On the Vocalic Values of Bactrian ⟨ο⟩: Josiah Medin Despite the fact that the phonemic inventory of the Bactrian language, an Eastern Iranian language that was once one of the most important languages in Central Asia (Gholami 2014), has become less obscure in recent years, a number of uncertainties still exist. In her indispensable Selected Features of Bactrian Grammar, Saloumeh Gholami remarks that the phonetic values of the vowel ⟨ο⟩, particularly whether or not it transcribes [o], [u] or both, are still not well understood (ibid.). This is largely due to the defective nature of the Greco-Bactrian script, which did not consistently transcribe all of the vowel qualities of the language (ibid.). Thus, in order to settle the issue of which vocalic qualities are represented by Bactrian ⟨ο⟩, external linguistic sources such as loanwords in contemporary languages with more widely understood phonologies must be examined. The following short paper will be a review of some of such evidence which pertains to the phonetic values represented by Bactrian ⟨ο⟩. Evidence for the usage of ⟨ο⟩ to transcribe schwa: It is clear from a number of words that omicron is occasionally used to transcribe schwa. The word παδοφρασο ‘punishment’ derives from Old Iranian *pāti-frāsa-, with the medial *-i- possible reflected as ⟨ο⟩ (Gholami 2014), but see below. Since no sound change in Bactrian is known where *i becomes **u or **o between two syllables containing long *ā, and ⟨ο⟩ is inconsistently used to transcribe the reflex of short *i in this environment (ibid.), as in the word αδοριγδο ‘to deny’, from *ati-rixta, an interpretation of ⟨ο⟩ as schwa in certain contexts is the only tenable view. This means that both ⟨ο⟩ and ⟨α⟩ may reflect schwa, as ⟨α⟩ is known to transcribe [ə] in words such as ασπορο < *us-pṛna (ibid.). There is, however, a difficulty in interpreting the word παδοφρασο, since the sound change *i > ə does not occur consistently (ibid.). In this word, ⟨ο⟩ may transcribe epenthetic schwa, since medial clusters of -δφρ- or even -τφρ- do not exist in any known Bactrian sources (ibid.). A schwa interpretation of some instances of ⟨ο⟩ is fairly certain, however, since there is no tenable etymology for the word ηβοδαλο, the Hephthalite autonym, that would yield a medial rounded vowel (cf. Gharib 1995 and Kurbanov 2010) and the medial cluster -βδ- is quite well-attested in Bactrian, discounting the notion that ⟨ο⟩ exists to mark a distinction between **-vd- and **-vð-. This indicates that the medial ⟨ο⟩ in ηβοδαλο was likely pronounced, and since it is unlikely to have been [o] or [u] (once again see cf. Gharib 1995 and Kurbanov 2010), it is likely that it transcribes schwa as in αδοριγδο, which is probably the result of etymological *i. Furthermore, many loanwords from ending in a consonant in Tocharian B show final syllable stress, indicating that they were likely borrowed with a final schwa, which then dropped out independently in Tocharian (Tremblay 2005). Since many of these words relate to urban life, it is likely that they were borrowed during the period of Kushan influence on the Tarim Basin, which is roughly contemporaneous with the oldest Bactrian texts (Sims-Williams & Cribb 1996, Beckwith 2009). From this, it can be inferred that in the earliest stages of written Bactrian, final ⟨-ο⟩ was likely still audibly heard as schwa (Tremblay 2005). For example, the unattested Bactrian form *λασταγγο is thought to be the source of Tocharian B lastāṅk ‘executioner’s block’ (ibid.). The Tocharian B form has final stress, which is only possible in the language if the word has an underlying final schwa which is no longer pronounced, due to Tocharian B’s rule of stress retraction (Hackstein 2017). Once again, since these loanwords likely date from the era in which the Kushan Empire expanded into the Tarim (Beckwith 2009), they are likely contemporary with some of our earliest length Bactrian sources. Thus, there must be some contexts in which ⟨ο⟩ transcribes schwa, including at the end of words, which is not assumed by all linguists studying Bactrian (cf. Gholami 2014). In some words, ⟨-ο⟩ seems to alternate with ⟨-α⟩, the old feminine ending, in word-final position (Gholami 2014). In this environment, ⟨ο⟩ may represent schwa, due to the existence of variant spellings for words such as λιζα/λιζο ‘citadel’. It is, however, possible, that at this time, the feminine ending was in the process of disappearing from the language and was included by some scribes and not by others (ibid.). As aforementioned, It is likely that the omicron was pronounced as a word-final [ə] in these words, at least during the time of the earliest substantial sources, and consequently the vowel represented by word-final -α in the variant spellings must have been distinct. Thus, it is far more likely that, instead of a textual variant that attests to two different ways to transcribe the same pronunciation, words such as λιζα/λιζο were losing their feminine endings as the final ⟨-α⟩ became no longer productive. Thus, two pronunciation variants of the word likely existed, one with a final -α continuing the then very archaic feminine ending, and one with the orthographic ending ⟨-ο⟩ common to almost all other nouns at this time (ibid.), which was likely being generalized to even those rare forms which originally had a feminine ending. These orthographic variants cannot be used as evidence for the notion that final ⟨-ο⟩ represented a pronunciation identical to ⟨-α⟩ because these textual variants only occur with a small number of words, all of which are feminine (ibid.). What this likely indicates is that final ⟨-α⟩ represented a different vowel in early written Bactrian than ⟨-ο⟩, as the Tocharian evidence presented earlier shows that final ⟨-ο⟩ was still likely pronounced into the historical period (Tremblay 2005), and word-final *-ā, which is an old feminine ending, regularly yields as ⟨-α⟩ in some of the oldest texts (Gholami 2014). Against a Bactrian Phoneme [o]: Despite the fact that Bactrian ⟨ο⟩ frequently transcribes schwa, it is certain that the same Greek letter was also used for one or more rounded back vowel phonemes which were distinct from long [oː] (ibid.). Etymological *au is usually reflected with a Greek omega ⟨ω⟩ (representing *[oː]) but is occasionally transcribed ⟨ο⟩, such as in the word σαβολο ‘jar’ < *sapauda- (ibid.), with the conditioning factors yielding the probably different vowel quality being rather unknown. This may reflect a long vowel [uː] that was spelled identically to its short equivalent, or it may be evidence of a phoneme [o] since the diphthong *au yields a long *[oː] in most other cases (ibid.). The latter of these two claims will be examined in the following paragraphs. This word was loaned into Tocharian B as sapule, presumably from an earlier form of the word that retained a final schwa (Bernard & Chen 2022), or from the oblique σαβολε. Since this word was evidently loaned after lambdacism occured in Bactrian, the vocalic value is unlikely to have been significantly different from that of the Bactrian spoken around the time of the Surkh Kotal and Rabatak inscriptions. Indeed, François De Blois (2013), based on a study of loanwords in the language which do not show the sound change, dates lambdacism in Bactrian and related languages to the 4th century B.C., only five centuries before the period of Tocharian-Bactrian contact brought about by the expansion of the Kushan Empire (Beckwith 2009). Thus, it can be said that the rounded back vowel phoneme represented by ⟨ο⟩, in this word was, at the time of very early contact with speakers of Tocharian B, something like *[u(ː)], as Tocharian B had in its sound system both diphthongal /aw/ and monophthongal /o/ (Hackstein 2017), neither of which were used to transcribe the vowel. It is, however, theoretically possible that the monophthongization of the *au diphthong in Old Iranian was not yet underway, leaving a Pre-Tocharian B form *sapəwle with later *əw > u independently in Tocharian, mirroring the Bactrian development in some environments. This may be the case because, at the phonemic level, high vowels in Tocharian can be analyzed as sequences of the mid central vowel and a glide (ibid.). Even still, this evidence from Tocharian shows that ⟨ο⟩ from etymological *au was pronounced either as a diphthong like [əw] or a monophthongal [u] or [uː] by the time of Tocharian-Bactrian contact sometime after the occurrence of lambdacism in the 4th century B.C. (De Blois 2013). Additionally, if the Tocharian B form goes back to oblique σαβολε, as aforementioned, this would better explain the final vowel, as final schwa in Bactrian is usually reflected with final stress in Tocharian B, with an underlying final schwa which prevents stress retraction (cf. Tremblay 2005). This could indicate a borrowing more contemporary with the most extensive written texts in Bactrian. In this case, the phonetic value of ⟨ο⟩ in this word would unambiguously be [u(ː)] in the place and time where Bactrian and Tocharian B were in contact, likely during the Kushan period (Beckwith 2009), despite the fact that diphthongal *au is often known to become *o in many disparate languages. Also of seemingly special interest is the Bactrian word μολο ‘wine’, which was loaned as *[mul] in Early New Persian (Lurje & Yakubovich 2017). Unfortunately, due to the lack of distinction between short [u] and [o] in Early New Persian (Miller 2012), a word containing short [o] in Bactrian may have been loaned with the vowel [u] as well. An early form of this word was also borrowed into Tocharian B, and is reflected as mālo (Bernard & Chen 2022), once again dating to a period after lambdacism was complete in Pre-Bactrian. While it is not impossible that the first omicron in this word retained a value [o] in the Kushan period, as the Tocharian word is clearly from Pre-Bactrian and an umlaut-like sound change were *a was rounded when followed by an *u in the next syllable seems to be the cause of the later Bactrian form, it is more likely that this word was indeed pronounced [mul] by the time of the first extensive Bactrian texts. Certainly, while contact between the Tocharian speakers and Bactrian speakers increased in the second and third centuries A.D. (Beckwith 2009), trade along the silk road was regularly occurring much earlier. The presence of the final vowel -o (which cannot be from the oblique μολε) in the Tocharian B form establishes that the loanword is specifically from a form of Bactrian that is significantly older than that of the Surkh and Rabatak inscriptions, as the only vestiges of final vocalic elements that remained in these texts was the archaic and no longer productive feminine ending -α, the oblique -ε, which would have yielded **māle in Tocharian B, and the final schwa, which, if present, likely would have yielded **māla or **māl (Hackstein 2017). Thus, it can be established with confidence that this loanword is very old, and cannot be used to argue the presence of a phoneme [o] in written Bactrian. This is certainly true because five centuries is more than enough time for an original *a affected by rounding umlaut to become [u], especially if a concurrent sound change *o > u was also occurring. Indeed, such a sound change is known from the history of Bactrian. The word χϸονο, from Greek χρόνος, was loaned into both Tocharian B (as kṣuṃ) and into Khotanese as kṣūṇä (Tremblay 2005). The corresponding forms in both languages unambiguously show an [u] vowel. Given that the first vowel in the Greek source of the Bactrian word was certainly a mid vowel (Teodorson 1979), some kind of sound change in Bactrian from *o > u would be required to explain the Tocharian and Khotanese forms, as, unlike in σαβολο, the vowel transcribed by the first omicron in χϸονο does not derive from an etymological diphthong *au. As such, some kind of *au/*əu value does not work to explain the [u] in Tocharian. Rather, the only possible explanation for the forms in Khotanese Saka and Tocharian B is that the Bactrian word was pronounced with a high vowel [u], probably something like *[xʃun(ə)], with the Khotanese Saka form possibly preserving the final schwa of the (early) Bactrian form. This quite clearly shows that a historical sound change from *o > u took place in Bactrian, and renders it unlikely that any instances of ⟨ο⟩ in written Bactrian represent [o]. From the evidence presented, it is clear that ⟨ο⟩ generally represents an u-like vowel in Bactrian, however there is still some uncertainty regarding whether or not it had contrastive long and short forms. Multiple different etymological sources for ⟨ο⟩ are known. Clearly, ⟨ο⟩ can transcribe a vowel derived from *au as in σαβολο, though *au usually gives ⟨ω⟩ (Gholami 2014), and it may also transcribe a vowel derived from *a…u through rounding umlaut and raising. The most common etymological source for ⟨ο⟩, however, is from Proto-Iranian *u, as in the form ιογδιγο ‘pair’, from *yuxta-ka (ibid.), which may serve as another piece of evidence that an [u]- like pronunciation for every appearance of the vowel (excluding the instances when it represents schwa) is more likely. Other Uses of the Letter ⟨ο⟩: Finally, it is also known that the letter ⟨ο⟩ was used for other representational purposes, such as indicating length in the number αταο *[ataː], or possibly representing *[uː] when doubled (ibid.). Unfortunately, since the written record of Bactrian is fairly scant, the aforementioned less common orthographic uses of omicron are more difficult to investigate, since, for example, long [uː], if it exists, may be only attested in a few words, such as οο ‘that’ (ibid.) or the aforementioned σαβολο, which likely shows long in its second syllable [uː] due to the fact that it derives from *sapauda- (ibid.). Indeed, related languages, such as Middle Persian and Khotanese frequently collapse the Proto-Iranian *ai and *au diphthongs into long monophthongs (Cantera 2017), which gives this interpretation some credence, as does the fact that, under most conditions, PI *au is reflected as ⟨ω⟩, a long vowel (Gholami 2014). The precise cause of the raising from an expected **ω to *[u(ː)], is, however, beyond the scope of this paper. Conclusion: The evidence presented above shows that, in Bactrian, the two short vocalic values of ⟨ο⟩ are likely [u] and [ə]. Thus, Bactrian’s vowel system was in some ways quite similar to that of Early New Persian, with which it was contemporary in its later years (Lurje & Yakubovich 2017), as both lacked a short vowel phoneme *[o]. A number of uncertainties remain, however, given the fairly scant attestation of the language and the seeming inconsistencies in the reflexes of certain Proto-Iranian sounds. This stands as a very fascinating topic on which much more research should be conducted in the future, in order to gain a more comprehensive picture of the phonology of this very interesting ancient language. Works Cited: Beckwith, C. (2009). Empires of the Silk Road: A History of Central Eurasia from the Bronze Age to the Present. Princeton: Princeton University Press. https://doi.org/10.1515/9781400829941 Bernard, C., & Chen, R. (2022, March 18). A fall into the pit: Tocharian B Koṣko, Koṣkīye. Indo-Iranian Journal. Retrieved October 13, 2022, from https://www.academia.edu/74020079/ A_Fall_into_the_Pit_Tocharian_B_ko%E1%B9%A3ko_ko%E1%B9%A3k%C4%ABye De Blois, F. (2013), “Bactria, Bāxδī-, Balx”, Commentationes Iranicae: Sbornik statej k 90-letiju V. A. Livšica, Sergej Toxtas’ev/Pavel Lurje, Sankt-Peterburg: 268-271. Cantera, A. (2017). "33. The phonology of Iranian". Volume 1 Handbook of Comparative and Historical Indo-European Linguistics, edited by Jared Klein, Brian Joseph and Matthias Fritz, Berlin, Boston: De Gruyter Mouton, 2017, pp. 481-503. https://doi.org/10.1515/9783110261288-033 Gharib B. (1995) Sogdian dictionary. Tehran, Iran: Farhangan publications. p. xvi Gholami, S. (2014). Selected features of Bactrian grammar. Harrassowitz. Hackstein, O. (2017). 75. The phonology of Tocharian. In J. Klein, B. Joseph & M. Fritz (Ed.), Volume 2 Handbook of Comparative and Historical Indo-European Linguistics: An International Handbook (pp. 1304-1335). Berlin, Boston: De Gruyter Mouton. https://doi.org/10.1515/9783110523874-030 Kurbanov, Aydogdy (2010). The Hephthalites: Archaeological and Historical Analysis (PDF) (PhD). Berlin: Berlin Freie Universität. Retrieved 5 September 2012 – via Freie Universitat Berlin Repository. Lurje, P., & Yakubovich, I. (2017). The myth of Sogdian lambdacism. Zur Lichten Heimat, 319–342. https://doi.org/10.2307/j.ctv11qdw34.30 Miller, C. (2012). Variation in Persian Vowel Systems. Orientalia Suecana, 61, 156-169. Scarborough, M. J. (2021). Bactrian χϸονο ‘(calendar) year, (regnal) year.’ Journal of the Royal Asiatic Society, 1–9. https://doi.org/10.1017/s1356186321000079 Sims-Williams, N. (2000): Bactrian Documents I: Legal and Economic Documents, Studies in the Khalili Collection III, Corpus Inscriptionum Iranicarum II, 4: The Nour Foundation in Association with Azimuth Editions and Oxford University Press Sims-Williams, N. & Cribb, J. (1996): "A New Bactrian Inscription of Kanishka the Great", Silk Road Art and Archaeology, 4, 1995–6, Kamakura. 75–142. Stein, G. J. (2015). The war-ravaged cultural heritage of Afghanistan: An overview of projects of assessment, mitigation, and preservation. Near Eastern Archaeology, 78(3), 187–195. https:// doi.org/10.5615/neareastarch.78.3.0187 Teodorsson, S.-T. (1979). Phonological Variation in Classical Attic and the Development of Koine. Glotta, 57(1/2), 61–75. http://www.jstor.org/stable/40266467 Tremblay, X. (2005). Irano-Tocharica et Tocharo-Iranica. Bulletin of the School of Oriental and African Studies, University of London, 68(3), 421–449. http://www.jstor.org/stable/20181951 Wei, R. & Li, H. & Wenkan, X. (2013). The separate origins of the Tocharians and the Yuezhi: Results from recent advances in archaeology and genetics.