CC BY
18ИЗ ИСТОРИИ слов. этимология
Avestica, Sogdo-Yagnobica, and Indo-Europaica
Schwartz M.
Шварц М.
The phonological form of two late Avestan words, problematic from the Indo-Iranian viewpoint, will be explained through an examination of Middle and New Iranian, and particularly Yaghnobi, with consequences for Avestan textual history. A Proto-Indo-European form will also be given and its semantics analyzed.
Key words: Avestan phonology, Indo-Iranian etymology, Yaghnobi, Sogdian, Pashto, Vendidad, Magi, Proto-Indo-European etymology
Мы рассматриваем в аспекте фонологии два позднеавестийских слова, которые представляются проблематичными с индоиранской точки зрения. Объяснения приводятся на основе рассмотрения средне- и новоиранских языков, в частности ягнобского, с последствиями для истории авестийских текстов. Кроме того, приводится праиндоевропейская форма и анализ ее семантики.
Ключевые слова: авестийская фонология, индоиранская этимология, ягнобский, согдийский, пашто, Вендидад, маги, протоиндоевропейская этимология DOI 10.37892/2313-5816-2021-1-199-205
I am pleased to offer this brief study to Professor Joy Edelman in recognition of her important scholarly service to Iranian linguistics in general, and specifically her work on Central Asiatic languages.
Avestan ganti- (nominative gaintis 'stench, odor', Widewdad 7.56 ad HadoxtNask2.25) and gantuma- 'wheat', are phonologically problematic for their -t-, for which an attempt at explanation will be given here, proceeding from ganti- to gantuma-.
Avestan ganti- is immediately irreconcilable with Old Indic gandha-, -gandhi-. Mayrhofer [1992: 461] affirms their connection, but remarks that their relationship is uncertain.
Cheung [2007: 103-104] derives Avestan ganti- and its Iranian cognates from a root *gant 'to smell badly, stink', and comments that the Iranian root (and Sanskrit gandh-) can hardly be of Indo-European origin, and he further states that Greek dennos 'a blaming, a reproach' is "not compelling" as a cognate, and that what he calls "a strange dental «alternation»" in the Indic and Iranian roots "points to borrowing". It may be said for now that this refuge in substratism, positing a borrowing of a verb for so basic and familiar a concept as 'to stink' does not especially command credence; an Indo-European origin of the verb in question will be upheld at the conclusion of this study, with its quest for an explanation of the phonic irregularity, from a broader view of the Iranian languages.
Such a view was attempted for the solution of ganti- by Szemerenyi [1951: 163 = 1991: 1809]. In an article positing a Sogdian linguistic influence on the Avesta, Szemerenyi tried to explain ganti- as attesting a Sogdicism. Suffice it to note that although Szemerenyi's other Avestan candidates for Sogdicism fail to convince, his explanation of ganti-, while flawed, deserves consideration for its helpful suggestiveness. Szemerenyi lists Old Indic gandha- 'odor', Pahlavi gannak [i.e. gn'k=gannag; cf. Parthian gandg 'foul' M.S.], Christian Sogdian yantäk [i.e. Estrangela Syriac script ynt'q M.S.] 'böse', Balochi gandag/y 'schlecht, böse', and Pashto yandal 'Ekel empfinden'. Szemerenyi sees the foregoing forms as all based on Iranian gand- and not gant-, since Balochi preserves original -nt- e.g. in ant 'they are' and dantan = Persian dandan 'tooth', and Southwest Iranian develops only original -nd- into -nn-, preserving Old Iranian -nt- as -nd-.1
1 This is not contradicted by Manichean Middle Persian dn'h, once thought to mean 'toothache' [Henning 1958: 98], but actually meaning 'harm, destruction' < *ati-na0a-; see [Schwartz 2002: 236 with fn. 27].
Szemerenyi continues: "Nevertheless, Avestan preserves the form ganti- 'übler Geruch, Gestank' to be sure in two later passages." Finding it difficult to assume an Old Iranian alternation gant-/gand- (Indo-European t/dh), Szemerenyi concludes: "I prefer to regard the Av. form as a Sogdian spelling represented also in the Chr. Sogdian form quoted above, i.e. t stands for the voiced dental stop."
It is difficult to imagine how (and why) a Syriac-script spelling of a Christian text, dating, moreover, from long after the period of composition of the Avesta, influenced a form found in the Zoroastrian Widewdad. Furthermore, Szemerenyi's "t stands for the voiced dental stop" is phrased obscurely.
We must first note that Christian Sogdian attests ynt (also Sogdian script ynt) 'stench'. Moreover, Manichean Sogdian attests ynd'q 'evil'; and ynd'ky qrynyy, ynd'kry 'evildoer' are found alongside ynt'k etc. Similarly Christian nbnt and nbnd = Manichean nßnd 'with, accompanying', Vband; Christian snq and sng = Manichean sng 'stone' < *asanga-; Chr. znq and zng = Man. zng 'species, kind' < *zanaka-; Chr. ympn and ymbn = Man. ympn and ymbn; Chr. ntwxs and ndwxs = Man. ntwxs 'to strive', Vtuxs; etc. etc. In Sogdian script, which was used for writing the Sogdian language before the employment of the Syriac and Manichean alphabets, we find for words like the foregoing only -nt-, -nk-, and -np-, even though the stops may have been pronounced voiced, as in the Chr. and Man. examples.
The explanation of these spellings is as follows: (1) The Old Iranian voiced stop phonemes /b/, /d/, and /g/ in simple environments became spirantized to Sogdian [ß] ([v]), [8], and [y], whereas immediately after nasals (and z) in clusters, the Old Iranian voiced stops were represented as p, t, and k. In addition, the Old Iranian voiceless stops became voiced by the immediately preceding nasals. Thus there no longer existed distinct phonemes /b/, /d/, and /g/, although these sounds, whose reflexes after nasals merged with the post-nasal outcomes of old /b/, /d/, and /g/, may
have been phonetically (but not phonemically) pronounced as voiced after nasals, perhaps in free alternation with voiceless articulation.
One must also take into consideration the influence of spellings in Sogdian script on Christian Sogdian orthography, in which indication of voicing of nasal clusters is rarer than in Manichean orthography, where this influence also figures, and whose orthography is in general more conservative (cf. remarkably Man. dry < Sogd. script dry, but Chr. sy 'three'). There are parallelisms in the orthography of Modern Greek, whose phonological history is in relevant details parallel to that of Sogdian.
In Yaghnobi, proceeding from the background situation, only voiceless stops are found after nasals in relevant words, e. g. sank 'stone', vant- 'to bind', buyunt 'pocket' (*apagunda-), zunk 'knee' (*zanuka-), etc., except where a stop is voiced under the influence of cognate words in Tajik. The matter is discussed in detail by Xromov [1973: 128-130], where [129] I. Gershevitch is also quoted with regard to relevant Sogdian data. Significantly for our main subject, Gershevitch affirms the antiquity of the Yaghnobi situation, and cites (solely) as parallels outside of Yaghnobi the Avestan words ganti- and gantuma-.
As for Avestan gantuma- 'wheat', an origin in older Iranian *ganduma- suggest that they try to fix the gap in the Old Indic word godhuma-, which, as noted by Mayrhofer [1992: 498-499], who cites Hittite kant- and Arabic hint-, is due to some folk etymology. However, for an underlying Iranian *ganduma-, note Balochi gandum, gandim, and Manichean Middle Persian gnwm (vs. Pahlavi gndwm). Morgenstierne [2003: 31] suggests that Pashto yan'am 'wheat' may show "pretonic loss of -d- from -nd-"; cf. Khotanese ganama-. This leaves for discussion the Avestan form, together with Yaghnobi yantum = Chr. Sogd. yntm.
I propose that our two focal Avestan words, occurring as they do in late texts, reflect a stratum of composition within a Sogdian area in which a speech prevailed which shows the idiosyncrasies discussed above for Yaghnobi, and hence -t- from expected *-d-.
The case for such a scenario is strengthened by the fact (which I recall from my student days with W. B. Henning) that the Widewdad, in which one of our words occurs, has for 'to give' not the expected Vdä, but fra-^bar, whence Sogdian /9ßar-/, as the ordinary word for 'to give' (= Yaghnobi tifár-), although comparable forms occur in Khwarezmian and Khotanese. If the Widewdad is indeed, as many Iranists have proposed, a Magian or Magianized work (note as a likely West Iranian trace the similarity of Wid. 2.20-36 to the Mesopotamian deluge legend), the tentative suggestion given here may constitute a marginal addition to the dossier of the Sogdian Magi (for which see [Grenet, Azarnouche 2007]).
Since it has been shown that an Indo-Iranian etymon *gandh-'smell, odor', the earlier question of the etymon's having a Proto-Indo-European origin (rejected, together with a unified Indo-Iranian proto-form, by Cheung) merits reconsideration. Toward this end, a view of linguistic change is offered which not merely observes formal phonic regularity, but also offers an awareness of scenarios of semantic evolution, informed by attested patterns of the development of meaning.
The basic materials for the relevant etymology were in fact provided by Pokorny [1959: 466-467], which will here be redacted with modifications and reshaped, positing V* gwedh (with a nasal infix stem), meaning 'to slam, knock forth/away, push at/away, punch, repel', and then expressions of repulsion, disdain, and loathing, a semantic match of Greek dénnos (< *gwéndhno- ?) 'a reproach, a blaming' and Pashto forms yand- 'to dislike, censure, criticise', yand 'эпа 'a censure, blame', cf. yända 'disliked, loathsome'. Obviously via 'loathsome, foul', these Pashto words are compared by Morgenstierne [2003: 31] with Balochi gandag 'bad', Persian gand 'stench', and the Afghan place-name Ghandak, a village reeking from sulphurous coal beds. Via 'odor', Old Indic generalized the meaning of gandhá-, -gandhi- to '(any) smell, fragrance'. Possibly relevant too are examples of etymologically different forms, noted by Pokorny, in which 'stoßen, zusammenstoßen' give 'stench, malodorous substance'.
From the root sense of 'slam, knock away, punch at', Lithuanian has gendù, gésti 'Schaden nehmen, verderben, zugrunde gehen', pagadas 'Verderben'. Non-infixed derivatives of the same root may now be recognized in Iranian: Christian Sogdian yS- 'a wound, scourge, beating', whose meaning provides more semantic specificity for Avestan gaSa-, hitherto taken as 'Verderben, Unheil' or the like, whose adjective apayaSa- would be verbal, 'that which scourges, afflicts violently'.
Our quest for the explanation of Avestan ganti- (and gantuma-) has taken us via various later languages, and most importantly Sogdian and Yaghnobi, with illumination of Iranian cultural history, and finally to Proto-Indo-European, with an example of the interest of Iranian for its reconstruction, and here our investigation ends2.
References
Cheung J. Etymological Dictionary of the Iranian Verb. Leiden, Boston, 2007.
Grenet F., Azarnouche S. Where are the Sogdian Magi? // Bulletin of the Asia Society, 2007, 21: 159-177.
Henning W. B. "Mitteliranisch" // Handbuch der Orientalistik 1/IV/1. Leiden, 1958, 20-130.
Mayrhofer M. Etymologisches Wörterbuch des Altindoarischen, I. Heidelberg, 1992.
Morgenstierne G. A New Etymological Dictionary of Pashto. Compiled and Edited by J. Elfenbein, D. N. MacKenzie, and N. Sims-Williams. Wiesbaden, 2003.
Pokorny J. Indogermanisches etymologisches Wörterbuch, I. Bern, 1959.
Schwartz M. "Qumrän, Turfän, Arabic Magic, and Noah's
Name" // R. Gyselen (ed.), Charmes et Sortilèges, Magie et Magiciens. Res Orientales XIV. Bures-sur Yvette, 2002, 231-238.
2 I thank Nicholas Sims-Williams and Michael Weiss for kindly supplying bibliographical material and comments.
Szemerenyi O. "Sogdicisms in the Avesta" // F. Altheim (ed.), Aus Spätantike und Christentum. Tübingen, 1951, 153-166 = Szemerenyi O. // P. Considine and J. T. Hooker (eds.), Scripta Minora IV. Innsbruck, 1991, 1799-1812.
Xromov A. L. Jagnobskij jazyk. Moscow, 1972. [Хромов А. Л. Ягнобский язык. Москва, 1972.] (In Russ.)
Schwartz Martin
Professor of Iranian Studies at the University of California,
Berkeley, USA
Шварц Мартин
Калифорнийский университет
Беркли, США
martz@berkeley.edu
