.II: 191, 1903: 75, Isaev 1987: 623, Christol 1990: 40. 14 See Thordarson 1989: 464, Cheung 2007: 126 s.v. *had. 15 Tentatively mentioned by Weber 1983: 85 (“analoger Ausgleich wie etwa *ah-amit Vollstufe und sekundärer Thematisierung (oder eventuell auch Konjunktivformen)”). 16 See Jügel 2015: 128–132 for discussion and references. 17 Isee no way of determining whether this change influenced the generalization of the optative ending in prs.ind.2pl -etæ (Miller 1903: 70) or vice versa (Miller Oss. St.II: 191–192, 1903: 75 [ajtæ < a+etæ], Weber 1983: 85fn.3), or is an independent innovation. OPEN ACCESS 24 CHATREŠŠAR 2/2021 1987: 623). Weber (1983: 86) took afurther step in positing (y)sty < *asti as the original 3sg form,18 which was then ousted by new copula forms of pronominal origin (< PIr.*aiša-, *awa-, *aya-; see §3) and shifted to 3pl function. Against this four objections may be raised: (6) a. the remodeling of the plural but not the singular on the basis of the 3sg would to my knowledge be without parallel (cf. ModP 1sg hast-am, 2sg hast-ī, Polish 1sg jest-em, 2sg jest-eś); b. the absence of i-umlaut in (y)sty < PIr. *asti is contradicted by festæg/ fistæg ‘pedestrian; rolled-up (sleeve)’ < PIr. *pasti-, which rather points to cop.3sg je(s)/i(s) < POss. *es(t) < *aist < *asty as the regular development of PIr. *asti (§3); c. the -yof intransitive preterites, which are composed of preterite participle + copula (e.g. prt.1pl cyd-ystæm, 2pl cyd-ystut, 3pl cyd-ysty ‘went’), and the underlying -yof prf.1pl festæm, 2pl festut, 3pl festy ‘turned out to be, became’ (fe- < fæ-y-; cf. prf.1sg fæ-dæn, 2sg fæ-dæ, 3sg fæ-c-i(s)) argue against aprothetic origin (Cheung 2002: 142); d. there is no evidence for confusion of the 3sg and 3pl at any point in the history of Oss. (Cheung op. cit.). As seen by Bielmeier (1977: 162–163), the Iplural forms rather go back to PIr. *staH- ‘stand’ (< PIE *steh2-) and have split off from the paradigm of D istun, Istyn ‘stand (up)’.19 Since the reduplicated present of this root was thematic already in PInIr. (cf. Ved. tíṣṭhati, YAv. hištaiti, OP impf aištatā),20 the 1pl and 2pl forms would have developed in the same way as those of ordinary presents, whence (y)stæm, (y)stut (cf. kænæm ‘we do’, kænut ‘you.pl do’). As for 3pl (y)sty with its peculiar short ending, this cannot directly reflect PIr. *hištanti, since word-final *-nti should have given POss. *-ndy > D -j, I-nʒ.21 It can 18 The idea goes all the way back to Bopp (1847: 48), but he later abandoned it and took (y)sty to be a“Verstümmelung” of (y)stync (op. cit.80n.39; see below, fn. 21). 19 Cf. Thordarson 1989: 477, 2009: 6, Cheung 2002: 142, Korn 2011: 56fn.10, 2020: 482. The connection was proposed already in the 19th century: see Bopp 1847: 48, 80n.39 (wrongly including 1sg dæn, 2sg dæ), Müller 1864: 528–529 (wrongly comparing ModP hastīm, hastīd, hastand). Copular forms from ‘stand’ occur frequently in western IE languages (Irslinger 2019) but are otherwise not found in Iranian, where PIr. *staHhas become grammaticalized in imperfective and continuous constructions (Benveniste 1966, Korn 2020: 481–482). 20 As also in Latin sistō and (with secondary *-ye/oinflection) Old Irish ·sissedar. These, along with Greek athematic ἵστημι, may be derived from aPIE h2e-conjugation present *stí-sth2- (Jasanoff 2003: 130–131). 21 Cf. D æfsoj, Iæfsonʒ ‘yoke’ < POss. *(æ)fsandy < OIr. *spān-ti-; also D fij, finʒ(æ), Ifynʒ ‘nose’ < POss. *findy, borrowed from Northwest Caucasian (cf. Abkhaz a-pɨnc’a, Abaza pɨnc’a). The parallel change of OIr. *-ti > POss. *-dy > D -j, I-ʒ is robustly supported, e.g. in insæj/ssæʒ ‘twenty’ < POss. *insædy < PIr. *winćati; kuj/kwyʒ ‘dog’ < POss. *kudy < PIr. *kutī- ‘bitch’. See Benveniste 1959:75, Christol 1990:14; for further examples, see Cheung 2002: 98–9, Kim 2007:60–4.— Suggestions that the final cluster was simply lost (Bopp 1847: 80n.39 “verOPEN ACCESS RONALD I.KIM 25 however continue ashortened byform *hištani, generalized from rapid speech. This would have become *istainy and then POss. *istiny, with i-umlaut followed by the regular change of *ai > *ibefore nasals; cf. (7) and for the latter (8).22 POss. *istiny would in turn have become (y)styj (attested in 19th-century sources), with evolution of wordfinal *-ny > I-j as in the noun-forming suffix -ojnæ/-oj (9), and finally modern I(y)sty. (7) PIr. *kani- → *kanikā > *kainiča > POss. *kinʒæ > kinʒæ/čynʒ ‘daughter-in-law, bride’ (MSo. knc ‘girl’; cf. Av. kainiiā-, kainīn-, Zazaki keynek)23 (8) PIr. *wain- ‘look at’ (GAv. vaēna-) > POss. *win- > winun/wynyn ‘see’ PIr. *-aina- (suffix forming adjectives of material; Av. -aēna-, OP -aina-, MP -ēn-) > POss. *-in > -in/-yn, e.g. ɣæd-in/qæd-yn ‘wooden’ to ɣædæ/qæd ‘wood’ PIr. *abi-šaiθnī- → *abišai(θ)nyā > *avišinya > POss. *æfsijnæ > æfsijnæ/æfsin ‘landlady, mother-in-law, housewife, hostess’ (9) PIr. *-āna- + *-yā- > *-ānyā- > POss. *-anyæ > -ojnæ/-oj, e.g. æncojnæ/æncoj ‘rest, support’, sagojnæ/sagoj ‘pitchfork’ (to æncon ‘light, easy’, sag ‘deer; *horned’)24 An analogous explanation underlies subj.3pl -oncæ/-oj, which should go back to PIr. thematic subj.3pl *-ānti (OAv. -āntī, YAv. -āṇti). The Iron ending reflects areduced allomorph *-āni, which became POss. *-any > I-oj by the change seen in (9). Digor on the other hand continues *ānti → *-ānθi → POss. *-ancæ > -oncæ, with replacement of *t with *θ after the 2pl just as in prs.ind.3pl *-anti → *-anθi → POss. *-unc(æ) > -uncæ/ -ync or cop.3pl *hanti → *hanθi → POss. *æncæ > D æncæ (§4). There is thus some basis for assuming avariant *hištani of *hištanti ‘they stand’, which by sound change yielded the Iron copula (y)sty ‘they are’. 6. CONCLUSION Most of the individual points discussed in this article have been previously proposed in the literature, but only when taken together does one obtain acoherent picture of the evolution of the copula from Proto-Iranian to the present. Although stümmelt”) or dropped to avoid homophony with 3pl stync ‘stand’ (Miller 1903: 76, Isaev 1987: 623) are entirely ad hoc. 22 For additional examples of *ai > *ibefore nasals, see Miller 1903: 18, Cheung 2002: 17, Kim 2003: 60–62. 23 On PInIr. *kaniand its derivatives, see EWAia I: 297–298 s.v. kanyā-. For the suffix *-ičā, see GMS: 39 §247, Testen 1994: 300–303; pace Abaev Dict. I: 607–608 (*kantī), Morgen - stierne 1962: 161–162, Cheung 2002: 13, 199 (< *kanīčī-). 24 See Cheung 2002: 136; on the cognate So. suffix [-ānī] forming action nouns, e.g. B ’krt’ny(h), M ’kt’nyy, C qt’nỵ [(ə)ktānī] ‘sin’ < *krtānyā to *krta- ‘done’, see Sims-Williams 1981: 16–17, 19. OPEN ACCESS 26 CHATREŠŠAR 2/2021 afew forms are phonologically (nearly) regular continuants of their PIr. sources, namely 1sg (d-)æn, 2sg (d-)æ, 3sg jes/is < PIr. *ahmi, *ahi, *asti (§§2, 3) or 3pl D æncæ < *hanθi ← PIr. *hanti (§4), the most notable changes are of wider crosslinguistic interest, such as the grammaticalization of pronominal stems in *ta- (> 1sg d-æn, 2sg d-æ), *hau-/*awa- (> Iu), and *aya- (> D æj) and the creation of asuppletive paradigm with forms of PIr. *staH- ‘stand’ in Ipl (y)stæm, (y)stut, (y)sty. The former has numerous parallels within and beyond Iranian, whereas the latter is isolated in Iranian but has arisen independently in IE languages of medieval and modern western Europe. The extremely scanty premodern attestation of Ossetic naturally precludes judgments as to chronology, but it is interesting that none of the changes posited here is exactly paralleled in Sogdian or Yaghnobi, the Iranian languages often considered to be most closely related to Ossetic,25 nor farther afield in e.g. Khotanese or the Pamir languages. Given the evidence already in Sarmatian proper names for loss of *h,26 which would have largely obscured the synchronic relation between OIr. prs.ind.1pl *hmah(i), 2pl *sta and prs.subj.1pl *ahamah(i), 2pl *ahaθa, the generalization of subjunctive forms in D an, ajtæ is likely to have occurred in Alanic or even Sarmatian times. The copular use of pronominal *tamay likewise be an old innovation, since this pronoun does not otherwise survive in Ossetic. On the other hand, the copular use of *staH- ‘stand’ may have begun already in Proto-Ossetic but ousted the inherited plural forms only in Iron, giving rise to asuppletive paradigm. The grammaticalization of new pronominal stems in the 3sg, *hauand/or *awain Iron and *ayain Digor, naturally took place after the dialects had begun to diverge. Acknowledgments 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). REFERENCES 25 See Bailey 1945, Sims-Williams 1989: 1970, Kim 2003: 67–68 and passim; the most famous isogloss is generalization of the originally collective suffix *-tāas plural marker (Oss. tæ, So. -t’, -t, Yagh. -t). Recent studies have arrived at amore nuanced picture of the position of Oss. within Iranian, however; see Wendtland 2009: 185 (Table 16) and passim. 26 Cf. retention of *h in Χανάκης (Pantikapaion; < OIr. *hana- ‘old’, cf. Av. hana-) vs. loss in ζαρίων (Tanais, 220 AD; < OIr. *hazahra- ‘thousand’, cf. YAv. hazaŋrəm), φθαίμακος ~ φθείμακος (Tanais, 3rd c. AD; < OIr. *haft-am-yāka- ‘seventh’, cf. Oss. ævdæjmag) and see Harmatta 1951: 307–308, 1970: 93–94, Bielmeier 1989: 241. Abaev Dict.: Абаев, В.И.Историкоэтимологический словарь осетинского языка [Historical-Etymological Dictionary of the Ossetic Language], Том I: A–K’ (1958), Москва & Ленинград: Издательство Академии Наук СССР. Том II: L–R (1973), Том III: S–T’ (1979), Том IV: U–Z(1989), Ленинград: «Наука» (Ленинградское отделение). ТомV: Указатель [Index], Москва: Российская Академия Наук, Институт Языкознания. Axvlediani, G. [Ахвледиани, Г.] (1948) Загадочный элемент д в осетинском дӕн, дӕ OPEN ACCESS RONALD I.KIM 27 [The mysterious element d in Ossetic dæn, dæ], in: Сообщения АН Грузинской ССР 9/1, 79–82. (Reprinted in Сборник избранных работ по осетинскому языку [Collection of Selected Works on Ossetic], Книга 1, Тбилиси: Издательство Тбилисского Государственного Университета им. Сталина [1960], 155–159.) Bailey, H.W. (1945) Asica, in: Transactions of the Philological Society [1945], 1–38. Benkato, A. (2020) Multiple copularization in Sogdian and Yaghnobi, in: Transactions of the Philological Society 118/1, 172–191. Benveniste, É. (1959) Études sur la langue ossète [Société de Linguistique de Paris, Collection Linguistique LX], Paris: Klincksieck. —. (1966) Le verbe stācomme auxiliaire en iranien, in: Acta Orientalia 30, 45–49. Bielmeier, R. (1977) Historische Untersuchung zum Erbund Lehnwortschatzanteil im ossetischen Grundwortschatz [Europäische Hochschulschriften, Reihe XXVII: Asiatische und Afrikanische Studien 2], Frankfurt am Main— Bern— Las Vegas: Peter Lang. —. (1989) Sarmatisch, Alanisch, Jassisch, und Altossetisch, in: R.Schmitt (ed.), 236–245. Bopp, F. (1847) Die kaukasischen Glieder des indoeuropäischen Sprachstamms, Berlin: Dümmler. Cheung, J. (2002) Studies in the Historical Development of the Ossetic Vocalism, Wiesbaden: Reichert. —. (2007) Etymological Dictionary of the Iranian Verb [Leiden Indo-European Etymological Dictionary Series 2], Leiden— Boston: Brill. —. (2008) The Ossetic case system revisited, in: Evidence and Counter-evidence: Essays in Honour of Frederik Kortlandt, Vol.1: Balto-Slavic and Indo-European Linguistics [Studies in Slavic and General Linguistics 32], Amsterdam— New York: Rodopi, 87–105. Christol, A. (1990) Introduction àl’ossète: éléments de grammaire comparée, in: LALIES: Actes des sessions de linguistique et de littérature 8, 7–50. GMS: Gershevitch, I. (1954) AGrammar of Manichean Sogdian [Publications of the Philological Society 16], Oxford: Blackwell. Harmatta, J. (1951) Studies in the language of the Iranian tribes in South Russia, in: Acta Orientalia Academiae Scientiarum Hungaricae 1/2–3, 261–314. —. (1970) Studies in the History and Language of the Sarmatians, Szeged: Szegedi nyomda. Irslinger, B. (2019) More tales of two copulas: the copula systems of Western European languages from atypological and diachronic perspective, in: R.I.Kim (ed.), Diachronic Perspectives on Suppletion [Studien zur historisch-vergleichenden Sprachwissenschaft 13], Hamburg: Baar, 27–76. Isaev, M.I. [Исаев, М.И.] (1987) Осетинский язык [Ossetic], in: V.S.Rastorgueva (ed.), Основы иранского языкознания. Новоиранские языки: Восточная группа [Foundations of Iranian Linguistics. New Iranian Languages: Eastern Group], Москва: «Наука», 537–643. Jasanoff, J.H. (2003) Hittite and the IndoEuropean Verb, Oxford— New York: Oxford University Press. Jügel, T. (2015) Die Entwicklung der Ergativkonstruktion im Altund Mitteliranischen. Eine korpusbasierte Untersuchung zu Kasus, Kongruenz und Satzbau [Iranica 21], Wiesbaden: Harrassowitz. Kim, R.I. (2003) On the historical phonology of Ossetic: the origin of the oblique case suffix, in: Journal of the American Oriental Society 123/1, 43–71. —. (2007) Two problems of Ossetic nominal morphology, in: Indogermanische Forschungen 112, 47–68. —. (2020) The numerals ‘one’ to ‘ten’ in Ossetic, in: H.Bichlmeier— O.Šefčík— R.Sukač (eds.), Etymologus: Festschrift for Václav Blažek [Studien zur historisch-vergleichenden Sprachwissenschaft 14], Hamburg: Baar, 257–265. —. (2022) The prehistory of Ossetic verbal inflection (I): present indicative and imperative, in: D.M.Goldstein— S.W.Jamison— B.Vine (eds.), Proceedings of the 32nd Annual UCLA Indo-European Conference, November 5th, 6th and 7th, 2021, Hamburg: Buske, 153–171. OPEN ACCESS