What is the difference between a dragon and a /lʊŋ³⁵/?

« previous post | next post »

Today is the Lunar New Year's Day, and it's the Year of the Dragon / /lʊŋ³⁵/ . As such, a kerfuffle is stirring in China and the English-speaking world regarding the English translation of lóng ⿓ / 龙 / 竜 (J), which is usually "dragon".

I will begin with the pronunciation of the word.  In MSM, it is lóng (Hanyu Pinyin), lung2 (Wade-Giles), lúng (Yale), long (Gwoyeu Romatzyh [the configuration of GR tonal spelling for this syllable indicates second tone), лун (Palladius).  They all represent the same MSM syllable.  I will not list the scores of other topolectal pronunciations for Cantonese, Shanghainese, Hakka, Hokkien, Xiamen / Amoy, Sichuan, etc., etc. and their dialects and subdialects.

Some patriotic and political sources in the Sinosphere assert that the dragon is fearful, frightful, ferocious, and ugly, while — so they claim — the lóng ⿓ / 龙 is auspicious and adorable.  (I don't believe that, since the typical Chinese dragon has horns, claws, fangs, gaping whiskered maw, glaring eyes, snake-like body with scales, etc., not a creature you'd like to meet in the middle of the night (or even in the daytime; some have saurian or bat-like wings, and so forth.)  Those who pettify and prettify the lóng ⿓ / 龙 are asking us to stop calling it a dragon and start calling it, not "long" (which some people are already doing, but it's not catching on, for obvious reasons), but "loong".  That's not gaining traction either, because a doubled vowel often informally indicates a low, dipping third tone (e.g., Shaanxi), while lóng ⿓ / 龙 is second tone.  If a variant spelling from "long" is desired for lóng ⿓ / 龙, it would be better to write "lorng", since "r" in some informal tonal romanizations is used to indicate a rising second tone (e.g., horng 紅 ["red"]).


Before moving on to the etymology and glyphology of lóng ⿓ / 龙, let us take a brief look at the reconstructions of earlier stages of the phonology of lóng ⿓ / 龙.

  • Middle Sinitic (ca. 600 AD): ljowng
  • Old Sinitic (ca. 600 BC:
(BaxterSagart): /*[mə]-roŋ/
(Zhengzhang): /*b·roŋ/



From Proto-Sino-Tibetan *m-bru(ŋ/k) (dragon; thunder). Cognate with Tibetan འབྲུག ('brug, dragon; thunder). The STEDT database also lists (OC *ɡ·ruːŋ, “thunder; sound of thunder”) and (OC *bruːɡ, “hail”) as cognates. Also compare (OC *brɯŋs, “sound of thunder”) and 霹靂 (OC *pʰeːɡ reːɡ, “thunder”).

This word is found in many languages of the region. Compare Proto-Hmong-Mien *-roŋ (dragon) (White Hmong zaj), Proto-Vietic *-roːŋ (dragon) (Vietnamese rồng), Vietnamese thuồng luồng (serpent-like monster), Khmer រោង (roong, year of the dragon), Thai มะโรง (má-roong, dragon; year of the dragon), Lao ມະໂລງ (ma lōng, year of the dragon), perhaps also Old Turkic [script needed] (*-lan, suffix denoting a wild, predatory animal) (Turkish aslan (lion), kaplan (tiger), yılan (snake)).



Shang Western Zhou Spring and Autumn Warring States    
Bronze inscriptions Oracle bone script Bronze inscriptions Bronze inscriptions Chu slip and silk script Qin slip script    

Pictogram (象形) – originally a serpent with prominent whiskered mouth and eyes.

Current form developed in large seal script, with serpent’s body on right (tail at upper right, legs on right), whiskered/fanged mouth at lower left, and eyes/crown at upper left. Left side was subsequently simplified and abstracted, with some influence of and /. Note that existed as a traditional variant dating back to large seal script, and figures a dragon seen face-on, rather than curled around.

(The above three sections are based on Wiktionary)

For comparative purposes, let us look at the etymology of "dragon".

From Middle English dragoun, borrowed from Old French dragon, from Latin dracō, dracōnem, from Ancient Greek δράκων (drákōn, a serpent of huge size, a python, a dragon), probably from δέρκομαι (dérkomai, I see clearly).


The word dragon entered the English language in the early 13th century from Old French dragon, which, in turn, comes from the Latin: draco (genitive draconis) meaning "huge serpent, dragon", from Ancient Greek δράκων, drákōn (genitive δράκοντος, drákontos) "serpent". The Greek and Latin term referred to any great serpent, not necessarily mythological. The Greek word δράκων is most likely derived from the Greek verb δέρκομαι (dérkomai) meaning "I see", the aorist form of which is ἔδρακον (édrakon). This is thought to have referred to something with a "deadly glance", or unusually bright or "sharp" eyes, or because a snake's eyes appear to be always open; each eye actually sees through a big transparent scale in its eyelids, which are permanently shut. The Greek word probably derives from an Indo-European base *derḱ- meaning "to see"; the Sanskrit root दृश् (dr̥ś-) also means "to see".


As soon as I saw that last sentence concerning the IE root *derḱ-, I immediately thought of the Indian religious concept of Darshana, also spelled Darshan, i.e., Sanskrit darśana दर्शन ("showing, appearance, view, sight") or Darshanam (darśanam), which is the auspicious sight of a deity or a holy person.  (source)



Dragons and lóng ⿓ / 龙 resemble each other sufficiently closely (compare the prows of Viking longships and even the dragon-shaped bows of modern Swedish ships) that their names constitute serviceable translations one for the other.  Transcribing lóng ⿓ / 龙 into English "loong" would cause more problems and confusion than translating it as "dragon".

Selected readings

[Thanks to Chau Wu]

A note on the shared iconography of the dragon and the lóng ⿓ / 龙 / 竜

As I prepared this post, I came increasingly to observe striking parallels in the imagery of west Eurasian dragons and east Asian lóng.

Take, for example, the Dacian draco military standard / banner, as in this detail from Trajan's Column, linked to other images of this device.  It has symbolic attributes that lead one to associate it with lóng from the first millennium BC and later periods.

The Dacians (/ˈdʃənz/; Latin: Daci [ˈdaːkiː]; Greek: Δάκοι,[2] Δάοι,[2] Δάκαι[3]) were the ancient Indo-European inhabitants of the cultural region of Dacia, located in the area near the Carpathian Mountains and west of the Black Sea. They are often considered a subgroup of the Thracians.[4] This area includes mainly the present-day countries of Romania and Moldova, as well as parts of Ukraine,[5] Eastern Serbia, Northern Bulgaria, Slovakia,[6] Hungary and Southern Poland.[5] The Dacians and the related Getae[7] spoke the Dacian language, which has a debated relationship with the neighbouring Thracian language and may be a subgroup of it.[8][9] Dacians were somewhat culturally influenced by the neighbouring Scythians and by the Celtic invaders of the 4th century BC.


See Adrienne Mayor's important post on X which shows mounted, helmeted Sarmatian warriors bearing such dragon standards. 

The Sarmatians (/sɑːrˈmʃiənz/; Ancient Greek: Σαρμάται, romanizedSarmatai; Latin: Sarmatae [ˈsarmatae̯]) were a large confederation of ancient Eastern Iranian equestrian nomadic peoples of classical antiquity who dominated the Pontic steppe from about the 3rd century BC to the 4th century AD.


These equestrian, highly mobile Indo-European peoples could easily have transmitted such imagery long (!) distances across the continent.

The dragon-lóng iconographic interface and the transferral of symbolism pertaining thereto is sufficiently significant to warrant the writing of a major paper on this subject.  I would welcome such a study (replete with appropriate illustrations) for consideration of publication in Sino-Platonic Papers.


  1. AntC said,

    February 10, 2024 @ 5:56 am

    originally a serpent …

    Right on cue: Serpent handle found in northern Taiwan last year could be 4000 years old

    The [archaeology] team said it resembles a cobra, which in ancient times served as "a tool for humans to communicate with divine and ancestral spirits."

    The stylised shape resembles those early Pictogram forms Prof. Mair illustrates.

    Possibly more info via the Facebook links in the article. Presumably (?) this artefact was brought to Taiwan much later, during Chinese Imperial times.

  2. Victor Mair said,

    February 10, 2024 @ 8:24 am

    Wow, AntC!

    Synchronistically, draconically supercalifragilisticexpialidocious!!!!!

  3. Philip Anderson said,

    February 10, 2024 @ 10:41 am

    Although the Western dragon and the Chinese dragon evoke different responses, and belong in different stories, they are essentially the same beast, especially when considering the Germanic “worms” rather than the High Medieval, fire-breathing, winged beasts. Beowulf’s bane breathed though, a development from poison-breathing.
    The possible connection with Aslan (C.S. Lewis took the name from Turkish) is interesting, although it would have to have been borrowed after the Chinese ‘l’ developed, but still early into Turkic.

  4. Peter B. Golden said,

    February 10, 2024 @ 11:27 am

    In the Rasūlid Hexaglot, which consists of several multilingual vocabularia (Arabic, Persian, Turkic [in several dialects], a dialect of Byzantine Greek, a dialect of Armenian and Mongol (as spoken in Ilkhanid Iran), in the Arabic-Persian-Turkic-Mongol section it has Mongol. lu "dragon," Turkic ïlan balïq (lit. "snake fish," cf. Middle Qïpchaq yïlan balïq), Pers. nahang "crocodile, alligator, water dragon"), Arab. al-timsāḥ "crocodile." The Hexaglot was compiled by the Rasūlid ruler of Yemen, al-Malik al-Afḍal al-'Abbās b. 'Alī (r. 1363-1377) on the basis of live informants, see "The King's Dictionary. The Rasūlid Hexaglot: Fourteenth Century Vocabularies in Arabic,Persian, Turkic, Greek, Armenian and Mongol," ed. P.B. Golden (Leiden: Brill, 2000): 221 (f. 199-C, 12). Lu "dragon," clearly a loanword in Old Turkic from Chinese (see Clauson, "An Etymological Dictionary of Pre-Thirteenth Turkish," Oxford: Clarendon Press, 1972: 763) probably entered Turkic (and thence Mongol) with the adoption of the twelve-year animal cycle calendar.

  5. Doctor Science said,

    February 10, 2024 @ 4:17 pm

    The biggest difference between eastern and western dragons is that we have eye-witness descriptions of eastern dragons. I found it translated in Mark Elvin's The Retreat of the Elephants. The observations were made by the Chinese scholar Xie Zhaozhe (謝肇淛, 1567–1624) and recorded in his work Wu Za Zu (五雜俎, ca. 1592), or "Five-Fold Miscellany".

    The event took place when he was approximately twelve years old, in the East China Sea, probably en route to Okinawa:

    I journeyed in 1579 with my paternal grandfather, …when he was in charge of the official travel arrangements [for the commissioner to the Liuqui Islands]. We were halfway there when a typhoon arose. Thunder, lightning, rain, and hailstones all fell upon us at the same time. There were three dragons suspended upside down to the fore and aft of the ship. Their whiskers were interwound with the waters of the sea and penetrated the clouds. All the horns on their heads were visible, but below their waists nothing could be seen.

    In another account of the same experience, Xie Zhaozhe says that the dragons were

    suspended upside down from the edges of the clouds, and still more than a thousand feet above the water, which rose boiling like steam or smoke to conjoin with the clouds, the people seeing the dragons with minute particularity.

    For me, as I said when I first posted about this in 2019, this raises the question: when did educated people in China *stop* seeing dragons, and why. Which dragons disappeared first, river dragons or storm-cloud dragons, or did it happen at the same time? Did officials in the provinces stop forwarding reports of dragons to the capital because the capital became less likely to believe them, or did the provincial officials get fewer reports to forward? These questions have the advantage of being answerable, if someone is willing to dig through the enormous number of surviving gazetteers and other reports from the Ming and Qing dynasties.

    Xie Zhaozhe's report proves that up through the end of the Mind dynasty, at minimum, dragons weren't *legendary* in China, they were *rare*: there was something(s) that people very occasionally saw that they agreed was a dragon.

    Because it's weird, when you think about it, that the Chinese zodiac has 11 perfectly mundane and real animals … plus the dragon. The Babylonian or "Western" zodiac is fundamentally astronomical, and is a mish-mash of creatures or even objects from very ancient myths. However the Chinese zodiac was assembled (I can't seem to find any reliable references), it's eleven-twelfths real, normal animals that could be found in China a couple thousand years ago … plus dragons.

  6. Chris Button said,

    February 10, 2024 @ 4:55 pm

    (Baxter–Sagart): /*[mə]-roŋ/

    Thai มะโรง (má-roong, “dragon; year of the dragon”),

    Since the má- component of the Thai form seems to be native to it, I wonder if Baxter & Sagart have some other evidence for their *[mə]- prefix in Old Chinese?

    Takashima's dicussion in his Bingbian concordance of conflation/overlap between and 龍 is interesting. *laŋʲ (giving *laɲ) and 龍 *raŋʷ are somewhat hard to reconcile phonologically, but when all the comparative forms across the region are included (to which we could add the Mon-Khmer forms for rainbow/dragon in Shorto's comparative dictionary), perhaps the discrepancy results from different languages and time depths.

  7. Chris Button said,

    February 10, 2024 @ 4:59 pm

    Takashima's dicussion in his Bingbian concordance of conflation/overlap between XX and 龍 is interesting. XX *laŋʲ (giving *laɲ) and 龍 *raŋʷ …

    It looks like the character I was trying to write in the XX slots above did not appear. It should be the phonetic component found in modern characters like 嬴 and 贏 (i.e. minus the 女 and 貝 respectively)

  8. Elizabeth Barber said,

    February 10, 2024 @ 5:10 pm

    Chinese dragons, almost certainly travelling on textiles, strongly influenced what Westerners thought a "dragon" must look like. An early Viking-era rock drawing at Ramsund, Sweden, illustrates the much older tale of Sigurd slaying the gold-guarding dragon Fafnir by stabbing him from below. (Several other incidents unmistakably from that tale are also depicted within the curl of the dragon's body, including Sigurd sucking his burnt thumb after testing to see if the dragon's heart is sufficiently roasted.) But the artist clearly had no idea what a dragon actually looked like: it is merely a long ribbon with a head! In short, a snake (draco; cf. the constellation). Presently, however, about the time that Chinese textiles (identifiable by the presence of bombyx mori fibers) start making it all the way across Eurasia to NW Europe, depictions of mythical Germanic dragons start to look very much like Chinese ones. (For an illustration of the rock art, plus much else about early European "dragons", see Fig. 43, p. 239 of WHEN THEY SEVERED EARTH FROM SKY: HOW THE HUMAN MIND SHAPES MYTH [Princeton, 2004].)

  9. Victor Mair said,

    February 10, 2024 @ 7:08 pm

    From Mehmet Olmez:

    Chinese long (龍) appears in runic inscription as ulu, in Uyghur manuscripts mainly as luu.

    For arslan (not aslan, it is a secondary form (?) of arslan, and only (?) in Turkish, without an -r-, maybe there is more material in Sevortian (1974).

    Marcel Erdal describes arslan and similar words with °lan as opaque.

    arslan occurs first in Old Uyghur texts (maybe from the end of 9th century?) from Turfan, Chocho etc. It was not attested in Old Turkic runic inscriptions.

    asrlan ('rsl'n, with one aleph) with the Uyghur script

  10. Lucas Christopoulos said,

    February 10, 2024 @ 7:11 pm

    Qin and Greek Dragons were both represented as sorts of large Pythons

    Kholikos Dragon:

    Qin massive bronze-Pythons-Dragons

  11. Lucas Christopoulos said,

    February 10, 2024 @ 8:06 pm

    For more Dragons, in Europe, the legendary Phoenician Kadmos (Κάδμος) from Tyr, founder of Thebes, and brother of Phoenix (Φοῖνιξ), son of king Agenor (Ἀγήνωρ), killed the Dragon, whose teeth will give birth to the Spartes (not the Spartans). Kadmos is traditionally said to have brought the Phoenician alphabet to Greece.

  12. TK Mair said,

    February 10, 2024 @ 9:51 pm

    Maybe very pedestrian or even a throw away comment, but, what is imperfect about calling this beast a Chinese Dragon? Just as above someone commented about a Viking Dragon.

  13. Jonathan Smith said,

    February 10, 2024 @ 10:41 pm

    OK what wikipedia editor(s) decided Mandarin (pinyin) -ong had nucleus [ʊ]…

  14. Victor Mair said,

    February 11, 2024 @ 7:55 am


  15. Chris Button said,

    February 11, 2024 @ 7:56 am

    @ T K Mair

    Fair enough, but don't we still at least need to connect 虹 "rainbow"?

    @Jonathan Smith

    Standard analysis, no?

  16. David Marjanović said,

    February 11, 2024 @ 10:34 am

    The possible connection with Aslan (C.S. Lewis took the name from Turkish) is interesting, although it would have to have been borrowed after the Chinese ‘l’ developed, but still early into Turkic.

    Other Middle Sinitic loanwords in Proto-Turkic have been identified; Proto-Turkic wasn't spoken terribly long ago.

    arslan occurs first in Old Uyghur texts (maybe from the end of 9th century?) from Turfan, Chocho etc. It was not attested in Old Turkic runic inscriptions.

    Nothing more lion-like than a tiger has lived in that region since last ice age; AFAIK, lions occur in Buddhist literature, and I'd be surprised if there aren't any in Manichean literature.

    OK what wikipedia editor(s) decided Mandarin (pinyin) -ong had nucleus [ʊ]…

    The claim isn't [ʊ], it's /ʊ/, i.e. that whatever the actual sound is is most easily represented as part or all of an /ʊ/ phoneme that Standard Mandarin is hypothesized to have. Phonemes are language-specific hypotheses.

    The actual sound is in the middle between [o] and [u], making it equidistant between [o], [ʊ] and [u]. I think the reason Pinyin – unlike all other transcriptions except GR – goes for o is to indicate that it's not [ʊ] but a fully back vowel (if at the same height as [ʊ]).

    Personally, I prefer the two-vowel analysis of Standard Mandarin, so that long is /lwŋ/.

  17. David Marjanović said,

    February 11, 2024 @ 10:35 am

    To be on the safe side, I should perhaps add that by "Proto-Turkic" I mean the last common ancestor of all attested Turkic languages. The Turkic branch must have been distinct from whatever its closest relative is for a long time (a few millennia) before that.

  18. Jonathan Smith said,

    February 11, 2024 @ 11:12 am

    "The claim isn't [ʊ], it's /ʊ/"

    No, as I discovered, the more general wikipedia approach turns out to involve "near-close near-back rounded vowel" [ʊ]; see, e.g., the so-titled article — where the value in (pinyin) -ong is nonetheless described correctly as "[f]ully back; height varies between mid and close depending on the speaker."

    Given fully-back-ness (as you say, "in the middle between [o] and [u]"), normal practice for individual languages is to use /o/ or /u/, certainly not /ʊ/. (The actual Mandarin value being more regional than individual — e.g., the Beijing value is higher and the Taipei value lower [but pinyin uses -ong and Wade-Giles/Yale -ung… go figure.])

    It is largely moot anyway since languages like Mandarin have very few coda consonants and thus lack the sorts of minimal contrast paradigms that make vowel phonemes make sense to begin with. This means that the question of how to represent e.g. (pinyin) na nai nao ne nei ni nia niao nie niu nou nu nü nüe nuo "phonemically" leaves a lot of room for pedantic dicking around. Suffice to say native intuition is that more than two entities are involved here.

  19. Nogos said,

    February 11, 2024 @ 2:41 pm

    Can Sanskrit नाग also be regarded as the Chinese version of dragon?

  20. Matteo Compareti said,

    February 11, 2024 @ 3:15 pm

    the identification of Chinese long with "Western" dragon is an invention of the first Western scholars who saw dragons in Islamic (mainly Persian) book illustrations from the time of the Ilkhanids. In this period, the style of Song artists but adopted by Yuan artists started to appear as "Chinese style" in the entire Mongol Empire that connected China with Iran, Central Asia. etc. It is the Chinese iconography that entered Iran, not its identification nor function. Ilkhanid artists adopted Chinese flames, mountains, clouds, phoenix (Fenghuang) and dragon (long). Western scholars (mainly European ones) have been in contact with Islamic book illustrations earlier for a long time and only some centuries later they entered in contact with Chinese long. When they saw the latter, they called it dragon because it called to their mind the Chinese iconography of Islamic dragons but it was actually the contrary; Muslim artists adopted it from the Chinese in the Mongol century and they are actually the same from an iconographic point of view even if in the West we call it dragon or variants and in China it is long. They both shared the same "ofidic" nature but were colossal, can talk and fly. Something similar happened with the phoenix and Fenghuang (both are birds, can speak make magic, etc.) but in origin they are two different fantastic animals. I think I wrote somewhere about this but I cannot check right now because I am hospitalized.

  21. Chris Button said,

    February 11, 2024 @ 4:27 pm

    Back in the early 1930s, L.C. Hopkins wrote a series of the three articles on dragon in Chinese: "Where the rainbow ends: An introduction to the dragon terrestrial and dragon celestial" in 1931; "The dragon terrestrial and dragon celestial" in two parts in 1932. There is also a comprehensive study "Chinese dragon names" by Michael Carr (1990) that references all three.

    A couple of notes from Hopkins' studies: the oracle-bone forms for 虹 "rainbow" clearly show a 龍 head at each end (the 虫 is what remains of the dragon now); there is an etymological link with 隆. In that regard, the evidence for a velar cluster in Old Chinese is far better than for a bilabial cluster.

  22. SusanC said,

    February 11, 2024 @ 5:46 pm


    The name of the Chinese dragon appears, untranslated, in the name of a CPU (MIPS compatible) from China. (I.e. n English you'ld typically call this piece of computer technology loongson rather than resort to its translation, dragon core).

  23. Finistis said,

    February 11, 2024 @ 5:52 pm

    Should the word "phenix" be used to translate the Chinese term of "鳳凰‘’ (Fènghuáng)?

  24. Victor Mair said,

    February 11, 2024 @ 6:34 pm

    From Alan Kennedy:

    Interesting to note distinctions between the long and the mang on Chinese 'dragon' robes, for example.

    The number of claws is the main visual difference, 5 for the long, 4 for the mang.

    An article related to this difference:


    The Gift of a Python Robe: The Circulation of Objects in "Jin Ping Mei" on JSTOR

    Sophie Volpp, The Gift of a Python Robe: The Circulation of Objects in "Jin Ping Mei", Harvard Journal of Asiatic Studies, Vol. 65, No. 1 (Jun., 2005), pp. 133-158


    VHM: During the next few days, I will continue to post on various aspects of the dragon / long.

  25. maidhc said,

    February 11, 2024 @ 6:34 pm

    Xie Zhaozhe's description sounds a lot like waterspouts.

  26. Victor Mair said,

    February 11, 2024 @ 6:37 pm

    atmospheric rivers?

  27. ~flow said,

    February 12, 2024 @ 2:55 am

    > Jonathan Smith said,
    > OK what wikipedia editor(s) decided Mandarin (pinyin) -ong had nucleus [ʊ]…

    I believe it's not advantageous to become overly precise / picky over this matter for the simple reason that phone{t|m}ic transcriptions in general and the IPA specifically are always somewhat broad (ITSO 'never arbitrarily precise', like a frequency distribution chart could be made), and natural language has always some regional, personal and random variation. It's not like the IPA symbols each denote a mathematical dimensionless point in the space of human language sounds; rather, they denote entire small volumes (or areas in the 2D visualization) that may sometimes overlap. To my ears [ʊ] is an adequate symbol for the vowel I hear in 龍; OTOH, [lʊŋ] is not the only possible transcription and not the final word on the typical sounds that can be heard when native speakers utter that word (for one thing one would have to talk about whether the final consonant is really a nasal *stop* as would be indicated by the symbol chosen here). Experience seems to tell me that high precision in transcription is a great thing for an in-depth discussion of the phonetics (and the phonology) of a language, but not practical or usable when we want to obtain a quotidian, pedestrian transcription, i.e. a reasonable systematic characterization of the basic building blocks that go into the articulatory and auditory phenomenon that is spoken language. We often say 'bricks' even if those are more specifically 'cinders', 'mud bricks' or 'clinkers'. That's fine.

  28. maidhc said,

    February 12, 2024 @ 4:47 am


    Victor Mair said: atmospheric rivers?

    No, I mean tornados over water. A catastrophic but well-documented phenomenon.


  29. ~flow said,

    February 12, 2024 @ 5:26 am

    > Jonathan Smith said,

    > It is largely moot anyway since languages like Mandarin have very
    > few coda consonants and thus lack the sorts of minimal contrast
    > paradigms that make vowel phonemes make sense to begin with.

    This point is very true and bears repetition: many of the tools that linguists used to use to developed the modern concept of "The Phoneme" really work best in your typical European well-behaved language, one which tells you that because "Wälder" is an inflected form of "Wald", therefore its final sound, though it comes out like [-t], is *really* *really* **in fact and truly** a /-d/, case closed. OK maybe except for the fact that when children learn to write German they always have to repeat to themselves inner dialogues along the lines of "mmmh…. ein [valt], zwei … mmmh … [veldɐ] … ah! (writes: 'Wald, Welda', but at least the 'd' is right)". In Mandarin we don't have such grammatical patterns so we have to make reasonable(?) assumptions, look for similar(?) sounds, and judge models ('phonolgical solutions') by their relative simplicity(?), beauty(?) and naturalness(?)—vague qualities all of them, none of which are hard-and-fast, and none of which would survive until lunch at a physicists' conference (except for particle physics maybe).

    I have for these reasons become very skeptical about the epistemological status (fancyish for "how can we be sure this thing is real") of phonological models, and think of them as tools—tools to think about, tools to memorize, tools to picture and describe, tools to master the sound systems of languages. Pick up another tool and see how well that one works in what parts of a given language, what description, what new or modified picture of the sound system arises from that choice. That's it; there may be *some* overlap with what native speakers think or how their brains work, but it's not a given—these are unproven claims that for the most part will have to remain unproven, and are therefore largely maybe-interesting but ultimately moot, at least the part where we make bold claims that *now* we're cooking with gas, *now* we know how man's mind ticks. There are way too many not-so-shabby but mutually incompatible models out there for many well-described languages. If two theories of facts are mutually incompatible, then we either lack the full picture, or at least one of them is wrong. Or they're just handles that give you different ways to see things.

    > This
    > means that the question of how to represent e.g. (pinyin) na nai
    > nao ne nei ni nia niao nie niu nou nu nü nüe nuo "phonemically"
    > leaves a lot of room for pedantic dicking around. Suffice to say
    > native intuition is that more than two entities are involved here.

    So I don't know what the native intutition about these syllables is, or how we can experimentally get close to that shy roe that is intuition. When you ask speakers of European languages about their mother tongue, they will stubbornly insist on answers in terms of Latin letters, and, in my experience, even suggest that entirely unnatural pronunciations are how we really should and really do speak ("ge-hen, es heißt geeeee-hänn!"—no it does not). So how do you get a taste of untarnished phonetic intuition when orthography is in the picture?

    That said, I believe the "two entities" that got mentioned by David Marjanović refer to phonolgical models of Mandarin that have two nuclear vowels, and beyond that two or three glides, some consonants, and either a zero element and a rigid syllable structure or no zero element and a variable syllable structure (these last two possibilities being good examples for models that are equally good but mutually exclusive, so personal taste will become the happy arbiter).

    Now, assume a model with a fixed syllable structure with four positions: 1234; two vowels: a high one ㄛ and a deep one ㄚ (either only possible in position 3); two glides: u, i (either or both only possible in positions 2 and 4, or both in 2); two nasals: n, ŋ (either one only allowed in position 4); and a zero element, ○ (any position). The claim is that we can write all finals of Mandarin core-system syllables—with the exclusion of some foreign words, some onomatopoetic novelties, some interjections and all of erhua (rhotacized syllables)—by merely writing exactly three symbols and by observing that if position 3 has the zero element, then either positions 2 or 4 or both must be taken by a non-zero element.

    With these rules, na nai nao ne nei ni nia niao nie niu nou nu nü nüe nuo then become

    na n○ㄚ○

    nai n○ㄚi

    nao n○ㄚu

    ne n○ㄛ○

    nei n○ㄛi

    ni ni○○ = n○○i

    nia niㄚ○

    niao niㄚu

    nie niㄛ○

    niu niㄛu

    nou n○ㄛu

    nu nu○○ = n○○u

    nü ni○u = nu○i (?)

    nüe niuㄛ○ (?)

    nuo nuㄛ○

    That's quite compelling, innit? We do have some artificial problems that are produced by the model (not the data): we have to decide whether ni○○ or n○○i or both is 'correct' (in the case ni○u vs nu○i, the former is nicer because the initial consonants will pattern with ü as they do with i, not with u). Also, the aberrant rule that i und u can *both* occupy position 2 is seemingly solely needed for nüe (at which point we also note that there's no PY *nüei, *nüeu, *nün, *nüng—coincidence??). Some of these problems vanish when you delete the zero element from the notations, some remain.

    I can promise that the two-vowel model gets better when a more systematic and complete survey of Mandarin syllables is done, and that some open questions are not easily answered (such as: is ü elementary in Mandarin or a compound i+u?—it's complicated). As sketched out above, the model is rough and only so-so (plausible and elucidating), but I believe it at least can explain what phonologists were thinking when they created it.

    Coming back to native intutitions, Chinese has a famous and thorny phonetic / phonological tradition on its own, starting at least with the rhyme books but really predated by the 'sounds-like' glosses in the Shuowen-Jiezi (100 CE). It is thorny because it's hard to understand, not only for stupid Westerners but also for wise Easterners (who didn't hesitate to label this branch of knowledge 苦学, the Bitter Teaching, a pun on 訓詁学?). It is also informed by Indian teachings and Yinyang theory, so how much of an untarnished document of the Chinese psyche's feelings towards language sounds it reveals is debatable.

    But I can reveal an interesting, relevant historical tidbit: above I've chosen the symbols /ㄛ/ for the Yin and /ㄚ/ for the Yang vowel, mainly to make clear that these are not [e] and [a] as heard in Italian. They both comprise quite a range of clearly discernible qualities; for /ㄚ/ those include [a, ɑ, æ] and, for /ㄛ/, [ɛ, ɤ, ə] plus, outrageously, [ɔ]. Pinyin has no qualms to write 'a' in both 'lang' and 'lian', and 'e' in both 'lei' and 'le' (as different as BrE. 'way' and 'were'); but 'e' for the nuclear sound of 'nuo?'—"nue", really? That, at the time, was deemed too foreign, apparently, although one language reformer famously asserted that "when Italians can use 'ci' for -tsh- and Hungarians can use 's' for -sh-, then of course we Chinese have a right to use q, x, c as we see fit" (quoted from memory).

    Turns out the alphabet that ㄛ and ㄚ are sourced from, the Zhuyinfuhao, had *just a single letter* for all of [ɛ, ɤ, ə] and, outrageously, [ɔ], at least until 1920, when ㄛ was relegated to the unrounded [ɛ, ɤ, ə] and a new letter, ㄮ, later changed to ㄜ, was to take up the responsibility to represent the rounded [ɔ]. If that doesn't convince everyone that the Two Vowel Theory for Mandarin is what native speakers *really* think about their language, I don't know what will.


    Introduction of ㄜ mentioned here: https://en.wiktionary.org/wiki/%E3%84%9C

  30. Lasius said,

    February 12, 2024 @ 5:29 am

    @Doctor Science

    Because it's weird, when you think about it, that the Chinese zodiac has 11 perfectly mundane and real animals … plus the dragon.

    I don't know how old the zodiac associations are, but at some point the dragon might have been a perfectly normal animal as well.


  31. James Wimberley said,

    February 12, 2024 @ 6:27 am

    Nobody so far has mentioned the obvious theory that dragons are dinosaurs. China has IIRC several regions rich in dinosaur fossils. Some species had impressive fangs, others horns, others wings, and yet others long tails – though none had all of them. The pop eyes and whiskers would not have survived, though there is increasing evidence for feathers, absent from the mythical contemporary ones. All we need to do is imagine Bronze Age miners with lively imaginations and luck. Perhaps earlier, as Neolithic humans dug mines for high-quality flint.

  32. Lasius said,

    February 12, 2024 @ 7:39 am

    @James Wimberley

    Very unlikely. Even if neolithic Chinese people would have found Dinosaur fossils they wouldn't have been able to properly reconstruct them a giant reptiles. Few fossils are complete enough to do that.

    See here for similar discussion:


  33. ~flow said,

    February 12, 2024 @ 8:48 am

    I want to add some thoughts: In connection to the theory that Chinese dragons were inspired by dinosaur fossil finds:

    The Chinese of later ages are great admirers and collectors of rare and strange artifacts and formations; do we know of any collection with relevant fossils?

    As for whether and how the Chinese interacted with fossils, Wikipedia (https://en.wikipedia.org/wiki/Oracle_bone) has the following to say: "During the 19th century, villagers in the area who were digging in the fields discovered a number of bones, and used them as "dragon bones" (Chinese: 龍骨; pinyin: lóng gǔ), a reference to the traditional Chinese medicine practice of grinding up Pleistocene fossils into tonics or poultices." The pleistocene is roughly the past 2,5 million years so no dinosaurs, but at least fossils that got connected to dragons. Of course, this being Wikipedia and an article about Oracle bones, not fossils in China or dragons per se, nothing tells us that they didn't hit upon the occasional fossil from much earlier times.

    It has been theorized that people in Greece and places got their concept of a cyclops from fossils, more precisely, the fossilized skulls of elephants and or mammoths; these are recognizable as skulls, huge in size, and have a hole in the forehead; of course, the mammoths and elephants in the European part of the Mediterranean died *much* later (ten-ish thousand years and less) than the dinosaurs (over 66mya, factor of over a thousand).

    What the Chinese certainly did have living around them or still in cultural memory was big animals, including snakes and maybe knowledge about reptiles like monitor lizards. But ancient depictions of long creatures (read that as you will) looked nothing like lizards, they look like snakes.

    There were similar theories in the 1970s that the European dragon is a reminiscence of dinosaurs, and that maybe early man had encounters with the last exemplars of these giants that had manged to hold out in what today we would call remote areas. But the lack of fossil records, the violence of the Chicxulub impact and the distance of well over 60my put this theory to rest I'm afraid. Are there any stories, myths, physical remains of dinosaurs in Europe, Mesopotamia or Egypt? I know of zero.

    One key stone of the argument from the 70s that maybe, just maybe the appearance of the dragon as a mythical animal in cultures 'around the world' may be due to a 'cultural imprint' from encounters of early man with late dinosaurs, but it looks to me now that if the Chinese and the European dragon have one thing in common is their late appearance in the history of mankind (less than 2kya), and their possible relatedness through intercultural exchange (the 龍学生 リュウガクセイ of old?).

    Do we have any depiction of dragons that look like huge lizards from other regions like Mesoamerica? Because if the dragon is theorized to take, in its modern form, its inspiration from living or dead dinosaurs, shouldn't we expect similar finds from over the world to have inspired similar myths?

    Oh, and one last thing. "China has IIRC several regions rich in dinosaur fossils" yes, and surprisingly rich depots they have, to the point that digging out a skeleton means some guys from the black market will visit you, it's become an entire industry in its own right. BUT that's fairly recent, isn't it?

  34. Chris Button said,

    February 12, 2024 @ 9:24 am

    To my knowledge, no-one has ever postulated /o/ as the phoneme in question. If they had, it would probably have resulted from association with pinyin spelling. [ʊ], or as it used to be written [ɷ] (e.g. in Y.R. Chao's time), is simply the phonetic representation for the standard "default" speaker and continues to be used today.

    I've always liked Taiwan's bopomofo too. My teacher at college was from Taiwan and exposed me to it right away.

  35. KeithB said,

    February 12, 2024 @ 9:40 am

    I find it interesting that Elliot, in _Pete's Dragon_ morphed from the clearly European (though goofy) dragon in the animated version to the much more Chinese dragon in the "live" [CGI] version.

  36. 2047 said,

    February 12, 2024 @ 10:06 am

    @ Doctor Science

    There are two accounts mentioned in Strange Tales from a Chinese Studio about dragon. One is also happened during the voyage to Okinawa.

  37. Victor Mair said,

    February 12, 2024 @ 1:51 pm


    I know about "waterspouts"

    I was just making a topical quip about the atmospheric rivers that have been afflicting California — as described by several of my West Coast friends who suffered at their expense.

  38. Jonathan Smith said,

    February 13, 2024 @ 10:06 am

    Re: pronunciation, well we can agree anyway that "/lʊŋ³⁵/" and similar are intended as phonetic representations despite "/" — any phonemic-ish system is going to identify the vowel in 'dragon' with the one in (pinyin) lu2 'stove'… or OK the one in lou2 'building' if one wants to be different. Maybe the latter choice is better…

    Phonetically, while who cares of course, [ʊ] remains a weird choice here IMO — kinda the whole point of "[ʊ]" is "not as high/back as [u]". But I guess it depends on certain unstated habits/preferences regarding how to deploy I(Phonetic~Phonemic)A.

    Basically agree with @flow's thoughts; there are definitely *aspects* of a two-main-vowel treatment of Mandarin that are appealing. But in e.g. Zhuyin, we find of course many simplex symbols alternating (that is used "minimally contrastively") after say consonant n-. It is thus a bit awkward to suggest that e.g. ni / na differ in some fundamental structural way.

  39. Adrienne Mayor said,

    February 13, 2024 @ 1:24 pm

    British paleontologist Kenneth Oakley (1965 and 1975) noted that some features of the traditional Chinese dragon resemble the shapes of Pliocene/Pleistocene animals, for example, the distinctive antlers of the dragon match those of an extinct deer.

    Note that dragons in all cultures are composite creatures and can arise from the storytelling imagination–no fossils are required to imagine a hybrid monster or fabulous animal, such as a dragon or a beaked quadruped. But once the story and description is known, then discoveries of fossils of remarkable size or shape might be explained by reference to the story of a dragon or griffin. We can never know which came first, the story or the observation of fossils.

    Also note that the dinosaur skeletons in the Gobi desert are some of the most exquisitely preserved fossils in the world, usually fully articulated and often eroding out in a standing position. They are not broken and jumbled like the Miocene-Pleistocene fossils of the Mediterranean region. And nests and eggs are conspicuous.

  40. Philip Taylor said,

    February 13, 2024 @ 2:31 pm

    Returning, if I may, to the very first comment :

    The [archaeology] team said it resembles a cobra, which in ancient times served as "a tool for humans to communicate with divine and ancestral spirits."

    would it be safe to assume that "communicat[ion] with divine and ancestral spirits" was accomplished by the simple expedient of allowing (or encouraging) the cobra to bite the enlightenment seeker ?

  41. Chris Button said,

    February 13, 2024 @ 3:52 pm

    Regarding a vertical vowel system for Mandarin, it is probably worth noting that such a system is a phonological entity rather than a (broad or narrowly transcribed) phonetic entity that will need a triangular vowel system of sorts.

    But rather than just existing as an artificial abstraction of little real worth, a vertical vowel system can influence how humans categorize syllables. Cross-linguistic comparisons of rhyming are testimony to that. In my opinion, any analysis of Old Chinese is essentially impossible without acknowledging it.

  42. /df said,

    February 17, 2024 @ 9:26 pm

    To elaborate on the point mentioned by @~flow, one might indeed consider that ancient inhabitants of the Far East were either familiar with or received reports of very large monitor lizards like the Komodo "dragon" and especially the crocodile monitor. From the linked WP article: 'In New Guinea, the lizard is … described as an evil spirit that "climbs trees, walks upright, breathes fire, and kills men'. This reptile has a characteristic long tail that it coils (horizontally on the ground) when standing on its hind feet for a better view. Somewhat surprisingly, people (with room for their own reptile house) keep these as pets, and, less surprisingly, your striking infant pet will quickly mature into an unapproachable predator (see also tigers).

    Or traditions of contacts with Megalania, a sort of triple-size Komodo dragon that failed to adapt to the arrival of homo sapiens in Australia, may have survived the millennia.

    Western dragons of the classical era, like top δράκων Typhon, lack clawed feet, and are more like the sort of monster that might emerge from a volcanic burrow. Apparently the same applies to African dragons despite the disturbing prevalence of crocodiles in the continent.

    I would say that this supports the idea that Chinese dragons flew West, along with the griffin/griffon/gryphon/protoceratops.

  43. Philip Taylor said,

    February 18, 2024 @ 9:02 am

    Sadly the intended hyperlink to http://markwitton-com.blogspot.com/2016/04/why-protoceratops-almost-certainly.html in the immediately preceding comment does not function as intended in neither of my two primary browsers (Seamonkey, Firefox) and also fails in Microsoft Edge. The code reads :

    <p><a href="http://markwitton-com.blogspot.com/2016/04/why-protoceratops-almost-certainly.html" rel="nofollow ugc">http://markwitton-com.blogspot.com/2016/04/why-protoceratops-almost-certainly.html</a></p&gt;

  44. Victor Mair said,

    February 18, 2024 @ 9:19 am

    re: previous comment by Philip Taylor

    "does not function as intended in neither of my two primary browsers"

    I would say "does not function as intended in either of my two primary browsers"

  45. Philip Taylor said,

    February 18, 2024 @ 10:11 am

    So would I, normally, but unfortunately I amended my earlier version when I tested in MS Edge, and the error crept in during the amendment — mea culpa, mea culpa, mea maxima culpa.

  46. /df said,

    February 18, 2024 @ 10:52 am

    My link, which I assume is that mentioned, was to a very gr****n-esque image of a bit of protoceratops at WP, because, inter alia, I hadn't read the Witton article at that time. (And the link works fine for me in SeaMonkey)

    The Witton article is an interesting counterpoint but the concept of ancient chimera generation presented seems to be as if a court artist was responsible rather some imaginative guys (most likely) at the far end of a graphical version of Chinese (yes) whispers. To that extent one might admit that the astonishing gr****n-esqueness of the linked pic could be coincidental.

  47. Vampyricon said,

    February 20, 2024 @ 6:59 pm

    @Finistis, I believe the phoenix has exerted enough memetic pressure on the Chinese equivalent that they can't really be considered separate anymore. At least, that's what I've heard.

    In other news, it's always something of a pet peeve of mine to see "Middle Chinese" equated with the THS (tshet hjunH system). I believe only a minority believe the THS reflects a real language, so if "Middle Chinese" (or "Middle Sinitic", as per Dr. Mair) is to be analogous to other Middle Languages, it will have to represent language(s) spoken circa 600 AD. The only such reconstruction I know of is W. South Coblin's reconstruction of the Sui-Tang Chang-An dialect, in which our dragon is reconstructed as */luoŋ/.

  48. Victor Mair said,

    February 23, 2024 @ 8:48 pm

    Should have added this long ago:

    How (exactly) to slay a dragon in Indo-European? PIE *bheid-{h₃ég w him, k w ŕmi-}

    Benjamin Slade
    Historische Sprachforschung / Historical Linguistics
    Bd. 121 (2008), pp. 3-53 (51 pages)


    In this paper I present evidence for a formula associated with the Indo-European dragon-slaying myth, Proto-Indo-European [PIE]*bheid-{h₃égwhim, kwŕmi-} 'split serpent/worm'. This formula is derived via an examination of the verbal collocations which frequently occur in the context of the Vedic dragon-combat; these involve not only √han- 'slay', but also the semantically more specific verbs √bhid- 'split', √vraśc- 'tear, cut, split', and √ruj- 'break'. Not only are these latter three verbs employed in describing the dragon-slaying itself, but they also often appear describing actions linked to the dragon-combat (e.g. the releasing of the waters/cows), and in both cases co-occur with forms of √han-. Vedic is found to provide robust evidence for the reconstruction of PIE *bheid- {h₃égwhim, kwŕmi-}, which is supported by data from Iranian and Germanic. Though not as widely distributed as PIE *gwhen- h₃égwhim 'slay serpent' (attested for instance in Vedic áhann áhim '(he) slew the serpent') – a formula discussed in great detail by Watkins (1987, 1995) – *bheid- {h₃égwhim, kwŕmi-} 'split serpent/worm' is semantically more specific, and therefore more distinctive, than *gwhen- h₃égwhim, thus lending additional support for Watkins' thesis that there exists a distinctively Indo-European dragon-slaying myth, and serving to further characterise the nature of that myth.

RSS feed for comments on this post