Abstract
This paper is a new analysis of the ‘Rosetta Stone’ for the decipherment of the extinct Pyu language once spoken in what is now Upper Burma. The two pillars collectively known as the Kubyaukgyi (a.k.a. Myazedi) inscription from c. 1112 CE provide two copies of the same text in four languages: Old Burmese, Old Mon, Pali, and Pyu. I present a critical edition of the text based on newly taken photographs using RTI (Reflectance Transformation Imaging) technology. I provide English glosses and translations of all versions of the text found on the better preserved pillar (A) and notes on the phonology of the Pyu text. As online supplementary material I provide readings of the four faces of the less well preserved pillar (B), a glossary of all words in the Pyu text including their equivalents in the other three languages of the inscription, and an apparatus of readings from previous scholarship. Unlike previous Western scholarship on the Kubyaukgyi, this paper incorporates the findings of earlier Japanese studies of the Kubyaukgyi inscription. It also takes into account recent developments in Mon and Pyu language studies.
1 Introduction: Previous studies of the Kubyaukgyi inscription1
The extinct Trans-Himalayan (a.k.a. Sino-Tibetan) language Pyu was spoken in what is now Upper Burma during the first millennium and early second millennium CE. For an overview of current knowledge about Pyu civilization, see Stargardt (1990) and Griffiths et al. (2017).
The Pyu language is known only from written sources of two types: inscriptional texts in the Pyu script and a limited number of transcriptions in Middle Chinese. The corpus of inscriptions, published by the École française d’Extrême-Orient is available online.2 Griffiths et al. 2017 provides an inventory of the former and offers a few examples of the latter. With two exceptions I will mention below, I refer to Pyu inscriptions by the inventory numbers from thse two sources: e.g., PYU 1 is the first Pyu inscription in the inventory. The Chinese material has yet to receive detailed examination.
The most studied and most famous of all Pyu texts are PYU 7 and PYU 8, collectively known as the
The two pillars were found in
The top parts of the OB and Pali text were partly transliterated in Forchhammer (1892) on the basis of rubbings made before the bottom piece of the B pillar was fully excavated. That first OB reading was marred by spellings influenced by modern Burmese orthography: e.g., OB ryā ‘hundred’ was transliterated as if it were modern Burmese rā. Tun Nyein was the first to translate the OB text into English in Nyein (1899); a first translation of the Pali translation into French was published a few years later in Beylié (1907).
The OM text was unexamined until Blagden (1909) which was the first decipherment of any text in OM. In addition to transliterating and translating the A version of the OM text into English with extensive line-by-line commentary, Blagden also includes the first complete transliteration of the Burmese text and a transliteration of the Pali lines missing from Forchhammer (1892). Blagden (1910a) provides corrigenda for the OB readings of this pioneering effort and a reading of the B version of the OM text.
Those two articles were then followed by Blagden (1911) on both versions of the Pyu text. Blagden’s third article is the foundation of Pyu linguistics. No one had ever studied the Pyu language before, and all other work on the Pyu section of the Kubyaukgyi has been heavily indebted to Blagden’s second breakthrough in Southeast Asian decipherment following his pioneering study of OM. Prior to Blagden, what he referred to as “the fourth text of the Myazedi inscriptions” had “variously been conjectured to be in some old form either of Assamese, Tibetan, Cambojan, or Shan” (Blagden 1911: 365).
Instead of taking a top-down approach with an a priori hypothesis about the language of the text, Blagden adopted an agnostic bottom-up approach “to study the text itself, in both copies, compare it with the parallel versions [in other languages] and endeavour to analyse it as far as possible.” He began by matching names and Indic loanwords in the various versions to identify the characters of the mystery script, and proceeded to read and gloss other words on the basis of his interpretation of the script and of the other texts. He makes no attempt to be comprehensive by claiming he understood every word of the mystery language. Instead he builds solid cases for just twenty-six words, noting that some of them were similar to OB and, in one, case, OM. He concluded that
we have before us a specimen of a language of Burma, not some distant and foreign tongue. Moreover, the language must have been in some kind of contact with Talaing [i.e., OM]: the Talaing loanword and the peculiar letter ḅ necessitate that inference. […] I think the language of our text may with much probability be ascribed to the neighborhood of Prome, and it is not an extravagant conjecture to suggest that it may have been the language of the Pyu (or Pru) tribe which is said to have inhabited that region at an earlier period. […] What is quite certain is that the language of our text (though assuredly not a mere dialect of Burmese) is either a Tibeto-Burman one or has been deeply modified by some member of the Tibeto-Burman family. (pp. 381–382)
All subsequent scholars have adopted Blagden’s identification of the language as Tibeto-Burman and his use of the term Pyu for it (albeit with reservations in some cases).
Blagden’s first paper on Pyu ends with a transliteration of the Pyu text of column A supplemented with variants from column B with English translations interspersed. Throughout this paper and all his works, Blagden maintained a rare degree of honesty about his limitations; at the very end and in Blagden (1912) he printed yet more corrigenda for his 1909 and 1910 papers on the other texts of the Kubyaukgyi inscription.
Cœdès (1911) praises Blagden (1911), writing that for “ce texte qui était resté jusqu’ici rebelle à tout essai d’interprétation” Blagden offers “un déchiffrement complet et très satisfaisant”. His only quibble concerns Blagden’s interpretation of the date in the Pyu text; he proposes that the Pyu used akṣaras as numeral symbols.
Blagden (1914) is a defense of a “much more literal” interpretation of OB spelling as “practically phonetic”. Although scholars now take transliterations of OB for granted, in Blagden’s day even a noted Orientalist such as his critic St. John (1914) could fall into the trap of anachronistically projecting modern Burmese pronunciation onto OB spellings: e.g., rejecting the evidence for a medial /l/ no longer present in modern Burmese.
The most extensive single work on the Kubyaukgyi inscription is the first part of the first volume of Epigraphia Birmanica which covered all four sides of both pillars. An article was devoted to each language of the inscription. Each article contains transcriptions and side-by-side photographs of rubbings of both the A and B versions of each text followed by a new English translation. The photographs are difficult to make out; they are superceded by those in Luce and Tin (1934) which remain unsurpassed in print.
Duroiselle (1919a), taking up more than sixty percent of that volume of Epigraphia Birmanica, contains extensive notes on its Burmese transliteration system and on individual OB words from a comparative perspective. Although Duroiselle (1919a: 31–35) justified OB medial /l/ at length, he oddly excluded it from his section on transliteration. His “Index of Burmese words explained” is unfortunately not accompanied by a comprehensive glossary like those of the Mon and Pyu articles in that issue.
Duroiselle uses the inscription “to rectify the chronological errors of the Burmese chronicle Mahāyāzawin
Blagden (1909), on the other hand, regarded the beginning of Kyanzittha’s reign as 1085 CE, implying he thought Kyanzittha died in 1113 CE. Blagden (1910b: 855) is uncertain whether
the 1628 of the Myazedi record is 1085–1086 AD. I am not even positive that it represents the year of the death, and not the accession, of Kyanzittha, that monarch of so many irreconcilable dates.
His later publications on the Kubyaukgyi inscription do not touch the issue of dating; they merely translate the Pyu numerals for dates without converting them into Gregorian equivalents.
Although the Kubyaukgyi inscription has often been assigned the date 1112 or 1113 CE (e.g., in Aung-Thwin (2005: 179)), none of its faces mention when it was written, and Duroiselle did not date it.
Blagden (1910b) “hesitate[s] to put a date” to it, although he did not agree with Fleet, who thought the inscription “is not a synchronous one; that is, that it was framed and engraved, not when the acts registered by it were performed, but a considerable time afterwards” (Blagden and Fleet 1910).
Blagden (1910b) hypothesizes that
a recent expansion of Burmese rule had brought neighbouring alien races under its sway, and that the prince who performed the act of piety recorded in these inscriptions was anxious that it should be commemorated in a manner which would be understood by all the more important sections of the population comprised in the Burmese empire. But would anyone, after a lapse of many years, have thought it worth his while to draft and set up in four different languages a statement of the fact that a long deceased prince had made a votive offering on behalf of a long deceased king? I do not think so: surely the principle of cui bono applies strongly to such a case of this.
He concludes that the Kubyaukgyi inscription “must be dated somewhere about the time of Kyanzittha’s death” without providing a precise year.
Duroiselle (1919b), the section of Epigraphia Birmanica dealing with the Pali text, is far briefer than the preceding section dealing with the OB text. Duroiselle (1919b) does not even carry a title; all it offers beyond the bare bones of transcriptions, photographs, and an annotated translation, was the text reorganized in metrical form.
The remaining two sections of Epigraphia Birmanica are Blagden’s refinements of his earlier work on the OM and Pyu texts of the Kubyaukgyi. Both Blagden (1919a) and Blagden (1919b) contained lexica of all OM and Pyu words other than names in those texts. Those lexica are still reliable from the perspective of present scholarship. Jenny and McCormick (2014) from nearly a century later more or less agrees with all but one of Blagden’s interpretations of the OM words in the Kubyaukgyi,4 and my own lexicon of the Pyu (§ 5) is a refinement and expansion of Blagden’s work rather than a drastic revision.
Blagden (1919b) was to be Blagden’s swan song on Pyu:
So far as appears at present, the prospects of Pyu epigraphy are not very promising, and unless much additional material is discovered in the future, it does not seem likely that any great progress will ever be made in the study of this obsolete language.
Although Blagden moved on to other OM texts beginning with part 2 of Epigraphia Birmanica, he never touched Pyu again.
Following Blagden’s departure from the field, there was a quarter-century-long void in Kubyaukgyi studies that ended with Shafer (1943). Shafer builds upon Blagden’s work on the Pyu faces of the Kubyaukgyi and other Pyu texts,5 reprinted in Blagden (1917), and Blagden’s reading of PYU 1 quoted in Ko (1911).
Shafer’s article has ten sections. Three overlap with what Blagden had already published: (1) epigraphy with the first detailed examination of the usage of the Pyu character Z6 which Blagden (1911: 385) had briefly mentioned as “unexplained marks”, (8) a transcription with the first word-for-word translation as well as a more natural translation, (9) a Pyu-English vocabulary. The other sections of Shafer’s article examine the Pyu text from a linguistic perspective for the first time: (2) compares Pyu grammatical words with the corresponding words in OB, OM, and Pali; (3) is a survey of the Indic loans in all three non-Pali texts of the Kubyaukgyi and the Pyu urns. Shafer concluded that there were three strata of Indic loans in Pyu, an older and a newer layer preserving final a and a third layer without stem-final a via OB or OM. He drew a line between what he called Old Pyu and New Pyu on the basis of the different strata of Indic loans and grammatical differences between the Kubyaukgyi and the older urn texts described in (7); (4) provides eight sets of sound correspondences between Pyu and other Trans-Himalayan languages: primarily Written Burmese, “Old Bodish” (i.e., Classical Tibetan, not Old Tibetan), and “Lucei” (i.e., Lushai, a.k.a. Mizo). (5) is a brief discussion of prefixes with a focus on numerals; (6) is a slightly less brief comparison of Pyu and Karenic vocabulary; (7) is a survey of Pyu grammar with notes on parts of speech and a list of differences between Kubyaukgyi and pre-Kubyaukgyi Pyu; (10) is a summary of the above including a list of Indic-to-Pyu sound conversion laws.
The Kubyaukgyi caught the interest of scholars again a decade after the Second World War.
Nishida (1955) contains tables of the OB characters and rhymes and a list of OB consonant clusters attested in the Kubyaukgyi and an annotated word-for-word translation of the OB text into Japanese.
Nishida (1956) presents sound correspondences between the rhymes of the OB of the Kubyaukgyi and those of Proto-Tibeto-Burman and Written Burmese with references to Written Tibetan and modern Burmese dialects. Nishida also provides an inventory of OB phonemes including retroflex consonants (!) based on his analysis and a classification of OB suffixes.
Both Than Tun (1958) and Tha Myat (1958) provide Burmese transliterations of the Pyu text and English translations of the Pyu text of the Kubyaukgyi. Than Tun also gives a word-for-word English translation and was the first to include Z in an eyecopy of the text. Tha Myat was the first to write a Pyu-Burmese glossary of words in the text. Tha Myat also provides reproductions of rubbings which were clear and legible but unreliable. The symbol Z seems to have been blacked out except in A11, A24, A26, and B11. All instances of subscript consonants were also apparently blacked out. Those symbols were still clearly present when the pillars were photographed in 2016 for this study, so they should have appeared in rubbings from the previous century. Tha Myat (1958) was reprinted as part of Tha Myat (1963) without the glossary. Tha Myat’s glossary has many dubious definitions apparently based on phonetic similarities to WB words.
Luce and Shin (1969–1970) place the Kubyaukgyi into historical context, explaining why it has the four languages that it does and why Kyanzittha, a Burmese king, wrote so much in Mon rather than his native language:
For the 11th century, we have to imagine the present proportion of Burmese and Mons reversed: a small minority of conquering Burmans, large numbers of native Mons; among the Burmans, only a few literates, mostly in Kyaukse and the capital; among the Mons, an old evolved literature, worthy vehicle for the arts, Buddhism and government. The first necessity for a united Burma was a common written language. The only possible alternative then to Mon was, not Burmese, but Pyu. Pyu, though venerable, was now archaic, and its peculiar script a curiosity. In numbers, too, and range, the Pyu were doubtless far inferior to the Mon. In seeking to impose the Mon written language on the peoples of Burma, Kyanzittha had reason enough: but other considerations, I suspect, may have influenced his choice. Like many another conqueror in history, the victor of the Mons was vanquished by their culture.
Luce went further than Blagden by stating a specific date for the erection of the Kubyaukgyi: “It was doubtless built in or about 1113 A.D., shortly after Kyanzittha’s death.”
Aung-Thwin (2005) challenges Luce’s views, denouncing them as part of what Aung-Thwin called the “Mon Paradigm”. In Aung-Thwin’s alternative paradigm, written OB preceded written OM in Burma, the OB on the Kubyaukgyi was not one of the very earliest, much less the first, attestation of the Burmese language, and Kyanzittha’s choice of OM for his inscriptions was an idiosyncratic aberration without long-term consequences. Aung-Thwin regards the OM on the Kubyaukgyi erected after Kyanzittha’s death as a last gasp of the language “as a medium for [Burmese] royal communication”.
On the issue of chronology, Aung-Thwin noted that the name Kyanzittha did not actually appear in the Kubyaukgyi and suggested that
if another calculating era, such as that used in Thailand was intended, or if the date was meant to represent a yet-to-be-completed year, then the reign of this king must be changed accordingly and calculated with 543 BC (hence, to 1083 AD). Since the inscriptions also state that the king had ruled for twenty-eight years, it means the original of the two Kubyaukgyi stones had to have been inscribed thereafter, dating the Kubyaukgyi to 1111, not 1112, as conventionally given. The second stone with its newer-looking script could, of course have been inscribed much later than either date, an issue not yet discussed in Burma Studies.
It is not clear which pillar has the “newer-looking script”. The question of which pillar came first has also not yet been discussed in Burmese studies. Blagden (1909: 66) and Duroiselle (1919a: 1) both regarded the texts on the B pillar as “replicas” of those of the A pillar, but neither stated their reasoning.
The first volume of the Ancient Burmese Inscriptions series (N.A. 1972: 1–7) contains readings of the A and B versions of the OB and Pali texts of the Kubyaukgyi in its first pages.
At some point prior to his death in 1979, Luce (n.d.b) transliterated the barely legible fragments of an OB text at the bottom of the B version of the Pali text. To the best of my knowledge, this has been the only study of that text until now.
Beckwith (2002a) reprinted entries from the vocabulary in Shafer (1943) with minor changes and the suggestion that Pyu aṁ “represents a vowel different from [a]” which “was perhaps closer to [e]”. Griffiths et al. (2017: 146–147) later found a spelling alternation in PYU 16 confirming Beckwith’s suggestion. I return to that suggestion in § 4.5.3.
Sawada (2002–2006) offers color photographs, transliterations, word-for-word English glosses, and English translations of all four sides of Kubyaukgyi pillar A. It also contained color photographs of all four sides of pillar B, but only the OB text of that pillar had a transliteration and an English translation without word-for-word glosses. This website was the first attempt to provide morpheme-by-morpheme glosses for all four languages. The glosses and translations are based on Duroiselle (1919a), Duroiselle (1919b), Blagden (1919a), and Blagden (1919b).
Kato (2005) translated the Pyu text of the A pillar into Japanese with word-by-word glosses. His interpretation of the Pyu script incorporated several novel features, the most noteworthy being his equation of ḅ, d…ṃ, and g…ṃ with implosives [ɓ ɗ ɠ]. He then compared Pyu with Karen which also has implosives. He observed that visarga in Pyu corresponded to Kato (2005)’s Proto-Karen tone 2, whereas the absence of visarga almost always corresponded to Haudricourt (1946)’s Proto-Karen tones 1 and 3. The final glottal stop that he reconstructed for several Pyu words written with an anusvāra corresponded to a Karen final glottal stop. However, Katō’s Pyu-Karen correspondences are difficult to evaluate because they are largely dependent upon his idiosyncratic, unexplained reconstructions of Pyu phonology and semantics.
Krech (2012) is even bolder than Katō while also lacking in substantive argumentation. Krech declared his article to be “the outset of a methodological theory of how to reconstruct ancient languages” (p. 121). But in fact he spent more time criticizing his predecessors than proposing a testable theory.
Unlike previous scholars who looked at the Kubyaukgyi Pyu text with reference to other Pyu inscriptions, Krech viewed that text as an isolated example of what he called “Myazedi Pyu”, regarding other Pyu texts as potentially being in other languages without demonstrating any differences between those texts and the Pyu text of the Kubyaukgyi. Solely on the basis of the Kubyaukgyi, Krech declared that
Myazedi Pyu seems to have been either (i) a Yipho-Naxi-Burmese language with some important contact influence from Kuki-Chin or (ii) it was originally a Kuki-Chin language that has been deeply modified by some member of the Yipho-Naxi-Burmese group (most notably Mranma).
Krech does not provide any evidence that would justify either of these classifications of “Myazedi Pyu”. Given Krech’s statement that “the narrower we can identify the genetic affiliation of a certain language the less arbitrary the lexical identifications will tend to be,” it is likely that his glosses for the Kubyaukgyi are rooted in his assumptions about the position of Pyu in the Trans-Himalayan family. However, like Katō, Krech does not explain how he arrived at his glosses. Moreover, Katō and Krech even supply conflicting glosses for words whose meanings eluded most of their predecessors: e.g.,
ḅa doṃ (line 1)
Blagden: (no gloss)
Shafer: (no gloss)
Tha Myat: ‘nibbāna’ from Pali pada which actually means ‘foot’ or, by extension, ‘unit’ (e.g., of verse) but not nibbāna itself.7
Than Tun: (no gloss)
Katō: ‘to profess faith’ + ‘eminent’, both modifying ḅaṁḥ ‘eminent person’
Krech: ‘Buddhist teachings’
Despite the methodological problems with his study, Krech should be credited with being the first to produce tables of the script of the Kubyaukgyi. He was also the first to transliterate the symbol Z that was overlooked in previous studies with the exceptions of Blagden (1911: 383, 385) and Shafer (1943: 318) and even generally blacked out in Tha Myat’s rubbings.
In contrast with Katō and Krech, Yabu (2006) is on firmer ground in two senses; he deals only with the far better understood OB text of the Kubyaukgyi, and he refrains from speculations. He translates the OB text into both word-for-word and natural Japanese and supplies transliterations into both the Latin and modern Burmese scripts. Like Nishida, he uses the OB text primarily as a source of OB-WB sound correspondences, though he also provided notes on grammatical morphemes and expressions extinct in modern Burmese.
Jenny and McCormick (2014), a handbook article on OM, contains word-for-word and natural English translations of the OM text of the Kubyaukgyi up to the middle of line 27 as a sample of the language. This most recent interpretation of the OM text differs little from Blagden’s final reading nearly a century earlier.
Jenny (2015) glossed OM, OB, and Pyu versions of a single line of the Kubyaukgyi to compare what he interprets as permissive causatives in the three languages.
Win (2016: 166–174), a collection of readings of many Pyu inscriptions, reproduces rubbings of the Pyu text of pillar A from Tha Myat (1958) and contains a new eye copy and Burmese transliteration of the Pyu text of pillar B. Like Than Tun and Tha Myat before him, Sein Win did not follow Blagden’s now century-old distinction between Pyu b and ḅ; both are b in transliteration, though they are distinct in his eye copy.
Griffiths et al. (2017) presents the archaeological context for the Pyu corpus, a history of the study of Pyu inscriptions, and “a methodology and notation for analyzing and representing Pyu inscriptional materials that can be applied to future research”. Although the only Pyu text it covered in detail was PYU 16, this study does mention the Kubyaukgyi at numerous points because of its great importance in Pyu studies: e.g., it cited Pyu words from the B pillar in its arguments in favor of interpreting the interlinear symbols of the Pyu script as subscript syllable-final consonants. Although previous scholars had observed those symbols in various Pyu texts, Griffiths et al. (2017) was the first study to note their presence in the Kubyaukgyi.
Miles and Hill (2018) describe the benefits of Reflectance Transformation Imaging (RTI) technology for reading Pyu inscriptions. This study demonstrates how RTI can bring out details that were difficult to see or absent from previous reproductions of inscriptions. Among its examples of such details were the subscript consonants of the B pillar of the Kubyaukgyi.
Apart from Sawada (2002–2006) and Jenny and McCormick (2014), there has been no work on the OM text of the Kubyaukgyi since Blagden, and there has never been an in-depth study of the Pali text. This is perhaps understandable since there are older and longer OM texts, and the Pali text is but a drop in the vast sea of Pali literature, whereas the OB text is one of the earliest in the language, and the Pyu text is one of the very few in that language with counterparts in other languages and is therefore a major key to the decipherment of Pyu.
2 Objectives of the present study
My study differs from its predecessors in three ways.
First, my readings are based on RTI images made in 2014 and 2016. Unlike the photographs and rubbings used by previous scholars, RTI images can be viewed with simulated lighting using a number of filters to bring out details and exclude noise for more accurate readings. My study is the first to incorporate Arlo Griffiths’ identification of subscript final consonants in the B version of the text. The acknowledgement of those consonants has great consequences for Pyu etymology; they strengthen some proposals for cognates while weakening or ruling out others.
Second, as part of the supplementary material, I have compiled the first five-way glossary of the Pyu text of the Kubyaukgyi. Although I claim no expertise in OB, OM, and Pali, I have synthesized the work of my predecessors and colleagues to use those languages to decipher Pyu. Whenever possible, I have explictly matched Pyu words with their counterparts in the other languages.
Third, also as part of the supplementary material, I have compiled the first apparatus for all the texts of the Kubyaukgyi to facilitate the comparison of all readings from 1892 to the present.
3 The text of the Kubyaukgyi inscription (A = PYU 7)
3.1 The Old Burmese text of A (PYU 7)
Reading by Arlo Griffiths, Julian K. Wheatley, and Marc Miyake; glosses by Marc Miyake.
1.
‖
śrī
‖
namo
buddhāya
‖
purhā
skhaṅ·
sāsanā
°anhac·
ta-
glory
reverence
Buddha.dat.sg
Buddha
lord
religion
year
one
Glory! Reverence to the Buddha! One thousand six hundred and
2.
c·
thoṅ·
khrok[·]
ryā
nhac·
chāy[·]
het·
nhac·
lon·
one
thousand
six
hundred
two
ten
eight
year
elapse
twenty-eight years of Lord Buddha’s religion
3.
liy·
brī
rakā
‖
°īy·
°arimaddanapur·
maññ·
su
praññ·
prox
complete
sbo
this
pn
be.named
attr
city
having elapsed, in this city named Arimaddanapura [Pagan],
Nishida (1955: 26) regards liy· as “expressing the feeling that an action is approaching the direction of either the writer or the speaker”.
4.
nhik·°ā
‖
śrī
tribhuvanāditya
dhamma-rāj·
maññ·
su
ma-
loc
hon
pn
dharma-king
be.named
attr
king
there was a king named Śrī Tribhuvanāditya
nhik·°ā is a combination of the locative marker nhik· and the dative marker °ā. nhik·°ā cannot be translated literally as ‘in to’, so I regard it as a single synchronic unit.
5.
ṅ·
phlac·
°e°·
‖
thiv·
maṅ·
°e°·
pāy·
mayā
ta-
king
be
rls
that
king
gen
dear
wife
one
Dhammarāja. That king’s dear wife
6.
c·
yok·
su
kā
trilokavaṭaṁsakā
devī
one
clf
person
top
pn
queen
was named Queen
7.
maññ·
°e°·
‖
thiv·
pāy·
mayā
sā
tamuleh·
rā[ja]-
be.named
rls
that
hon
wife
son
as.for
pn
Trilokavaṭaṁsakā. As for that queen’s son, he
8.
kumār·
maññ·
°e°·
‖
thiv·
maṅ·
kā
kyon·
suṁ
rvoh·
pn
be.named
rls
that
king
top
slave
three
village
was named Rājakumāra. That king gave the queen
9.
teh·
pāy·
mayā
°ā
piy·
°e°·
‖
thiv·
pāy·
ma-
emph
hon
wife
dat
give
rls
that
hon
wife
three slave villages. That queen
Nishida (1955: 32) regards teh· as an accusative marker. It could be interpreted that way here and in B11, though not in other contexts.
10.
yā
syī·
kha
rakā
‖
thiv·
pāy·
mayā
tan·chā
nhaṅ’·
thi-
wife
die
pst
sbo
that
hon
wife
ornament
com
that
died, and the king gave that queen’s ornaments
The word syī· to be read as sīy· [Eds.].
11.
v·
kyon·
suṁ
rvoh·
su
nhaṅ’·
teh·
thiv·
pāy·
mayā
that
slave
three
village
person?
com
emph
that
hon
wife
and those slaves, the people of the three villages, again
The function of su is unclear. It cannot be an attributive marker since it neither follows a verb nor precedes a noun. It cannot be a nominalizer since rvoh· ‘village’ is already a noun. Perhaps it is ‘person’ as in B6.
teh· may be an accusative marker. See B9.
12.
sā
°a
sā
rājakumār·
maññ·
so
°ā
maṅ·
piy·
tuṁ
son
dat
son
pn
be.named
nmlz
dat
king
give
do.again
to the queen’s son named Rājākumāra.
sā unexpectedly appears twice. Duroiselle (1919a: 40) regards the first sā as an error that was partly crossed out. He believes the passage was intended to read mayā °· sā ‘queen gen son’, though the virāma · has been accidentally omitted from mayā °a·. He notes that a virāma is also missing from A27. Unfortunately the passage in B corresponding to his reconstructed mayā °· sā is damaged, so the other pillar cannot confirm his guess.
13.
°e°·
‖
thuiv·
maṅ·
°anhac·
nhac·
chāy·
het·
nhac·
maṅ·
mū
bri
ru-
rls
that
king
year
two
ten
eight
year
king
reign
pfv
sbo
That king reigned twenty-eight years.
14.
y’a
°e°·
‖
siy·
[kha]mū
nā
su
rhov·
nhik·
teh·
‖
thuiv·
rājaku-
sbo
rls
die
about.to
ill
attr
time
loc
emph
that
pn
When he was ill and about to die, the queen’s son
I transliterate ruy’a with a final vowel because it is missing a vowel-cancelling visarga ·. The correct spelling first appears in B17.
15.
mār·
maññ·
su
pay·
mayā
sā
mimī
keiv·
muy·
so
maṅ·
pn
be.named
attr
hon
wife
son
self
acc
raise
attr
king
named Rājakumāra remembered the favors of
16.
grī
klaññjo
°ok·mi
rakā
‖
rhuy·
°ati
su
purhā
skhaṅ·
°a
great
favor
remember
sbo
gold
all
attr
Buddha
lord
image
the great king who raised him. He made a pure gold image
17.
chaṅ·
plu
ruy’·
°e°·
nhap·
liy·
su
rhov·
teh·
°iy·
si-
image
make
sbo
rls
approach
prox
attr
time
emph
this
man-
of Lord Buddha, and when he approached [the king], he
Julian Wheatley (p.c.) notes that in modern Burmese, nhap· means ‘to bring [food] to a proper consistency’. That meaning of nhap· does not fit here, so nhap· has either undergone a major semantic shift since OB or A17 contains an unrelated homophone.
Unable to identify a convincing modern reflex of OB nhap· or any other passages with nhap· in a similar context, Duroiselle (1919a: 40) regards it as a hapax legomenon and equates it with °upasaṅkamma ‘approach.abs’ on the Pali face of the Kubyaukgyi. I have followed Duroiselle here.
18.
°·
min’·
°e°·
‖
°iy·
rhuy·
purhā
kā
ṅa
skhaṅa
°aphei°·
°ati°·
ky-
-ner
speak
rls
this
gold
Buddha
top
1sg.poss
lord
for
1sg.hum
slave
spoke thus: “As for this gold Buddha, I, [your] slave, have made
19.
on·
plo°·
su
teh·
‖
kyon·
su[ṁ]
rvoh·
°atui°·
kyo-
servant
make
nmlz
emph
slave
three
village
1sg.hum
slave
it for my lord. As for the three villages
su seems to be nominalizing the clause like modern colloquial Burmese tā.
Nishida (1955: 25) seems to regard teh· as a copula, though it is possible that the copula なり is intended as a translation and not as a word-for-word gloss. Nishida has extensive endnotes throughout his edition of the text but leaves this word unmarked, so I can only take his apparent gloss at face value.
20.
n·
ṅa
skhaṅ·
piy·
su
saññ·
kā
°iy[·]
rhuy[·]
pur[h]ā
°ā
°atui°·
kyo-
slave
1sg.poss
lord
give
nmlz
top
top
this
gold
Buddha
dat
1sg.hum
slave
of slaves my lord gave [me], I gave [those] slaves to this
The WB reflexes of OB saññ· and kā are polysemous, so there is disagreement on how they function here. Nishida (1955: 30) interprets saññ· kā as ‘because’, literally ‘this’ + ‘from’, and reads the line as ‘Since my lord gave three villages of slaves, I gave those slaves to the lord Buddha’. However, the WB reflex of OB kā, ka in the sense of ‘from’, refers to times, places, and events, not causes. Both WB ka and saññ· can function as topic markers. Hence I follow Sawada (2002–2006) who regards saññ· kā as a pleonastic double topic marker sequence. However, Yabu (2006: 31) glosses saññ· kā as a topic-subject marker sequence which is also possible since WB ka can also serve as a subject marker.
The vowel-cancelling visarga · has been accidentally omitted from °iy and rhuy. The correct spelling of the latter, rhuy·, is attested elsewhere: e.g., A16.
21.
n·
piy·
ye°·
‖
thiv·
rhov·
teh·
maṅ·
nhac[·]klui°·
rakā
koṅ·
slave
give
rls
that
time
emph
king
pleased
sbo
good
gold Buddha. At that very moment, the king was pleased, and
22.
lheṅ’·
teh[·]
koṅ·
lheṅ’·
teh·
min’·
ruy’·
°e°·
saṅgrī
?
emph
good
?
emph
speak
sbo
rls
master
said, “That is indeed good! That is indeed good!” The lord
I am unsure how to translate lheṅ’·. Duroiselle (1919a: 40) regards lheṅ’· as the fusion of “a verbal euphonic suffix” le with “colloquial aspiration” (i.e., voicelessness) and a “future suffix” °am’·. Even if one ignored the problem of the mismatching initials, that etymology is still phonetically impossible, as final /m/ and /ŋ/ were distinct in Old Burmese. A contraction of le and °am’· should have resulted in ˟°lem’· with m’·, not lheṅ’· with ṅ’·.
Nishida (1955: 30–31) regards lheṅ’· as an early spelling of WB lhuiṅ’· ‘to be numerous, abundant’.
I follow Duroiselle (1919a: 40) and Nishida (1955: 25) who both regard teh· as an exclamatory marker. Nishida (1955: 28) views WB taññ·ḥ ‘indeed’ as the descendant of OB teh·. However, Yabu (2006: 31) regards lheṅ’· teh· as a unit ～ダロウ ‘would be’.
23.
mahāther·
‖
saṅgrī
muggaliputtati[ssa]tther·
‖
saṅgrī
su-
pn
lord
pn
lord
pn
Mahāthera, the lord Muggaliputtatissatthera, the lord Su-
24.
medhapaṇḍit·
‖
saṅgrī
brahmapāl·
‖
saṅgrī
brahmadi-
pn
lord
pn
lord
pn
medhapaṇḍita, the lord Brahmapāla, the lord Brahmade-
25.
v·
‖
saṅgri
son[·]
‖
saṅgrī
saṅghasena
vara-paṇḍi-
pn
lord
pn
lord
pn
best-pun-
va, the lord Sona, the lord Saṅghasena, best of pun-
26.
t·
‖
thuiv·
skhaṅ·
tui°·
°amhok·
teh·
maṅ·
[r]iy·
son·
°e°·
dit
that
lord
pl
presence
emph
king
water
pour
rls
dits, in the presence of those lords, the king poured water.
27.
‖
thiv·
brī
rakā
thuiv·
rājakumār·
maññ·
su
pay·
mayā-°a
sā
that
complete
sbo
that
pn
be.named
attr
hon
wife-gen
son
After that was finished, the queen’s son named Rājakumāra
The virāma is missing from °a as in A12. It is present in the corresponding °a in B28.
28.
thiv·
rhuy·
purhā
thāpanā
ruy’·
°e°·
°iy·
rhuy·
°athot·
mū
so
that
gold
Buddha
enshrine
sbo
rls
this
gold
spire
do
attr
enshrined the gold Buddha and made a cave-pagoda with
29.
kū
plo°·
°e°·
‖
plo°·
brī
rakā
°iy·
kū
purhā
lhot·
cave-pagoda
make
rls
make
complete
sbo
this
cave-pagoda
Buddha
dedicate
this gold spire. Having made that, when he dedicated
30.
su
rhov·
nhik·
teh·
‖
sak·munalon·
tac·
rvoh·
‖
rapā-
attr
time
loc
emph
pn
one
village
pn
this cave-pagoda and its Buddha, he brought the one village of Sakmunalor,
31.
y·
tac·
rvoh·
‖
hen·buiv·
tac·
rvoh(·)
‖
°iy·
kyon·
pn
one
village
pn
one
village
this
slave
the one village of Rapāy, the one village of Henbuiv—these
32.
suṁ
rvoh·
yo
ruy’·
°e°·
‖
thuiv·
rājakumār·
maññ·
su
pay·
three
village
bring
sbo
rls
that
pn
be.named
attr
hon
three villages of slaves. The queen’s son named Rājakumāra
33.
mayā°·
sā
°iya
kū
purhā
°ā
riy·
son·
ruy’·
°e°·
°iy’·
sei-
wife
son
this
cave-pagoda
Buddha
dat
water
pour
sbo
rls
this
way
poured water for this cave-pagoda and its Buddha, and
I transliterate °iya with a final vowel because it is lacking its vowel-cancelling visarga ·. Perhaps the visarga has been omitted because there is no space for it under the u of ruy’· of the previous line. The correct spelling °iy· is in the following line.
34.
°·
min’·
°e°·
‖
°iy·
ṅā
°amho°·
kā
sarvvaññutaññā-
way
speak
rls
this
1sg.poss
deed
top
omni-
spoke thus: “May this deed of mine be the cause
35.
ṇ·
prajññā
ra
°am’·
sū
°akroṅ·
phlac·
ciy’·
teh·
‖
ṅa
science
wisdom
get
fut
attr
cause
be
caus
emph
1sg.poss
of my obtaining omniscience and wisdom. After
36.
noṅ·
°ā
ṅa
sā
laññ·goṅ·
‖
ṅa
mliy·
laññ·goṅ·
‖
ṅa
°achu-
after
dat
1sg.poss
son
be.it
1sg.poss
grandson
be.it
1sg.poss
relatives
me, be it my son, be it my grandson, be it
37.
y·
laññ·goṅ·
‖
sū
tac·thū
laññ·goṅ·
‖
°iy·
purhā
relatives
be.it
person
other
be.it
this
Buddha
my relative, be it another person, if they
38.
°ā
ṅā
lhū
kha
su
kyon·
°anhip·°acaka
teh·
mū
mu-
dat
1sg
offer
pst
attr
slave
ill.treatment
emph
do
if
poorly treat the slaves that I offered to this Buddha,
39.
kā
‖
°arimittiryā
purhā
skhaṅ·
°a-phu
ra
ciy·
‖
=
‖
if
pn
Buddha
lord
not-behold
get
caus
may they not get to behold the Lord Buddha!”
3.2 The Old Mon text of A (PYU 7)
Reading by Arlo Griffiths and Marc Miyake; glosses by Marc Miyake.
1.
‖
śrī
‖
namo
buddhāya
‖
śrī
‖
sās·
kyek
buddha
tirley·
glory
reverence
Buddha.dat.sg
glory
religion
sacred.being
Buddha
lord
Glory! Reverence to the Lord Buddha! Glory!
2.
kuli
°ār·
moy·
lṅim·
turov·
klaṁ
ḅār·
cvas·
diññcām·
cnām·
last
go
one
thousand
six
hundred
two
ten
eight
year
When the religion of the Lord Buddha had lasted for 1628 years
3.
tuy·
‖
ḍe[y·]
[ḍu]ṅ·
(°a)rimaddanapur·
vo°·
smiṅ·
śri
̄tribhuvanādi-
pfv
in
city
pn
this
king
hon
pn
in the city of Arimaddanapura [Pagan], its king was Śrī
4.
tya
dhammarāj·
das·
‖
gna.kyek·
smiṅ·
goḥh·
moy(·)
tri-
pn
dharma-king
be
queen
king
that
one
pn
Tribhuvanāditya Dhammarāja. One of the king’s queens was
5.
lokavaṭaṁsakā
devī
°imo°·
‖
kon·
gna.kyek·
goḥ-
pn
queen
name
child
queen
that
named Trilokavaṭaṁsakādevī. The son of that queen was
6.
h·
rājakumār·
°imo°·
‖
smiṅ·
goḥh·
kil·
ḍik·
pi
tvā-
that
pn
name
king
that
give
slave
three
vil-
named Rājakumāra. The king gave three villages of slaves
7.
ññ·
ku
gna.kyek
goḥh·
‖
kāl·
gna.kyek·
goḥh·
cuti
-lage
obl
queen
that
time
queen
that
die
to the queen. When the queen died,
8.
°ār·
‖
‘°ut·
kiryā
gna.kyek·
goḥ
ku
ḍik·
pi
tvāññ·
goḥ
go
all
apparel
queen
that
obl
slave
three
village
that
the king gave all the queen’s possessions and all three
9.
smiṅ·
tun·
kil·
ku
kon·
gna.kyek·
ma
°imo°·
rājaku-
king
again
give
obl
child
queen
rel
name
pn
of the villages of slaves again to the son of the queen, who was named
10.
mār·
goḥ
‖
smiṅ·
goḥh·
kmin·
ḅār·
cvas·
diññcām·
cnām·
tuy·
pn
that
king
that
reign
two
ten
eight
year
pfv
Rājakumāra. The king reigned for 28 years,
11.
[kā]l·
smiṅ·
goḥ
‘jey·
ññan·
scuti
‖
kaun·
gna.kyek
ma
°i-
time
king
that
sick
near
irr.die
child
queen
rel
name
and when he became sick, approaching death, the queen’s son named
12.
mo°·
rājakumār·
goḥ
[m]ir·nas·
guṇ·
ma
smiṅ·
°iññcim·
name
pn
this
remember
virtue
rel
king
feed
Rājakumāra remembered the virtu[ous things] that the king did to sustain
13.
jirku
kin[d]aṁ
kyek·
thar·
moy·
°ār·
tu[ḅ]ok·
smiṅ·
mu-
body
build
sacred.thing
gold
one
go
offer
king
inform
him. He cast a golden Buddha image and went to offer it to the king, informing
14.
nas·
rov·
vo°·
‖
kyek·
thar·
vo°·
°ey·
ḍik·
pa
raṁ-
inform
manner
this
sacred.thing
gold
this
1sg
slave
make
portion
him thus: “This golden Buddha image I have made on your behalf,
15.
po°·
tirla
ḍik·
pi
tvāññ·
ma
tirla
kil·
ku
°ey·
goḥh·
portion
lord
slave
three
village
rel
lord
give
obl
1sg
that
[mi]lord. Those three villages of slaves which you gave me,
16.
°ey·
ḍik·
kil·
ku
kyek·
vo°·
tirla
°anumodanā
da°a·
1sg
slave
give
obl
sacred.thing
this
lord
approve
foc
I give to this image. May you approve, [mi]lord.”
17.
‖
kāl·
goḥ
smiṅ·
sḍik·
gap.pumas·
thic·
°ā
thic·
°ā
smiṅ·
p·
time
that
king
pleased
pleased
good
xcm
good
xcm
king
do
Then the king was pleased and saying, “well done, well done,” gave his
p· is an error for pa which appears correctly in B23.
18.
sādhukār·
‖
kāl·
goḥh·
tirla
poy·
mhā[the]r·
‖
ticā-
approval
time
that
lord
1pl
senior.monk
lord-
approval. Then [in the presence] of our lord, the Senior Monk, the lord-teacher
Shorto (1971: 149) derives ticār·, a title for “(leading) monks”, from tirla ‘lord’ + °ācār· (< Sanskrit ācārya) ‘teacher’.
19.
r·
muggaliputtat(i)ssatther·
‖
ticār·
sumedhapaṇḍit·
‖
ti-
teacher
pn
lord-teacher
pn
lord-
Mugaliputtatissathera, the lord-teacher Sumedhapaṇḍita, the
20.
[cā]r
brahmapāl·
‖
ticār·
brahmadiv·
‖
ticār·
son·
teacher
pn
lord-teacher
pn
lord-teacher
pn
lord-teacher Brahmapāla, the lord-teacher Brahmadeva, the lord-teacher Son,
21.
‖
ticār·
saṅghasena
vara-paṇḍit·
‖
kinta
tirla
lord-teacher
pn
best-pundit
before
lord
and the lord-teacher Saṅghasena, best of pundits, before these lords,
22.
ta
goḥ
smiṅ·
cut·
ḍek·
han·
ti
‖
blaḥ
goḥ
kon·
gna.kye-
pl
that
king
put
water
loc
soil
end
that
child
queen
the king poured water on the soil. After that, the son of the queen,
blaḥ is a verb ‘to come to an end’ (Shorto 1971: 279) that may be related to the homophonous noun blaḥ ‘that which precedes or is finished’ (Shorto 1971: 278) found as an adverb in isolation or followed by other morphemes such as goḥ. Shorto does not regard the two blaḥ as related; he even provides distinct lists of Mon-Khmer cognates for them.
23.
k·
ma
°imo°·
rājakumār·
goḥ
ket·
kyek·
thar·
goḥ
queen
rel
name
pn
that
take
sacred.thing
gold
that
queen, who was named Rājakumāra, took the gold image
24.
thāpanā
kandaṁ
guoh·
cloṅ·
thar·
[v]o°·
‖
kāl·
busac·
kye-
enshrine
build
cave-pagoda
spire
gold
this
time
dedicate
sacred.thing
and enshrined it, building this cave-pagoda with the golden spire. When he dedicated this image,
25.
k·
guoh·
vo°·
kon·
gna.kyek
goḥ
ket·
sak·muna-
sacred.thing
cave-pagoda
this
child
queen
that
take
pn
and cave, the queen’s son brought from the villages of
26.
lor·
moy·
tvāññ·
‖
rahay·
moy·
tvāññ·
‖
ññaḥh·
(gir°u-)
pn
one
village
pn
one
village
pn
Sakmunalor, Rapāy, and Ñaḥgir-
Blagden (1909: 809) regards ññaḥh· (gir°u)y· (A) and ññaḥh· gir°uy· (B) as errors for *hen·buiy· which is much closer to OB hen·buiv·.
27.
p·
|
moy·
tvāññ·
‖
°a’ut·
ḍik·
pi
tvāññ·
goḥ
cut·
ḍe(k·
ku)
pn
one
village
all
slave
three
village
that
put
water
obl
‘uy, all the slaves of the three villages, and poured water for
28.
kyek·
thar·
ma
māpanā
hin·
goḥ
vo°·
rādhanā
rov·
(vo°a)
sacred.thing
gold
rel
enshrine
for
cave-pagoda
this
pray
manner
this
the gold image that he had enshrined for this cave, [and] prayed thus:
māpanā is an error for thāpanā in B influenced by the preceding ma.
29.
‖
sinraṅ·
°e°·
vo°·
°or·
dap·
het·
ku
gvo°·
sarvvaññ(uta)-
deed
1sg
this
opt
be
cause
obl
attainment
omniscience
“May this deed of mine be a cause for the attainment of omniscience!
30.
ññāṇ·
‖
kon·
°ey·
laḥ
‖
cov·
°ey·
laḥ
‖
ku(lo)
omniscience
child
1sg
or
grandchild
1sg
or
kinsman
Be it my child or my grandchild or my kinsman
31.
°ey·
laḥ
‖
ññaḥ
c’eṅ·
laḥ
‖
yal·
pa
X
°upadrov·
ku
ḍ(i)-
1sg
or
person
other
or
if
do
violence
obl
ser-
or [any] other person, if he do violence to the slaves
32.
k·
ma
°ey·
kil·
ku
kyek·
vo°·
yaṅ·
ññir·ññāc·
kye-
vant
rel
1sg
give
obl
sacred.thing
this
emph
sight
sacred.
whom I am giving to this very image, may he
33.
k·
trey·
mettey·
laḥ
°or·
ḍeh·
go°·
‖
0
‖
thing
sacred.being
pn
proh
opt
he
get
not get sight of Metteya the most holy!
3.3 The Pali text of A (PYU 7)
Reading by Arlo Griffiths and Marc Miyake, with corrections by Aleix Ruiz-Falqués.
Stanzas are numbered with Roman numerals in parentheses: e.g., ‖ śrī ‖ (I) buddhādikaṁ in line 1 indicates that stanza 1 begins with buddhādikaṁ.
1.
‖
śrī
‖
(I)
buddhādikaṁ
vatthu-varaṁ
namitvā
puññaṁ
kataṁ
yaṁ
jina-sā
glory
buddha.beginning.m.acc.sg
object-excellent.m.acc.sg
bow.abs
merit.acc.sg
do.ppp.n.acc.sg
rel.n.acc.sg
conqueror-
Glory! After bowing down to the excellent objects [of respect] beginning with the Buddha, I shall
2.
sanasmiṁ
°anādikaṁ
rājakumāra-nāma-dheyyena
vakkhā-
dispensation.n.loc.sg
perpetual?.3sg.n.acc
pn-name-assigning.m.ins.sg
speak
speak of the perpetual … by the one called Rājakumāra in the Conqueror’s dispen-
3.
mi
sunātha
me
taṁ
‖
(II)
nibbānā
loka-nāthassa
°aṭha-vī-
.fut.1sg
hear.imp.2pl
1sg.acc
3sg.n.acc
nirvāṇa.abl.sg
world-lord.gen.sg
eight-
sation. Listen to me! A thousand six hundred twenty-
As is the case in general in the orthography of Burmese Pali manuscripts, here and throughout both the A and B Pali texts, ṭṭha is written ṭha.
4.
sādhike
gate
sahasse
pana
vassānaṁ
cha-sate
cāpare
ta-
twenty.plus.loc.sg
go.ppp.n.loc.sg
thousand.loc.sg
and
year.gen.pl
six-hundred.loc.sg
and.later.loc.sg
thus
eight years having thus gone since the nirvana of the lord of the
5.
thā
‖
(III)
°arimaddana-nāmasmi
pure
°āsi
maha-bbalo
rājā
thus
pn-name.loc.sg
city.loc.sg
be.aor.3sg
great-power.m.nom.sg
king.nom.sg
world, in a city named Arimaddana was a great and mighty King
6.
tibhuvanādicco
°udiccādicca-vaṁsa-jo
‖
(IV)
tass’
ās’
e-
pn.nom.sg
exalted.sun-lineage-born.m.nom.sg
3sg.m.dat
be.aor.3sg
one
Tribhuvanāditya, born of the exalted solar lineage. He had
7.
kā
piyā
devi
sā
tilokavaṭaṁsikā
hi-
.f.nom.sg
beloved.f.nom.sg
queen.nom.sg
3sg.f.nom
pn.f.nom.sg
desiring.
a beloved queen Tilokavaṭaṁsikā, desirous
N.B. devi is indeed the reading, although devī is expected. [Eds.]
Tilokavaṭaṁsikā for Tilokāvataṁsakā. The retroflex ṭ is also in the OB and OM texts. The -i- before kā is unique to the Pali text.8
8.
tesī
kusalā
sabba-kiccesu
pana
rājino
‖
(V)
ta-
welfare.desiring.f.nom.sg
skillful.f.nom.sg
all-duty.loc.pl
and
king.m.gen.sg
3sg.f.gen
of others’ welfare, and skillful in all the affairs of the king.
9.
-ass’
ās’
eko
suto
rājakumāro
nāma
nāma-
…
.be.aor.3sg
one.m.nom.sg
son.m.nom.sg
pn.m.nom.sg
quot
name.n.abl.sg
She had a son named Rājakumāra,
10.
to
°amacco
rāja-kiccesu
byāvato
satimā
name
minister.m.nom.sg
king-duty.loc.pl
zealous.m.nom.sg
mindful.m.nom.sg
a minister, zealous, prudent, and wise in the work of
11.
vidū
‖
(VI)
°adā
gāma-ttayaṁ
tassā
deviyā
so
ma-
wise.m.nom.sg
give.aor.3sg
village-triad.n.acc.sg
3sg.f.dat
queen.f.dat.sg
3sg.m.nom
king.
the king. The king, always devoted, gave the
12.
hīpati
pasanno
sabbadā
dāsa-paribhogena
bhuññjituṁ
nom.sg
devoted.m.nom.sg
always
slave-property.inst.sg
use.inf
queen three villages with the right to use the slaves.
13.
‖
(VII)
°aniccatā-vasaṁ
tassā
gatāya
pana
deviyā
rā-
impermanence-control.m/n.acc.sg
3sg.f.gen
go.ppp.f.gen.sg
and
queen.f.gen.sg
king
And the queen having departed due to the force of impermanence [i.e., the queen having died],
14.
ja
rājakumārassa
°adā
gāma-ttaya
puna
‖
(VIII)
°aṭha-vīsa-
.m.nom.sg
pn.dat.sg
give.aor.3sg
village-triad.n.acc.sg
again
eight-twen-
the king reassigned the three villages again to Rājakumāra. Having ruled
°aṭha- for °aṭṭha-.
15.
ti-vassāni
rajjaṁ
dhammena
kāriya
māranantika-rogassa
-ty-year.n.acc.pl
kingship.n.acc.sg
righteousness.m.ins.sg
do.abs
death.ending-illness.m.gen.sg
the kingship with righteousness for twenty-eight years, the ruler of men
16.
vasaṁ
patte
narādhipe
‖
(IX)
saranto
dhamma-rājassa
mahantaṁ
gu-
control.m/n.acc.sg
reach.ppp.m.loc.sg
man.ruler.m.loc.sg
remember.prs.ptcp.m.nom.sg.
righteous-king.m.gen.sg
great.m.acc.sg
vir
was stricken by a fatal illness. Remembering the great virtue accumulation of
17.
ṇa-saññcayaṁ
kāretvā
satthuno
bimbaṁ
sabbasovaṇṇa-
-tue-accumulation.acc.sg
make.caus.abs
teacher.m.gen.sg
image.n.acc.sg
all-gold
the righteous king, he had an image of the Teacher made entirely
18.
yaṁ
subhaṁ
‖
(X)
gahetvā
taṁ
mahantena
sakkārena
sumānaso
.n.acc.sg
beautiful.n.acc.sg
take.abs
3s.n.acc
great.m.ins.sg
reverence.m.ins.sg
joyful.m.nom.sg
out of gold. Taking it with great reverence, [being] joyful,
19.
°upasaṅkamma
rājānaṁ
°āha
cintitam
attano
‖
(XI)
bhāgaṁ
katvā-
approach.abs
king.dat.pl
say.prf.3sg
thought.acc.sg
self.gen.sg
part.m.acc.sg
do
approaching the king, he told [him] his thoughts. “I made
20.
n-idaṁ
satthu-bimbaṁ
sovaṇṇayaṁ
subhaṁ
°akāsiṁ
vo
va-
.abs-this.n.acc.sg
teacher-image.n.acc.sg
golden.n.acc.sg
beautiful.n.acc.sg
make.aor.1sg
2pl.dat
excellent
this beautiful golden image of the Teacher on your behalf. Rejoice,
21.
raṁ
puññaṁ
sāmi
tumhe
’numodatha
‖
(XII)
gāma-ttayaṁ
pi
vo
.n.acc.sg
merit.n.acc.sg
lord.voc.sg
2pl.nom
rejoice.imp.2pl
village-triad.n.nom.sg
also
2pl.ins
[mi]lord, at this excellent [work of] merit! Even the three villages
22.
sāmi
pubbe
dinnan
tu
me
°ahaṁ
°imass’
eva
munindassa
demi
ta-
lord.voc.sg
in.the.past
give.ppp.n.nom.sg
nonetheless
1sg.dat
1sg.nom
this.m.dat.sg
only
sage.chief.m.dat.sg
give.prs.1sg
3sg
[mi]lord gave me in the past now I give to this very chief of sages. And
23.
ññ
cānumodatha
‖
(XIII)
°evaṁ
vutte
mahīpālo
roge-
n.acc
and.rejoice.imp.2pl
thus
say.ppp.loc.sg
king.m.nom.sg
illness
rejoice at that!” When speaking thus, the king, his
24.
nātura-mānaso
sādhu
sādhū
ti
vatvāna
tuṭha-hattho
.m.ins.sg.afflicted-mind.having.m.nom.sg
good
good
quot
say.abs
pleased.ppp-delighted.ppp.m.nom.sg
mind afflicted by disease, having said, “Good, good!”, was pleased and delighted,
tuṭha-hattho for tuṭṭha-haṭṭho.
25.
pamodito
‖
(XIV)
dayāparo
mahāthero
thero
muggali-
rejoiced.ppp.m.nom.sg
compassion.supreme.m.nom.sg
great-thera.m.nom.sg
thera.m.nom.sg
pn
and rejoiced. The supremely compassionate great thera Muggali-
The name here spelled ‘Muggali’ is more commonly written ‘Moggali’.
26.
puttako
sumedhatta
sumedho
ti
laddha-nāmo
ca
paṇḍito
‖
pn.m.nom.sg
wisdom.n.abl.sg
pn.m.nom.sg
quot
obtain.ppp-name.m.nom.sg
and
paṇḍita.m.nom.sg
‖
puttaka, and a paṇḍita, who, on account of his wisdom (sumedhattā), obtained the name Sumedha (Wise).
The Pali text of B (PYU 8) reads more correctly sumedhattā sumedho, with a long vowel.
27.
(XV)
brahmapālo
tathā
brahmadevo
sampanna-sīlavā
sono
.pn.m.nom.sg
likewise
pn.n.nom.sg
succeed.ppp-virtuous.m.nom.sg
pn.m.nom.sg
Brahmapāla, likewise Brahmadeva the successful and virtuous, Sona
28.
bahu-ssuto
saṁghasenavho
vara-paṇḍito
‖
(XVI)
°etesaṁ
pa-
much-learned.m.nom.sg
pn.called.ppp.m.nom.sg
excellent-paṇḍita.m.nom.sg
this.3pl.gen
and
the much-learned, [and] the excellent paṇḍita called Saṅghasena. And in
29.
na
bhikkhūnaṁ
saṁmukhā
so
su-mānaso
jalaṁ
pātesi
katvāna
sa-
and
monk.m.gen.pl
in.front
3sg.m.abl
good-minded.m.nom.sg
water.n.acc.sg
fall.caus.aor.3sg
do.abs
witness
front of these monks, he, [being] joyful, poured the water, now taking
30.
kkhin
tu
vasudhā-talaṁ
‖
(XVII)
tato
so
taṁ
mahāmacco
bimbaṁ
sova-
.n.acc.sg
now
earth-surface.n.acc.sg
then
3sg.m.nom
3sg.n.acc
great-minister.nom.sg
image.n.acc.sg
gold
the earth’s surface as [his] witness. Then the minister, having had the
31.
ṇṇayaṁ
subhaṁ
patiṭhāpiya
kāresi
guhaṁ
kaññcana-thūpikaṁ
‖
.n.acc.sg
beautiful.n.acc.sg
establish.caus.abs
make.caus.aor.3sg
cave-pagoda.f.acc.sg
gold-spired.f.acc.sg
beautiful golden image established, had a gold-spired cave-pagoda made.
32.
(XVIII)
katvāna
maṅgalaṁ
buddha-patimāya
guhāya
ca
°akās’
evaṁ
paṇī-
do.abs
ceremony.n.acc.sg
buddha-image.f.dat.sg
cave-pagoda.f.dat.sg
and
make.aor.3sg
thus
aspiration
Having perfomed a ceremony for the Buddha image and the cave-pagoda, weary of existence,
33.
dhānaṁ
nibbinno
bhava-saṅkate
‖
(XIX)
karontena
mayā
°etaṁ
yaṁ
pu-
.acc.sg
weary.ppp.nom.sg
existence-created.loc.sg
do.prs.ptcp.ins.sg
1sg.ins
this.n.acc.sg
rel.n.acc.sg
merit
he made the following aspiration: “May this merit that has been accumulated
For saṅkate read saṅkhate.
The word here written paṇīdhāna is more typically paṇidhāna, with a short -i-. Perhaps this shortening is metri causa, to keep with the usual ˘ ¯ ¯ ¯ (pa-nī-dhā-naṃ) of the first foot of the stanza.
34.
ññaṁ
taṁ
samācitaṁ
hotu
sabbaññuta-ññāṇa-pativedhā-
.n.acc.sg
3sg.n.acc
accumulate.ppp.n.acc.sg
be.imp.3sg
omniscience-wisdom-attainment
and performed by me be the cause of the attainment of omniscience
pati is a spelling variatn of paṭi.
35.
ya
paccayo
‖
(XX)
yattakā
tu
mayā
dāsā
gāma-ttaya-nivā-
.f.dat.sg
cause.nom.sg
however.many.m.nom.pl
but
1sg.ins
slave.m.nom.pl
village-triad-dwelling
and wisdom! But however many slaves dwelling in the three villages
36.
sino
dinnā
guhāya
sovaṇṇa-patimāya
mahesi-
.m.nom.pl
give.ppp.m.nom.pl
cave-pagoda.f.dat.sg
gold-image.f.dat.sg
great.sage
were given by me to the cave-pagoda [and] the golden image of
37.
no
‖
(XXI)
putto
me
vā
paputto
vā
°añño
vā
pana
ññā-
.m.gen.sg
son.nom.sg
1sg.gen
or
grandson.m.nom.sg
or
other.m.nom.sg
or
and
kinsman
the great sage—if any man, whether a son or a grandson
38.
tako
yo
koci
pāpa-saṁkappo
naro
°assaddha—
.m.nom.sg
rel.m.nom.sg
whoever.m.nom.sg
evil-thought.m.nom.sg
man.m.nom.sg
unbelieving—
or a kinsman, with evil thoughts and an unbelieving
39.
mānaso
‖
(XXII)
kareyy’
upadduvaṁ
tesaṁ
dāsānaṁ
so
narādhamo
mind.m.nom.sg
do.opt.3sg
oppression.acc.sg
m.dat.pl
slave.dat.pl
3sg.m.nom
man.vilest.m.n.sg
mind, oppresses those slaves, then may [that] vilest of men
upadduvaṁ is an error for upaddavaṁ.
40.
metteyya-dipadindassa
dassanaṁ
nâthigacchatū
pn-two.foot.lord.m.gen.sg
sight.acc.sg
not.attain.imp.3sg
not attain the sight of Metteya, lord of bipeds!”
athi- is an error for adhi-
41.
ti
‖
0
‖
quot
3.4 The Pyu text of A (PYU 7)
Reading by Arlo Griffiths, Julian K. Wheatley, and Marc Miyake; glosses by Marc Miyake.
1.
1
‖
siri
‖
dathagaṃda
ḅa-doṃ
ḅaṁḥ
ḅiṁḥ
pduṃ.sgu.daṃḥ.ḅa
tva
1000
[600]
1
glory
tathagāta
nirvāṇa
hon
rls
enter?
tmp
1000
600
Glory! Since the Tathāgata … noble nirvāṇa, one thousand six hundred
2.
20
hraṁ
°o
sniḥ
ḅiṁḥ
tvaṃṁḥ
tha-daṃṁ
//
yaṁ
tiṁ
priḥ
rimadhanarbu
°o
rmi
ḅiṁḥ
si
//
20
eight
poss
year
rls
elapse
pfv
this
loc
city
pn
nmlz
be.named
rls
cop
twenty-eight years have elapsed. This was in the city named Arimaddanapura [Pagan].
3.
sri
tribhuvaṃnadiṃṁtya
dhama-raja
°o
rmi
ḅiṁḥ
si
//
°o
doṃḥ
ḍaZ
ḅaṁḥ
°o
rvaṃḥ
ma-
hon
pn
dhamma-king
nmlz
be.named
rls
cop
poss
beloved
or
lord
poss
ruler?
wife
There was a righteous king named Śrī Tribhuvanāditya. His beloved, or the ruler’s
4.
yaḥ
triḍogavaṃdasaga
deṃviṃ
ḅiṁḥ
si
°o
rmi
//
pau
°o
saḥ
rajaguma
ḅiṁḥ
wife
pn
queen
rls
cop
nmlz
be.named
that
poss
son
pn
rls
wife, was named Queen Trilokavaṭaṁsakā. Her son was named
5.
si
°o
rmi
//
°o
vaṁ
traḥ
kra
hoḥ
ḅiṁḥ
paṁḥ
toḥZ
//
pau
ḅaṁḥ
mayaḥ
ḅiṁḥ
hi
ta-daṃṁ
be
poss
name
poss
ln
slave
village
three
rls
give
pfv
that
hon
wife
rls
die
pfv
Rājakumāra. To her [the king] gave three villages of slaves. That queen died.
6.
ma[ya]ḥ
°o
dra
traḥ
kra
hoḥ
ḅiṁḥ
paṁḥ
tḅaḥ
ḅaṁḥ
mayaḥ
°o
saḥ
rajaguma
°o
vaṁ
//
wife
poss
personal.item
slave
village
three
rls
give
again
hon
wife
poss
son
pn
poss
ln
[The king] gave [his] wife’s personal items and the three villages of slaves again to the queen’s son Rājakumāra.
7.
pau
ḅaṁḥ
tdaṃḥ
sniḥ
rpu
hraṁ
biṁḥ
ta
daṃṁ
//
ḅiṁḥ
sriḥ
ḅiṁḥ
hniṁḥ
hḍiṁḥ
hi
°o
mtu
duṃ
that
hon
king
year
twenty
eight
rls
place
pfv
rls
reign
rls
sick
nmlz
die
poss
point
tmp
That king was in place [i.e., ruled] for twenty-eight years. Having reigned, having become sick, [he] was
8.
roḥ
//
pau
ḅaṁḥ
mayaḥ
°o
saḥ
rajaguma
ḅiṁḥ
si
°o
rmi
//
°o
diṃṁ
rls.cop
that
hon
wife
poss
son
pn
rls
cop
nmlz
be.named
3
acc
at the point of death. The queen’s son was named Rājakumāra.
9.
ḅiṁḥ
mtau
ma
pau
tdaṃḥ
to
°o
kḍeḥ.troḥ
diṃṁ
ḅiṁḥ
mdauṃ.haḥ.ḍaḥ
daṃṁ
//
pau
ḅaṁḥ
rls
raise
rel
that
king
righteous
poss
favor?/virtue?
acc
rls
remember
pfv
that
hon
[He] remembered the favors?/virtues? of the righteous king who raised him. He
10.
ḅudha
°o
chaḥ.bo
bradima
tha
[tlu]
ḅiṁḥ
se
kyaḥ
//
pau
ḅaṁḥ
ḅudha
biṁḥ
tuḥ
buddha
poss
likeness
image
golden
entirely
rls
make
caus
that
hon
buddha
rls
offer
caused a pure gold image in the likeness of Buddha to be made. He offered the Buddha
11.
thmuḥ
ḍoḥ
yaṁ
na
ḅi[ṁḥ]
tdiṃḥ
toḥZ
//
yaṁ
ḅaṁḥ
ḅudha
tha
ḅaṁḥ
raḥ.saḥ
biṁḥ
presence
loc
this
manner
rls
say
pfv
this
hon
buddha
golden
hon
on.behalf.of
rls
when in [the royal] presence and said thus: “I made this golden Buddha
12.
se
ma
ḅuḥ
ḅaṁḥ
°o
vaṁ
paṁḥ
ce
choḥZ
//
yaṁ
traḥ
kra
hoḥ
ḅiṁḥ
paṁḥ
make
nmlz
do
hon
poss
ln
give
irr
xcm
this
slave
village
three
rls
give
on behalf of [my] lord, and I shall give it to him!” [The king] gave [him] the three villages
13.
ma
ḅuḥ
//
yaṁ
baṁḥ
hra
tha
°o
vaṁ
paṁḥ
cheZ
//
pau
ḍoḥ
baṁḥ
tdaṃḥ
ḅiṁḥ
kiṁ-
nmlz
do
this
hon
sacred.image
golden
poss
ln
give
irr
that
in
hon
king
rls
pleased
of slaves, [and he] would give [them] to this golden Buddha image. At that point the king was
14.
-pha
daṃṁ
ḅiṁḥ
ṅa
ha
pra
choḥ
ha
pra
choḥ
ḅiṁḥ
si
//
pau
°o
doṃḥ
traḥ
ḅaṁḥ
pleased
pfv
rls
say
good
do
xcm
good
do
xcm
rls
be
that
poss
after
?
hon
pleased and said, “Well done! Well done!” After that the ? lord
15.
mahaṭhe
/
traḥ
ḅaṁḥ
mugaṃḍubudadisaṭhe
/
traḥ
ḅaṁḥ
⟪su⟫medhabadiṃṁ
pn
?
hon
pn
?
hon
pn
pn
Mahāthera, the ? lord Muggaliputtatissatthera, the ? lord Sumedhapaṇḍita,
16.
/
traḥ
ḅaṁḥ
vrahmaba
/
traḥ
ḅaṁḥ
vradaṃyoḥ
/
tra[ḥ]
ḅaṁḥ
su
/
traḥ
ḅaṁḥ
?
hon
pn
?
hon
pn
?
hon
pn
?
hon
the ? lord Brahmapāla, the ? lord Brahmadeva, the ? lord Sona, the ? lord
17.
sagaṃsi
vaṃrabadiṃṁ
//
pau
traḥ
ḅaṁḥ
sagha
tvo
°o
hṅa.diṁ
duṃ
tdaṃḥ
pn
best.paṇḍita
that
?
hon
saṅgha
pl
poss
presence
tmp
king
Saṅghasena, best of paṇḍitas,—when in the presence of the ? saṅgha, the joyful
18.
tu
ḅaṁḥ
ḅiṁḥ
cha
toḥ
tduṃ
//
pau
ḅiṁḥ
ta-daṃṁ
mayaḥ
°o
sa[ḥ]
raja-
joyful
hon
rls
pour
pfv
water
that
rls
pfv
wife
poss
son
pn
king poured water. That having been done, his wife’s son, whose
19.
guma
ḅiṁḥ
si
°o
rmi
ma
[//]
ḅiṁḥ
stabana
[b]udha
tha
ḅi(ṁ)ḥ
se
goṃ
°o
sto
tha
ḅi(ṁḥ)
pn
rls
cop
nmlz
be.named
nmlz
rls
enshrine
buddha
golden
rls
make
cave-pagoda
poss
spire
golden
rls
name was Rājakumāra, enshrined the golden Buddha, made a golden spire of a cave-pagoda, and
20.
ta-daṃṁ
//
pau
goṃ
°o
hḍ[ī]
ḅiṁḥ
saṁḥ
[r]oḥ
//
[sa]manarḍo[ḥ]
kra
taṁ
/
rabai
kra
[ta]ṁ
[/
j]i[ṁ]-
pfv
that
cave-pagoda
poss
dedication
rls
perform
when
pn
village
one
pn
village
one
pn
put [things in place]. He performed the dedication for that cave-pagoda. The one village of Sakmunalor, the one village of Rapāy, the one
21.
vuḥ
kra
taṁ
//
yaṁ
traḥ
kra
hoḥ
diṃṁ
ḅiṁḥ
diṃṁ
daṃṁ
//
yaṁ
ḅaṁḥ
mayaḥ
(°o
saḥ)
pn
village
one
this
slave
village
three
acc
rls
assemble
pfv
this
hon
wife
poss
son
village of Jiṁvuḥ—he assembled these three slave villages. This son of the queen,
22.
rajaguma
yaṁ
goṃ
ḅu[dha]
°o
vaṁ
tduṃ
ḅiṁḥ
chai
ta-daṃṁ
//
yaṁ
na
ḅiṁḥ
diṃṁ
ch[o]
(//
yaṁ)
pn
this
cave-pagoda
buddha
poss
ln
water
rls
pour
pfv
this
manner
rls
pray
quot
this
Rājakumāra, poured water for this cave-pagoda Buddha. [He] prayed thus, “May this
23.
ma
gaṁḥ
pra
ḅuḥ
saveññudeñña
breñña
ḅiṁḥ.ḅiṁḥ
paṁḥ
che
naḥ
tiṁ
pḍaṁḥ
paZ
//
rel
1sg
do
do
omniscience
wisdom
self
give
irr
cause
loc
base
may.be
which I do be the basis for a cause to give myself omniscience and wisdom!
24.
yaṁ
tra
tiṁ
mtu
knaṁḥ
duṃ
gi
saḥ
ḍa
/
gi
pli
la
gi
sruḥ
ḍaZ
mra.ja.hṅa
ḍa
yaṁ
this
slave
loc
regard
fut
tmp
1sg.gen
son
or
1sg.gen
grandchild
or
1sg.gen
kinsman
or
other.person
or
this
In regards to these slaves in the future, whether it be my son or my grandchild or my kinsman or another person, suppose someone
25.
(ḅu)[dha]
°o
vaṁ
gaṁḥ
hḍiṁḥ
toḥ
ma
diṃṁ
/
ga
hñiṁ.chi
ga
bro.pdaṃ
ma
taḥ.ṅuḥ
ḅuḥ
//
buddha
poss
ln
1sg
dedicate
pfv
nmlz
acc
either
ill.treatment
or
?
rel/nmlz
?
do
//
performs either ill treatment or oppression (?) which is evil (?) and unbelieving (?) upon those whom I have dedicated to this Buddha.
26.
yaṁ
baṁḥ
ḅudha
°arimedeyaṃ
daṃṁ
ḅaḥ
kdiṃ.kchiṁḥ
tiṁ
tmu
ma
paṁḥ
che
choḥ
//Z
this
hon
buddha
pn
?
not
sight
loc
presence
?
give
irr
xcm
May [they] not get the sight of this Buddha Āriyametteyya and be permitted in [his] presence!”
4 Phonology of the Pyu text
The phonology of the Kubyaukgyi text has characteristics distinguishing it from the phonology of all other Pyu texts in the corpus other than PYU 39 whose Pyu and Old Mon texts refer to the year 441 = 1078 CE, three and a half decades before c. 1112 CE.
Shafer (1943: 316) was the first to suggest that the Kubyaukgyi text was in Late Pyu whereas earlier Pyu texts were in Old Pyu. Shafer differentiated between the two stages of Pyu on the basis of two criteria:
-
Late Pyu had grammatical differences from Old Pyu
-
Late Pyu had borrowings from Old Mon and Old Burmese absent from Old Pyu
Shafer (1943: 357) also speculated that “If we had more common lexical comparisons from the two periods, some phonetic change might perhaps be observed.”
4.1 (Sesqui)syllabic structure
No changes in syllable structure are observable between Old an Late Pyu. The minimal Old and Late Pyu syllable is /CV/ with an optional preinitial /C./ and coda:
/(C.)CV(C)/
I use the term preinitial to refer to any subsyllabic element which may have been either a single consonant or a consonant followed by a minimal vowel (e.g., a schwa).
Blagden (1919b: 60) was the first to suggest that Pyu was sesquisyllabic:
It [Pyu] allowed a limited number of combinations of two consonants as initials, the second member of such combinations being a semi-vowel or a liquid. But even in such cases it is not quite certain that a short neutral vowel may not have been inserted in between the two consonants, for the spelling often uses conjuncts in combinations (such as td) where such a vowel must of necessity be introduced. It shares this peculiarity with the contemporary Môn usage, a usage that has survived in modern Môn also.
Although such a vowel must have separated t and d and other sequences of homorganic stops, it is not possible to determine in many other cases whether such a vowel existed or was optional: e.g., was mt pronounced as [mət] with a short unstressed vowel, [m̩t] with a syllabic nasal, and/or a true cluster [mt]? The lenition of consonants following preinitials (§ 4.3.2) suggests that /./ was a vowel like [ə] in most or even all cases. Nonetheless, I prefer not to guess when a vowel was present or when a consonant may have been syllabic. I separate all preinitials from initials with a period /./, leaving open the question of how /./ was phonetically expressed.
The Indic script of Pyu is hence not an unambiguous guide to phonological structure. Often all that can be said for sure is that a group of segments was written as an akṣara, a consonant symbol with or without other symbols in ‘orbit’ around it. Since (1) it is often not possible to determine whether an akṣara represents a sesquisyllable or a syllable and (2) there is no convenient cover term for both sesquisyllables and syllables, I will use the term ‘akṣara’ to refer to Pyu phonological units written as single akṣaras. Strictly speaking, ‘akṣara’ should only refer to graphic units rather than phonological units, but I hope the reader will forgive my looser usage.
4.2 Preinitials
The preinitial inventory remains unchanged from Old Pyu:
Table 1
Inventory of Pyu preinitials
|
/k./ |
|||
|
/r./ |
|||
|
/s./ |
/t./ |
/n./ |
|
|
/p./ |
/m./ |
Periods in phonological reconstructions separate preinitial consonants from initial consonants. These periods may or may not have corresponded to short unstressed vowels.
Six of these seven preinitials are also attested in PYU 37, the only known Middle Pyu text. The absence of /s./ is probably accidental due to the short length of the text (45 akṣaras).
Spellings sometimes may indicate the lenition of preinitial stops after vowels:
-
/°o t.ra/ > °o dra (A6) but °o tra (B6)
-
/ga p.ro(C) p.da(C)/ > ga bro pdaṃ (A25, B27)
It is not clear whether this inconsistency reflects phonetic variation: e.g., was it possible to pronounce /°o t.ra/ with either [t] or [d], or was that sequence always pronounced with [d] even though /t.ra/ could be phonemically spelled as tra?
4.3 Initials
Unlike the preinitial inventory which at least remains graphically intact, the initial inventory of Late Pyu has undergone phonetic shifts and phonemic mergers.
4.3.1 Simple initials
I regard one shift as potentially diagnostic for dating texts: the fortition of Old Pyu initial /l/ to Late Pyu retroflex ḍ /ɖ/,9 possibly via a retroflex ḷ /ɭ/ in an intermediate Middle Pyu stage. Retroflex ḍ is unique to the three Late Pyu texts (7, 8, 39), and retroflex ḷ is unique to PYU 37, which I tentatively regard as the only Middle Pyu text.
‘Middle Pyu’ is a shaky category since
-
37 cannot be dated; it may be contemporary with Old Pyu or Late Pyu
-
37 is the only Pyu text found near modern Nay Pyi Taw, so its retroflex ḷ may reflect an unique dialectal development rather than an intermediate stage between Old and Late Pyu
-
ḷ only appears in three distinct akṣaras in PYU 37: ḷo, ḷiṁ, and pḷaṁḥ. Out of these three akṣaras, only pḷaṁḥ resembles a word with a meaning found in another text: Old Pyu plaṁḥ (16.1b, 2b, 2C) > Late Pyu pḍaṁḥ (7.23, 8.24) ‘base’. However, there is no guarantee that pḷaṁḥ in PYU 37 also means ‘base’; it may be an unrelated word.
-
PYU 37 has two apparent exceptions to the sound change /l/ > /ɭ/: pla(ṁ), possibly ‘four’, and le ‘?’. The first akṣara may be a conservative spelling of Middle Pyu /p.ɭä/ corresponding to Old Pyu plaṁ; the second akṣara may either be a conservative spelling of Middle Pyu /ɭe/ corresponding to Old Pyu le or a loanword postdating the Old Pyu /l/ > Middle Pyu /ɭ/ shift.
The Kubyaukgyi has two apparent exceptions to the sound change /l/ > ḍ /ɖ/: tlu ‘entirely’ (A10 and B10) and pli ‘grandson’ (A24) corresponding to pḍi ‘id.’ (B25). The first exception is less certain than the first since the reading tlu is only tentative.
tlu is unique to the Kubyaukgyi. It may be an archaic spelling for /t.ɖu/ from an earlier †tlu not attested elsewhere in the corpus. It may also be a loanword postdating the fortition of l.
pli ‘grandson’ (A24) is not a loanword. Although its spelling is identical to that of Old Pyu pli ‘grandson’ (16.4A), it may represent /p.ɖi/ like pḍi (B25).
A more speculative phonetic shift is the fortition of Old Pyu voiceless dental hl /l̥/ to Late Pyu retroflex hḍ /D/, likely to be the voiceless counterpart of ḍ /ɖ/. hl is unique to Old Pyu, and hḍ is unique to Late Pyu. However, only one of the three Late Pyu hḍ-words in the corpus has a clear ancestor in Old Pyu: the Late Pyu nominalizer hḍiṁḥ (A7, B8) is from Old Pyu hliṁḥ ‘id.’ (16.3d). There are no Middle Pyu words with †hḷ corresponding to Old Pyu hl and Late Pyu hḍ, though their absence may simply reflect the brevity of the only known Middle Pyu text (37). Cṭha (37), the only instance of ṭh in a non-Indic akṣara in the corpus, may contain the Middle Pyu reflex of Old Pyu hl /l̥/.
Table 2
Inventory of Pyu simple initials
|
/°/ |
/h/ |
||||||||
|
/k/ |
/g/ |
hṅ /ŋ̊/ |
ṅ /ŋ/ |
||||||
|
/c/ |
ch /cʰ/ |
j /ɟ/ |
hñ /ɲ̊/ |
(ñ /ɲ/) |
y /j/ |
||||
|
ḍ /ɖ/ |
hḍ /D/ |
hr /r̥/ |
/r/ |
||||||
|
/s/ |
/t/ |
th /tʰ/ |
/d/ |
hn /n̥/ |
/n/ |
||||
|
/p/ |
ph /pʰ/ |
ḅ /ɓ/ |
/b/ |
hm /m̥/ |
/n/ |
v /w/ |
The unusual spelling of hḍ /D/ is difficult to interpret phonetically. I use the non-IPA symbol /D/ to avoid committing to a specific interpretation. The h of the ligature may be interpreted at face value as preaspiration. On the other hand, that h may merely be a graphic retention of the h of Old Pyu hl; it might not indicate that the phoneme was stil preaspirated or voiceless in Late Pyu.
The symbol /°/ may represent either a zero initial or a glottal stop. Areal tendencies would suggest that /°/ was a glottal stop, but there is no strong evidence for such an interpretation: e.g., words in which preinitials precede °.
The following gaps in the initial inventory may be accidental due to the brevity of the text: kh /kʰ/. hy /j̊/. hl /l̥/. hv /ʍ/. All four are represented elsewhere in the corpus: e.g., /kʰ/ and /j̊/ are in PYU 39 which mentions the date 1078 CE and therefore probably slightly predates the Kubyaukgyi.
Aspirates are of low frequency in Pyu (Shafer 1943: 348), so it would not be surprising if not all aspirates were represented in a short text.
Initial stops of words may lenite when in close juncture with a preceding word ending in a sonorant. See § 4.4.
4.3.2 Complex initials
These are all the possible combinations of preinitials and initials in the text.
I do not include rḍ, rb, and rv-ṃ because their superscript r represents vowel retroflexion rather than a preinitial /r./. See § 4.5.3.
Voiced stop phonemes lenited to fricatives after preinitials: /g d b/ > g-ṃ d-ṃ v-ṃ [ɣ ð v]. Since voiced stops also lenite to fricatives between sonorants (§ 4.4.1), their lenition after preinitials suggests that they were between sonorants in that environment: i.e., an unwritten, nonphonemic presyllabic vowel and the vowel of the main syllable. Lenition is more likely between vowels than between a consonant and a vowel.
In theory /ɟ/ would lenite to y-ṃ [ʝ] after preinitials. The lack of examples in the Kubyaukgyi may reflect the brevity of the text and the low frequency of palatal stops in Pyu. kyaṃḥ in PYU 17.5 and 177 may be /k.ɟa(C)H/ with a /ɟ/ that lenited after a presyllable.
The near-total absence of voiceless unaspirated stop phonemes after preinitials in the Kubyaukgyi may either be an artifact of a small corpus or a clue to the phonemic structure of Pyu.
I initially tried to explain away the variation between kiṁ pa (B14) and kiṁ pha (A13) by proposing the nonphonemic aspiration of unaspirated stops preceded by /k./: [kʰC] /k.C/ (cf. Khmer /kC/ [kʰC]) or [xCʰ] /k.C/ (cf. Korean *kC- > [Cʰ] Vovin (2003: 85), possibly via *xCʰ-).10
But I would expect [kʰp] to be written with an aspirated ˟kh as in Khmer. Similarly, I would expect [xpʰ] to be written as ˟khph or ˟hph.11 Perhaps presyllabic vowels could be indicated with ˟iṁ: ˟khiṁ p ˟khiṁ ph ˟hiṁ ph. However, the actual spellings in the Kubyaukgyi are kiṁ p and kiṁ ph.
The parallels with Khmer and Korean break down when trying to explain the /k./-less variation between hñiṁ ci (A25) and hñiṁ chi (B27).12 Neither Khmer nor Korean have a vowel like iṁ /ï/ between the consonants in their respective synchronic and diachronic aspiration processes. I am unaware of any precedent of consonants aspirating between vowels. On the contrary, in English, aspirated initial stops lose their aspiration in noninitial position: e.g., compare typical [tʰɪpɪkl̩] with prefixed atypical [ʔejtɪpɪkl̩].
Table 3
Inventory of Pyu complex initials
|
/k./ |
/t./ |
/n./ |
/p./ |
/m./ |
/r./ |
/s./ |
|
|---|---|---|---|---|---|---|---|
|
/g/ |
sg-ṃ /s.g/ |
||||||
|
/cʰ/ |
kc |
||||||
|
/ɖ/ |
kḍ /k.ɖ/ |
pḍ /p.ɖ/ |
|||||
|
/tʰ/ |
mt /m.tʰ/ |
st /s.tʰ/ |
|||||
|
/d/ |
kd-ṃ /k.d/ |
td-ṃ /t.d/ |
pd-ṃ /p.d/ |
md-ṃ /m.d/ |
|||
|
/n/ |
/k.n/ |
/s.n/ |
|||||
|
/pʰ/ |
kiṁp |
||||||
|
/ɓ/ |
tḅ /t.ɓ/ |
||||||
|
/v/ |
tv-ṃ /t.b/ |
||||||
|
/m/ |
tm, thm /t.m/ |
/r.m/ |
|||||
|
/j/ |
ky /k.j/ |
||||||
|
/r/ |
/k.r/ |
/t.r/ |
/p.r/ |
/m.r/ |
/s.r/ |
||
|
/l/ |
/t.l/ |
/p.l/ |
|||||
|
/w/ |
tv /t.w/ |
||||||
|
/h/ |
/n.h/ |
/m.h/ |
kch /k.cʰ/
kiṁph /k.pʰ/
The borrowing of Sanskrit sthāpana ‘establishment’ as stabana (A19) instead of ˟sthabana suggests that Pyu may have had a similar process of noninitial deaspiration: /s.tʰ/ was pronounced as [s(ə)t].
So as odd as the absence of voiceless stop phonemes is, that must be balanced with the oddity of proposing a rule of aspiration in intervocalic position.
I predict that voiceless stop initials voice between preinitials and vowels just as they do between sonorants elsewhere (§ 4.4.1) since at least some preinitials must have been followed by nonphonemic neutral vowels: e.g., /C.tV/ > [CədV]. There are Old Pyu spellings that could be interpreted as examples of that proposed phenomenon: e.g. kgoy· /k.koj/ [kəgoj]? ‘?’ (20.8), pjauy·13 /p.coj/ [pəɟoj]? ‘?’ (20.2), tdit·ṁ /t.tït/ [tədɨt]? ‘?’ (16.2), and pban·ḥ /p.panH/ [pəbanH]? ‘?’ (20.3). However, the degree to which lenition occurred in Old Pyu is still unclear. Perhaps the above sesquisyllables had unlenited voiced initials /g ɟ d b/.
The following alternation in spelling may indicate that written disyllables beginning with Ciṁ- may have been /C.CV(C)/ sesquisyllables: kiṁpha (A13–14) 
kiṁpa (B14) for /k.pʰa(C)/ ‘to be pleased’
Old Mon has both monosyllabic and disyllabic spellings of sesquisyllables: e.g., pdar 
pudar for /p.dɔr/ ‘to shade’ (Shorto 1971: xix, 246). If the Pyu corpus were larger, it might contain monosyllabic spellings like †kpha and/or †kpa for ‘to be pleased’.
Sequences like thm may have been pronounced as an aspirated stop followed by a voiced sonorant [tʰm] as in Khmer, as an unaspirated stop followed by a voiceless sonorant [t(ə)m̥] with or without an intervening neutral vowel, and/or as an aspirated stop followed by a voiceless sonorant [tʰm̥].
In any case, the following pair of words presumably sharing a common root may demonstrate that neither aspiration nor voicing was phonemic in combinations of preinitial stops with sonorant initials: . thmuḥ /t.mu(C)H/ ‘presence (?)’ (A11) 
tmu /t.mu(C)/ ‘presence (?)’ (A26).
The loss of distinctive voicing and aspiration in certain positions may be a Late Pyu innovation. Not enough is known about earlier Pyu spelling variations and semantics to determine if, for instance, Old Pyu thṅam·ḥ ‘?’ (20.4) and tṅam·ḥ ‘?’ (25.3) represent the same phonemic form /t.ŋamH/ or are a minimal pair /t.ŋ̊amH/ and /t.ŋamH/.
4.4 Intervocalic syllable-initial consonants
The following table excludes intervocalic consonants within sesquisyllables: e.g., the /d/ of tduṃ /t.du(C)/ [təðu(C)] ‘water’.
The table also excludes clusters of codas followed by initials or preinitials: e.g., /t t.j/ in tribhuvaṃnadit·ṃṁtya /t.ri bu ba na dït t.ja/ ‘Tribhuvanāditya’ (B2). I am unable to list all such clusters because codas were unwritten in most of the text.
I regard rḍ and rb as single consonants and not as /r./-consonant sequences since their superscript r represents vowel retroflexion rather than a preinitial /r./.14 See § 4.5.3.
Table 4
Inventory of Pyu intervocalic syllable-initial consonants
|
/h/ |
g /k/ |
g-ṃ, gh /g/ |
|||||
|
j, y-ṃ /ɟ/ |
ñ /ɲ/ |
y /j/ |
|||||
|
ḍ, rḍ /ɖ/ |
/r/ |
||||||
|
/s/ |
ṭh, th /tʰ/ |
d /t/ |
dh, d-ṃ /d/ |
/n/ |
|||
|
ḅ /ɓ/ |
b, rb /p/ |
bh, v-ṃ, rv-ṃ /b/ |
hm /m̥/ |
/m/ |
v /w/ |
4.4.1 Lenition of voiceless stops
Initial voiceless stops /k t p/ lenite to voiced g d b [g d b] between sonorants: /t.ri ɖo ka va ta(N) sa ka/15 > triḍogavaṃdasaga (A3; from Sanskrit Trilokavaṭaṁsakā), /si ri ta tʰa ga ta/ > siri ‖ dathagaṃda (A1/B1; from Pali siri + Sanskrit/Pali tathāgata), /s.tʰa pa na/ > stabana (A19; from Sanskrit sthāpana). I would expect the palatal voiceless stop /c/ to similarly lenite to j voiced [ɟ] in intervocalic position between sonorants, but there are no clear examples: e.g., the j of mra.ja.hṅa ‘other person’ (A24/B26) might have been either a lenited /c/ or a nonleniting /ɟ/ after an unwritten voiceless stop /C̥/.
As the voicing of initial /t/ after /si ri/ demonstrates, lenition of voiceless stops may also occur in word-initial position in close juncture.16
4.4.2 Nonlenition of voiceless aspirated stops
Voiceless aspirated stops do not lenite between sonorants: e.g., /ta tʰa ga ta/ becomes [datʰaɣada] after a vowel, not †[dadaɣada]. This is parallel to the behavior of Korean voiceless aspirated stops: contrast Korean 可動 /katoŋ/ [kadoŋ] ‘mobile’ with lenition of unaspirated /t/ and 可痛 /katʰoŋ/ [katʰoŋ] ‘lamentable’ without lenition of unaspirated /tʰ/. This is not parallel to the behavior of modern Burmese voiceless aspirated stops which do lenite between vowels: e.g., /jù/ ‘take’ + /kʰɛ̰/ pst + /
The following alteration suggests that voiceless aspirated stops may optionally lose their aspiration between sonorants:
hñiṁ.chi (A25) 
hñiṁ.ci (B27) for /ɲ̊ï.cʰi(C)/ ‘ill treatment’
Although such deaspiration may also occur after preinitials (§ 4.3.2), /ɲ̊./ would be an anomalous preinitial, as no other preinitial is palatal or a voiceless sonorant.
Voiceless /s/ also does not lenite between sonorants. This also has a parallel in Korean whose /s/ also does not lenite in that position. It would be inconsistent for /s/ to have an allophone [z] that was not indicated in the script given how all other lenited allophones were indicated. Once again, this does not have a parallel in modern Burmese whose /s/ does lenite between vowels: e.g., /bəmà/ ‘Burma’ + /səgá/ ‘speech’ =
4.4.3 Lenition of voiced stops
Voiced stops /g ɟ d b/ lenite to fricatives g-ṃ y-ṃ d-ṃ v-ṃ [ɣ ʝ ð v] between sonorants: /ta tʰa ga ta/ > dathagaṃda (A1/B1; from Sanskrit/Pali tathāgata), /°a ri me de ɟa/ > °arimedeyaṃ (A26/B28; from Pali Ariyametteyya with yy hardened to [ɟɟ] (Masica 1991: 169), /t.ri ɖo ka ba ta(N) sa ka de bi/ > triḍogavaṃdasaga deṃviṃ (A4/B4; from Sanskrit Trilokavaṭaṁsakā + Sanskrit/Pali devī ‘queen’). This lenition has no parallel in modern Burmese whose voiced obstruents do not weaken in intervocalic position. Modern Burmese [ð] is a lenition of voiceless /
Intervocalic Indic v often corresponds to v-ṃ in Pyu borrowings. I analyze all these instances as /b/ even though they are etymologically from v.18 Might the Pyu have been borrowing words from Indic speakers who had already shifted v to b by the early centuries CE? Although Masica (1991: 202) includes the shift in a list of New Indo-Aryan innovations, Whitney (1896: 18) says that “[f]rom an early period in the history of the [Sanskrit] language, but increasingly later, b and v exchange with one another, or fail to be distinguished in the manuscripts.” I symbolize the phonetic value of v-ṃ as [v] following Indic romanization, though the actual value may have been [
4.4.4 Lenition of voiced stops written as voiced aspirates
Orthographic voiced aspirate stops may have represented unaspirated voiced stop phonemes that also lenited as fricatives: e.g., /saŋ ga/ from Sanskrit and Pali saṅgha appears as both sagha and sagaṃ on the same line (A17/B17). If gh is an etymological spelling disguising an unaspirated /g/, then by analogy perhaps other etymological spellings with voiced aspirate stop characters also disguise unaspirated voiced stop phonemes: e.g., ḅudha /ɓu da/ with /d/ leniting to [ð] (A10/B10; from Sanskrit/Pali Buddha), and tribhuvaṃnadit·ṃṁtya /t.ri bu ba na dït t.ja/ with /b/ leniting to [v] (B2; from Sanskrit Tribhuvanāditya). If bh represents /b/, then the spelling tribhuvaṃnadit·ṃṁtya contains two different spellings of a lenited /b/: etymological bh and nonetymological v-ṃ with a subscript dot absent in Indic spellings of /v/.
There are no examples of the rare Indic consonant jh in the Pyu corpus.
4.4.5 The lenition chain
The aforementioned changes that occur to stops between sonorants19 (§ 4.3.2, § 4.4.1, § 4.4.3, § 4.4.4) can be summed up in terms of a chain: voiceless aspirate stop > voiceless unaspirated stop > voiced stop > voiced fricative. Perhaps voiced stops were the first to change, leaving a gap to be filled by voiceless unaspirated stops. Voiceless aspirated stops were beginning to fill the gap left by voiceless unaspirated stops. If Pyu had continued to exist, voiceless aspirated stops might have ceased to exist between sonorants.
4.5 Vowels
There are seven vowel phonemes in the text: /a ä i ï u e o/.
4.5.1 Absence of vowel length
There is no phonemic vowel length. Indic loanwords do not even have orthographic long vowels in the text: e.g., Sanskrit/Pali Rājakumāra corresponds to Pyu rajaguma.
There is only one definite instance of the vowel symbol ī in the text (hḍī, B20); the vowel symbol in the corresponding word in A (hḍ[ī], A20) is less certain. I follow Griffiths et al. (2018) which reads rmi for mī in Blagden (1911).
It is unlikely that the orthographic long vowel ī in hḍī represented a long vowel phoneme /iː/ because the Pyu consistently borrowed the Indic long vowel ī as an orthographic short vowel i: e.g., Sanskrit/Pali devī corresponds to Pyu deṃviṃ.
Moreover, ī is rare in the Pyu corpus as a whole. It appears in hypercorrect spellings of Sanskrit siddham ‘success’ with an unnecessary long ī and in a handful of non-Indic akṣaras: three instances of vīṃṁ (27.4, 27.5, 27.7), and one each of tīṁ (26.3) and nīl· (27.2). If nīl· is a borrowing of Sanskrit/Pali nīla ‘dark blue’,20 it would be the only Indic loanword in which original long ī was retained in the Pyu spelling. In any case, a true /iː/ would have had a broader distribution.
The function of the vowel symbol ī is unknown; it may have been an orthographic device to distinguish homophones or a ‘Indicism’: an Indic spelling feature imposed on non-Indic words.
Although most previous scholars have distinguished between u and ū in the text, I follow Griffiths et al. (2017: 79–80) who interpret what was previously read as ū as an allograph of u instead of a distinct grapheme. There are up to twenty-seven potential instances of a true ū in the Pyu corpus, but some may be questionable,21 over two-thirds (20/27) are concentrated in three texts,22 and none are in the Kubyaukkgyi. There is only a single Indic loanword in the corpus containing an ū corresponding to an Indic ū: pūja ‘sacrifice’ (20.4). As with ī, the limited distribution of ū suggests that it did not represent a distinct long vowel phoneme.
4.5.2 Phonetics of vowels with Indic parallels
The identification of /a i u e o/ on the basis of the spellings of Indic loanwords with those vowels is generally straightforward, though the precise phonetics are elusive: e.g., there is no way to tell if Pyu /o/ was [o] or [ɔ], etc.
One complication is the apparent fronting of /a/ before /ɲ/: /sa(r) wa ɲu ta ɲa(n)/ > saveñudeña (A23/B24; from Sanskrit sarva- + Pali -ññŭ̄- + Pali -ññāṇa), /p.ra ɲa/ > breña (A23/B24; from Sanskrit prajñā). There is only one instance of a before ñ in the corpus: the name sri bañaṇa /s.ri pa(C) ɲa na/ (74.3) from the Sanskrit honorific śrī + Pali paññāṇa ‘wisdom’. This spelling may (a) be phonemic, (b) be influenced by Indic, (c) predate /a/-fronting, (d) be in a dialect without /a/-fronting, and/or (e) have an unwritten coda after /pa/ that blocked /a/-fronting.
Although /o/ is written as both o and au, this allography does not reflect allophony. See § 4.5.6.
4.5.3 Vowels without Indic parallels
There are at least two Pyu vowels without Indic parallels, namely /ä/ is a low front vowel spelled aṁ, and . /ï/ is a nonfront, nonlow vowel spelled iṁ. Beckwith (2002a) was the first to identify aṁ as a low front vowel. See Griffiths et al. (2017: 146–147) for further reasoning behind interpreting aṁ and iṁ as oral vowels rather than as nasalized vowels, vowel-nasal sequences, or vowels with tone marks. I use non-IPA symbols to avoid committing to specific phonetic values. /ä/ could have been [ɛ] 
[æ] 
[a], and /ï/ could have been [ə] 
[ɨ] 
[ɤ] 
[ɯ].
The Old Pyu spellings paṁ[ḥ] (16.5A) and [peḥ] (16.4C) ‘to give’ suggest that /ä/ may have been confused with /e/ as early as the sixth century CE in the Sriksetra dialect. It is improbable that such confusion could have persisted for centuries; it is likely that the two had merged into a single phoneme /e/ by the twelfth century CE. The aṁ 
e alternation in the cognate verbs saṁḥ ‘to do’ (A20/B20) and se ‘to make’ (A10/B10) is suggestive of such a merger. Nonetheless I will phonemicize all instances of aṁ in the Kubyaukgyi as /ä/ because I am not certain such a merger had occurred in the Pagan dialect of Late Pyu.
4.5.4 Retroflex allophones of vowels
Unusual spellings of loanwords with rCV-akṣaras corresponding to foreign /CVr/ suggest that /r/ may have conditioned the allophonic retroflexion of a preceding vowel: rimadham·narbu /ri ma(t) da(m) na pu(r)/ [ri mat dam na puʳr] for Arimaddanapura (B2), rvaṃra-badiṃṁ /ba ra pa(n) dï(t)/ [vaʳ ra ban ðït] for varapaṇḍita (B17), samanarḍoḥ /sa(k) ma na ɖo(r)H/ [sak ma no ɖoʳ(r)H] for Sakmunalor (A20/B20).
4.5.5 Retroflex vowel phonemes?
It is possible to interpret all instances of written superscript r plus vowel as retroflex vowel phonemes: e.g., rmi (A2) 
miṅ· (B2) 
rmiṅ· (B3) ‘to be named’ might have been /miʳŋ/ rather than /r.miŋ/. The spelling without r would reflect the absence of a preinitial /r./. *rCV could have shifted to /CVʳ/ in Late Pyu as in Tangut (Miyake 2012: 248). The use of superscript r in an Indic script to represent retroflexion in vowels has a roughly contemporary parallel in the Tibetan transcription of Tangut (Chung-pui 2008: 147).
But it is also possible that miṅ· is simply an isolated error for rmiṅ· rather than evidence for the loss of an original preinitial.23 Hence I prefer to continue to analyze Pyu rCV sequences as /r.CV/ if they do not corresponding to foreign /CVr/ sequences.
I could analyze Pyu rCV sequences that do correspond to foreign /CVr/ sequences as /CVʳ/ in word-final position.24 However, at this point I will be conservative and reconstruct an unwritten word-final /(r)/. If a final /r/ conditioned vowel retroflexion, there must have been a /CVr/ [CVʳr]-stage at some point, regardless of whether /r/ was later lost or not. So it is safest to reconstruct /r/, even if it was already gone by the early 12th century.
4.5.6 Interpreting diphthongs in the Pyu script
There are two Indic vowel symbols that do not stand for unique Pyu vowel phonemes: ai and au. ai appears in only two words as an alternative spelling of /aj/: rabai (A20/B21), a foreign toponym spelled with āy· in OB and OM, and chai (A22/B23) ‘to pour’, also spelled cha (A18/B18) with an unwritten /j/. One might expect au to be an alternative spelling of /aw/, just as ai is an alternative spelling of /aj/. However, there are no OB or OM words ending in /aw/ corresponding to Pyu au. Moreover, there are no Pyu words with au alternating with a or av· spellings.
The actual function of au is not parallel to that of ai. au is an allograph of o used atop U-shaped consonant symbols: p m s. Although there are no examples of au atop the U-shaped consonant symbol h in the Kubyaukgyi, the consonant-vowel combination hau can be found in other texts: 3, 20.3, 25.6 (2×), 27.6, and 179. o is used atop all other consonant symbols. This pattern of complimentary distribution holds throughout the corpus with only a few exceptions, viz. gauṃ (16.2A) instead of goṃ ‘cave’ (< Sanskrit/Pali guhā ‘id.’), pto ‘?’ (17.5) instead of †ptau, so ‘?’ (39.7) instead of †sau,25 and hoḥ ‘three’ (A6, A12, A21, B6, B21) instead of nhoḥ as in other lines of the Kubyaukgyi (A5, B5, B13) and in other texts (20.1, 73) or hauḥ as in 3.26
As the fricative phonemes /s h/ have nothing in common with the labial phonemes /p m/ besides being written with a similarly shaped character, I regard the use of au as a strictly graphic phenomenon without phonological or phonetic significance. In other words, I do not believe that au represented a phoneme distinct from /o/ or an allophone of /o/ after /p m s h/. See Griffiths et al. (2017: 80–81) for more on the graphemes au and o, including images of variant shapes of the latter.
4.6 Codas
From Blagden (1911) onward, scholars have almost unanimously rejected the reconstruction of codas for Pyu. Two exceptions are Diller (1996: 240–241) who interpreted the dots of the Pyu script as “a sort of shorthand” for a “system of eight final consonants” and Kato (2005) who interpreted some but not all instances of anusvāra as a glottal stop corresponding to a glottal stop in Karen. Neither Diller nor Katō regarded the subscript consonants of the Pyu script as symbols for codas. According to Bradley (1979: 537), Gutman (1976: 113) interpreted the subscript consonants as musical notation for a gourd reed-organ. Although Burmese scholars have long included the subscript consonants in their transcriptions and have come close to recognizing their true function, Griffiths et al. (2017: 84–88) is the first to make an explicit, detailed case for interpreting them as codas on epigraphic and etymological grounds. I transliterate them with a middle dot · in this study.
There are ten subscript codas attested in the corpus, but none appear in the A text, and only three appear in the first three lines of the B text: -ṅ· /ŋ/, -t· /t/, and -m· /m/. Codas not written in the Kubyaukgyi are in parentheses in the table.
There is no reason to believe that the presence or absence of written codas was determined by context. The words sni[ṅ]·ḥ ‘year’ and rmiṅ· ‘to be named’ that appear with a coda ṅ· /ŋ/ in B2 and B3 reappear in subsequent lines of B in similar contexts but without subscript consonants. It is hence not the case that the Kubyaukgyi was a lipogrammatic text composed in such a way that words with codas were avoided from the third line onward. The text almost certainly has a normal distribution of codas that is disguised by the orthography.
It is unlikely that /ŋ t m/ were the only codas in 12th century Pyu. There is one potential case of a coda written with a vowel symbol: /j/ in /aj/ written as ai in chai /cʰaj/ ‘to pour’ (A22/B23). Although ‘to pour’ is not found outside the Kubyaukgyi, the coda /j/ is written in other texts as a subscript consonant y·.
If a Pyu word in the text is attested with codas in other texts, I supply that coda in parentheses in phonological forms: e.g., tdaṃḥ ‘king’ (A7) appears outside the Kubyaukgyi as tdav·ṃḥ (16.2d), so I phonologize it as /t.da(w)H/ with an unwritten coda /(w)/.
If a Pyu syllable in the text corresponds to a closed syllable in another language, I supply a possible coda in parentheses in phonological forms: e.g., /ba ra pa(n) dï(t)/ for vaṃrabadiṃṁ (A17, B17) on the basis of Pali varapaṇḍito and OB and OM varapaṇḍit·.
Table 5
Inventory of Pyu maximal potential codas
|
(k· /k/) |
ṅ· /ŋ/ |
|
|
(y· /j/27) |
||
|
(r· /r/) |
||
|
t· /t/ |
n· /n/ |
(l· /l/) |
|
(p· /p/) |
(m· /m/) |
(v· /w/) |
The parentheses not only indicate the fact that the coda is unwritten but also indicate the possiblity that the coda may have become something else or been lost by the 12th century. Including such supplied codas raises the total number of codas in the text to ten: /(k) ŋ t (n) (p) m j (r) (l) (w)/. That inventory is identical to the inventory of codas in neighboring Old Mon minus the palatals /c ɲ/.
/r/-codas may have been lost in Late Pyu after conditioning vowel retroflexion. See § 4.5.4.
If a syllable with an unidentified possible coda is followed by a lenited initial, I write that coda as /(C̬)/ with a subscript voicing symbol. /(C̬)/ represents a voiced coda /(ŋ) (n) (m) j/ that might condition lenition. The three examples are: /sa(ŋ) ga si ba ra pa(n) dï(t)/ > sagaṃsi vaṃra-badiṃṁ ‘Saṅghasena best-pundit’ (A17), /ho(m)H dï/ > hoḥ diṃṁ ‘three acc’ (A21/B21), and /ɓä(j)H p.ra m̥a pa(l)/ > ḅaṁḥ vrahmaba ‘hon Brahmapāla’ (A16/B16). The last example is open to question since it is not certain that v represents a lenited /p/. Lenited p was normally written as b, not v which might have been a hypercorrect attempt at a learned spelling.
The absence of examples of lenition after /(r) (l) (w)/ is not necessarily evidence for their inability to trigger lenition in the initial consonants of following words.
/(r)/ is either in phrase-final position, before consonants that cannot be lenited (/ɓ °28/), or before a preinitial consonant (/t./) whose lenition is optional.
/(l)/ is only represented by a single word each in phrase-final position (vrahmaba /p.ra m̥a pa(l)/).
As for /(w)/, although /t.da(w)/ is a common word in the text, it may be in environments where lenition might not have been obligatory: e.g., before adjectives: tdaṃḥ to /t.dawH to/ ‘king righteous’ instead of †tdaṃḥ do.
Moreover, not all instances of lenition may have been indicated in writing. The possible borrowing of Pyu kḍeḥ.troḥ /k.ɖe(ŋ)H t.roH/ ‘favor’ as OB klaññjo ‘id.’ with j corresponding to Pyu /t.r/ after /ŋ/ may indicate that the preinitial onset /t/ was pronounced with voicing not reflected in the Pyu spelling. If /t/ was voiced in ‘favor’, it might also have been voiced after the final /(w)/29 ‘king’ in /t.dawH to/, lit. ‘king righteous’, in spite of its spelling t.
I do not reconstruct parenthetical codas if (1) a syllable in a word appears as an open syllable in texts with subscript consonants or (2) if a syllable corresponds to an open syllable in another language. There is no evidence to suggest that Pyu developed secondary codas like Maru30 or added nonetymological codas to loans like OM which required all stressed syllables to be closed.
If there is no evidence for or against a coda, I write /(C)/ to indicate the possibility of an unidentified coda. Many instances of /(C)/ probably correspond to zero, whereas specified codas in parentheses have a high probability of corresponding to an actual coda in the past if not in the 12th century CE.
4.7 Suprasegmentals
There are two written suprasegmentals: ḥ and Z. I represent both with capital letters in phonemic notation: /H Z/.
4.7.1 /H/: segment, phonation, or tone?
Diller (1996) expressed great skepticism toward the idea of tonal notation in Pyu and other Southeast Asian languages prior to tone-marking in the first Thai orthography of the 13th century.31 If one accepts his arguments, one could interpret Pyu ḥ as a segment /h/ as in Sanskrit, Old Mon, or Old Khmer. Sequences of sonorant subscript consonants with Pyu ḥ could be interpreted as representing voiceless sonorants: e.g., m·ḥ would represent a voiceless nasal /m̥/. However, voiceless sonorant codas are unusual though not without areal parallels in Kri (Enfield and Diffloth 2009: 15–26). Moreover, it is doubtful that ḥ remained a segment /h/ or that voiceless sonorants remained voiceless until the extinction of Pyu. Consonants after (C·)ḥ could lenite in the Kubyaukgyi: e.g., the accusative marker /dï/ appears as diṃṁ with lenited d-ṃ [ð] after ḥ: kḍeḥ.troḥ diṃṁ ‘favor acc’ (A7/B9–10), and hoḥ diṃṁ ‘three acc’ (A21/B21). ‘Three’ ended in a labial nasal, as indicated by spellings in other texts: e.g., nhom·ḥ32 (20.1). The lenited initial d-ṃ [ð] of diṃṁ after hoḥ ending in an unwritten nasal indicates that the (m·)ḥ coda of ‘three’ could not have been voiceless. There would be no phonetic motivation for voiceless consonants to voice after voiceless codas. It is more likely that an original final segment /h/ had conditioned phonation and/or a tone before being lost by Late Pyu: /Vh/ > /V/ + nonphonemic phonation/tone + /h/ > /V/ + phonemic phonation/tone. Similarly, it is more likely that voiceless sonorant codas (symbolized here as /C̥/) voiced after conditioning phonation and/or a tone: /VC̥/ > /VC̥/ + nonphonemic phonation/tone > /VC̬/ + phonemic phonation/tone. Lenition is more likely after voiced /V(C̬)/ than after voiceless /h/ or /C̥/. I could try to defend /h/ by having voiced stops lenite to fricatives to assimilate to a preceding /h/. But I could not come up with a similar defense for voicing after voiceless codas (/C̥/). I do not know of any language in which /VC̥C̥V/ is pronounced as [VC̥CV].
Hence I conclude that by the 12th century, h no longer represented a segment /h/, and that written combinations of subscript sonorant consonant symbols and ḥ represented voiced /C̬/ rather than voiceless /C̥/. In this view, the onset of the accusative marker /dï/ lenited after the /o/ of kḍeḥ.troḥ ‘favor’ and the /m/ of hoḥ ‘three’. But what did ḥ stand for if it no longer stood for /h/ or the voicelessness of sonorants? The evidence at present does not point unambiguously to either a phonation or a tone. Nor does it suggest any possible phonetic value for a phonation or a tone.
I choose to represent the unknown suprasegmental value of ḥ as /H/ with a non-IPA symbol to avoid committing to any specific phonetic value. Although I write this H after vowels and codas in phonemic and even phonetic forms, that is merely a written convention modelled after the transliteration of Pyu; vowels and codas were the actual final segments of syllables: e.g., the first syllable of kḍeḥ.troḥ /k.ɖe(ŋ)H t.roH/ ‘favor’ probably ended in /ŋ/, and the second syllable ended in /o/. I am not comfortable with this practice and would not object if others arbitrarily replaced H with an accent or some other nonletter symbol when referring to Pyu phonemic and phonetic forms.
4.7.2 /Z/
The most enigmatic symbol in the Pyu text may be Z which is found nowhere else in the corpus. Although Blagden (1911: 383, 385) observed its presence in his pioneering work on the Pyu language, he was unable to “attach any definite meaning” to it and regarded it as “unexplained”. Shafer (1943) was the first to attempt to explain it:33
One of these is a slightly upward curve over certain letters. It occurs over la (Blagden’s da [and ḍa in this study]) (A, and perhaps B, 3) and over the third la [= ḍa in this study] of a series of four (A 24; Face B damaged). In all other occurrences it is found on the last word before the punctuation marks ‖: toḥ (AB 5, 11), choḥ (AB 12, A 26-B 29), pa (A 23; Face B damaged above the letter), except in AB 13, where it occurs over both the words paṁḥ che which are the last words before the punctuation marks. In view of its occurrence over words already provided with the vowel o and possibly e, it is improbable that it denotes a vowel. From its position just before the punctuation marks in all cases except where it occurs over la [= ḍa in this study], which is a conjunction, we may conclude that it is either some additional mark of punctuation or that it was not an inherent tone but one induced by the sentence structure, perhaps something like the English falling tone or the French rising tone at the end of a sentence. Although the writer is not able to define it more precisely, there seems to have been no uncertainty in the mind of the author or of the stone cutters as to its use, as there was in the case of ‖ and |, as A and B faces agree in its use where the latter has not been damaged.
Nearly seven decades later, Krech (2012) was the first to include Z in a romanized transliteration of the text. He agreed with Shafer that “the positions of its occurrences suggest that this sign represents prosodic or intonational features” (p. 153).
I can add little to Shafer and Krech’s remarks beyond noting the presence of Z atop two more akṣaras preceding ‖: roḥZ in B8 and maZ in B19.
Without discovering any further instances of Z, it will be impossible to say much more about its possible functions, and its phonetic value is likely to remain forever lost unless a Pyu text on intonation is discovered.
4.8 Syllabic division
I assume that each written akṣara represented a spoken syllable. Without metrical evidence, there is no way to determine if any written vowels or even whole akṣaras were silent.
I use spaces to divide syllables in phonological forms. These spaces do not necessarily coincide with morphemic boundaries.
There is a strong correlation between syllables and morphemes as would be expected from a Trans-Himalayan language. However, there are also polysyllabic native expressions which may or may not be divisible into syllabic morphemes if their structure were better understood.
Supplementary materials
An annotation transcription of the inscriptions on pillar B together with a comprehensive comparative glossary is provided in the supplementary materials. See
Abbreviations and conventions
Languages
| OB |
Old Burmese in the modified Indological transliteration for Pyu from Griffiths et al. (2017) |
| OC |
Old Chinese in the reconstruction of Baxter and Sagart (2014b). |
| OM |
Old Mon in the modified Indological transliteration for Pyu from Griffiths et al. (2017) |
| OT |
Old Tibetan in the Indological system recommended by Hahn (1996) |
| WB |
Written Burmese in the modified Indological transliteration for Pyu from Griffiths et al. (2017) |
| WT |
Written Tibetan in the Indological system recommended by Hahn (1996) |
Glossing abbreviations not in Leipzig conventions
| aor |
aorist |
| attr |
attributive |
| emph |
emphatic |
| fin |
finite |
| hon |
honorific |
| hum |
humilific |
| ln |
locative noun |
| pn |
personal or place name |
| ppp |
past passive participle |
| rls |
realis |
| sbo |
subordinator |
| tmp |
temporal |
| xcm |
exclamatory marker |
Textual conventions
| . |
possible morphemic boundary in glosses; do not confuse with /./ for potential presyllabic vowels in phonological forms |
| [ ] |
uncertain reading |
| ( ) |
editorial restoration of lost text; see below for parentheses in phonemic forms |
| ⟪ ⟫ |
scribal insertion |
| { } |
scribal deletion |
| ? |
illegible akṣara |
| C |
illegible consonant element of an akṣara |
| + |
lost akṣara |
| ◊ |
punctuation space |
| Z |
intonation marker |
Unattested forms are each preceded by one of three symbols
| * |
reconstructed |
| † |
hypothetical and presumably correct |
| ˟ |
hypothetical and (presumably) incorrect |
Non-IPA phonemic symbols
| /C/ |
any consonant |
| /C̥/ |
voiceless consonant |
| /(C)/ |
optional consonant; in coda position indicates an unknown final consonant that may have been unwritten |
| /H/ |
the unknown suprasegmental (phonation or tone?) corresponding to ḥ in Pyu orthography |
| /V/ |
any vowel |
In the course of his duties as a postdoc on the project ‘Beyond Boundaries: Religion, Region, Language and the State’ (ERC Synergy Project 609,823 ASIA, 2014–2020), funded by a grant from the European Research Council, Marc Miyake wrote this article in 2018–2019. It was accepted subject to minor revisions in 2019, but Marc has not subsequently had the opportunity to finalize it for publication. Lest this major study fail to see the light of day, the final revisions were in the end overtaken by Nathan Hill, Arlo Griffiths, and Julian K. Wheatley. Remarks by the editors are placed in footnotes and indicated with [Eds.]. In addition to due acknowledged owed the ERC, the editors would also like to acknowledge the support of The Robert H.N. Ho Family Foundation for the project ‘From Vijayapurī to Śrīkṣetra? The Beginnings of Buddhist Exchange across the Bay of Bengal as Witnessed by Inscriptions from Andhra Pradesh and Myanmar’, which supported the work on Pyu of Arlo Griffiths and Julien Wheatley. We would also like to thank Mathias Jenny for carefully reviewing the treatment of the inscription’s Mon face and for pointing out other mistakes, and to thank Aleix Ruiz-Falqués for similarly improving the treatment of Pali.
The name [gu˩bjawʔtɕi˥] has been romanized in a variety of ways: e.g., as ‘Kubyaukkyi’ (Blagden 1909 and Duroiselle 1919a), ‘Kubyauk-gyi’ (Luce and Shin 1969–1970), etc. Although my own preference is for ‘Gubyaukkyi’, a compromise between a phonetic transcription and a graphemic transliteration, I use ‘Kubyaukgyi’ for consistency with the most recent works on Pyu (Griffiths et al. 2017). That spelling is not very different from the K-spellings in the earlier literature. Switching to a G-spelling might confuse nonspecialists.
Jenny and McCormick (2014: 549) read as a single word raṁpo°· ‘portion’ what Blagden regarded as two words, raṁ ‘to help’ and po°· ‘for’. Although raṁ is attested by itself, po°· “is still not quite satisfactorily explained, inasmuch as it has not been found elsewhere or correlated with a modern form” (Blagden 1919a: 54). Shorto (1971) has no entry for po°·.
Blagden (1913–1914) covers four Pyu urn inscriptions (PYU 3–6).
Shafer (1943: 318) has no special term for Z which he described as “a slightly upward curve over certain letters.”
Naturally, verse unit is pāda, not pada. Note that sometimes in Pali the word amataṃ padaṃ “the deathless place” is used unequivocally as a synonym of nibbāna [Eds.].
The occurrence of -ika- for -aka- is perhaps intended to make the word appear more obviously feminine, cf. m. upāsaka ‘lay follower’ vs. upāsikā ‘female lay follower’, or indeed perhaps -ika- in Pali for Skt. -aka- is general. Cf. Pa. Anāthapiṇḍika vs. Skt. Anāthapiṇḍada. [Eds.]
The editors wonder whether there is scope for the possibility that ḍ represented an implosive as in Old Mon [Eds.]
In Khmer, any preinitial stop conditions the aspiration of a following stop, whereas in Korean, it is only preinitial *k that conditioned the aspiration of a following stop.
˟kh and ˟h would be the closest available characters for [x] since the Pyu script has no character ˟x.
I could claim that *k.ɲ- became hñ. That might explain the absence of ˟kñ in the corpus. However, I would then have to explain why the corpus contains /k./ before other sonorants: e.g., why does km still exist? It would be odd if *k.- devoiced *ɲ- but not other nasals. The lack of ˟kñ is more likely to be a random gap in the initial inventory or the corpus since palatals are not common Pyu initials.
Ideally I would have wanted to provide an example with cj to parallel the others, but no such akṣara exists. I doubt Pyu had a preinitial /c./ except in marginal words. There are only two examples of cC in the corpus.
The editors are unsure of Miyake’s reasons for positing vowel retroflection [Eds.].
/(N)/ represents a possible nasal coda with an unknown point of articulation.
Another possibility is that siri dathagaṃda was preceived by the Pyu as a single syntagma [Eds.].
Note that this Burmese lenition does not occur following schwa [Eds.].
Miyake’s analysis here would have benefited from greater explanation [Eds.]
Including unwritten vowels after preinitials.
This would be an anomalous borrowing since the Pyu typically retained the final a of Indic loanwords.
For instance, an akṣara in 20.1 may either be pū or °a. In that particular case, the °a-reading results in a plausible Indic loan °a sa ma from Sanskrit/Pali asama ‘incomparable’, but in other instances the context is too unclear to point toward one possibility or another.
Although two of those three texts (20 and 32) are long, a potential ū only appears once in two other long texts (25 and 27), and does not appear at all in other long texts such as 16. This suggests that ū may be a trait of a particular scribal tradition. The matter requires more investigation.
I assume that the sixth-century spelling rmiṅ· (16.2A) reflects an actual preinitial /r./.
It would be bizarre to analyze nonfinal rvaṃra- as /baʳa/ with two adjacent vowels. Such vowel sequences with or without retroflexion are unknown in Pyu. I would prefer the phonemicization /bara/ with secondary retroflexion of the first /a/ even if Pyu had retroflex vowel phonemes.
This akṣara is difficult to read and may be something else.
The n in the nh-ligature is tiny and could have been accidentally omitted. If so, the choice of vowel allograph reflects that missing n.
/j/ is not in parentheses because the coda /j/ is in the Kubyaukgyi, though it is written with a vowel symbol ai for /aj/ instead of a subscript consonant y·.
/°/ may be a zero initial rather than a consonant. In either case, it cannot lenite.
Although I write the suprasegmental /H/ after /(w)/, the actual final segment of ‘king’ is /(w)/, as /H/ is not a segment.
See Burling (1967: 59–61) on how Maru developed final stops in *open syllables.
Diller’s arguments concerning Pyu specifically are aimed against Luce’s interpretation of ṁ ṃ and the final stops as tone markers. To some extent he anticipated the correct interpretation of the final stops. [Eds.]
The omission of the preinitial /n./ in A21/B21 may be accidental. The n of the ligature nh is small and was overlooked until Griffiths et al. (2017).
I have converted Shafer’s notation into the Corpus of Pyu Inscriptions system to facilitate the identification of akṣaras with Z.
References
Aung-Thwin, Michael. 2005. The Mists of Rāmañña: the legend that was Lower Burma. University of Hawai‘i Press, Honolulu.
Baxter, William H. and Laurent Sagart. 2014b. Baxter-Sagart Old Chinese reconstruction, version 1.1. Online document.
Beckwith, Christopher I. 2002a. A glossary of Pyu. In Christopher I. Beckwith, editor, Medieval Tibeto-Burman languages: proceedings of the 9th Seminar of the International Association of Tibetan Studies, Leiden 2000, Brill’s Tibetan studies library, 159–161. Brill, Leiden
Beylié, Louis de. 1907. Prome et Samara: voyage archéologique en Birmanie et en Mésopotamie. Publications de la Société française des fouilles archéologiques. Ernest Leroux, Paris.
Blagden, C.O. 1909. The Talaing inscription of the Myazedi Pagoda at Pagan, with a few remarks on the other versions. Journal of the Royal Asiatic Society of Great Britain and Ireland, 1017–1052.
Blagden, C.O. 1910a. A further note on the inscriptions of the Myazedi pagoda, Pagan, and other inscriptions throwing light on them. Journal of the Royal Asiatic Society of Great Britain and Ireland, 797–812.
Blagden, C.O. 1910b. The early use of the Buddhist era in Burma. Journal of the Royal Asiatic Society of Great Britain and Ireland, 850–856.
Blagden, C.O. 1911. A preliminary study of the fourth text of the Myazedi inscriptions. Journal of the Royal Asiatic Society of Great Britain and Ireland, 365–388.
Blagden, C.O. April 1912. Two corrected readings in the Myazedi (Talaing) inscription. Journal of the Royal Asiatic Society of Great Britain and Ireland, 486–487.
Blagden, C.O. 1913–1914. The “Pyu” inscriptions. Epigraphia Indica, 127–132.
Blagden, C.O. October 1914. The Myazedi inscriptions. Journal of the Royal Asiatic Society of Great Britain and Ireland, 1063–1069.
Blagden, C.O. 1917. The “Pyu” inscriptions. Journal of the Burma Research Society, 7(1):37–44.
Blagden, C.O. 1919a. The Mon or Talaing face of the Myazedi inscription at Pagan. Epigraphia Birmanica, 1(1):53–58.
Blagden, C.O. 1919b. The Pyu face of the Myazedi inscription at Pagan. Epigraphia Birmanica, 1(1):59–68.
Blagden, C.O. and J.F. Fleet. 1910. The revised Buddhist era in Burma. Journal of the Royal Asiatic Society of Great Britain and Ireland, 474–481.
Blagden, C.O. and Taw Sein Ko. 1911. Early use of the Buddhist era in Burma. Journal of the Royal Asiatic Society of Great Britain and Ireland, 209–216.
Bradley, David. 1979. Speech through music: the Sino-Tibetan gourd reed-organ. Bulletin of the School of Oriental and African Studies, 42(3):535–540.
Burling, Robbins. 1967. Proto-Lolo-Burmese. Indiana University, Bloomington.
Chung-pui, Tai. 2008. Xīxiàwén Fójīng cánpiàn de Zàngwén duìyīn yánjiū. Dissertation, Chinese Academy of Social Sciences.
Cœdès, George. 1911. Review of C.O. Blagden: “A preliminary study of the fourth text of the Myazedi inscriptions”. Bulletin de l’Ecole française d’Extrême-Orient, 11(1):435–436.
Diller, Anthony V.N. 1996. Thai orthography and the history of marking tone. Oriens Extremus, 39(2):228–254.
Duroiselle, Charles. 1919a. The Burmese face of the Myazedi inscription at Pagan. Epigraphia Birmanica, 1(1):1–46.
Duroiselle, Charles. 1919b. (Transliteration of the Pali face of the Myazedi). Epigraphia Birmanica, 1(1):47–52.
Enfield, N.J. and Gérard Diffloth. 2009. Phonology and sketch grammar of Kri, a Vietic language of Laos. Cahiers de Linguistique—Asie Orientale, 38(1):3–69.
Forchhammer, Emil. 1892. Inscriptions of Pagan, Pinya and Ava, deciphered from the ink impressions found among the papers of the late Dr. E. Forchhammer. The Superintendent, Government Printing, Rangoon.
Griffiths, Arlo, Bob Hudson, Marc Miyake, and Julian Wheatley. 2017. Studies in Pyu epigraphy and Pyu language, I: state of the field, edition and analysis of the Kan Wet Khaung Mound inscription, and inventory of the corpus. Bulletin de l’Ecole française d’Extrême-Orient, 43–205.
Griffiths, Arlo, Marc Miyake, and Julian Wheatley. 2018. Corpus of Pyu inscriptions. Online resource. URL http://hisoma.huma-num.fr/exist/apps/pyu/works/.
Gutman, P.C. 1976. Ancient Arakan. Dissertation, Australian National University.
Hahn, Michael. 1996. Lehrbuch der klassischen tibetischen Schriftsprache. Indica et Tibetica Verlag, Swisttal-Odendorf.
Haudricourt, André-Georges. 1946. Restitution du karen commun. Bulletin de la Société de Linguistique de Paris, 103–111.
Jenny, Mathias. 2015. The far west of southeast asia: ‘give’ and ‘get’ in the languages of myanmar. In N.J. Enfield and Bernard Comrie, editors, Languages of mainland Southeast Asia: the state of the art, Pacific Linguistics 649, 155–208. Mouton de Gruyter, Berlin.
Jenny, Mathias and Patrick McCormick. 2014. Old Mon. In Mathias Jenny and Paul Sidwell, editors, The handbook of Austroasiatic languages, 519–552. Brill, Leiden and Boston.
Kato, Atsuhiko. 2005. Pyū-go to Karen-kei shogengo no ruijisei. URL http://www.sfs.osaka-u.ac.jp/user/burmese/Pyu_Karen.pdf.
Ko, Taw Sein. 1911. The linguistic affinities of the Pyu language. Journal of the Burma Research Society, 102–109.
Krech, Uwe. 2012. A preliminary reassessment of the Pyu faces of the Myazedi inscriptions at Pagan. In Nathan W. Hill, editor, Medieval Tibeto-Burman languages, volume 4 of Brill’s Tibetan studies library, 121–169. Brill, Leiden.
Luce, Gordon H. and Ba Shin. 1969–1970. Old Burma–early Pagán. Number 25 in Artibus Asiae, Supplementum. Published for Artibus Asiae and the Institute of Fine Arts, New York University [by] J.J. Augustin, Locust Valley, N.Y.
Luce, Gordon H. n.d. Notebook K, portfolio plate inscriptions. Title: G.H. Luce Papers. Notebooks J-L / G.H. Luce. Author: G.H. Luce, (Gordon Hannington), 1889–1979. Subjects: Inscriptions—Burma Language: Burmese Identifier: System number 015992609 Notes: Notebook J, Miscellaneous Unlisted Inscriptions, including transcriptions mainly from Pagan, the Mandalay Palace Shed and Sagaing, with an index of transcriptions in Portfolios I–V, giving site, date etc., pp. 2322–2699. Notebook K, Portfolio Plate Inscriptions, including transcriptions mainly from Pagan, with some extra loose folios, pp. 2700–3073. Notebook L, Listed Inscriptions, including transcriptions mainly from the Mandalay Palace Shed, pp. 3078–3263. Reproduction note: Microfilm. London: British Library, 26mm. Shelfmark(s): Asia, Pacific & Africa IOR.Neg.7207 UIN: BLL01015992609.
Luce, Gordon H. and Pe Maung Tin. 1934. Inscriptions of Burma. Oxford University Press, Oxford.
Masica, Colin. 1991. The Indo-Aryan languages. Cambridge University Press, Cambridge.
Miyake, Marc. 2012. Complexity from compression: a sketch of pre-Tangut. In Irina Popova, editor, Tanguty v central’noj Azii: sbornik statej v čest’ 80-letija prof. E.I. Kyčanova, 244–261. Oriental Literature, Moscow.
Miyake, Marc, James Miles and Nathan W. Hill. 2018. The use of Reflectance Transformation Imaging in the recording and analyses of Burmese Pyu inscriptions. Archaeological Research in Asia, 130–138.
Nishida, Tatsuo. March 1955. Myazedi hibun ni okeru chūko Biruma-go no kenkyū (Studies in the ancient Burmese language through the Myazedi inscriptions) (1). Palaeologia, IV(1):17–32.
Nishida, Tatsuo. March 1956. Myazedi hibun ni okeru chūko Biruma-go no kenkyū (Studies in the ancient Burmese language through the Myazedi inscriptions) (2). Palaeologia, V(1):22–40.
Nyein, Tun. 1899. Inscriptions of Pagan, Pinya and Ava: translation with notes. The Superintendent, Government Printing, Rangoon.
Nyeiin Maung, U. (ed.). 1972. Rheḥ hoṅḥ Mranmā kyokcā myāḥ pathama twai. Archaeological Directorate, Rangoon.
Sawada, Hideo. 2002–2006. Stone and ink inscriptions of Burmese script. Online document.
Shafer, Robert. 1943. Further analysis of the Pyu inscriptions. Harvard Journal of Asiatic Studies, 7(4):313–366.
Shorto, H.L. 1971. A dictionary of the Mon inscriptions from the sixth to the sixteenth centuries. Oxford University Press, London.
St. John, R.F. St. Andrew. 1914. The inscriptions of the Myazedi pagoda, Pagan, Burma. Journal of the Royal Society of Great Britain and Ireland, 1058–1063. URL http://www.jstor.org/stable/25189253.
Stargardt, Janice. 1990. The ancient Pyu of Burma, vol. 1: Early Pyu cities in a man-made landscape. PACSEA in association with the Institute of Southeast Asian Studies, Singapore, Cambridge.
Tha Myat, U. 1958. Mracetī Pyū krokcā. Pr. Democracy Publishing, Rangoon.
Tha Myat, U. 1963. Pyu reader. U Hla Din, Rangoon.
Tun, Than. 1958. A course on Pyu inscriptions. Burma Historical Commission, Rangoon.
Vovin, Alexander. 2003. Once again on lenition in Middle Korean. Korean Studies, 27:85–107. URL http://www.jstor.org/stable/23719571.
Whitney, William Dwight. 1896. A Sanskrit grammar. Breitkopf and Härtel, Leipzig.
Win, Sein. 2016. Pyū akkharā cā pe poṅḥ khyup. Department of Archaeology and National Museum, Ministry of Culture, Rangoon.
Yabu, Shirō. 2006. Old Burmese (OB) of Myazedi inscription in OB materials. Osaka University of Foreign Studies, Osaka.
