Of knots, pimples, and Sinitic reconstructions

« previous post | next post »

A couple of months ago, we talked about gēda 疙瘩, which is one of those very cool, two syllable Sinitic words, neither of whose syllables means anything by itself (i.e., not only is it a disyllabic lexeme, it is also a disyllabic morpheme).  Furthermore, gēda 疙瘩 is highly polysemous, with the following meanings:  "pimple; knot; swelling on the skin; lump; nodule; blotch; a knot in one's body or heart (–> hangup; problem; preoccupation)".

See "Too hard to translate soup" (9/2/18).

It has always been my impression that, when you have such disyllabic morphemes / lexemes that are widely current in Sinitic, some of them stretching back to BC times, chances are good that they may have entered the lexicon from an external or local source (i.e., they are not, at least in the beginning, mainstream, elite terminology, and may not even originally have been Sinitic). Moreover, they are often the result of dimidiation (of complex phonological clusters).  However, since these words tend to be more on the vernacular or colloquial side of language, they are not well documented in writing.

The primacy of their phonology as lexically and morphemically disyllabic is further borne out if they have variant writings, which is often the case, as with gēda 疙瘩, e.g., 疙垯, 疙墶, 疙搭, 紇噠, and many other forms — at least twenty (see below).  As I often tell my students, sound is more important than symbol (characters) in Chinese, whereas — because they are so complicated and visually captivating — most people think just the opposite, that characters are primary and sounds are secondary.  Nothing could be further from the truth.

After reading the "Too hard to translate soup" post, Bob Ramsey wrote to me:

What's most interesting to me is what you started the chain with, that is, the observation that geda is one of those colloquial, two-syllable Mandarin words that is clearly a single morpheme that can't be easily broken down etymologically into separate syllables. Analytically, it looks pretty opaque. But what I thought was still more interesting is that, in various forms, it's found in the colloquial speech of (most of?) the other Sinitic languages. That means it isn't some neologism peculiar to Mandarin. I wonder how old it is, though. Any attestations to be found in older written records (maybe not even written with the same characters)?

"Geda" is very widespread in the Mandarin group of topolects.  The following massive lexicon lists occurrences of "gēda tāng 疙瘩汤" ("dough drop soup") or variations thereof in scores of places all across the vast range in which types of Mandarin are spoken:  Chén Zhāngtài 陈章太 and Lǐ Xíngjiàn 李行健, ed., Pǔtōnghuà jīchǔ fāngyán jīběn cíhuì jí 普通话基础方言基本词汇集 (Collection of the Foundational Vocabulary of Mandarin Basic Topolects) (Beijing:  Yuwen chubanshe, 1996), 5 vols., vol. 4, p. 2930.

"Geda" is found in Cantonese, where it is pronounced gat6daap3 and means "lump; swelling; bump".  It's possible that it was borrowed into Cantonese from Mandarin, and I honestly don't know how widespread "geda" is in the other Sinitic languages.

Although I suspect that "geda" could have been circulating in the spoken language centuries before, the earliest written attestations of the word occur in two medical texts of the Jurchen Jin Dynasty (1115-1234), one dating to 1174, where the word signifies "pimple" or "lump / swelling of the flesh".  Thereafter, from the Mongol Yuan Dynasty (1271-1368) onward, "geda" is widely used in drama, fiction, and other types of vernacular writing.  Moreover, in addition to the physical meanings, it embraces a number of extended, metaphorical connotations, such as "a knot in one's body or heart (–> hangup; problem; preoccupation); complexity", etc.  See Wáng Jīng 王晶, "'Gēda' de cíyì hé yòngfǎ kǎochá '疙瘩'的词义和用法考察" ("An investigation on the meaning and use of 'geda'"), Guìzhōu shīfàn dàxué xuébào (shèhuì kēxué bǎn) 贵州师范大学学报(社会科学) (Journal of Guizhou Normal University [Social Sciences]), 3, cum. 146 (2007), 101-104.

Here I wish to add the observation that medical texts are often where vernacular terms first show up in the Sinitic written record.  The reason this is so is that medical practitioners were in close touch with illiterate citizens who had real physical problems for which the specialists and the patients had terms to identify and describe them that may originally have existed strictly in the oral realm, conceivably for quite some time before someone finally felt the need to write them down.  The other main source of vernacular vocabulary in Sinitic is Buddhist tales and scriptures.  Zhū Qìngzhī 朱庆之 and I have been working on a large dictionary of Middle Vernacular Sinitic for more than two decades.  When it is finally finished in a couple of years (we hope), the sources of early written vernacular will be clearly and fully documented for the first time.

A representative gathering of citations of "geda" through successive periods starting from the Mongol Yuan may be found in Hànyǔ dà cídiǎn 漢語大詞典 (Unabridged dictionary of Sinitic).  Because this morpheme / lexeme is written in so many ways employing a plethora of characters, the entries are spread around in different places in this huge 12 volume dictionary, but they are all brought together for easy comparison in Victor H. Mair, An Alphabetical Index to the Hanyu Da Cidian (Honolulu:  University of Hawaii Press; Shanghai:  Hanyu Da Cidian chubanshe, 2003), p. 339cd. Here we see that "geda" is written twenty different ways.  Moreover, the semantic classifiers (semantophores, radicals) used for the syllables range from "sickness; illness" to "textile; fabric", "earth", "leather", "flesh", "food", "hill", "hand", and "mouth" — the latter two being used to indicate sound, not meaning.

A recently published multivolume dictionary of premodern vernacular terms is Bái Wéiguó 白维国, ed., Jìndài Hànyǔ cídiǎn 近代漢語詞典 (Dictionary of Recent Sinitic), 4 vols. (Shanghai:  Shanghai jiaoyu chubanshe, 2015).  In vol. 1, pp. 604b-606, it lists scores of different ways of writing "geda", provides precise definitions indicating subtle variations in usage, and gives textual citations illustrating each particular form and meaning.  The takeaway from all this generous data is that "geda" exploded into the vernacular lexicon during the Mongol Yuan Dynasty in the late 13th to mid-14th centuries, became solidly rooted during the 15th and 16th centuries, and an essential part of Mandarin vocabulary from the 17th century up to the present.  As indicated above, along the way "geda" — which began as a reference to a lump or swelling of the flesh or a pimple, boil, etc. on the skin — took on increasingly abstract and metaphorical connotations.  Judging from all of the physical manifestations of "geda" early on, its basic meaning must have been LUMP, CLUMP, KNOT, NODE, NODULE, KNOB.

Now, looking more closely at the sounds of "geda", I cannot find any Middle Sinitic (MS, circa 600 AD) reconstructions of 疙瘩 (with the "sickness; illness" semantophore), so I'll start with Cantonese (Jyutping) gat6 daap3, Min Nan (POJ) kit-tap, and Wu (Wiktionary) keq taq (T4).

Next, let us look at the MS reconstructions of some other characters that have been used to write "geda" and extrapolate from there.

紇噠 is a variant orthographical form of 疙瘩.  The MS reconstruction for 紇 would be ɦet̚ (Zhengzhang Shangfang) or ɣət̚ (most other reconstructions).  The MS for 噠 (judging from that for 達) would be dɑt̚.  Together that yields ɦet̚ / ɣət̚ dɑt̚.  But the regular MS reconstruction of the phonophore of the second character of 疙瘩 would be something like tăp̚ (Karlgren; there are slight differences of opinion concerning the exact quality of the vowel).

"Gēda 疙瘩" reminds me of another disyllabic Sinitic morpheme, gūlu 軲轆 ("wheel"), which also has a number of variant forms, e.g., 轂轆 軲轤.  For our present investigation on "geda", it is suggestive that "gulu" has a relationship with Tibetan འཁོར་ལོ ('khor lo, “wheel”) and Proto-Indo-European *kʷékʷlos, as proposed by Robert S. Bauer, “Sino-Tibetan *kolo 'Wheel',” Sino-Platonic Papers 47 (Aug. 1994), 1-11 (free pdf); see also Wiktionary.

My favorite word of this sort, gāngà 尴尬 (simpl.) 尷尬 (trad.) / 尲尬 ("awkward"), is discussed at length in "GA" (8/6/17).  See also Hànyǔ dà cídiǎn 漢語大詞典 (Unabridged dictionary of Sinitic), 2.1581b and 12.476a.

Given what we know about the origins of gūlu 軲轆 ("wheel") and gāngà 尷尬 / 尲尬 ("awkward"), I think we are justified in looking for a non-Sinitic or local source of gēda 疙瘩 ("lump; clump; knot; knob; node; nodule; pimple")

My first thought is that gēda 疙瘩 may ultimately have derived from Proto-Indo-European *gnod- (to bind), compare Latin nōdus and its Romance descendants or PIE *ned- (see J. P. Mallory and Douglas Q. Adams, The Oxford Introduction to Proto-Indo-European and the Proto-Indo-European World [2006]), with cognates in Celtic, Italic, Germanic and Indo-Iranian.  As for how IE could have impacted Sinitic, it is clear that IE languages have been interacting with Sinitic since the Bronze Age, and we have seen plentiful evidence for that in the Readings listed at the bottom of this post and elsewhere.

Looking at the section for "knot" (9.192) in Carl Darling Buck, A Dictionary of Selected Synonyms in the Principal Indo-European Languages:  A Contribution to the History of Ideas (Chicago:  The University of Chicago Press, 1949), p. 551, I see that most words in this group begin with kn- or just n- and end with a dental, -t or -d — e.g., knot, nodus, etc., but there's a small group of languages (Dutch, OHG and MHG) that have words that begin in kn- and end in a labial, -p or -pf — e.g., knoop, knopf, and cf. English "knob".

Don Ringe cautions:

There is no reconstructable root *gnod-, and there is no evidence that Lat. nōdus ever had an initial *g-.  The Lat. word is usually connected with Old Irish nascaid ‘(s)he binds’ (*nad-ske/o-), verbal noun naidm, and with Goth. nati, OE nett, etc.; a more or less exact cognate is Old Norse nót 'net'.  Since gn- survives in Irish and becomes kn- in Germanic, this whole cognate set is incompatible with a PIE initial gn-.  The main reasons for preferring this connection to the other are (1) the ON word and (2) the fact that we get broader support for a PIE root.

Although I am not necessarily advocating a connection with IE words for "knot, knob, node, nodule", etc., I'm not completely abandoning that possibility either.

Melanie Malzahn states:

'Knot' seems to belong to a large Group meaning knotty / knobby / protuberant things and this group is usually said to be Germanic only and derived ultimately from the  word for 'knee'.

Here is the entry from Kluge, Etymologisches Wörterbuch der deutschen Sprache (25. Auflage), which is the Standard reference work, and a recent study on such n-stems by Guus Kroonen, The Proto-Germanic n-stems : A study in diachronic morphophonology, Leiden/Brill 2011.

Kluge, Friedrich

Etymologisches Wörterbuch der deutschen Sprache 2012


Sm std. (8. Jh.), mhd. knote, knode, ahd. knoto, knodo, as. knotto ‛ Knoten an Fäden usw., Verdickung an Halmen usw.’, mndd. knutte, (geminiert)

Ebenfalls die geminierte Form zeigt ae. cnotta (ne. knot), zu dem ae. cnyttan ‛stricken’ (ne. knit), ndd. knütten ‛Netze knüpfen’ gehört. Als dritte Variante gehört hierher anord. knútr ‛ Knoten ’, anord. knúta f. ‛Knöchel’; also nebeneinander *knuþ-/knud-, *knut(t)- und *knūt-. Die Gruppe kann in den Bereich ‛verdickte Gegenstände mit Anlaut kn- ’ gehören (Knolle); es kann sich aber auch um eine Erweiterung zu ig. *ǵenu- ‛Knie’ handeln, vgl. gr. góny n. ‛Knie, Gelenk, Knoten an Halmen’; entsprechend knobeln und Knochen, Knöchel. Knoten als seemännisches Maß der Schiffsgeschwindigkeit ist entlehnt aus ne. knot und bezieht sich auf die Zahl der Knoten , die in  einer bestimmten Zeit von der Logleine abgelaufen sind. Verb:


Verweis: Knopf, Knüttel.


Schüwer, H. Niederdeutsches Wort 17 (1977), 115-123; Google Scholar

knoten ; Adjektiv: knotig.

Lühr (1988), 281f. (anders); Google Scholar

Röhrich 2 (1992), 860f. Google Scholar

*knupo, *knuttaz 'knot'

  • *knufia(n)-: Icel. hnudi, -ur m. 'knob, hump'1134
  • *knutton-: Icel. hnuta, Far. knuta f. 'bone'1135
  • *knutta-\ ON knutr m. 'knot, knag', Icel. hnutur m. 'knot'1136,Far. knutur m. 'knot, lump'1137
  • *knuttan-: Icel. hnotti m. 'tussock, ball'1138 (^ hnjota 'tostumble' ^ hnjoti, -ur m. 'bump'1139], MLG knutte m. 'knot (of flax]', MDu. knutte m. 'knot of flax', OE cnotta m. 'knot'^ *knuttjan-: OE cnyttan w.v. 'knot', E knit
  • *knufran-, -on-: Icel. hnodi m., hnoda n. 'ball, clew'1140, OHG chnodo m. 'knuckle', Swi. Ja. xnodd1141, Visp. xnodo1142 m. 'id.'
  • *knufifian-: G Cimb. knotto m. 'rock', EDu. knodde 'nodus,nexus'
  • *knuton-: Icel. hnota 'clew, vertebra', Far. knota f. 'bone'
  • *knudan-: OHG chnoto m. 'knot', G Knoten 'id.'
  • *knattu-: ON knǫttr m. 'ball, knob'

Most of the material points to a paradigm *knufio, *knuttaz, *knudeni, which seems to be derived from PIE *gnu- with the same ton-suffix that must be reconstructed for e.g. *klfyo, *klittaz 'burdock' (p. 235] and *klufio, *kluttaz (p. 272], The original nominative *knufio is directly continued by Icel. hnudi 'knob', the genitive *knuttaz by Icel. hnotti 'tussock, ball' and OE cnotta 'knot'. This original genitive was replaced by *knuppaz in a secondary paradigm that underlies Cimb. knotto 'rock' and EDu. knodde 'node'.

Finally, OHG chnoto 'knuckle' seems to preserve the consonantism of the locative *knudeni.

Fully parallel to other *,gnu-derivatives, the paradigm of *knufio, *knuttaz further seems to have contained an a-graded root variant, viz.

1134 Boðvarsson 393. 1135 Poulsen 609.
1136 Boðvarsson 394. 1137 Poulsen 610. H38 Boðvarsson 393. 1139Boðvarsson 392. "4° Ibid.

" « Stucki 70. 1142wipf 41.

As a counter to my hypothesis that "geda" might somehow be related to IE, I queried whether it might come from some Altaic language.  So far that avenue doesn't look promising.

Juha Janhunen:

If the basic meaning is 'lump' etc. I cannot identify any possible source in either Mongolic or Tungusic. I will continue thinking about it. – Tungusic has the word gida / geda (Manchu gida), meaning 'spear'.

Christopher Atwood:

Mongolian words for “pimple”: Kh. güwdrüü/göwdrüü (< UM. gübdürüü). But there’s lots of other words, such as güwee which is “hillock” but idiomatically also pimple, and (Kh.) batga (< UM. badq-a). Inner Mongolian dictionaries give for geda 疙瘩 the following equivalents:  bilcau, bilcuu, onci, bulduruu, gübdürüü, and adaruu all with various nuances.


P.S.:  As the next installment in this series, I am planning a Yuletide surprise in December.

[Thanks to Doug Adams, Jim Mallory, Don Ringe, Wolfang Behr, and Dotno Pount]


  1. AG said,

    November 12, 2018 @ 6:31 pm

    wait… would this be the same root as gnocchi, Knödel, and noodle?

  2. Victor Mair said,

    November 12, 2018 @ 6:34 pm

    I asked Nikita Kuzmin to check this out in Tangut and Tibetan. Here are his findings (he presented the data to me in neat charts, but I cannot maintain them in the comment box here):

    I checked the term "geda" in Tibetan. It is impossible to trace this word in Tangut, since there is no "English – Tangut" or "Mandarin-Tangut" dictionaries, but only "Tangut-Mandarin/English".

    I checked all the terms: "knot", "node", "lump", "clump" in Tibetan, but there is nothing that sounds similar to "geda". Below you can see some dictionary entries:


    "knot" (Pronunciation – "dü pa")


    JH-ENG, JV, IW, RY knot

    JH-SKT {MSA}grantha; {MSA}pAMzu

    OT [1379] 1. (tha mi dad pa) mdud pa 'debs pa'am/ rgyag pa/ … thag pa chad pa de rnams mdud pa mdud nas mthud pa/ … 2. mdud pa brgyab pa'i ming/ … skud par mdud pa rgyag pa/ … bal thag de gnyis mdud pa bcings nas mthud pa/ … lus nang gi rtsa mdud/ … ser sna'i mdud pa/ … ol mdud/ … sbrul mdud sbrul bshig …

    DM Skt. grantha. 'ties' (a particular state of, or synonym for, the klezas). M.Vy. 2145.

    IW (Tha mi dad pa),, ; 1) [tie a] *, connection of strands/ nadis; 2) tie together. *, bond

    RY {mdud pa, mdud pa, mdud pa, mdud} intr. v.; 1) *. 2) bond

    "lump" (Pronunciation – dok-dok)


    OT [1447] 1) snying po/ … don dag rdog rdog snying por 'dril thub pa dgos/ … 2) ril ril/ … gser rdog rdog gcig … spags rdog rdog gcig … 3) lham gyi sna mgo'i yar tsam gyi phar phyin/ …

    JV piece, block, chunk, concrete, lump, nugget, solid

    IW 1) essence; 2) [small solid] pieces/ lumps [Gser rdog rdog gcig…Spags rdog rdog gcig]; 3) Lham gyi sna mgo'i yar tsam gyi phar phyin). 1) snying po; 2) ril ril [small solid] pieces/ lumps; 3) lham gyi sna mgo'i yar tsam gyi phar phyin solid, bulky, bump/ rise on a flat surface, straightforward [statements/ speech]

    RY small solid pieces

  3. David Marjanović said,

    November 12, 2018 @ 7:04 pm

    Given the dates and places of first attestation, I would definitely look in Tungusic and Mongolic first, but while the consonants fit Mongolic well enough, the vowels really don't. Assuming the word was used in northern Sinitic for several hundred years before it was first written down, do Iranian, Tocharian or Tangut have anything to offer?

    "Gēda 疙瘩" reminds me of another disyllabic Sinitic morpheme, gūlu 軲轆 ("wheel"), which also has a number of variant forms, e.g., 轂轆 軲轤. For our present investigation on "geda", it is suggestive that "gulu" has a relationship with Tibetan འཁོར་ལོ ('khor lo, “wheel”) and Proto-Indo-European *kʷékʷlos, as proposed by Robert S. Bauer, “Sino-Tibetan *kolo 'Wheel',” Sino-Platonic Papers 47 (Aug. 1994), 1-11 (free pdf); see also Wiktionary.

    Two potential problems:

    1) Bauer's "*kolo" is not an actual reconstruction arrived at by the comparative method; such a thing is not yet possible, because many of the sound correspondences between the branches of Sino-Tibetan (Transhimalayan, Tibeto-Burman, whatever) simply haven't been worked out yet. (As an example, this paper presents, near its end, a newly discovered correspondence between one vowel feature of just two branches; that's a major breakthrough and only came out this year.)

    2) posted separately to get past the spam filter

  4. David Marjanović said,

    November 12, 2018 @ 7:06 pm

    2) Vaguely [kul]-like words are globally widespread as onomatopoeia for rolling. German kullern "to roll downhill in a funny, harmless way" (said of small round objects) is nowhere near cognate with PIE *kʷel-. Start here for a taste of it – for starters, some kind of "*qwəl-" with roundish meanings is all over the Pacific Northwest, where wheels were altogether unknown until Lewis & Clark or thereabouts.

    In other words, such words are no more reliable than finding that cuckoos tend to be called "cuckoo" – that English word, or its vowels at least, was remade after the Great Vowel Shift; it's not cognate with German Kuckuck (final -ck irregularly missing in various dialects).

  5. Victor Mair said,

    November 12, 2018 @ 8:30 pm

    From David Dettmann:

    I don't have any ideas on the origins of where that word "geda" came from, but other references in my culinary encyclopedia with that word for "roundish" vegetables (疙瘩菜 or 疙瘩白), indicated those names are used in the Northeast (东北).

  6. Jonathan Smith said,

    November 12, 2018 @ 8:34 pm

    "As an example, this paper presents, near its end, a newly discovered correspondence between one vowel feature of just two branches; that's a major breakthrough and only came out this year"

    Gong Xun's paper (which has not come out) is about Rgyalrong Q- : Tangut K- "Grade 1", etc.; is that what you mean? That is (apparently) one branch, and a consonant feature. Or do you mean Tangut -j- : Old Chinese *-j- (= "Type B") from Gong Hwang-Cherng (2007: 450) (four examples)? That is not newly discovered, and at this point can't be said to constitute a "major breakthrough". As for the further possibility of a connection between the former correspondence and the latter one… debatable and framed in such cautious terms by the author for good reason.

  7. Victor Mair said,

    November 12, 2018 @ 9:03 pm

    From Douglas Adams:

    The best I can do for 'knot' in Tocharian is TchB meske, also 'joint, chain' (as [Chinese] cash knotted on a string), the exact phonological cognate of Lithuanian ma~zgas 'knot.' If there were a cognate of knot in Tocharian, I would expect it to begin with kn- in TchA and n- in TchB (cf. knaa- 'know' in A and naa(n)- in B).

  8. Rodger C said,

    November 13, 2018 @ 7:59 am

    ON kngttr m. 'ball, knob'

    Surely you mean knǫttr.

    [VHM: fixed typo]

  9. Andy said,

    November 13, 2018 @ 10:30 am

    I've no idea if it could be relevant, or if it can be reconciled with the Chinese forms, but there is 'gaNDa' in Sanskrit (the capitals denote retroflexes), which refers to the cheek and various swellings and excrescences. No thoughts on the etymology though.

    I can't seem to link to the actual entry, but here is the dictionary:

  10. Chris Button said,

    November 13, 2018 @ 12:57 pm

    I wonder if Mongolian хавдах (havdah) "swell" has anything to do with it?

  11. Victor Mair said,

    November 14, 2018 @ 12:16 am

    From Timothy May:

    From Lessing’s Mongolian dictionary:

    Gede means nape or the back of the neck.

    Gedes means bumpy or uneven.

    Gedeng can also mean bumpy or uneven.

    Göbdürigüü means elevation, wart, blister, pimple.

    Göbdürigüde- measn to have pimples.

    Gübegegen means hillock, knoll, mound

    Güber-e is a bruise or welt caused by lashing.

    So I think Gedes or Göbürigüü might be what you are looking for.

  12. Chris Button said,

    November 14, 2018 @ 10:41 am

    I'm hoping that someone with some knowledge of Proto/Old Mongolian will chime in to provide some examples of how some possible comparanda might have sounded around that time to compare with the Middle Chinese forms. Comparing modern Mandarin with modern Mongolian is unfortunately unlikely to yield much.

    One thing of interest is the vacillation between the voiced velar fricative /ɣ/ (presumably going back to earlier /g/) and the velar nasal /ŋ/ in the Middle Chinese forms of 紇/疙. Given that pre-nasalization is an articulatory mechanism for maintaining voicing in obstruents most particularly found with velar /g/ such that it may be phonetically [ᵑg] and may ultimately merge with /ŋ/ (which is incidentally one of the reasons I dislike the Baxter-Sagart Old Chinese nasal "N" prefix since I think a better account for the prenasalization can be made on phonetic grounds), I wonder if we should be looking for a voiced /g/ onset in the source word? Having said that, the typological rarity of the uvular plosive /ɢ/ (which is incidentally one of the reasons to be suspect of Baxter & Sagart's reconstruction of such a phoneme in Old Chinese) and the nasalized uvular /ɴ/ makes me wonder if some kind of original uvular initial /q/, /χ/, /ʁ/ in the source word under a particular conditioning environment might not lie behind the alternation? I wonder what the situation is in Old Mongolian in terms of velar and uvular consonants?

  13. Victor Mair said,

    November 14, 2018 @ 11:24 am

    I'm planning on making a separate post on Khitan, a Mongolic language from a millennium ago, that may provide some helpful suggestions for further research.

  14. Rodger C said,

    November 15, 2018 @ 8:03 am

    Whoa, I completely missed another OCR error: "Bo9varsson" and "Bo8varsson" should of course be "Boðvarsson."

    [VHM: fixed now]

RSS feed for comments on this post