Submissive woman or bound slave: interpreting oracle bone forms as a Rorschach test, part 2
« previous post | next post »
Throughout my research and teaching career, I have always emphasized that, when it comes to genuine etymology of Sinitic, what matters are the sounds and meanings of the constituent etyma, going all the way back to the fundamental roots. The shapes of the glyphs used to write the eyma in question are far less important than the sounds and meanings. In fact, discussion of the shapes of the glyphs is often more of a distraction than a benefit to understanding what the true etymologies of given etyma are. We demonstrated that by the sharp disagreements we had over the meanings of the shapes of the ancient glyphs / forms / shapes of such a simple / definite / concise lexeme / morpheme as "woman; female". That is why the sound nǚ and its attendant meaning "woman; female" are more important for Sinitic etymology than is the the three-stroke character 女, albeit the latter derived from more complicated and difficult to explain / interpret forms.
It is for these reasons that I have so strongly supported the works of Axel Schuessler (e.g., ABC Etymological Dictionary of Old Chinese [2007]) and will close this post by concentrating on the sound and meaning of 女 rather than its shape.
By a stroke of good fortune, Rachel Boughton happened to read and comment on part 1 of this post. After a couple of zoom conversations and numerous e-mail exchanges, I was delighted to elicit her views on the deeper significance of 女.
Rachel is a Jungian Analyst who previously has given a lot of thought to 女 / "woman ; female" and has used her findings in her research, teaching, and treatment. She is also a koan teacher and translator with a concentration on the legacy and words of women, though she sometimes takes an interest in a male teacher like Old Zhaozhou, who was a nature mystic, even though his record focuses on his snappy quips.
So that we better understand Rachel's purview and expertise, and its relevance for these posts about women in premodern times, I will quote from her a brief introductory note about koans:
A koan is a teaching phrase or story, sometimes pithy, sometimes long, sometimes from a conversation that was transformational to somebody. The words of a koan become a part of a meditation practice in such a way that the original insight can be shared by the practitioner through their own "aha!" moments. Sometimes this is done in a kind of apprenticeship with a teacher. There are classics used in the west, books like the wumenguan (gateless gate) and the biyanlu (blue cliff record), with many very similar translations, that are the texts for what the Linji/Rinzai school considers a complete curriculum. Finishing that curriculum with its 250 or so koans is usually part of the training for transmission as a Zen Roshi ("old teacher; old master"). Women are hardly ever mentioned in those 250 koans though, despite women having been seminal teachers and some of their records existing in Chinese archives.
What follow are some comments by Rachel on early forms of 女:
Also I am curious about the elbows bent which is the traditional neolithic goddess figurine's usual gesture, sometimes seen cradling the breasts, so ubiquitous that it looks like it might be a ritual gesture for a woman.
Both as a Jungian analyst and as a koan exegete, Rachel has devoted a great deal of attention to 女 ("woman; female"), but even she is unwilling to affirm with absolute confidence what those early forms represent.
So much for the shape of the 女 glyph.
Before closing, what can we say about the sound and meaning of the etymon?
Pronunciation 1
MSM nǚ
"woman; female"
"daughter"
- Old Sinitic (ca. 1250 BC)
- (Baxter–Sagart): /*nraʔ/
- (Zhengzhang): /*naʔ/
From Proto-Sino-Tibetan *naq (“woman”). Compare Tibetan ཉག་མོ (nyag mo, “woman”) (Hill, 2019).
Pronunciation 2
MSM nǜ
(obsolete on its own in Standard Chinese) "to give one's daughter in marriage; to give as wife"
- Middle Sinitic (601 AD): nrjoH
- Old Sinitic (ca. 1250 BC)
- (Baxter–Sagart): /*nraʔ-s/
- (Zhengzhang): /*nas/
Exoactive derivation of pronunciation 1 (Schuessler, 2007).
Selected readings
- "Submissive woman or bound slave: interpreting oracle bone forms as a Rorschach test" (1/20/26)
- "Women's script wins in the end" (1/19/26)
- "Misogyny as reflected in Chinese characters" (12/25/15)
- "Womanless" (11/11/21) — profound panoply of 100 sinographs with the 女 ("woman") semantophore by a woman artist
Chris Button said,
January 25, 2026 @ 12:48 pm
The velar coda -ɰ at the end of 女 is essential. Same as the one in 巫.
Jonathan Smith said,
January 26, 2026 @ 7:10 pm
second glyph shown is chronologically earlier than the first i.e. better to say development is in the opposite direction. and both of these are from bronze inscriptions. "feet and knees on the same level" or so i.e. "kneeling" is typical in the bone inscriptions, which generally speaking are a couple or three centuries earlier.
Axel Schuessler said,
January 26, 2026 @ 8:29 pm
(1) The inscriptional graphs: I see three different versions, all have the positions of the arms in common. One is the figure kneeling, one the figure crouching, finally the figure standing; in this case it is the graph for 人, basically a vertical line, with the arm replaced by the woman's arms.
(2) A few thoughts to comments by Chris Button:
(a) Why is Pulleyblank's velar coda -ɰ essential?
This coda corresponds to Li Fang-kuei's *-g, which is Karlgren's *-g transferred from the 鐸 duó-bu (*-ak ~ Karlgren's *-ag, actually now almost universally *-aks) applied to the 魚 yú-bu rime group. In the duo-bu the *-g alternates with *-k (各 *krâk vs. 路 *râks — Karlgren *ak vs. -ag). It is clear (cf. Baxter's work) that the yú-bu had open syllables in *-a in all MC tones (including 女巫), while the *-ag from the *-ak group all have qusheng, i.e. had final *-aks. Thus Li's and Pulleyblank's postvocalic finals in the yú-bu did not exist.
(b) 女 OC *nraʔ (Zhengzhang *naʔ) has a medial /r/ whereas suspected Tibeto-Burman cognates do not, cf. Tibetan nya (with palatalization). Baxter and most others insert this /r/ to account for the MC retroflex initial (ṇjwo´). However, this hypothesis has resulted in an OC medial *-r- epidemic which looks suspicious. Thus retroflexion may have other causes, the -r- is simply placed there by default. Michel Ferlus (Linguistics of the Tibeto-Burman Area 22.2, 1999:5) has suggested that nü is related to Mon-Khmer languages, cf. Khmer kmbraʔ (i.e. /kmraʔ/) 'wife', to account for the -r-.
Nü *nraʔ ends in a glottal stop, accounting for the MC shangsheng. This feature is a Chinese innovation with no Tibeto-Burman counterpart (unless one prefers the Khmer connection). Human relations terms tend to end in this feature: 父母弟姐子. Some scholars try to find connections of such a glottal stop with Tibeto-Burman final velars, e.g. *nraʔ cognate to the marginal Tibetan dialect form nyag 'woman'; to reconcile such forms, Nathan Hill, for example, postulates a ST form like *naq. There seems no need for such final *-q.
(c) Chris Button speculates that nü might be related to nú 奴 *nâ 'servant, slave. M. Ferlus has proposed this in 1999 (see above). I would exclude an etymological connection. Shuowen says that nú refers to male as well as female servants; as to the graph, Xu Shen claims 从女从又. However, it seems that originally yòu 'hand' was the radical, nü was the phonetic, which comes in handy, as being partially suggestive, because a servant could ALSO be a female.
(d) A cognate of nü could be nú 孥 *nâ 'wife and children'.This is similar to the Written Tibetan expression ma-smad (i.e. s-ma-d) 'mother and her children'.
(e) As to the origin of English wench, counterexamples of such semantic developments seem to be more common, cf. maiden > maid.
Chris Button said,
January 26, 2026 @ 9:04 pm
*-aks > *-aɰ and merges with *-aɰ.
Very much agreed. Unfortunately the vowel lengthening that gave alternations like 荼 *láɰ vs. 茶 *láːɰ doesn't work here. But 女 is still more phonologically complex than 奴.
Ferlus follows (and cites) Pulleyblank's lead here. The proposal is that the sense of 奴 came from 女 via 孥, which is indeed the opposite of the proposal that 奴 was the original sense of 女.
Chris Button said,
January 26, 2026 @ 10:48 pm
Or more specifically, *-aks > *-aɰʰ and merges with *-aɰʰ (from *-aɰs )
Victor Mair said,
January 26, 2026 @ 11:11 pm
From Bob Bauer:
As for your strong suspicion that 奴隸 nu2 li4 “slave” may be a borrowing from a non-Sinitic language, the bisyllabic word’s n-/l- alternation of initial consonants and its particular meaning make me think it is a good candidate for ultimately being a loanword, so I do agree with you. However, at the moment I don’t have any data to support this idea. I can imagine that both the word and the social practice of slavery could have been borrowed from a non-Sinitic speech community with which the early Han Chinese had come into contact.
Victor Mair said,
January 26, 2026 @ 11:21 pm
From Guillaume Jacques:
It is true that 奴 is rare in pre-Han texts, but we find at least one prominent example in the Lunyu:
微子去之，箕子為之奴，比干諫而死。孔子曰：「殷有三仁焉。」
Unger thought that 奴 nu < *nˤa could derive from 努 nuX < *C.nˤaʔ (converted to Baxter and Sagart 2014), which wouldn't be impossible, but the morphological derivation is unclear. A very speculative cognate would be the root *hnaa "work" in proto-Kuki-Chin: WORK2 / JOB1 PKC *hnaa F. Lai hnàa ‘work, job’; Mizo hnàa ‘work, job, task, employment, occupation’; Tedim na3 sem3 ‘work’; Thado Kuki nàa ‘work’; Sizang na sep ‘work’. ------ VHM: Will probably ring some bells with Chris Button.
Victor Mair said,
January 26, 2026 @ 11:30 pm
From Julie Lee Wei:
'
Nu Li 奴隸
Nu 奴
• Old Chinese (Wiktionary)
(Baxter–Sagart): /*nˤa/
(Zhengzhang): /*naː/
Wiktionary says etymology uncertain. Gives the following:
Etymology
Uncertain.
• Possibly cognate with Mru nar (“servant”) & Awa-Khumi Chin tana (Löffler, 1960);
• Ferlus (1999) relates this to 女 (OC *naʔ), which has semantic parallel, especially among foreign loans (e.g. 嬯 (OC*dɯː, “servant, slave woman”) which is possibly from Austroasiatic).
• Unger (1990) groups 奴 (OC *naː) as well as 努 (OC *naːʔ, “to tense, to exert”), 弩 (OC *naːʔ, “crossbow”), & 怒(OC *naːs, “angry”), in a word-family with the basic meaning "tense", hence 奴 (OC *naː)'s meaning "press into service".
Wikipedia:
Mru, also known as Mrung (Murung), is a Sino-Tibetan language of Bangladesh and Myanmar
I think above may be cognate with Sanskrit:
Sanskrit for “woman”
Sanskrit नारी • (nārī) f (Urdu spelling ناری)
1. woman
2. wife
Declension
Declension of नारी (fem ī-stem)
Etymology
From Proto-Indo-Iranian *Hnā́riH (“woman”). Cognate with Avestan (nāirī, “woman”), Middle Persian n'(y)lyk' (nārīg, “woman, lady wife”).
**** see below
Li 隸
Wiktionary
• Middle Chinese: lejH
• Old Chinese
(Baxter–Sagart): /*[r]ˤə[p]-s/
(Zhengzhang): /*rɯːds/
“servant”. (Wiktionary)
Wiktionary gives no etymology for 隸。
I’ve found none so far.
*****
Victor Mair said,
January 26, 2026 @ 11:36 pm
From Sanping Chen:
Individually, both characters have a pre-Qin origin. Used as a binome, it would be an open question. One may try to search for the word's appearance in the Taisho canon?l
Axel Schuessler said,
January 27, 2026 @ 1:02 am
Concerning Guillaume Jacques:
My teacher Ulrich Unger had explained the word family nú in a seminar in the early 1960, which I find convincing. Interesting the possible Kuki-Chin cognates.
Concerning Julie Lee Wei's Sanskrit idea:
Skt. nārī 'woman' is derived from naraḥ 'man'. However, Vedic gnā 'wife, goddess' = Greek γυνή, βανα (from something like IE *gʷn̥ā; English queen) comes to mind, — just one of many OC ~ ancient Greek or IE look-a-likes. Long ago I played with the idea if some OC "medial *-r-" could not rather reflect a lost prefix like *g-, hence nü < gnaʔ. But there is no basis for this, of course.
Chris Button said,
January 27, 2026 @ 6:26 pm
Pulleyblank's suggestion that 奴 came from 女 via 孥 goes back to his 1956 article on chattel slavery.
I didn't record that one. The Thado reflex shows it to be a different word from *ʰna¹ "ear" though!
Without the velar coda, how do you account for the phonetic relationship of 呶 with 奴 or the likely etymological relationship of 女 with 娘?
Pulleyblank (1996) proposed a w- prefix with a casual observation that this "incidentally would bring the word very close" to the IE root!
Axel Schuessler said,
January 27, 2026 @ 7:47 pm
Depends on one's view of the OC syllable, I suppose. I see no problem with an OC that had many words and syllables that ended in a vowel, with an addition of a consonant on occasion within a word family. Pulleyblank had this unique idea that all OC syllables ended in a consonant.
On niáng 娘, South Coblin wrote (A Compendium of Phonetics in Northwest Chinese, 1994, p. 389), that this word does not occur until the Tang period and is probably a fusion of 女+郎。
Chris Button said,
January 27, 2026 @ 8:42 pm
Li Fang-Kuei (1945) assumes a shift of -g > -ɣ > -ɯ. Pulleyblank does not assume the earlier stopped phase and treats Li’s vocalic -ɯ as the velar approximant -ɰ.
Whether you treat the approximant as a vowel or consonant comes down to phonological analysis rather than phonetic distinction. A familiar example from English would be /-aj/ versus /-ai/ or /-aw/ versus /-au/.
The -ɰ ~ -ŋ alternation occurs elsewhere (e.g., 于 vs 彺) and has a plausible phonetic explanation. So, I'm not sure such a proposal is needed.
In addition, even if we exclude 孃 for 娘 since its earlier meaning was different, there appears to have been a 婦娘 "Lady 娘" in the oracle bones.
Jonathan Smith said,
January 28, 2026 @ 12:06 am
Randomly,
* the medieval 'mother' word is supposedly strictly written "孃" to begin with. Whereas (sometimes?) homophonous, equally medieval 'lady' was written "娘". Now both are written '娘" in e.g. Mandarin, or maybe better to say these lexemes have merged.
* that some character (antecedent of) "女" + (antecedent of) "良" exists in early inscriptions writing some name is immaterial. We need *words*, since characters are created from the same components over and over again. The above words would seem to be medieval.
* Coblin's idea of 'lady' as a contraction is good. Whereas Vovin & McCraw (2011) propose the 'mother' item (and others e.g. die1 re: which note die1 niang2 are ever a pair) are from Old Turkic-ish… they write OT ana 'mother' and believe -ŋ could reflect ana-ŋ 'mother-2psp.'
* Interesting is e.g. Taiwanese (I don't know offhand about other Min) because the 'mother' item is distinct from 'lady': niâ 'mother' in e.g. a-niâ vs. niû(~niô) 'lady' in e.g. ko͘-niû 'young lady' and meaning wife of the doctor, the priest, the teacher etc. Generally both are written "娘".
* the above are just nasalized syllables in the Hokkien manner, so on the face of it hard to say re: earlier coda. Niang2 'mother' in e.g. Shandongese is also just nasalized :D Could an original loan from Turkic just have had a nasalized vowel mayhaps?
* re: Li 1971 etc., he was maybe first to expand *-g and the like to make the traditional rhyme groups look alike-er, but I don't think he quite meant this literally. General thought is if you have some solid Chinese-to-"TB" comparison with open vowel, then need -g to make your Old Chinese rhyme groups look right, then need to have it disappear to make "Middle Chinese" and all the modern languages look right… that -g might not be the best idea.
Axel Schuessler said,
January 28, 2026 @ 1:05 pm
婦娘 means a woman 女 from the Liang 良 clan, just as emperor Wu Ding's consort was a woman 女 from the 子 clan, written 女+子 = 好. She should be referred to as Fu Zǐ, but of course she is conventionally known as Fu Hǎo.
于 vs 彺, actually *wa vs. *waŋʔ — How would Pulleyblank have explained all the derivations from 無 •ma: 亡靡蔑 et alia? (Just a rhetorical question).
Chris Button said,
January 28, 2026 @ 5:53 pm
Good point–I had forgotten about that. And Takashima does go with Fu Liang rather than "Fu Niang" in the Bingbian instance, but he does note that there "seems no hard evidence to choose one over the other."
Incidentally, there seems to be a 帚娘 and a 帚良.
In any case, 娘 as a vulgar form of 孃 makes sense since 娘 struggles in the 良 series in any case. So we should really be starting with 孃 if at all.
But all of that is probably irrelevant since -ɰ ~ -ŋ is a well-attested alternation across the OC lexicon. So why bother to look for an alternative explanation for the relationship of 女 with 孃/娘 in this one case?
Well, 無 vs 亡 is like 于 vs 彺 and all those other cases of -ɰ ~ -ŋ out there.
The problem is that by ignoring velar -ɰ, people have no way to account for the alternation.
Instead we get the usual approach of slapping a random affix on (in this case -ŋ) and claiming to have uncovered some vestige of Old Chinese morphology. In fact, the alternation seems more likely to have a phonetic explanation in what has been euphemistically known in the literature since Grierson (1922) as "spontaneous nasalization".
Chris Button said,
January 28, 2026 @ 6:32 pm
And what about the xiesheng relationship between 呶 with -ʁ and 奴 with -ɰ. The use of the latter as phonetic in the former is hard to make sense of with something like -w and -∅.
Victor Mair said,
January 28, 2026 @ 9:02 pm
See also this comment about Nu Li 奴隸.
Chris Button said,
January 28, 2026 @ 9:56 pm
Regarding spontaneous nasalization, the conditioning environment would preference Pulleyblank's alternative treatment of unfricated -ɰ as fricated -ɣ. In his 1977-8 article he comments on -ɣ accordingly: "At least by the Han period this must have been pronounced with very little friction."
That does then reopen (or rather "re-close"–pun intended) the question of a lack of open syllables. I tend to see -ɰ and -ɣ as allophones since the counterpart in onset position is exclusively ɣ- and the distinction between -ɰ and -ɣ is hardly stark (a single -ʁ symbol serves for approximant and fricative a little further back in the mouth)
The fricated -ɣ would be devoiced in coda position since its fricative nature and its syllable-final position would naturally render it unable to reliably maintain voicing. That would then create the high airflow conditions that John Ohala notes in various articles to trigger spontaneous nasalization (or perhaps rather the auditory perception of nasalization).
Jonathan Smith said,
January 30, 2026 @ 12:35 am
“But all of that is probably irrelevant since -ɰ ~ -ŋ is a well-attested alternation across the OC lexicon. So why bother to look for an alternative explanation for the relationship of 女 with 孃/娘 in this one case?”
No, at all. "Alternation" means that something is (~ seems like) a single lexeme but is realized in different "alternating" forms — conditioned, ideally, but maybe appearing this way vs. that way for inscrutable (dialectal? diachronic? other?) reasons. So as always, very annoyingly, we must "bother" to examine actual texts/languages.
When we do, we find re: Mand. nv3 女 'woman' vs. niang2 娘 'madam; mother' that they are in no sense whatsoever the same word, nor are their antecedents in any sense the same word at any earlier historical stage. Reference to "alternation" is literally nonsensical.
And re: (pesudo?) Mand. wu2 無 vs. wang2 亡, well… they appear in e.g. the Book of Odes with different meanings/distributions, the latter rhyming repeatedly with *-ang words. Same for yu2 于 vs. wang3 往. If one wishes for some reason to refer to "alternation" in such cases, one would need to point to… alternation.
And main point (not for the first time), even if properly applied on the basis of again hack cough examination of real languages/texts, "unconditioned alternation" is just an IDK-type description, not an insight.
Chris Button said,
January 30, 2026 @ 6:07 pm
The inverse of "spontaneous nasalization" before a fricative is "nasal effacement" before a fricative. Again, John Ohala has written a lot on these seemingly opposing phenomena.
For example, Spanish/Portuguese "pensar" (to think) represents the same word as "pesar" (to weigh), but they are now semantically differentiated too.
Chris Button said,
February 1, 2026 @ 1:28 pm
The more I think about this, the more I like Pulleyblank's (1996) similar proposal for a back dorsal prefix, in his case w-, that conditioned the retroflexion. His proposal that w- > ɣ-, which is acoustically similar to a retroflex rhotic, addresses the phonetics nicely. Personally I would favor something like ʁənàɰˀ
It reminds me, but is of course separate, from the Sanskrit-style k-conditioned retroflexion in 殺 "kill" (Pulleyblank 1965) and also in 剎 "temple", which has been connected by Samuel Martin (1991) with Sanskrit kṣétra “holy precinct”.
Regarding the word family, I've posted correspondences like the following elsewhere on Llog: 奴 "slave", 努 "exert", 怒 "anger", 能 "able" (another case of -ɣ/ɰ ~ -ŋ alternation in its dual EMC – j ~ -ŋ reflexes; note how the -ɣ/ɰ nicely accounts for the EMC -j too), 耐 "endure", etc. Compare the similar PIE semantic field found in Lithuanian vergas "slave"; work; irk / Irish fearg "anger"; energy; bulwark.
So perhaps–just perhaps–the character 女 originally meant "slave" (as suggested by its original graphic form, which from an inscriptional perspective is undeniably devoid of any feminine features regardless of what was later associated with it). An external word for "woman" then entered the language with a sound something like ʁənàɰˀ that happened to sound very similar to the sound of náɰ "slave" …
Chris Button said,
February 7, 2026 @ 4:28 pm
Actually, instead of hypothetical long-range external influence, I think we might be able to posit a vowel length alternation instead:
https://languagelog.ldc.upenn.edu/nll/?p=72747#comment-1638666