The Iranian Sound Change *w - > *γw - in the Indo-Iranian Borderlands and a New Etymology for Gāndhārī and Sanskrit guśura(ka) -

It is generally accepted that the etymology of the Gāndhārī and Sanskrit official title guśura(ka) - has to be sought within the Iranian sphere, but the details remain debatable. In this article, I first give an overview of recently discovered evidence for an early soundchangeof * w ->* γw -insomeIraniandialectsfromtheIndo-Iranianborderlands. Onthisbasis,Ithenproposetoderive guśura(ka) -fromadialectformsuchas* γwazurg / * γwuzurg / * γuzurg < * wazr̥ka - ‘strong’. Two by-products of this article are a new Bac-trian etymology for the Gāndhārī personal name G̱aṇavhryaka and some notes on the etymology of the Gāndhārī title sturaka -*.


Introductory Remarks
The Niya documents and some inscriptions in Gāndhārī and Sanskrit are our primary evidence for an official title guśura(ka)-.1 While it is generally accepted that guśura(ka)-is in origin an Iranian title, its exact etymology is still open to debate. The main aim of this article is to put forward a novel hypothesis on the origin of this title. This brief article is structured as follows. I introduce the available material on guśura(ka)-in §2 and critically evaluate older etymologies for this title in § 3. In §4, I present evidence in favour of a dialectal fortification of w-to γwin Iranian dialects from the Indo-Iranian borderlands. As part of this discussion, I also propose a new Bactrian etymology for the Gāndhārī personal name G̱ aṇavhryaka. Finally, I use the evidence of these dialect forms to present a new etymology for guśura(ka)-in §5. The conclusions are wrapped up in § 6.
In Kuča Prakrit, we find gośora (ckd 837, 3x), as the difference between the vowels -u-and -ohas largely been obliterated in the type of Kharoṣṭhī script used for this dialect. It is further possible that go (ckd 831, 2x) is an abbreviation for gośora (so Ingo Strauch apud Ching 2013: 66). In the Helagupta inscription in South Asian Gāndhārī (cki 564), a gen. pl. guśuraṇa is attested. 4 The -u-mostly gets a length mark in Sanskrit, but not always; see just below. For a tentative explanation of the length mark as due to folk etymology, cf. § 5. In this article, I use guśura(ka)as a shorthand to refer to the title under discussion.
Iran and the Caucasus 27 (2023) 285-298 gauśura officials' , itself parallel to Niya Prakrit guśuramahatva (ckd 216, 295, 415;cf. Lüders 1940: 544-546). The same title is once also found as far south as Sāñcī in a Gupta-period inscription, where we encounter it in the spelling gośūra- (Fleet 1888: 280). The oldest evidence comes from the Senavarma and the Helagupta inscriptions in Gāndhārī (cki 249, 564), both of which roughly date to the first century of our era (von Hinüber 2003: 7;Falk 2014: 4;Salomon 2020: 4). It seems therefore likely that this title originated at the nw border of the South Asian subcontinent and spread over a large area of Central and South Asia during the time of the Kuṣāṇa empire.5 This brings us thirdly and finally, to the duties of a guśura(ka)-. As likewise noted by Burrow (1937: 87), the Niya documents make it clear that a guśurahad inter alia judicial functions. In addition, there are also strong indications that a guśura-was closely connected to the army. One piece of evidence for this conclusion is found in Niya document ckd 478, where two guśuras are referred to as seniye 'army-people' .6 This passage is comparable to a section in the Senavarma inscription (cki 249 8g-9b), where the guśurakas and the closely connected sturakas are evidently military functionaries.7 More generally, guśura(ka)-may also have been an honorific title of the nobility.
With these things in mind, the etymologies previously suggested for guśura(ka)-can be critically evaluated in the following section.

A Critical Evaluation of Earlier Proposals on the Etymology
Both etymologies which have so far been proposed for guśura(ka)-are from the hand of Thomas Burrow.8 For both to work, Burrow surmises that guśura(ka)-5 Note also that the Senavarma inscription mentions the Kuṣāṇa king Kujula Kadphises. For the sake of completeness, I mention that Bailey (1950: 391-393)  232; emphasis mine). sturakehi will be discussed in more detail in § 5. 8 I leave aside the vague comparisons with the (Orkhon) Turkic title kül čur (so, very hesitantly, is borrowed from an Iranian source where *wi-had developed to *gu-, as e.g. in New Persian gudār 'crossing, passing' < *witāra-. According to Burrow, at least one of the Iranian-speaking groups in Gandhāra had undergone a similar change already at the beginning of our era. For this sound law, he refers to the name of the first Indo-Parthian king Gondophares (Gāndhārī Guduvhara; Gudaphar(ṇ)a), which is commonly derived from *winda-farnah-'he who finds glory' . One can compare the trilingual Sasanian inscription škz 26 (3rd c. a.d.), where Parthian Wyndprn corresponds to Middle Persian Gwndply and Greek Γυνδιφερ (cf. e.g. Schmitt 2016: 241). While the chronological gap makes it clear that these names in škz 26 cannot refer to the Indo-Parthian king with this name, they at least corroborate the traditional etymological analysis of Gondophares. Nevertheless, it would still be preferable to have more evidence for this strengthening of *w-. We will return to this question in § 4 and, for now, regard it as a possibility to reckon with.
Early in his career (1935: 781f.; 1937: 87f.), Burrow proposed to relate guśura(ka)-to Middle Persian wcyl /wizīr/ 'argument, decision' and especially New Persian wazīr 'counsellor, vizier' . According to Burrow, an agentive meaning 'decider' would also be seen in Avestan vīčira-, generally translated as 'he who decides' and which Burrow considers cognate with the Persian words. However, the absence of a meaning 'decider' for Middle Persian wcyl /wizīr/ is noteworthy,9 and the appurtenance of Avestan vīčira-to the Persian words is also no longer universally accepted (cf. Nyberg 1974: 210 f.;Ciancaglini 2008: 166). Besides, Burrow needs to make the unparalleled assumption that the vowel -iin the second syllable would have been assimilated to the -u-in the first syllable. Hence, while Burrow's first etymology is not impossible in itself, it would seem best not to accept wazīr as an etymological comparandum for guśura(ka)without further ado (pace e.g. Allon 2019: 24).
Iran and the Caucasus 27 (2023) 285-298 Ten years later, Burrow came up with a new etymology for guśura(ka)-, published as a personal communication by Bailey (1947: 149 f.;1950: 391-393); in this hypothesis, guśura(ka)-would ultimately derive from the Old Iranian title *wisah puθra-'son of the house > prince, nobleman' , known from e.g. Avestan visō.puθra-; Middle Persian / Parthian vispuhr and the Aramaic calque br byt' . This title generally referred to members of the royal family, such as brothers and cousins of the king (cf. e.g. Henning 1964; Colditz 2000: 328 ff.). One wonders, however, whether this meaning is so suitable for guśura(ka)-, as in the Niya documents, our most extensive source for this title, there seems to be no clear evidence for any close connection between the guśuras and the royal court.10 As regards the phonology, von Hinüber (2003: 29 f.) makes the fair criticism that the assumed loss of *-p-would be surprising.11 Hence, also Burrow's second etymology proves difficult to verify (pace e.g. Tremblay 2005: 430).
In brief, it seems fair to conclude, with Falk (2004Falk ( : 150 = 2013, that the title guśura(ka)-"is still not fully understood". Therefore, a new attempt to etymologize guśura(ka)-will form the topic of §5, but it will first be necessary to provide an overview of more recently discovered evidence for an early fortification of *w-to *γw-in the Indo-Iranian borderlands in § 4.

Bactrian and Gāndhārī Evidence for a Dialectal Sound Change *w-> *γw-
In this section, I return to the evidence for fortition of an initial *w-in possible Iranian donor languages in Gāndhārī because new data for this sound change has recently come to light in both Bactrian and Gāndhārī. One piece of evidence from Bactrian is the personal name γοραμβαδο /γurambādə/, which likely is a dialect form derived from *wr̥ θragna-pāta-'protected by Wr̥ θragna' 10 This is in contrast with another official title from the Niya documents, i.e., the kāla, because people carrying this title are occasionally (ckd 307, 331 (2x), 622, 634) referred to as a maharayaputra 'son of the great king' (cf. Burrow 1937: 82). 11 Note also that Burrow's idea implies that Ṣiṇā gušpūr and Burushaski gušpuhr, both 'prince' , are borrowed from a different Iranian source, which does not seem too probable. In his 1947 article, Bailey gives the following development for guśura-: *wisah puθra-> *wisē puθra-> *gusī̆vura-> *gusiura-> guśura-, but it is difficult to parallel all of these changes together. Similarly, the newly discovered Bactrian οισβορο* (in the personal name Οισβοροζινιιο 'under the care of the prince') and οισβοργο 'prince' (Sims-Williams 2007: 247; idem 2010: 109f.) do not appear to be compatible with guśura(ka)-, even when assuming dialectal strengthening to *γw-.
Further evidence comes from the Gāndhārī representation of the Bactrian royal name Οημο /wēmə/, possibly 'rock' < *waima-(cf. Sims-Williams 2010: 107, 110f.). This name is spelled in various ways in Gāndhārī and other Indic sources (cf. Falk 2009Falk = 2013, and two of the Gāndhārī attestations deserve to be discussed here in more detail. Building further on Falk's work, Allon (2019: 24f.)  For some considerations on the connection between Bactrian and some New Iranian languages of the region (but not Ormuri and Parāčī), cf. now Kreidl (2021). 14 This inscription used to be called the "grāmarakṣaka seal". However, as this name is based on an obsolete reading of the inscribed text, it should best be avoided. 15 This text is a monastic ledger from the first half of the second century (see Allon 2019: 6-9).
Iran and the Caucasus 27 (2023) 285-298 be compared with Bactrian Οημο Καδφισο. However, it is phonetically unlikely that an initial *w-became *γr-rather than the expected *γw-. Therefore, I prefer Stefan Baums's palaeographically equally plausible readings G̱ ematakhtuasa (cki 1073) and G̱ emakataph(*sas̱ a) (ckm 297) (see Baums/Glass 2002-). ⟨g ̱ ⟩ indicates a fricative [γ] in Kharoṣṭhī script and is thus, like ⟨gv⟩ in Gvaraza, a likely rendering of a foreign cluster *γw-.16 Alternatively, Falk (2020-2021: 131-134) wonders whether the strengthening could have happened within Gāndhārī in the same way as Germanic borrowings into Romance got their initial *w-strengthened to *gw-, e.g. in French guêpe 'wasp' . However, as Falk's proposal does not take the Bactrian examples of this type of strengthening into account,17 it seems more likely that this phonetic process had already happened in the donor language(s).
Another potential piece of evidence from Gāndhārī showing *w-to *γwhas remained unnoticed so far. The Gāndhārī inscription cki 150 contains a personal name G̱ aṇavhryaka. Because the element -vhryaka unambiguously points to a derivative of *friya-'dear' ,18 the Iranian origin of this name has never been doubted, yet the exact analysis of the element G̱ aṇa-is still unclear. Konow (1929: 150) tentatively suggested that it would be a noun meaning 'fight' derived from the Iranian verbal root *√gan 'to strike' , but he is unable to cite parallels for this analysis. Instead, I would propose that G̱ aṇavhryaka renders a Bactrian name *Γοανοφριιακο /γwanəfriyakə/ as a dialect form of *Οανοφριιακο /wanəfriyakə/. Both Οανο < *wana-'victorious' and Φριιακο < *friya-'dear' with hypocoristic -κο, are attested as personal names in Bactrian (Sims-Williams 2010: 98f., 144), and we could have here a mechanical combination of these two names. If this etymology of G̱ aṇavhryaka proves to be correct, this would thus be another example of *γw-being substituted with g ̱ -[γ]. 16 When looking at the available pictures, one could alternatively perhaps read Gvema-in both instances, but Baums's reading G̱ ema-still seems preferable. For the name G̱ aṇavhryaka, discussed below, **Gvaṇavhryaka seems totally excluded. 17 In contrast to Gāndhārī, initial w-is an integral part of Bactrian's phonological system, being written as ⟨ο⟩. 18 Cf
In either case, it is difficult to know for certain from which Iranian language sturaka-* was borrowed. However, Bactrian would be one plausible option, given that Senavarma reigned over the borderland between Afghanistan and Pakistan, probably at least partially Bactrian-speaking at the time. In addition, Sudaṣkana, the son of the reigning Bactrian king Kujula Kadphises, is one of the two persons in charge of these sturakas and guśurakas. At the linguistic level, one will then have to assume that the cluster -rg-was broken up by a svarabhakti vowel, i.e. -rəg-, either already in Bactrian itself (cf. perhaps the spelling στορογο) or when the word was borrowed.21 19 This inscription, together with Helagupta's one, counts among the oldest evidence for this title, so it is a good place to start (cf. §2). It is also because of this early date that, despite being a hapax, I attach considerable weight to the alternative form guśuraka-*. 20 As an alternative etymology, Falk (2004: 12 fn. 14 = 2013: 363 fn. 14) has compared sturaka-* to Niya Prakrit stora-'large animal' ← Bactrian (α)στωρο /(ə)stōrə/ or a cognate thereof (Sims-Williams 2007: 266; Schoubben 2022b: 345 f.). Falk presumably assumes that sturaka-* would refer to a 'groom' , but the Iranian word for 'groom' is *staura-pāna-(cf. Middle Persian stōrbān) instead of *staura-ka-. In a later article (2010: 75 = 2013: 246), Falk adopted the connection with Bactrian (α)στοργο / στορογο instead. 21 A typological parallel for this is found in spoken Dutch, where in a word like werk 'work' the cluster -rk-can be broken up by adding a subphonemic svarabhakti-vowel. Note in Iran and the Caucasus 27 (2023) 285-298 As regards guśura(ka)-, I suggest to compare the derivatives of Old Iranian *wazr̥ ka-'big ' (cf. Schmitt 2014: 278;Brust 2018: 305): Old Persian wazr̥ ka-;22 Middle Persian / Parthian wuzurg (wazarg / wuzarg);23 Pāzand guzurg (guzarg); New Persian buzurg etc. The same word is also found in Bactrian as οαζαρκο / οαζορκο. However, by the lack of attestations from the Kuṣāṇaperiod, it is difficult to say whether οαζαρκο / οαζορκο is inherited from Old Iranian or a later borrowing from Middle Persian.24 Sogdian wz'rk (Buddhist) / wzrg (Manichaean) is, in any case, most likely a later loanword from Western Iranian (cf. Sims-Williams/Durkin-Meisterernst 2012: 212). As a fundamental social term in the Iranian world (see Colditz 2000: 241 ff.), it is no surprise that this word was also adopted into languages in contact with Iranian, e.g. in Armenian, which has the word as vzowrk.
wuzurg (wazarg / wuzarg) is a frequent element of titles, and there are the wuzurgān 'the great people' , i.e. the nobility in Sasanian times (Colditz 2000: 254ff.). As Bailey (1985: 6) already noted, there is an interesting parallel to be drawn between wuzurgān and guśura(ka)-, as people carrying the latter title were probably also part of the nobility (cf. § 2). Depending on which etymological analysis of sturaka-* one favours, guśurakehi sturakehi ca in the Senavarma inscription could be interpreted as a hendiadys of two near-synonyms or it could refer antithetically to 'the nobility and the vassals/slaves' . If one would like to extend the comparison between wuzurgān and guśura(ka)-to the linguistic side too, as I suggest we do, one will have to assume that the seemingly Western Iranian word wuzurg travelled to Gandhāra at the beginning of our era. Such a spread would seem quite possible because of the Indo-Parthian rule in Gandhāra at the time.25 In order for the linguistic comparison to work, one would then need to postulate a dialect variaddition Sanskrit stavaraka-'silk garment' , which comes from an East Iranian, possibly Bactrian, cognate *stavragə of Middle Persian stabrag. Interestingly, de Jong (2003: 71) also suggests that the frequent occurrence of svarabhakti-vowels in Pāzand may be due to influence from the Gujarātī spoken by the scribes of these texts, although other explanations are possible. 22 The Persian is an old loanword with its -z-, but this should not concern us here. 23 See Durkin-Meisterernst (2004: 360)  Three assumptions are needed to compare *γwazurg / *γwuzurg / *γuzurg with guśura(ka)-. The first two of these are trivial, while the third one is admittedly more speculative, though not unreasonable in itself. First, in the names Gvaraza and G̱ ema / G̱ aṇavhryaka, *γw-would be rendered with respectively Gāndhārī gv-and g ̱ -. However, both gv-and g ̱ -are not standard initials in Gāndhārī,27 and as guśura(ka)-is not a personal name, one can expect more adaptation to native Gāndhārī / Sanskrit phonetics. As was already noted in § 2, Sanskrit variants of guśura(ka)-with initial go-, gau-and ga-have been discovered too, and this variation would be nicely explained as due to different attempts at rendering an un-Indic initial cluster as e.g. *γwa-in *γwazurg.28 Second, one again has to reckon with an additional schwa to break up the cluster *-rg-, in the same way as discussed above for sturaka-*.
Third, it seems necessary to assume that when this Iranian title was borrowed into Indo-Aryan, it was folk etymologically connected to śūra-'warrior, hero' given the military flavour of the title. One can compare Falk's ingenious idea (2010: 75 = 2013: 246) that the nw title bhaṭarag̲ a 'master' finds its origin in a descendant of Old Iranian *frataraka-which was folk etymologically contaminated with Middle Indo-Aryan bhaṭṭr̥ -< bhartr̥ -'chief, lord' . When assuming contamination with śūra-, one can, first of all, explain why guśura(ka)-is often written with a long -ū-in Sanskrit. At the same time, this folk etymology can have contributed to the otherwise somewhat unexpected rendering of Iranian -z-with Gāndhārī -ś-, in intervocalic position perhaps pronounced as a voiced palatal fricative [ʝ] (cf. Baums 2009: 137). Iranian -z-is mostly rendered with the Kharoṣṭhī sign , nowadays usually transliterated as ⟨z⟩, but in previous scholarship as ⟨jh⟩.29 However, there are also cases where -z-is ren-26 I do not want to imply any direct historical relationship between my reconstructed dialect form and the Pāzand form. Given our imperfect understanding of the linguistic situation in the Indo-Iranian borderlands at the beginning of our era, I keep my discussion deliberately vague, which is why I speak about a "dialect form" rather than trying to identify the donor language more precisely. As noted above ( § 2), there seems to be a connection between the title guśura(ka)-and the Kuṣāṇa empire, but this does not necessitate that the word is originally Bactrian because official titles spread very quickly and can survive the change of dynasties. Compare the attestation of gośūra-in an inscription from the later Gupta dynasty. 27 There is, so far, no other example of initial gv-, while -g ̱ -is most often found in intervocalic position as the outcome of lenition of older -k-or -g-and only rarely in initial position. 28 In addition, a mechanical application of Sanskrit vr̥ ddhi probably plays its role as well. 29 In my forthcoming dissertation on Niya Prakrit, I intend to include a detailed argumentation in favour of the older transliteration with ⟨jh⟩.
Iran and the Caucasus 27 (2023) 285-298 dered, at least initially, with -j-, e.g. in jenavida (ckd 506) ← Bactrian ζηνοβιδο /zēnəvidə/ < *zaina-pati-'lord of the armour' . If the same substitution was possible in intervocalic position too, guśura(ka)-could also come from an earlier form with *-j-further lenited to -ś-within Gāndhārī (for which cf. Burrow 1937: 6 f.). The same folk etymological association with śūra-can also be seen as a possible motivation for the existence of guśura-next to guśuraka-*. My etymology assumes that guśuraka-* is the more archaic form, which would imply that guśura-derives from a re-analysis of guśuraka-* as guśura-with an added -ka-suffix.30 As it happens, śūra-'warrior, hero' is often used as the second part of a compound in Sanskrit (e.g. raṇaśūra-'hero in war' , śr̥ ṅgāraśūra-'a hero in love-affairs'), and as the second part of compounded personal names (e.g. Somaśūra-or Samudraśūra-). By contrast, *-śūraka-does not seem to be attested, which may be explained by the diminutive and, hence, at times, pejorative function such a -ka-suffix can have in (Classical) Sanskrit. This suffix could well have been considered inappropriate for a word meaning 'hero' , and, mutatis mutandis, for a high-ranking military official as guśura(ka)-, which, as said above, may have been folk etymologically connected to these compounds in -śūra-.

6
Concluding Summary To wrap up, I have, as an alternative to Burrow's earlier etymologies ( § 3), presented a novel hypothesis on the linguistic origin of the Sanskrit and Gāndhārī title guśura(ka)-( §2). More concretely, I have argued in favour of a dialectal Iranian form *γwazurg / *γwuzurg / *γuzurg < *wazr̥ ka-'strong' , which could have been folk etymologically associated with śūra-'warrior, hero' when the word was borrowed into Indo-Aryan ( §5). As my proposal depends on an early dialectal strengthening of *w-to *γw-in Iranian dialects spoken in the Indo-Iranian borderlands, I have likewise included a more detailed discussion of the relevant evidence in support of this sound law ( § 4).

30
The absence so far of intermediary forms such as *guśurag ̱ a or *guśurae makes lenition less likely. It seems traditionally assumed that the -ka-suffix was added in Indo-Aryan.