1The pages below result from a confrontation between evolving Burmese script and some of its dialectal forms collected in the area where they are still spoken as they are written. Local reactions to differences between regional and national languages were often helpful.
2From the very first Burmese inscriptions (12th c.) two letters foreshadow the modern tone markers. During the following centuries, their use and shape underwent changes, step by step. But only an hypothesis can be formulated about the moment when they first stood for tones or for something that has pitch as one of its characteristic features.
3Both letters existed in Nagari script, source of so many Southeast Asian alphabets, among which the Mon and Burmese ones. One is the vowel carrier, classically transliterated by a among the Indianists and in Epigraphia birmanica (1919); the other one is the visarga, final h. Both, placed after a syllable, are devowelized by the virama above them, as is every final consonant, which means that the vowel carrier was treated graphically as a consonant without, paradoxally, its inherent vowel. In traditional Burmese alphabets, as well as in The Myanmar-English Dictionary, (1993, p. viii), this letter remained one of the 33 consonants; that is to say, it is supposed, as the 32 others, to ‘carry’ a graphically invisible vowel a, unless a virama suppresses it, so that a realistic transliteration of the so called a ought to be, when initial:
4အ – ’a –
and when final, in early lithic inscriptions:
5အ် – –’
6In lithic inss, this final letter was eventually written under the last letter of the syllable. Final h or its abbreviation, i.e., two dots or small circles after the syllable, after having nearly disappeared, reappeared in its abbreviated form. At last the subscript ’ was reduced itself to a small circle. Both had then acquired their modern shape. They were usual in printed works in the 19th c., but their general use, including in manuscripts, became systematic only in the 20th c.
7It is a system subdivided into three parts; in each of them, when the vowel of the syllable implies a tone by its graphism, no specific tonal marker is added; in other words, in modern Burmese rules for marking tones are different according to the vowel of the syllable.
- 1 The transliteration of Epigraphia birmanica will be followed here except for tonal marks, as near a (...)
8In old Burmese the vowels were written with letters inherited from short vowels or long ones and diphtongs of Nagari: in modern writing, creaky tone is either implied by a letter inherited from a short vowel or marked by a small circle under the last letter of the syllable, low tone (usually so called, actually level, non high) is implied by a letter inherited from a long vowel or written as a diphtong, and breathy tone is either implied by another long vowel, an inherited diphtong or marked by two small circles after the syllable1:
- 2 Similar to the virama in shape but not in effect.
10[Note 2 dans l’image = « with a superscript mark 2 »]
11If a nasal consonant is written at the end of the syllable, it is in the low tone, which is left unmarked; if it is a stop consonant, unvoiced, unaspirated, it implies the glottal stop considered as a tone, anyhow it is pronounced as a glottal stop if the syllable is not in close juncture with a following one.
12One feature is common to the three parts of that system: one tone is always left unmarked. Since the tonal markers were used systematically only very late and since they are only graphisms, one cannot rely on them alone to deduce the time when distinctive tones appeared in Burmese. First of all tones belong to orality and the question of their notation is a subsidiary one, moreover, notation might mean only awareness of tones already existing in the language, as Anthony V. N. Diller notes in his “Thai Orthography and the History of Marking Tone” (1996, p. 232 sq.).
- 3 Thus in the first syllable of disyllabic words, if totally unstressed, a vocalic and tonal neutrali (...)
13Tones are syllabic, in central Burmese as well as in other Burmese dialects, in the respect that several of their distinctive features concern the syllable as a whole; tones confer the syllable more or less resistance against areal influences (phonetic influences); in the syllable itself place and nature of stress vary according to the tone: stronger at the beginning of the syllable or at its end, even throughout the syllable; length of its nuclear vowel is not contrastive by itself but varies with the tone; as for pitch, another distintive feature of the tone, it is influenced by the phonemes of the syllable and by the pitch of syllables in close contact with it, influences which might produce allotones. In the flow of speech, tones are more audible in stressed syllables, while weak may be atonic, and this, in turn, depends on the word they belong to: on its morphology—mono or polysyllabic3—, on its syntactic status—verb, noun, adverbial sequence—, on its function—main or auxiliary, lexical item or grammatical one—. R.K. Sprigg pointed out this problem in his “Burmese Orthography and the Tonal Classification of Burmese Lexical Items” (1964) and Jörgen Rischell (2002: 87) notes a discrepancy between the tonal marking of the interrogation /`la/ and its realization due in fact to its constant place at the end of the sentence. In isolated syllables, a very rare situation, if not an artificial one, tones are supposed to enjoy full independance. In fact these suprasegmental units are contrastive for syllables segmentally identical, but under various conditions, at least they are lexically important. Let us consider them as tonemes, for the moment, but let us call them “tones” because it is shorter and easier.
14It is generally admitted that there are four tonemes: creaky, low, breathy, and glottal stop; three of them occur on open syllables and on syllables with either a nasalized vowel nucleus, or a nasalizing effect on the initial of the following syllable; the fourth toneme implies the existence of an unreleased glottal stop at the end of the syllable. Classifying glottal stop as one of the suprasegmental units means that the final unreleased stop of the syllable is considered as one of the distinctive features of its toneme. But if we consider it as a contrastive element belonging to the phonemic structure of the syllable and not to its tonemic structure, there would be three tonemes only. As for syllables ending with a stop consonant, glottal and unreleased, though they seem to have most often the same pitch as syllables with creaky tone, at least according to the analysis of tone contour and table of toneme contrasts: 54~53, by Nitasakorn Shiwaruangrote (2000: 92, 96), it is questionable to classify them as syllables with creaky tone plus a segmental glottal stop.
15Concerning the number of tonemes, it might be interesting to quote A Burmese Phonetic Reader written in 1925. Its authors were an English phonetician, Lilias E. Armstrong, and a Burmese student, Pe Maung Tin, who became a famous historian, grammarian and linguist; the student pronounced words, phrases and tales, which were the basis of the study; hereunder the phonetic transcription of the authors has been followed:
p. 20-22 (of the reprint) showing the awareness of allotones “in Burmese there are four tonemes… Toneme I has four members… the principal… is high and has a slight fall, the second is mid, slightly falling, the third is low… slightly falling… syllables bearing members of toneme I may be divided in two classes: (a) Those in which the vowel is <either> pronounced with medium length, <or> pronounced with… intermittent (creaky) voice and terminated by a weak closure of the glottis … ‘pja’ (to show), ‘sau’ (to wait), (b) those in which the vowel is pronounced with strong stress, extremely short, terminated by an abrupt closure of the glottis… ‘lauʔ’ (sufficient), ‘tahIʔ’ (to love).”
16Then “tonemes II and III” are called the low and the breathy tones, respectively.
18In fact, from 1898 onwards several Burmese authors numbered three tones, Taw Sein Ko: 1898, Shwe Zan Aung: 1916, Pe Maung Tin: 1925, Minn Latt, who changed his mind from four: 1962, to three: 1966. In 1982, Thein Tun numbered four tones. Most of the English, Japanese and Thai linguists, from 1936 onwards, consider there are four tones, but David Bradley, 1982, remarks (p.120) “that the ‘killed’ type (glottal stop) occurs only with a final stop, glottal in isolation and homorganic to the initial consonant of the following syllable in close juncture” and (p. 121) adds “it is possible to regard the ‘killed’ syllables as non-contrastive (as far as suprasegmental units are concerned): it is the only possibility in syllables with a final stop”.
19To conclude these preliminaries, in the examples presented below, my phonetic notation is not the most generally used, though, for tonemes, it is close to John Okell's (1969, 1995): same accents, but placed by him on the vowel, while I place them before or after the syllable:
creaky: syllable followed by ´
low: syllable without any marker
breathy: ` before the syllable.
20Now we shall try to trace the appearance of those contrastive and puzzling elements in Burmese script. Their first appearance, their modifications, periods of absence, reappearance and at last systematic use, are summarized in the Myanmar-English Dictionary (1993). According to this dictionary, p. x, xi,
21– ဟ် final h (written with devowelizer) appeared in Burmese writing from the very beginning, disappeared at the beginning of 13th c., reappeared in the form of two dots, present marker of the breathy tone, in the 16th c.;
– အ် final a , with the devowelizer, was used, first, after syllables with vowel ေ e (I shall transcribe here this final အ် a as ’, apostrophe, in order to avoid confusion with the real vowel, the inherent a, also inherent in Burmese as well as in other scriptures inherited from Nagari); it was subsequently subscript, applied to syllables with a final nasal, then reduced to a small dot, and it disappeared in the 16th c. till it became, much later, the regular mark of creaky tone. (p.XIII) At the very beginning of the writing, there were different letters for the short and long vowels of nagari origin: a, i, u and ā, ī, ū.
22Are these statements documented by the inscriptions?
23The “Inscriptions of Burma”, portfolio I (1933) published by Pe Maung Tin and Luce, and their epigraphic studies (1960, 1969) show that, as soon as the beginning of twelth century, syllables written with the nuclear vowels, inherent a or i, u, belonged to three types, according to the length of the vowel and the presence or absence of final h, short a being not visible but inherent to the consonant:
Ca, Cā, Cāh
Ci, Cī, Cīh
Cu, Cū, Cūh
and among them several syllables with short vowel have now creaky tone, several with long vowel have now low tone and some ending with devowelized h are now breathy. As for syllables with nuclear vowel e or those with nasal final, they belonged, at least, to two types according to the presence or absence of final ’ or subscript ‚ :
Ce Ce ’
among such syllables, several of those marked by — ’ (အ် on the line) or —, (အ under the last letter of the syllable) have now creaky tone.
24There are also syllables with long vowels ā, ī, ū of the inscriptions and syllables with a nasal final corresponding to syllables with low tone now.
- 4 The second syllable is now used alone and is written kyū.
25For instance, in the Chitsagon terracota, discovered by Duroiselle in 1925, 1926, and published in 1969 by G.H. Luce with U Wun commentaries, one can find
kra klū4 “reed”, krā “lotus”, ṅā “I”, phuhrā: “lord”, kā: “image”,
mi “mother”, nī “red”
tu “copy, portray”, phlū “white”, lū “man”, phūh “bud” or “bulbous root”
26and also short vowels followed by ’ :
kra klū“reed” is also written kra ’ klū , chu ’ “reward”,
mhya ’ “as many as, all”.
27When the syllable had a final nasal: ṅ, n, m, this last consonant marked as devowelized could not be followed by h or ’ :
tan chip, modern taṁ chip, “seal”, but than ryak, modern than: lyak, “palm jaggery” or kwam, modern kwam: , “betel”.
28There are various final stop consonants: k, t, p:
rhok “lemon”, klit, modern kyit, “Job's tears”, yut mā, still same nowadays, “inferior”, tan chip, modern taṁ chip, “seal”.
29In that first attempt to write Burmese, on humble terra cotta votive tablets, several spellings might occur for the same word:
phlu, phlū, phlū: “white”, kra, krā “lotus”, kri, krī “big”.
30However if we compare:
ṅā “I” to ṅa ’ mi “my mother”
we find already there the distinction between ṅā and ṅā ’, with the morphological significance (possessive case) of the modern creaky tone. Constant writing of “fruit” (modern sī: ) with short i after a name of plant (its botanical name) is puzzling, as if “fruit”, in this position, were an auxiliary term…
chi phūh “Kaempferia, or Gastrochilus pandurata, edible root”
corresponding to modern chit phūh: according to G.H. Luce and U Wun, in this case, there would be an evolution from short vowel syllable to modern glottal stop syllable, while it is usually the contrary.
32Last remark, in:
kra ’ klū “reed”, chu ’ “reward”, mhya ’ “as many as, all”
’ might seem redundant after a short vowel, or is it there because short vowel of the Nagari was not felt as sufficient, in 12th c., to transcribe such Burmese syllables? In fact such a redundance remained usual in the following centuries.
33The royal, lithic ins. of Kubyaukgyi (1113) seems more elaborate ; use of the subscript stop (–‚) under nasal final consonants appears as a new contrastive mark:
min‚ “to order”, nhaṅ, “with”
and also final a (—’) after ei :
’aphei ’ “for”, modern ’aphui..
34In the above examples it corresponds to modern creaky tone marker, but its use is far from being systematic.
35Still more consistency is attained in Dhammayangyi (Ajawlat) inscription, dated 1125 or 1126. It is not necessary to insist on this well known document, let us only point out the regular use of the previous contrastive marks and also regular use of ’ subscript —‚ under final y or w allowing to differentiate consistently syllables with nuclear vowel or diphtongue (?) corresponding to modern e and ui:
niy‚ “day”, modern ne. versus nīy “to dwell”, modern ne
pheiw‚ “for”, modern phui. , teiw, morpheme of plural, modern tui. versus keiw objective morpheme, modern kui ,
while the use of short and long i allowed another unexpected differenciation:
rīy “water”, modern re versus riy “to write”, modern re: .
36Hesitations in the evolution of Burmese writing were so important, specially concerning the vowels, that a contrastive marking based on the nature of vowels could not remain untouched. But what about the theory: survivance of final devowelized h to the beginning of 13th c., use of two dots from 16th c. onwards, survivance of final or subscript devowelized a to 16th c.?
37During the 13th c.11 “a”, final of syllable —’ or subscript—, is surviving:
38In Mingalazedi ins., 1285, not a single —’ or —‚ no —h, no —: but in Monywa ins., 1343:
39na:kā maṅ “king of Nagas”, modern nāga maṅ: ,
40where neither the abbreviation of h (two dots) after the first syllable nor its absence after the last one are corresponding with the modern spelling.
41The above examples, borrowed to Professor Than Tun (1966), show that marking of something corresponding to the modern creaky tone was systematic only when it had a morphological value, when it translated a syntaxic relation, and that it was more frequent with basic and recurrent term such as “this” or “day”.
42The Chinese offices of translation, flourishing under the Ming Dynasty (1368-1644), were equipped with vocabularies and bilingual texts, manuscripts meant for teaching Burmese to Chinese translators. These texts were compiled and printed with wooden blocks at the end of 17th c. They were studied by Roy Andrew Miller, who attributed the original ms to the 16th c. (1954, p. 372) Mss and printed compilation were mentioned or studied several times, from Abel Rémusat (1826) to Tatsuo Nishida (1972). The Burmese is translated into Chinese, traced in Burmese letters, by a Chinese hand (?), and its pronunciation is rendered through Chinese pictograms, the old marks: devowellized ’a and h, or their substitutes, are totally absent from the Burmese writing, but final consonants of closed syllables correspond pretty well with their ancestors of the lithic inscriptions, short and long vowels are rather more consistent with modern creaky and low tones than in the inss, and most of the words are quite intelligible through modern Burmese, provided you keep in mind the evolution of the Burmese pronunciation through the ages, and seek help from the dialects.
43The Alphabetum barmanorum seu Regni Avensis (1787) is a book where you could find, at the end of 18th c., printed Burmese letters, apart from the Chinese block printing. It shows, through its transcriptions, that modern T was still pronounced s and modern s (written c) “dura sicuti in zappa ab Italis vel sicuti a Gallis… c in dicere, facite”, as if still affricate, while written r was already pronounced y by many. In giving the value (“potestas”) of vowels, it points out that written —ai , as well as written —ā: are post–aspirated and transcribes them —œh, —áh, contrasting even the first one with an —œ written with “accent grave”, described as very short and interrupted, and an —œ written with “accent aigü”, which is compared to the vowels of French grêle and chaîne. That is all for tones, only the shape and position of their specific markers, the two dots and the subscript dot, are depicted.
44Around 1975 several Burmese palm leaves mss of the 19th c. were fortunately reproduced in Yangon, by Sarpay Beikman, under their original form, for instance: in 1974, a copy dated « 1871 » of Rāma watthu, in 1976, a copy dated 1863 of several manuscripts, including Amarapūra mrui. tañ samuiṅ:. Their writing shows a modern system almost achieved, tones specific markers are regularly used:
koṅ: “be good” kā: “as for” mrui. “town” rhe. “before”.
45A few redundant letters are still used:
kuiw objective particle sui. “towards” mrai: “always”.
- 5 Transcription follows the Burmese in the Anthology of Burmese Literature, vol. I , revised ed., 196 (...)
46If we consider poetry, the first dated Burmese poem was composed in Arakan in 15th c. (1458). It is called “Yakhaing minthami egyin” (Panegyric poem/lullabye for <the birth of> an Arakanese Princess). Usual climbing rhymes of Burmese poetry affect vowel nucleus and tone, as seen in the beginning of the first stanza:
kambhā … world
krā ṅā: pwaṅ. luik five lotus flowers, following (as?)
roṅ mruik ñāṅ phrū mental lights (?)
ṅā: chū …5 five…
and as pointed out by U Pe Maung Tin (1955: 24-25). Doctor Hla Pe’s prosodic analysis (1974: 4-5) of Kogan Pyo, dated 1526, shows the same rhyming of tones.
47In proverbs, vowels and tones often rhyme, but the rhymes are not climbing:
ma pe: bhai kui ma lui nhaṅ. “if not offered don’t desire”
ma me: bhai kui ma chui nhaṅ. “if not asked don’t speak”.
48Since in written Burmese the old characters foreshadowing the present marking of tones seem to have disappeared for centuries, we might try to fill the gap thanks to dialects keeping, in their oral or written form, archaic features belonging to different ages of the language, for instance to Arakanese and Marma dialects, which have the same poetical rules as Central Burmese, based on rhymes including tone, as I learned among the Marma in 1951-52, and in Burma proper from 1958 onwards.
49In fact, Arakanese culture was developed in the West of Burma, behind its mountain range, at least from the 14th c. onwards. There were always contacts between Arakan and Central Burma, sometimes temporary submission of Arakan to the central power, and definitive Burmese conquest of Arakan in 1785. As soon as the 17th and 18th c., periods of disorders, wars and ruin in Arakan, the population had started migrating massively to Bengal, where they are known as Marma.
50Today, differences between the Burmese dialects show archaisms, as well as regional evolution, among outsiders, and a drastic change in the central one, which became the national language.
51A modification in the meaning of an onomatopeic expression used in the most popular story, Medwe, shows how far the change might go; it is the noise imitating something breaking, spelt in all the written versions of the story (Marma, Arakanese, Central Burmese): krwat, pronounced krwɛʔ krwɛʔ in the two first dialects, but cuʔ cuʔ in the last one, at least in its southern part. Orally the only thing left is the glottal stop but the rest became so different that, while the Marma storyteller referred to crushing pieces of pottery as a cause of the noise, the Burmese one refers nowadays to crushing puffy, crisp cakes.
52Some differences between Marma and Burmese concerning allotones can be traced from Marma vocabularies and folktales collected in 1951-52.
53For instance, in the southern Circle of Marma, the tones had a specific contour at the end of a clause. They might fall and rise again, the voice might be interrupted, “creaky”, the nuclear vowel shortened or lengthened according to morphology, syntax, and individual dialectal evolution. Though Arakanese and Marma have the same number of tonal units as Central Burmese, one can notice in the realization of tones such morphological and syntactical influences as well as a dialectal difference.
54Let us begin with the Marma, who migrated from Arakan, then the Arakanese Province of Burma, to Bengal during 17th and 18th centuries, and specially at the end of 18th. In 1952, a Marma friend wrote to them in Marma script ((first column in the examples). The following examples show that Burmese tonal system is lexically valid:
55Listening to the tales, I added later my phonetic trancription (second column), but on the spot I transcribed the words I heard as I could: two vowels for a long one, vowel above the line for high pitch, on the line for low pitch (third column):
- 6 ပဓ် in Burmese, where the pronunciation of -(a)c and -uik are different, while it is the same in Ma (...)
Note 6 dans l’image : 10e ligne en partant de la fin.
56On the other hand, in the East of Burma, Intha and Tavoyan dialects present very archaic features, keeping the old kl and pl of the lithic inscriptions in their pronunciation and script, but reducing many disyllabic words of Central Burmese to their second syllable, as did sometimes other archaic dialects, and, progressing further than Central Burmese in reducing final stops to creaky tone.
57As human groups, both might have been separated early from the majority. Historical legends, concerning them, refer constantly to Alaungsithu, Burmese king of the 12th c., to his travels, military expeditions and pious deeds. It seems that they were involved in such expeditions not only once, during Alangsithu reign, but much later also; the two groups are now located very far from each other, each isolated enough from the environment—Shan or Burmese—to keep their common linguistic originality.
58Concerning tones, their originality is obvious on syllables where rhymes contrast according to another distribution than in Central Burmese, for instance when the Intha creaky tone corresponds to Central Burmese glottal stop. Admitting that in Central Burmese creaky tone originated from a final glottal stop, other final stops melting into a new glottal stop, Intha and Tavoyan went further in the process by reducing more final stops to creaky tone.
59Let us now quote examples of discrepancies between Central Burmese and Intha, from vocabulary and tales collected in 1968 in Yangon and in 1971 in Inchan ywa, a village on Inle Lake (first column: Burm., second: Intha):
60To Burm. glottal stop corresponds here Intha creaky tone but most of the time it is difficult to decide wether you hear glottal stop or creaky tone in Intha, you just catch a difference between the two pronunciations. In Intha it seems weaker, specially when the speaker tells a traditional story. On the other hand, the correspondence might be also inverted: creaky tone in Central Burmese corresponding to a checked stop in Intha:
61Low and breathy tones correspond between Intha and Central Burmese. In Intha these tones are often the only way to identify terms which, in Central Burmese, are also segmentally different, for instance:
62and the morphological or syntaxic use of tones corresponds also from one dialect to the other:
63Tavoyan is very similar to Intha :it still uses initials pl- and kl-:
64It uses shorter forms of words than modern Burmese:
65It uses, much more than Intha, creaky tone when there is a glottal stop in Burmese:
66though it can be found in some words:
67But allow me to quote an individual pronunciation (second column) of a Burmese sentence very usual as an answer to a question. I heard it repeatedly uttered by a true Burmese in Yangon:
68Of course it might be just the exception which confirms the rule…, or is it because this Rangonian is native from Maulmein (Mawlamyaing)?
69As for the correspondance with low and breathy tones of Central Burmese, or the regular use of morphological or syntaxical meaning of tones, Tavoyan is similar to Intha.
70Intha and Tavoyan form now two groups, separated by several hundreds of miles; their name (“sons of the lake”, inhabitants of Tavoy) refers to their geographical position, but their languages show clearly that once they formed only a single group; from the comparison of the two dialects, so similar, one may even deduce that their splitting is not old: it suggests how quickly national Burmese language is evolving and has been evolving before, so that it is now very far from regional dialects, while Intha and Tavoyan, in spite of the distance, are still very near from each other.
71Less clear, in both Intha and Tavoyan, is the age of their archaism. Some of their pronunciations belong to a remoted past: kl, pl, /pl, kl/ (which survived in Burmese writing only to the middle of 15th c.); while others are much less old: pronunciation /y/, /py/, /c/ of r, pr, kr. Perhaps their group—or groups—after a long period of isolation, have had again contacts with Burmese rulers, their army and their administration.
72To conclude, in every dialect of the Burmese branch it appears that phonetic possibilities are different for: A. syllables written with a final vowel (opened syllables), pronounced, nowadays, with an oral vowel as nucleus, and: B. all the others, written, most of the time but not always, with a final consonant (closed syllables) pronounced, nowadays, with a nasal sound somewhere (nasalized vowel for instance), or with a final stop consonant.
73In syllables of type B it is possible, nowadays, to hear a diphthong, or a vowel plus glide as syllable nucleus; also, syllables of type B. might, according to the nature of juncture with the following syllabe, nasalize or devoice the initial consonant of the latter. This is not possible for syllables of type A. Orally, nasality or glottal stop mark the difference between A and B.
74Then let us point out that, within type B., a supra-segmental feature, creaky, level or breathy, is distinctive between nasalized syllables segmentally identical, while between syllables ending with a checked glottal stop, segmentally identical, there is no possible distinction.
75Another feature remains to be pointed out. After a period of fluctuation, in the early 12th c., the syllables regularly written with final h as well as with final vowel carrier a in old Burmese, most of the time have respectively nowadays breathy and creaky tone; while a continuum between long vowels or short vowels of old Burmese and, respectively, low tone and creaky tone of modern spoken Burmese is not regular, though noticeable: graphism of vowels was varying so much and so often in lithic or terracotta inscriptions that such a continuum could not be expected. At least there is a general agreement between the dialects, with regular correspondances, which allows to identify semantic units.
76Two questions remain beyond the scope of this paper. As for these contrastive features, which appear now as suprasegmental units made of registar, contour, and realized differently according to the place of stress in the syllable, in the word, in the clause, and in the utterance, are they tones/tonemes or are’nt they? This question is as embarrassing as that of the gender of angels. The second question is: When did they become suprasegmental contrastive units? The answer requires a lot of further investigations and comparisons of data… concerning endangered languages.