We use cookies to distinguish you from other users and to provide you with a better experience on our websites. Close this message to accept cookies or find out how to manage your cookie settings.
Published online by Cambridge University Press: 24 December 2009
It has long been recognized that the letters n and z, whose initial and medial forms had already fallen together by the date of the ‘Ancient letters’, remained distinct in final position, -n being provided with a long horizontal or vertical tail. At a slightly later period in the history of the Sogdian script two more pairs of letters—still distinguished in all positions in the ‘Ancient letters’— fell together: ‘ain with r,1 and x (ḥīth) with γ (gimel). The rare letter ‘ain does not occur in final position; but both x and γ are comparatively common in all positions. It is therefore remarkable that it should never have been noticed that final x and y are regularly distinguished in exactly the same way as -n and -z. The distinction is clearest in the Buddhist manuscripts,2 in which final x has a long horizontal or vertical tail: — (= ‘horizontal -x’) or L (= ‘vertical -x’), while final γ has only a rudimentary tail pointing down-wards. There is some variation in the distribution of the two forms of -x. In some manuscripts, e.g. Vim. and P 2, both are used interchangeably. In most ‘horizontal -x’ is preferred: it is the only form found in VJ, SCE, Dhu., P 5, and P 6, while in Dhy. and P 3 ‘vertical -x’ is also occasionally employed
1 See BSOAS, xxxv, 3, 1972, 614–15.Google Scholar
2 I have consulted the British Museum MSS in addition to the published facsimiles.
3 These MSS may differ also in the representation of final γ: cf. below on P 7.77 βr'ywγ.
4 See AM, NS, xviii, 1, 1973, 99, on 1. 88 of fragment.Google Scholar
5 Once apparently ’βs'nγ, P 2.S34, probably due to lack of space at the end of the line.
6 Man. pyx, see Henning, apud Boyce, The Manichaean hymn cycles in Parthian, London, 1954, p. 122, n. 4.Google Scholar
7 The first part of ’rp(’)wx‘ pénétrant’ perhaps belongs to Skt. arpáyati ‘ insert, fix, etc.’. For a comparison with NP rabūxe see Henning, , BSOS, x, 1, 1939, 102.Google Scholar
8 It is worth noting the transoription of Skt. -h- by -x in the Nīlakaṇṭhadhāraṇī, ed. Gauthiot and de La Vallée Poussin, JRAS, 1912, 2, pp. 629–45.
9 The reading of the latter is not quite clear: Benveniste's wrγ‘nt (?) is also possible.
10 For its being a heavy stem cf. GMS, §§ 484 and 527.
11 cf. also NP š***x: šaγny (= Sogd. šnx), Horn, Grundriss der iranischen Philologie, I, 2, 70.
12 See Livshitz, , in Boyce, M. and Gershevitch, I. (ed.), W. B. Henning memorial volume, London, 1970, 258.Google Scholar
13 This spelling is in accord with the etymology (which I owe to Dr. Gershevitch) from Av. Z***mō huδ***nhō. Man. zmwxtwy (Henning, Onenialia, viii, 1939, 92) may be due to dissimilation.
14 Confirming the interpretation first offered by Gershevitch, CAJ, vii, 2, 1962, 89–90.Google Scholar
15 JRAS, 1912, 1, p. 350; Grammaire sogdienne, I, Paris, 1914–23, 12.Google Scholar
16 Being tacitly accepted in Reichelt's edition, and quoted with approval by, for instance, Henning, BSOAS, xi, 4, 1946, 717, and Livshitz, art. cit., 259.Google Scholar
17 JRAS, 1912, 1, p. 351.
18 ibid., 351–2.
19 There can bo no question of an ideogram undergoing phonetic changes. The p of RYPW ‘ten thousand’ (Aram, rbw = ribbō) may indicate that the constant use of p for [b] in Sogdian script was not a Sogdian innovation but an inheritance from the Aramaic-speaking scribes of a much earlier period: the development of intervocalic [b] into a spirant [v] which led to this usage was as much a feature of Aramaic as of Sogdian. Alternatively, according to a suggestion of Dr. Gershevitch, one may have to reckon with the influence of 1-LPW ‘thousand’ (Aram. 1-LP; still spelt thus in the ‘Ancient letters’), which in turn may owe its -W to RYPW. In either case, the change would be graphic rather than phonetic.
20 BSOAS, xi, 4, 1946, 717.Google Scholar
21 Henning, , AM, NS, xi, 2, 1965, p. 174, n. 30.Google Scholar
22 Sogdian examples include ČWRH ‘self’ (Aram, ‘neck’), and ZY ‘and; that’ (Aram. ‘that’ only), in which two distinct Old Iranian words have merged. Cf. Henning, loc. cit.: ‘the ideogram inevitably would follow the meaning of its Iranian equivalent’.
23 All the letters of the Aramaic alphabet occur In the Sogdian script of the ‘Ancient letters’ with the exception of ṭ and d: Gauthiot's hypothesis (Gramm. sogd., I, 7–8), according to which Sogd. δ would continue d as well as l, was rightly discarded by Henning (e.g. ‘Mitteliranisch’, Handbuch der Orientalistik, Abt. I, IV Bd., Iranistik, 1, pp. 60–1; cf. also Livshitz, art. cit., 261). It is therefore hard to accept Henning's derivation (apud GMS, § 13361) of B γδnyw ‘one another, mutual(ly), together’ from Aram, ḥd ‘one’ (there can be no question of it breaching the principle stated by himself, AM, NS, xi, 2, 1965, 174: ‘the question whether the Aramaic was once uttered with -f- or -p- [in the present context, -δ- or -d-] is immaterial for the orthography’). The plural of γδnyw is possibly to be recognized in Chr. γdnyty (‘we are united in our *common concern’ ?), as Henning himself envisaged, cf. Schwartz, M., Studies in the texts of the Sogdian Christians, Ann Arbor, University Microfilms, 1968, 149,Google Scholar in which case the word will not be an ideogram. In view of Pa***to jabla ‘together’ <*hača-dwityā (Morgenstierne, An etymological vocabulary of Pashto, Oslo, 1927, 14) one might consider a derivation from hada + anyam (Sogd. γ- from h- is attested at least in γyztwq’ ‘spittle’ <*hazduka-, cf. Bailey, , BSOAS, xx, 1957, 59); but the spelling γδn'yw (also γnδS(n)'yw) seven times in P 2 tends to destroy confidence even in the interpretation of the second half as anyam.Google Scholar
24 Livshitz, art. cit., 259.
25 Dodge, B., The Fihrist of al-Nadīm, New York and London, 1970, I, 33;Google ScholarRoss, E. D. and Gauthiot, R., ‘L'alphabet sogdien’, JA, xie Sér, I, mai–juin 1913, 521–33.Google Scholar
26 In Inscriptional Parthian and Choresmian script, as in Sogdian, d, r, and ‘ain all have the same shape, cf. Henning, , AM, NS, xi, 2, 1965, 171.Google Scholar
27 This marks a return to the reading proposed by Cowley, JRAS, 1911, 1, p. 163, at a stage when the script was not fully deciphered nor even the language identified, and accepted initially by Gauthiot, see JRAS, 1911, 1, p. 506.Google Scholar
28 Occasionally also ‘in’ as in the Manichean poem quoted by Henning, TPS, 1945, 151: this meaning has developed from ‘to’ as in the case of Chor.f- (OChor. ‘L) ‘in, at, on’ <abi‘to’.
29 So far as I know, neither this nor any other explanation of tym has previously been proposed.
30 Geiger, , ‘Etymologie des Balūčī’, ABAW, xix, 1891, 119. The semantic development assumed here from ‘until’ to ‘to’ is illustrated by, for instance, Pa***to ta ‘to’ (if, as seems probable, this is to be equated with NP tā, etc.) or English to (which belongs with Latin dōnec, etc.).Google Scholar
31 The etymology implied by Gershevitch, CAJ, vii, 2, 1962, 87.Google Scholar
32 For the semantic proximity of the notions ‘through’, ‘beyond’, and ‘until’ one may compare the use of ‘through’ in American English.
33 Which may contain the article’w.
34 GMS, § 1632.
35 Only Mugh document A 18, R 5.
36 Already attested in the ‘Ancient letters’.
37 By means of the equation ’c: c’- = ’δ:δ’- = ’t: X. The forms parāmā, parāfā (<pr<upari), though not attested in the ‘Ancient letters’, could be oited as a parallel.
38 Quoted GMS, § 1632, from T ii T 31 (= M 7440); otherwise only ’t’ kw (in an address written as an exercise, M 7392, see Boyoe, , Catalogue of the Iranian manuscripts…., Berlin, 1960, 131), for which of. OP yātā ā, MMP d' ‘w, Parth. yd 'w, perhaps also Pa***to wa-ta (‘to’, Morgen- stierne, EVP, 79). The relationship of ’t’ to ’t’ and t is probably that of the stressed form (perhaps originally used with the value of a conjunction ‘until’, or, as in the example, adverbially with another preposition) to the proclitio and enclitio forms appropriate to the preposition and postposition, cf. such doublets as kδ’/kδ and.’ty/’t.Google Scholar
39 For a parallel phenomenon of. Skt. prati, Gk. πρoτί the semantic identity of which with OIr. pati, Gk. dialectal πorί suggests that IE *proti was formed by a contamination of *potiwith *pro. (This explanation was arrived at independently by Dr. Gershevitoh and myself.)
40 Henning, , AM, NS, v, 1, 1955, 47.Google Scholar
41 Of which a good example is ’fčwr- + d’ ‘to surpass’ (Henning, , ed. MaoKenzie, , A fragment of a Khwarezmian dictionary, London, 1971, 17).Google Scholar
42 On the multifarious uses of NP tā see Lazard, , La langue des plus anciens monuments de la prose pertane, Paris, 1963, §§ 685, 827–36, 844–5.Google Scholar
43 No explanation of Bactr.aro so far proposed (cf. Morgenstierne, , BSOAS, xxxiii, 1, 1970, 129) is entirely satisfactory. ‘So that’ is amongst the meanings of Pahl., NP tā, and seems to be indirectly attested for Olr. *tā by the use of Imp. Aram, ‘d in this sense: for ‘so that’ <‘until’ cf. Syr. ‘ dm’ d-‘ so that’: ‘dm’ ‘until’, etc.Google Scholar
44 Dā is also found in Tati, and in Bal. (<MP ?) dāīn (on which see above) and danikarāhitherto’ (on which see Gershevitch, , Bulletin of the Iranian Culture Foundation, I, 2, 1973, 84).Google Scholar
45 Preserving its final -ā under the influence of tā.
46 The derivation of dā from OP yātā was suggested by Henning, in BSOAS, xii, 1, 1947, 52; no explanation for the loss of the initial syllable was offered there, however. Henning also derived Parth. yd ‘until’ from yātā via *yătā. Dr. Gershevitch points out that yd can equally well be compared with Av. yaδōṯ ‘idem’: clearly Old Iranian possessed, besides yātā, a number of synonymous derivatives of the relative stem, e.g. Av. yaδōit yahmāi, (Parth. yhm), yavaṯ.Google Scholar
47 Similarly Pahl. kū tā, NP ki tā ‘so that’ beside tā.
48 The Iranian origin of these expressions was not recognized by Driver, , Aramaic documents of the fifth century B.C., revised ed., second impression, London, 1965, 76 and 80. The explanation of kî adi as a loan-translation from Iranian is made likely by its late attestation (kindly confirmed for me by Mr. J. Kinnier Wilson).Google Scholar
49 The resulting coexistence of andā and dā may have provided the model for the shortening of NP andar to dar, which has not been sufficiently accounted for.
50 In this connexion, Dr. Gershevitch refers also to the ending of Av. vaēsmәnda (cf. Pokorny, IEW, 181) and vīsāδa, Yt. 13.49, as interpreted by Henning apud Boyce, BSOAS, xxxiii, 3, 1970, p. 521, n. 39.Google Scholar
51 Dr. Gershevitch proposes to regard OP yātā as a juxtaposition of the correlatives yā and tā, cf. Av. tā…yā… in Yt. 10.78 (see Gershevitch, The Avestan hymn to Mithra, Cambridge, 1959, 222). This explanation accounts well for the wide range of meanings of yātā, and of *tā, whether one regards the latter as a contamination of yātā and *dā, or merely an abbreviated form of the expression *yā… tā….
52 On the other hand, Chor. d’(s) cannot be derived from Olr. *dā, which would have given *8ā. The explanations offered above for MP dā as a contamination of tā with *yādā or andā could in theory be applied to the Choresmian forms, but the contamination would have to have occurred independently in the two languages, since it could not have taken place before the Middle Iranian stage. Thus it is only the derivation from *atā (or a similar form, * Vtā or *hVtā) which permits a direct equation of MP and Chor. dā, while it is only a connexion with *ati which adequately accounts for the meaning of Chor. d’, -da.
53 In addition to the individual suggestions attributed above, I have to thank Dr. Gershevitch for many illuminating discussions of the matters considered in the present note, in the course of which both the presentation and the substance of the argument have been modified in important respects.
