Tang (618-907) poetry in Min pronunciation

« previous post | next post »

Usually, though not always, when I Romanize Sinographs on Language Log, I do so using Modern Standard Mandarin (MSM), but that is misleading, because MSM is only one of countless different topolectal pronunciations that could be used (Cantonese, Shanghainese, Sichuanese, and so on and so forth).  MSM is particularly ill-suited for the Romanization of pre-modern literature, since — of all topolects — it is the most highly evolved (ergo youngest) and least like earlier stages of Sinitic.  In this post, I will use Southern Min pronunciation to give a sense of how different it is from MSM.

The Min Romanizations have been prepared by Conal Boyce using a Yale-like system he developed in 1975 in preference to Douglas-Campbell.

Douglas, Carstairs (1899) [1873]. Chinese-English Dictionary of the Vernacular or Spoken Language of Amoy (2nd ed.). London: Presbyterian Church of England.

 Campbell, W. (1913). A Dictionary of the Amoy Vernacular. Tainan: Ho Tai Tong.

Conal says he loves "the 'antiquity' and rich sound of Min."  When he was living in Taiwan during 1973-1976, Conal composed a Chinese Requiem (using texts from the Zhuang Zi [c. 369-286 BC]) and a choral composition (scored for chorus of sopranos, altos, basses, flute [doubling on piccolo], 2 clarinets, 2 bassoons, and double bass) called A Voice from the Late Tang.  Both were intended to be sung in Southern Min.  The two poems transcribed and translated below are part of the latter composition.  For those who are interested in the musical and Sinological aspects of these compositions, Conal has demo tapes, vocal scores, and other materials that he would be willing to share and may turn into a separate publication (contact information available upon request).

I also provide MSM Romanizations for the sake of comparison.

Here are two poems by the mid-Tang poet, Li He 李賀 (c. 790–791 – c. 816–817):

  1. Song of the Old Jade-Hunter

採玉採玉須水碧           Tsái ǥiỏk tsái ǥiỏk su súi pik
琢作步搖徒好色           Dok dzok bōh iaû dôh hóhn sik
老夫飢寒龍為愁           Ló hu gi hân liông wî tsiû
藍溪水氣無清白           Lâm ke súi kì ḇû tsing bἰk

夜雨崗頭食蓁子           Yā ú gong tiô sἰt jin dzú
杜鵑口血老夫淚           Dōh goan kóh hiet ló hu lūi
藍溪之水厭生人           Lâm ke ji súi iàm sing zîn
身死千年恨溪水           Sin sú tsien liên hūn ke súi

斜山柏風雨如嘯           Siâ san bik hong ú zû siàu
泉腳挂繩青裊裊           Dzoân giok goà sing tsing niáu niáu
村寒白屋念嬌嬰           Tsun hân bἰk ok liām giau ing
古臺石磴懸腸草           Góh dâi sἰk dìng hiên diông tsóh

Cǎi yù cǎi yù xū shuǐ bì
Zuó zuò bù yáo tú hào sè
Lǎofū jīhán lóng wèi chóu
Lán xīshuǐ qì wú qīng bái

Yè yǔ gǎng tóu shí zhēn zi
Dùjuān kǒu xuè lǎofū lèi
Lán xī zhī shuǐ yàn shēng rén
Shēn sǐ qiānnián hèn xīshuǐ

Sié shān bǎi fēngyǔ rú xiào
Quán jiǎo guà shéng qīng niǎoniǎo
Cūn hán bái wū niàn jiāo yīng
Gǔ tái shí dèng xuán cháng cǎo

Hunting for jade! Hunting for jade! Only crystal-emeralds will do,
For cutting into Shake-as-she-walks [merely to inspire lust].
For an old man hungry and cold, even [the] dragons must grieve.
The mist-hung waters of [Indigo Gorge are not clear and white].

On rainy nights, on the ridge of a hill, he sups on hazel-nuts,
Blood that wells from a cuckoo’s maw, the old man’s tears.
The waters of [Indigo Gorge are sated] with human lives.
Dead a thousand years, [those men] still loathe the torrents.

A steep hillside, wind in the cypress, whistle of rain,
On spring-dripping rocks he hangs from a rope, green curling and swirling.
Cold village, white thatched hut — he frets for [wife and child],
On ancient terraces, steps of stone, the Heartbreak grass.

  1. Poems about Horses (#11 of 12)

內馬賜宮人               Lōe má sù giong zîn
銀韉刺騏麟               Ǥûn jien tsì gî lîn
午時鹽阪上               Ngóh sî iâm hoán siōng
蹭蹬溘風塵               Jīng dīng kap hong dîn

Nèi mǎ cì gōng rén
Yín jiān cì qílín
Wǔshí yán bǎnshàng
Céngdèng kè fēngchén

A royal horse given to a palace lady,
Silver trappings embroidered with unicorns,
At midday, on that hill of salt,
A foundering steed [struggles] through wind and dust [to its death].

The translations are by Frodsham, with some slight departures indicated by square brackets.  See J.D. Frodsham, tr., The Poems of Li Ho (Oxford:  Clarendon Press, 1970), pp. 79, 72.  Frodsham republished his book in 1983 [North Point Press – San Francisco] with small revisions; the corresponding page numbers there are 63 and 58.  In praise of Frodsham, Conal says that his book of translations still boggles the mind:

— to think that he translated all of Li Ho's poems [I believe he covers the entire corpus], so many of which are highly allusive, switching allusions every two or three characters sometimes. And Frodsham added notes for every such twist and turn, all the way through!

When I asked Conal for further elaboration on why he chose to transcribe the Li He poems in Southern Min, he stated:

It was partly because of the rich tonal structure: I liked the idea of tailoring the musical contours to the Min tones. Also, it was partly a kind of youthful ‘rebellion’ against the prevalence of 國語 [VHM:  Guoyu — the "National Language", i.e., MSM] which is, after all, so ugly compared to most other so-called dialects which are suggestive of the 'real Chinese' of yore.

Although slowly attenuating as living languages, the synchronic and diachronic implications of the Sinitic topolects are still enormous.  When Conal was doing his original research and creation utilizing Southern Min, he wasn't thinking of the political dimensions of that language.  Now, nearly half a century later, his historical, literary, and linguistic-musicological investigations have taken on ideological implications that he hadn't anticipated.  As Bathrobe said in the first comment to "Confessions of an Ex-Hokkien Creationist" (9/20/16), "Language is such a political animal…"


Selected readings



  1. J.W. Brewer said,

    October 14, 2021 @ 6:14 pm

    Maybe this will never happen in the Sinitic world, but I am struck by the parallel of how pronunciation of Latin in Europe both evolved over the centuries/millennia and came to vary by region, such that by the Middle Ages no one tried to pronounce it the way it had been pronounced in the days of Julius Caesar and likewise no one worried too much that it was pronounced differently elsewhere in Europe. Then came the 20th century and, for the first time ever, schoolchildren were taught to pronounced Latin in a conjectural reconstruction of "proper" ancient pronunciation rather than whatever living evolved topolectal tradition had been handed down to them, such as https://en.wikipedia.org/wiki/Traditional_English_pronunciation_of_Latin. Prior to this, some topolects had more political/cultural power than others. Most notably the Italian topolect developed what you might call a Mandarin-like power in parts of Europe that were Roman Catholic, even though there was no particular basis to think post-medieval Italian pronunciation of the old Latin texts was any more "authentic" in a deep historical sense than the pronunciations found in the topolects of Late Latin current in Belgium or Portugal or Slovakia might be.

    But as I say, this all ended when Latin teachers felt sufficiently motivated to give up their regional topolects in favor of a scholarly reconstruction of the actual Middle Chinese — oops, I mean Latin — pronunciation.

  2. David Marjanović said,

    October 14, 2021 @ 7:12 pm

    ergo youngest

    I'm not sure I understand – do you mean the youngest in its current form, the last to undergo noticeable changes?

    Most notably the Italian topolect developed what you might call a Mandarin-like power in parts of Europe that were Roman Catholic

    No, just in the English-speaking ones. The Spanish-, German- and French-speaking parts continue to use native spelling-pronunciations for Latin (plus some concessions to the original stress placement in German), and the Slavic-speaking parts likewise except for allowing c as /k/ when something other than a front vowel follows.

    However, the German pronunciation of Latin was updated after the vowel changes between Middle and Early New High German, as was the spelling of German itself. This did not happen in English, where the traditional pronunciation of Latin applies the Great Vowel Shift.

  3. Terpomo said,

    October 14, 2021 @ 7:36 pm

    I get the impression that Classical Chinese read in Mandarin pronunciation is often too ambiguous to be understood clearly, but is that less the case in Hokkien because it's more conservative? Hokkien speakers who are familiar with Classical Chinese, can you understand a Classical Chinese text on hearing it out loud without major uncertainty about homophonic words?

  4. Scott P. said,

    October 14, 2021 @ 8:48 pm

    Maybe this will never happen in the Sinitic world, but I am struck by the parallel of how pronunciation of Latin in Europe both evolved over the centuries/millennia and came to vary by region, such that by the Middle Ages no one tried to pronounce it the way it had been pronounced in the days of Julius Caesar and likewise no one worried too much that it was pronounced differently elsewhere in Europe. Then came the 20th century and, for the first time ever, schoolchildren were taught to pronounced Latin in a conjectural reconstruction of "proper" ancient pronunciation rather than whatever living evolved topolectal tradition had been handed down to them

    This happened much earlier than the 20th century, according to Roger Wright. Vernacular pronunciations were maintained until the Carolingian period, when there was a reform movement in favor of standardization. Wright specifically identifies Alcuin of York, advisor of Charlemagne and himself not a native Latin-speaker, with the innovation of speaking the words as spelled and not according to vernacular pronunciation. That essentially 'created' written Romance as heretofore the written and spoken languages were the same, but now Latin was, as pronounced, incomprehensible.

  5. Chris Button said,

    October 14, 2021 @ 8:48 pm

    Interesting that he went with Min rather than one of the varieties of Chinese that would have evolved from the Middle Chinese of Li He’s time.

  6. Jerry Packard said,

    October 14, 2021 @ 9:23 pm


    Yes, the Min versions are less ambiguous. Even if I were not a Hokkien speaker we can suss out that there are virtually no homophones in the Min versions of the classical texts, as can be seen in the Hokkien-romanized versions of the above poems.

    That said, it should be pointed out that the classical texts (though less so with poetry) are rarely adequately rendered by a simple literary reading of the sinographs of the texts. Vernacular Min versions would radically depart from the textual versions, as would also be the case with, e.g., Cantonese.

  7. John Rohsenow said,

    October 14, 2021 @ 9:59 pm

    On a less esoteric note, when I was a student in the IUP Stanford Center
    in the late 1960s, we inevitably had to memorize some familiar T'ang
    poems. Being a linguist, when I read them (in Mandarin), I put the ru- sheng final stop tones in at the ends of the lines, much to the delight of Yang Taitai, who grew up in "Peip'ing" and probably heard them that way from her teachers more than half a century before.

  8. Jonathan Smith said,

    October 14, 2021 @ 11:55 pm

    A couple points which are well-known but (it seems to me) obscured in popular and no small number of more technical descriptions…

    * with respect to syllable codas — among other salient features — Cantonese and Hakka are indeed (generally) "conservative," while Min is (generally) decidedly not; SM, for instance, consists of lects which are at root highly innovating on this front (indeed, in descriptive terms, Mandarin-like or at least Jin-like), which divergent development was apparently arrested by the late medieval-ish importation of massive quantities of material from (relatively) northern, *conservative* (!!) (think Cantonese-like) varieties (cf. Chris Button's comment). It is these late medieval pronunciations which are traditionally employed in poetry like the above (a point related to but I think not identical with Jerry Packard's comment.)

    * same for tones really… for "running" tones, making up the balance of continuous speech, five phonological tones is a useful way of understanding Xiamen/Taiwanese-type varieties, for instance: A1/A2/B/C1/C2 (and Min sandhi systems are again of course an innovation.)

    Thus, re: David Marjanović's comment, "youngest" applied to Mandarin is at least not to be taken to mean that these lects are uniquely innovative / "most diverged" (which yes would be odd anyway)… as is usually the case, these branches have just changed in different ways — Mandarin's "retroflex affricates" ZH represent several merged "MC" categories, but Z stays separate ("conservative"!); core Min for its part instead collapses early "retroflex stops" (?) "TR" with the dentals "T" which for no good reason is often preferentially described as "conservative" rather than "innovative"… etc.

  9. Philip Taylor said,

    October 15, 2021 @ 6:00 am

    Could you explain, Victor, the scholarly interpolations such as [merely to inspire lust] and [Indigo Gorge are not clear and white] ? In V1L4, the interpolation (7 words) is longer than the text (5 words) — whence came the interpolations, and how were they derived ?

  10. Victor Mair said,

    October 15, 2021 @ 6:50 am

    Of course, all Sinitic lects have innovated in their own ways, some more, some less,


    Major innovations in Mandarin:

    palatalization of the velars, which I've written about several times on LL; a phenomenon that began in the north about three hundred years ago, and which has reached around Shanghai by now

    numerous features, not merely phonological, that arose through contact with Tungusic and Turkic languages

    overall loss of entering tones

  11. David Cowhig said,

    October 15, 2021 @ 7:30 am

    Thank you for this Dr. Mair! It recalls from me the days I audited Chinese Literature Department classes at Tunghai University in the early 1980s. The dean of the humanities college [wenxueyuan] , Professor Jiang Zhuqian, a mainlander from Fujian, told his Taiwanese students "Be proud that you speak Taiwanese! Taiwanese is much closer to the pronunciation of ancient China that is the Mandarin spoken today."

  12. Mark Meckes said,

    October 15, 2021 @ 10:47 am

    On the subject of different national pronunciations of Latin, I once heard the following anecdote (don't know how reliably): Supposedly at one of the 20th century Vatican councils, the caucuses of cardinals from different countries would sometimes wish to converse more-or-less secretly among themselves. Many of them could simply speak their native languages with little fear that a random passerby would understand, but this option wasn't available to the English-speaking cardinals. Instead, they spoke in Latin among themselves, since even though everyone around knew Latin, nobody else could understand their pronunciation of it.

  13. Chris Button said,

    October 15, 2021 @ 10:54 am

    It’s probably worth noting here Pulleyblank’s 1968 article “The rhyming categories of Li Ho”.

  14. PeterL said,

    October 15, 2021 @ 12:37 pm

    As to Min vs other "varieties of Chinese" (from Chris Button's comment); here are some examples:

    In some ways, Sino-Xenic pronunciations have been even more conservative; but they're also overlaid on languages that have very different underlying morphological characteristics and come from multiple times and locations in China – the most common in Japanese are from the Tang Dynasty – https://en.wikipedia.org/wiki/Kanji#On'yomi_(Sino-Japanese_reading)

  15. David Marjanović said,

    October 15, 2021 @ 12:46 pm

    one of the 20th century Vatican councils

    That would be the Second, seeing as the First was in 1870 and there hasn't been a Third.

    I haven't encountered that anecdote, but there's another which says the Latin of the English-speaking cardinals wasn't understood at the First Vatican Council.

  16. J.W. Brewer said,

    October 15, 2021 @ 6:50 pm

    In tension with David M.'s anecdote, there are Victorian travelers' tales in which Respectable Englishmen find themselves stuck for the night in some remote village in the Hapsburg-ruled Balkans in which none of the Bloody Foreigners speaks English and resolve the difficulty (since the R.E. naturally knows no Slovenian/Magyar/Ruthenian/whatever) by finding the parish priest and striking up a conversation with him in Latin. Of course you can harmonize this all by thinking about the differences between a situation in which someone is speaking his native idiolect (in his actual L1 or a specific L2) while consciously trying to minimize the chances of being understood by eavesdropping outsiders and a situation in which that same someone is speaking the same idiolect but very much trying to maximize the chances of being understood by the particular outsider he is trying to communicate with.

  17. Jonathan Smith said,

    October 16, 2021 @ 8:35 pm

    The idea that Mandarin is objectively relatively innovating could be true (one could, I suppose try to quantify the claim), but my main interest above was the tendency for such claims to bleed into the notion that (say) Taiwanese is a language of "rich […] antiquity," indeed "real Chinese," in stark contrast to "ugly" Mandarin. This is silly and unscientific to begin with, all the more so given that the words at issue referenced represent extrinsic material in TMN (a fact which remains synchronically transparent to speakers)… re: the poems, contrast, e.g, colloquial kûã ‘cold’ (cf. hân), nâ ‘blue’ (cf. lâm), hōo ‘rain’ (cf. ú), tsia̍ʔ ‘eat’ (cf. si̍t), lâng ‘person’ (cf. jîn), kʰa ‘foot’ (cf. kiok), and so on…

  18. Victor Mair said,

    October 17, 2021 @ 6:13 am

    "The idea that Mandarin is objectively relatively innovating could be true….".

  19. KIRINPUTRA said,

    October 18, 2021 @ 10:32 pm

    Just want to point out that Vietnamese is no less suited for this than Cantonese, and arguably a bit more suited than Taiwanese and Hokkien…

  20. legatrix said,

    October 19, 2021 @ 1:58 am

    On reconstructed pronunciation (J.W. Brewer's point), I wonder if a poet of today, were a time traveller from the year 2521 to materialize, would feel their poetry better reflected in 2521 English, or in the time traveller's attempt at reconstructed pronunciation of 2021 English. In any case, it would depend on any given poet's whim. The only thing I can conclude is that there can never be a knockdown argument against using reconstructed pronunciation, unless the reconstruction is such that it would be uninterpretable to a native of the past, which it may turn out to be for current OC, but surely not MC.

  21. Chuck said,

    October 19, 2021 @ 9:20 am

    @VHM, you write that MSM "is the most highly evolved (ergo youngest) and least like earlier stages of Sinitic." Evolutionary biologists avoid using "highly evolved" and instead use "highly derived", which means that the taxonomic unit in question shows a large number of differences from a (usually hypothetical) common ancestor, usually but not necessarily in comparison to other related taxa. When relationships are diagrammed, typically using a branching "tree", the highly derived taxa have longer branches, but are not necessarily younger or older than any other taxa.

    With regard to age, MSM might branch from the Sinitic "tree" relatively recently, but any still-extant descendants from that branch point would be the same "age" (neither younger nor older) as MSM, even if they are less derived and therefore more similar to a common ancestor.

    These points are somewhat pedantic, but "highly evolved" is often interpreted as "better" and I personally, and I suspect LL readers in general, would not want anyone to infer that MSM is "better than" any other topolect, or indeed that any one language is "better than" another (except Taiwanese, which is the best.

  22. Victor Mair said,

    October 19, 2021 @ 10:09 am

    "These points are somewhat pedantic…".


    Problem is that speakers of non-MSM languages believe that "less highly evolved" is better.

  23. J.W. Brewer said,

    October 20, 2021 @ 7:18 pm

    @legatrix: Of course in English we already have this issue because the language is pluricentric w/r/t pronunciation, including normative/prestige pronunciation. For Chaucer and Shakespeare you can find people interested in reconstructed historical pronunciation, but for Blake and Shelley and Tennyson and Whitman and Dickinson and Yeats etc., I read them aloud in my own regional American pronunciation, and a reader from Australia in his/her native pronunciation, and a Posh Brit in his/her RP pronunciation and so on, and we are all doing so on a basis of more or less aesthetic and cultural parity.

  24. Terpomo said,

    October 21, 2021 @ 12:20 pm

    @Scott P.
    It's interesting to imagine how Latin might sound in modern evolved pronunciations. I think I once wrote out and recorded the beginning of the Aeneid in hypothetical modern Spanish reflexes but I don't have it to hand now.
    @Jerry Packard
    It's not clear to me that how many homophonous terms there are in the poem proper is the appropriate measure, since the poem only contains a few dozen distinct words. What's more important is the existence of homophonic terms in the LS lexicon as a whole. And indeed I'm not talking about translations, but simply pronouncing the classical text aloud.
    Indeed, I've seen Vietnamese printings of Literary Sinitic texts which are facing-page translations with Vietnamese on one side and the original text in Sino-Vietnamese pronunciation on the other, with no characters in sight (or only the occasional parenthetical word, as in modern Korean), which is evidence that Sino-Vietnamese pronunciation retains enough distinctions to render LS intelligibly.

RSS feed for comments on this post