"Hunger" and" thirst" in Sogdian
«Голод» и «жажда» в согдийском
Sims-Williams N.
Симс-Уильямс Н.
In various Iranian languages, words for "hunger" and "thirst", "hungry" and "thirsty", etc. form rhyming pairs. Such pairs raise the question whether these words were always parallel or whether they were originally less similar and have converged as a result of mutual influences in phonology and word-formation. Here, as a contribution towards answering this question, the relevant words in Sogdian and some of the most closely related languages will be put under the microscope.
Key words: Sogdian, Iranian languages, etymology, etymological convergence, "hungry", "thirsty", "drink"
В различных иранских языках слова «голод», «жажда», «голод -ный», «испытывающий жажду» и т. д. формируют рифмующиеся пары. Наличие таких пар поднимает вопрос, были ли данные слова всегда параллельными или изначально они не были столь схожи и подверглись конвергенции вследствие взаимовлияний в фонологии и словообразовании. В качестве вклада в ответ на этот вопрос мы рассмотрим в данной статье «под микроскопом» соответствующие слова в согдийском и нескольких наиболее близких языках.
Ключевые слова: согдийский, иранские языки, этимология, этимологическая конвергенция, «голод», «жажда», «питие» DOI 10.37892/2313-5816-2021-1-220-230
It has long been observed [Hubschmann 1895: 92; Bartholomae 1925: 65-66; Benveniste 1933: 242] that some Iranian languages attest parallel or rhyming forms for the adjectives "hungry" and "thirsty", e.g. Sogdian cs'ntk [casande]: ws'nty [wasande], Pashto
tagay: wagay, Persian tisna : gusna. The same applies to the related nouns "hunger" and "thirst" and the verbs "to be hungry, to be thirsty". Nowadays such pairs can easily be collected using the word-lists in the Comparative-thematic dictionary of the New Iranian languages and dialects by Mohammad Hassan-doust [2011]. However, the parallelism is not always exact and sometimes seems to result from the gradual convergence of forms which were originally less alike. In this paper I would like to gather together the relevant words attested in Sogdian, including a few forms from texts that have not so far been published, in order to clarify their etymologies and, where possible, the stages of their development and possible etymological convergence.
1. dpz- 'hunger'
The earliest attested Sogdian word for "hunger", cognate with Parthian dbz and Yaghnobi devaz 'id.', is S[z-, which occurs twice in the 4th-century "Ancient Letters" in the phrase MN S[z or MN S[lz' 'from hunger'.1 The abl. sg. form S Sfiz, d[[z, Sfizh, C dbz' is frequently attested in later texts too. The fact that it belongs to a m. stem S[z-, dbz- is clear from the C nom. sg. dbzy and acc. sg. dbzw in the manuscript E27. However, S S[z' is also attested as nom. sg. in P22.3, and C dbz' as acc. sg. in E28/16, V15. These forms seem to indicate a f. stem. As I have suggested previously [Sims-Williams 2016: 70], the occasional transfer of S[z- from
In general, to save space, detailed references to the texts and editions will only be given in the case of forms not included in the standard Sogdian dictionaries [Gharib 1995; DMT III/2; Sims-Williams 2016], where one can also find keys to the conventional abbreviations of the names and signatures of Sogdian texts.—Other abbreviations: abl. = ablative, acc. = accusative, C = Christian (i.e. Syriac) script, DMT III/2 = Sims-Williams and Durkin-Meisterernst 2012, f. = feminine, GMS = Gershevitch 1954, m. = masculine, M = Manichaean script, nom. = nominative, S = Sogdian script, sg. = singular.—I take this opportunity to express my thanks to Agnes Korn for kindly reading this paper and saving me from several errors and solecisms.
m. to f. inflection may result from assimilation to csn 'thirst'. This is supported by the formation of ¿fiz'mwrt'y 'dead from hunger', which stands beside csn-mwrtk 'dead from thirst' in VJ 317-18.2 In passing we may note MacKenzie's attractive proposal [in Morgenstierne 2003: 46] to explain Pashto Iwaga f. 'hunger' as a conflation of *duaza- with wsgay 'hungry'. If this is correct, the f. gender of the underlying *du aza- may perhaps be secondary, as appears to be the case in Sogdian. In theory, one could attribute it to the influence of tmda f. 'thirst' < *trsna-ta- [Morgenstierne 2003: 81], but it is also possible that the source is some form which does not survive in modern Pashto.
2. ws- 'to be hungry' and *cs- 'to be thirsty' The Sogdian verb "to be hungry", unrelated to the noun Sfiz- discussed above, is attested by two forms, the first being C wsmsq 'I am hungry', 1st person sg. present durative, in E23.99. This clearly indicates a present stem ws-, as is confirmed by the well-attested adjective S ws'nty, C wsnty 'hungry', in origin a present participle in *-ant-a-ka- [GMS §1068]. The second is a strange 3rd person sg. present attested in the phrase S cs'yt PZY ws'yt 'is thirsty and is hungry' (SCE 29). The ending -()yt for expected *-ty in this rhyming pair displays an aberrant development, which is also attested by the form ftyS'yt 'is obstructed' (for expected *fiySty) < *badiat(a)i 'is bound', SCE 34, 415 [MacKenzie 1970: 51], and which is apparently a dialectal feature of this manuscript. Thus, despite appearances, ws'yt can be understood as belonging to the present stem ws- mentioned above, while cs 'yt can be attributed to a present stem *cs-.3 The latter is not
Gershevitch [1946: 140], followed by Wendtland [2006: 232-233], suggests that the first part of ôfiz'mwrt'y might be an ablative but cites no parallel for such an unusual formation. The vowel -a- in the personal name Py'fiy'rt may result from -a+a- in sandhi, see Sims-Williams [1992: 45].
Maggi [1995: 115] compares cs ' 'yt with Khotanese [tta](rr)ayata, later ttarraye 'thirsty', which is interpreted as the past participle of a denominative verb *ttarray- based on the adjective ttarraa- 'id.'.
2
directly attested, but is again implied by the adjective (former present participle) csntk 'thirsty'.
Etymologically, it has always been clear that *cs- and its derivatives belong to the Iranian root *tars, Indo-European *ters 'to dry up; to be thirsty', which has clear cognates in Hittite, Greek, Latin etc. [Mayrhofer 1992: 635-636; Rix et al. 2001: 637-638; Cheung 2007: 383-438]. The form cs'yt almost certainly implies a present stem in *-ia-, as does the Choresmian equivalent cy- 'to be thirsty' [MacKenzie 1971: 32a]. Sogdian *cs- and Choresmian cy- can thus be directly equated with Vedic trsya- 'id.', already attested in the Rgveda in the form of the participle trsyant- 'thirsty' (cf. Sogdian cs ntk). As noted by Gershevitch [GMS §§150, 277], the regular Sogdian development of initial *tr- to ca- probably implies an intermediate stage *tra-, a treatment strongly reminiscent of Avestan forms such as trafiia- 'to steal' (= Sogdian cf- [caf-]), if we may accept the reading and interpretation of this form by Hoffmann and Forssman [2004: 91; differently de Vaan 2003: 512-13].4 Gershevitch's proposed intermediate stage *tra- makes it possible to bring together under a single rule the change of initial *tr- to c- and that of final *-tr to -c in Sogdian zrwsc < *zara(h?)ustr 'Zoroaster' [GMS §278] as opposed to the preservation of non-tautosyllabic *-t.r- in forms such as the nom. sg. xwstry [xust.ri] 'camel'.5
In Sogdian, however, there is no known noun or adjective from which a denominative cs'"y (or wsy) could be formed. It remains unclear in exactly what conditions the spellings with tra may be considered regular in Avestan. De Vaan [2003: 513 n. 673] objects to the use of Middle Persian triftltruft 'stolen' as evidence for the linguistic reality of Avestan trafiia- [Hoffmann and Narten 1989: 73 n. 126], but the Sogdian and Choresmian forms provide even stronger evidence on this point.
Sogdian zrwsc [zrosc] (which is a heavy stem, as noted in DMT III/2, 233b, despite the statement to the contrary in GMS §399) can hardly derive from an Old Iranian *zaratustra- or zaraOustra-, whose intervocalic *tlO should not have been lost. If it is a native
4
Unlike *cs- 'to be thirsty', ws- 'to be hungry' does not belong to a root with a clear pedigree in the older Indo-Iranian languages, let alone in distant branches of Indo-European. Cognates are well attested in Iranian, however, the earliest being Avestan varasiia-, known only from the Frahang-i dim, where it is glossed by Middle Persian gursag 'hungry' [Klingenschmitt 1968: 132]. Johnny Cheung [2007: 423] lists most of the relevant forms under an assumed Iranian root *u ars, which, he says, 'appears to be a kind of «Reimbildung»' modelled on *tars. However, as Cheung himself notes, some forms in Persian and closely related languages, such as Middle Persian gursag just mentioned, seem to attest a variant *U ars, which would suggest that *u ars was not invented ex nihilo as a pendant to *tars, but is at the most an adaptation of a preexisting root. The co-existence of forms with apparent *urs- and apparent *urs- is explained by Klingenschmitt [2000: 208-210] on the basis of a double treatment of Indo-European *ks (and *kp) in Middle and New Persian as either s (probably genuine Persian) or s (North-Western Iranian or more generally non-Persic), thus making possible a connection of *urs-/urs- 'to be hungry' < *ulks- with Armenian k'aic 'hunger' < * sulks-. If one accepts this explanation, one cannot even claim that the root in question has been adapted to resemble *tars, since the s of *uars is in fact the regular outcome of *ks in all non-Persic Iranian languages. However, it is possible that the existence of the inherited *trsia- is responsible for the formation of a present stem in -ia- to the root *uars, as attested by Sogdian ws'yt and Choresmian 'wsy- 'to be hungry'. The latter is derived by MacKenzie [1971: 32a] and Samadi [1986: 222] from *ursja-, but the case of Choresmian ysy- 'to rejoice' < *grsia- shows that a derivation from *ursia- is equally possible.
form, it must derive from *zara-ustra-, where the prior element either never had a final -t or lost it in quasi-final position (type: Old Persian Daraya-vau-); alternatively, it may be borrowed, either from an Old Persian *zara-ustra-, or from an early Middle Iranian form such as *zarahustr (cf. Parthian zrhwst with h < *6).
3. csn 'thirst', *csnwq 'thirsty'
The oldest form of the noun 'thirst' is attested as csn' in both Buddhist and Christian texts: S csn' in VJ 814, Vim. 134, and in the compound csnmwrtk 'dead with thirst' already mentioned; C csn' in E27/102, V21, and E29/5, V19. M *csn' can also be reconstructed in the compound ws()[nd]t-c(s)[n] 'hunger and thirst' to be discussed below. A later form, with regular loss of s between c and a nasal [GMS §385-386], is C en in E27/69, R25, and M cn, which is attested in compounds with the word for "hunger", e.g. wsn-cn' in M121, V4 [DMT III/2: 205], and probably also as a separate word in M501j, B2 (unpublished, without clear context). It used to be assumed that csn' or cn' is a m. noun, from *trsna- according to Gershevitch [GMS §386], the form in - ' being the abl. sg., but it clearly does not function as an ablative in all contexts. The fact that no form is attested with an ending other than - ' suggests that it is in fact an uninflected f. noun like sy' 'memory, intelligence' or qty ' 'deed' [Sims-Williams 2014: 104].6 If so, it will correspond more closely to Vedic trsna- f. 'thirst' than to Avestan tarsna- (< *trsna-) m. 'id.'. Like the Avestan form, Khotanese ttarra- 'thirst' is a m. noun, but a f. equivalent seems to be attested by its derivative pattarra- 'drink' < *pati-trsna-,7 lit. 'anti-thirst'.
From the noun csn , cn , is formed an adjective meaning 'thirsty'. The two attested forms C csnwq and cn wq imply an underlying *csn wq. This is the only example listed by Gershevitch [GMS §1077] for an adjectival suffix -auk, but it is clear enough that it is merely an extended form of the possessive suffix -au < *-a-uan-
Admittedly, it is not clear why *trsna- did not result in a f. light stem *csn- (nom.-acc. sg. *csn', oblique cases *csny). A reconstruction
*trsna-ka- does not help, as this should have resulted in *csn 'k. Thus Maggi [1995: 115-117], though it is not clear to me whether he understands the prefix *pati- as having adversative force as I here assume following Skjsrv0 [2004: II, 295]. It also seems possible that the Khotanese denominative verb *ttarray- is not based on the adjective ttarraa- (see above, n. 3) but on the noun *ttarra-.
5
or *-a-uant- [GMS §1076]. Finally, from the adjective *csn wq a noun 'thirst' is formed with the common f. abstract suffix -ya-: S csn wky' in Huyadagman 1.28 [Reck 2006: 160], C csnwqy in E28/16, V15.
4. wsnt 'hunger', *wswq 'hungry'
Beside dbz-, a single Christian Sogdian text attests a second word for 'hunger': C wsnt (E29/5, V12, 19, 23). The same noun is probably attested in the compound M wsn-cn' 'hunger and thirst' (M121, V4), t being lost before c as it often is in the f. forms of past participles (e.g. 'wswytyy 'pure', f. wswyc).8 This interpretation receives confirmation from the variant spellings wysnd-cn (in the same text, M121, R1) and ws()[nd]t-c(s)[n'] (M900, R1), both passages so far unpublished. The closest cognate of wsnt 'hunger' and its variants is Parthian wsynd, which is attested both as an adjective 'hungry'— in this sense of course a present participle in origin, like Sogdian ws nty (above, §2) and Chores-mian wsynd 'hungry'—and as a noun 'hunger'. The same double function is attested in the case of Parthian tsynd 'thirsty; thirst', though ambiguity can be avoided by the use of the abstract suffix: tsyndyft 'thirstiness, thirst'. There seems to be no way of determining whether Parthian wsynd and tsynd simply attest the use of the adjective as a noun, or whether the nouns were originally distinguished by some feature which is no longer observable (e.g. a shift of accent or a suffix such *-(i)i a-).
An adjective C wswq 'hungry' is attested only once, in E27/102, V27, where it occurs beside csnwq 'thirsty'. This fact might suggest that wswq (i.e. [wasok] < *[waSauk], the vowel o being indicated by the point above w) is modelled on csnwq. In that case, however, one would rather expect *wsnwq, especially given that the -n- is actually present in the noun wsnt. It may therefore be better to
8 Thus Sims-Williams [2014: 104]. Gershevitch [GMS §386] cites wsn cn from this text, translating 'hungry and thirsty'. The text is not sufficiently well-preserved to exclude this translation, but the form of cn' is hardly that of an adjective.
interpret wswq as being formed with the same suffix as csnwq but from an otherwise unattested noun *ws- 'hunger'. Such a noun could derive from *ursa- < *ulks-eh2-, cf. Persian gurs 'id.' as explained by Klingenschmitt [1968: 132], or from *u rsi a-, a form implied by Choresmian wsykm. 'id.' (with added *-ka-) and 'wsy'wnd 'hungry' (with a possessive suffix comparable to that of Sogdian wswq).
Appendix: csnt 'drink'
The Sogdian noun meaning 'drink' is widely attested as S csnt, M csnd (with variant csndk), C csnt. The similarity to the family of words meaning 'thirst, thirsty' etc. is striking, but there can be no original connection: Sogdian cs nt 'drink', together with the verb ptcs-, pcs- 'to taste, to drink', no doubt belongs to a root *cas, cf. Persian casidan 'to taste', Pashto cx- 'to drink' etc., as rightly stated both by Rastorgueva and Edel'man [2003: 235-238] and by Cheung [2007: 35-36].9
A problem arises with Parthian tysng 'drink', variants tysng, tysyng, which Sundermann [1992: 123] derives from *trsna- and which Cheung [2007: 383] consequently lists under *tars 'to be thirsty'. Sundermann refers to Henning [1947: 43 with n. 1], where one finds this remark on tysng: "In M[iddle] Pers[ian] = thirsty, but in Parthian = drink (cf. Sogdian cs nt 'drink' beside csn- 'thirst')". It is not clear to me whether Henning intended to imply that the words meaning 'thirsty' and 'drink' are cognate or whether he was merely noting an interesting coincidence. In any case, although these words certainly belong to related semantic spheres, their meanings are hardly compatible. It is simpler, in my view, to suppose that Parthian tysng 'drink' belongs to *cas, with dissimilatory loss of the sibilant component of the initial [c] (i.e. [f]) before the following sibilant. I do not know of another example of this development in the rather small Parthian corpus,
9 The question whether *cas 'to taste, drink' is ultimately related to *cas 'to show, teach', on which these authors disagree, is not relevant here.
but it is common in Middle Persian, cf. tis 'something, thing' < *cis-cim [Sims-Williams 1982: 174-175], tasum 'fourth' < *caguma-[Hoffmann 1975: 186 n. 1], etc. Similar cases are attested in certain Tat dialects, e.g. düsmä 'spring (of water)' (~ Persian casma), düsirän 'to boil' (~ Persian jusidcn) [Suleymanov 2020: 79-80], and I suspect that further such examples can be found elsewhere in Iranian.10 It seems, in fact, that there is even an isolated case in Sogdian, where csnt 'drink' appears once in the spelling tsnt (i.e. [atsand] < *[tsand] < *[tasand], with regular development of a prothetic vowel before a double consonant): S xwrt tsnty xwrtö'r(t)[ 'he consumed food and drink', Ch/U6506, V3 [Kudara, Sundermann and Yoshida 1997: 121].
References
Bartholomae C. Zur Kenntnis der mitteliranischen Mundarten,
VI // Sitzungsberichte der Heidelberger Akademie der Wissenschaften, Phil.-hist. Kl. 1925-1926. 6. Heidelberg, 1925.
Benveniste E. Notes sur le fragment sogdien du Buddhadhyänasamädhisägarasütra // Journal Asictique, 1933, 223: 193-245.
Cheung J. Etymological dictionary of the Iranian verb. Leiden, Boston, 2007.
de Vaan M. The Avestan vowels. Amsterdam, New York, 2003. DMT III/2 — Sims-Williams N., Durkin-Meisterernst D. Dictionary of Manichaean Sogdian and Bactrian. Dictionary of Manichaean Texts III/2. Turnhout, 2012.
Gershevitch I. Sogdian compounds // Transactions of the Philological Society. 1945 [1946]: 137-149.
Gharib B. Sogdian Dictionary: Sogdian-Persian-English. Tehran, 1995.
GMS — Gershevitch I. A grammar of Manichean Sogdian. Oxford, 1954.
An anonymous reviewer of this paper kindly adds the example Tajik tusbera < Uzbek cucvara 'dumpling'.
10
Hassan-doust M. Farhang-e tatbiqi-mouzui-ye zabanha va guyesha-ye Irani-ye nou [Comparative-thematic dictionary of the New Iranian languages and dialects]. 2 vols. Tehran, 2011.
Henning W. B. Two Manich^an magical texts, with an excursus on the Parthian ending -endeh // Bulletin of the School of Oriental and African Studies, 1947, 12(1): 39-66.
Hoffmann K. Aufsätze zur Indoiranistik. I. Wiesbaden, 1975.
Hoffmann K., Forssman B. Avestische Laut- und Flexionslehre. 2., durchgesehene und erweiterte Auflage. Innsbruck, 2004.
Hoffmann K., Narten J. Der sasanidische Archetypus. Wiesbaden, 1989.
Hübschmann H. Persische Studien. Strassburg, 1895.
Klingenschmit G. Frahang-i öim. Edition und Kommentar. PhD dissertation. Erlangen, 1968.
Klingenschmitt G. Mittelpersisch // Indoarisch, Iranisch und die Indogermanistik. Wiesbaden, 2000, 191-229.
Kudara K., Sundermann W., Yoshida Y. Iranian Fragments from the Otani Collection. Text Volume. Kyoto, 1997.
MacKenzie D. N. The "Sutra of the causes and effects of actions" in Sogdian. London, 1970.
MacKenzie D. N. Supplement // A fragment of a Khwarezmian dictionary by the late W. B. Henning. London, 1971, 28-50.
Maggi M. The Khotanese Karmavibhanga. Rome, 1995.
Mayrhofer M. Etymologisches Wörterbuch des Altindoarischen. I. Heidelberg, 1992.
Morgenstierne G. A new etymological vocabulary of Pashto. Beiträge zur Iranistik 23. Wiesbaden, 2003.
Rastorgueva and Edel'man — Расторгуева В. С., Эдельман Д. И. Этимологический словарь иранских языков. Москва, 2003, Т. 2. (In Russ.)
Reck C. Reste einer soghdischen Version von Huyadagmän I in der Form eines Responsoriums zwischen Erwähltem und Hörer // Languages of Iran: Past and Present. Iranian studies in memoriam David Neil MacKenzie. Iranica 8. Wiesbaden, 2005 [2006], 153-163.
Rix H. et al. LIV. Lexicon der indogermanischen Verben. Zweite Auflage. Wiesbaden, 2001.
Samadi M. Das chwaresmische Verbum. Wiesbaden, 1986.
Sims-Williams N. Notes on Manichaean Middle Persian morphology // Studia Iranica, 1981 [1982], 10/2: 165-176.
Sims-Williams N. Sogdian and other Iranian inscriptions of the Upper Indus. II. London, 1992.
Sims-Williams N. Biblical and other Christian Sogdian texts from the Turfan Collection. Berliner Turfantexte 32. Turnhout, 2014.
Sims-Williams N. A Dictionary: Christian Sogdian, Syriac and English. Beiträge zur Iranistik 41. Wiesbaden, 2016.
Skj^rv0 P. O. This most excellent shine of gold, king of kings of sutras: The Khotanese Suvarnabhäsottamasütra. 2 vols. Cambridge MA, 2004.
Suleymanov M. A grammar of §irvan Tat. Beiträge zur Iranistik 46. Wiesbaden, 2020.
Sundermann W. Der Sermon vom Licht-Nous. Berliner Turfantexte 17. Berlin, 1992.
Wendtland A. Einige Bemerkungen zu den soghdischen Alten Briefen // D. Weber (ed.) Languages of Iran: past and present. Iranian studies in memoriam David Neil MacKenzie. Iranica 8. Wiesbaden, 2005 [2006], 229-238.
Sims-Williams Nicholas SOAS University of London London, UK Симс-Уильямс Николас Лондонский университет SOAS Лондон, Великобритания ns5@soas.ac.uk