Nose, iris, pupil
« previous post | next post »
Last week, a master's student went to the board to write the Chinese character for "nose" (bí 鼻), but forgot how to do so. There is no simplified version. The form of this character differs slightly between China and Japan: in China it is 鼻 and in Japan it is 鼻. Can you spot the difference?
Believe it or not, the top part of the character depicts a nose. Here's the small seal script form, about two millennia ago (the bottom part is the phonophore, which was added long after the top part was invented):
Glyph origin
Phono-semantic compound (形聲, OC *blids): semantic 自 (“nose”) + phonetic 畀 (OC *pids).
自 (OC *ɦljids) originally meant “nose” but came to be used to mean “self”, so the sense of “nose” has been replaced by 鼻 (OC *blids). Some scholars interpret 鼻 (OC *blids) as a combination of a nose (自 (OC *ɦljids)) and two lungs (畀 (OC *pids)).
Etymology
From Proto-Sino-Tibetan *bi (“nose”); compare Sichuan Yi ꅳꁖ (hnap bbit, “nose; snot”).
Alternatively, it may be from Proto-Sino-Tibetan *s-brit (“sneeze; nose; swallow”), whence Tibetan སྦྲིད (sbrid, “sneeze”), but there is no trace of r in Chinese (Schuessler, 2007).
In some modern lects, including Mandarin, Gan, Jin, Wu, and Xiang, and even in the literary layer of some Min dialects, the word reflects a form with final *-t. For example, in standard Mandarin, the word is pronounced bí (implying an old entering tone) instead of bì (the expected reflex from the departing tone in Middle Chinese). This is due to a phonological phenomenon in the northwest, either an early loss of *-s in the *-ts cluster before regular final cluster simplification occurred (Baxter, 1992), or a dialectal change from *-s to *-t (Pulleybank, 1998).
(source)
Here's what 自 ("nose; self; from, since") looked like on the oracle bones, approximately 3,200 years ago:
Originally a pictogram (象形) of a nose; in China (and East Asia) one points at one’s nose to indicate oneself, hence an ideogram (指事) of “self”.
The original meaning of “nose” has been replaced by 鼻.
(source)
My wife (and many other Chinese friends and acquaintances) actually emphatically pointed to her nose (placed the end of her index finger on the tip of her nose) when she would say, "Wǒ zìjǐ 我自己" ("I myself").
Misremembering how to write the character for "nose" in Chinese is one type of problem in mastering the script, but altogether forgetting the word for something essential and familiar is another.
The Mandarin word for "iris" is "hóngmó 虹膜" (lit., "rainbow membrane") and the word for "pupil" (of the eye) is "tóngkǒng 瞳孔" (lit., "pupil hole"), but people often cannot recall them or confuse them.
A side note about tóng 瞳 ("pupil") that has long fascinated me, viz., the character is composed of the semantophore for "eye" (mù 目) and the phonophore tóng 童 ("pupil", [i.e. "student"]).
The long list of common words that people have difficulty writing in characters includes "Sneeze, hiccup, cough" (12/19/13). Almost nobody I know can write all three of them, and only exceedingly few individuals (around 2 out of 50) are able to write the Chinese characters for "sneeze".
Selected readings
- "Heart-mind" (9/29/14)
- "Character Amnesia" (7/22/10)
- "Character amnesia revisited" (12/13/12)
- "Character amnesia redux" (4/22/16)
- "Spelling bees and character amnesia" (8/7/13)
- "Character amnesia and the emergence of digraphia" (9/25/13)
- "Dumpling ingredients and character amnesia" (10/18/14)
- "Character amnesia in 1793-1794" (4/24/14)
- "Japanese survey on forgetting how to write kanji" (9/24/12)
- "Kanji amnesia of the week" (7/25/20)
- "Aphantasia — absence of the mind's eye" (3/24/17)
Paul Garrett said,
February 3, 2022 @ 9:36 pm
Even under magnification, I cannot see the difference between the Chinese and Japanese versions of the character you introduce at the beginning. True, my eyes are not what they once were… Would you explain the difference, please?
Brian said,
February 3, 2022 @ 9:40 pm
> The form of this character differs slightly between China and Japan: in China it is 鼻 and in Japan it is 鼻. Can you spot the difference?
Either this is a trick question, or some piece of software between here and there couldn't tell the difference either. Both of these characters are the same (U+9F3B, to be specific).
Laura Morland said,
February 3, 2022 @ 9:41 pm
The top of the Chinese character is *every-so-slightly* less wide than the Japanese version?
If that's not it, I give up.
I love it that Chinese speakers, including your late wife, point to their noses as a metonym for their whole selves!
Akito said,
February 3, 2022 @ 10:35 pm
See the Japanese character in a Japanese font (e.g., MS Mincho) and the Chinese one in a Chinese font (e.g., SimSun), then the difference emerges in the last three strokes.
Victor Mair said,
February 3, 2022 @ 11:01 pm
Akito is right. In the Japanese form of the character, the two vertical strokes at the bottom cross through the horizontal stroke. In the Chinese form of the character, the two vertical stroked stop at the horizontal stroke.
Jim Breen said,
February 3, 2022 @ 11:35 pm
As others have commented, the two characters in Victor's log article were identical as they used the same Unicode codepoint (U+9F3B). What you saw depended on the font used by your browser (mine uses Japanese-flavoured glyphs.)
The page at http://www.unicode.org/cgi-bin/GetUnihanData.pl?codepoint=9f3b will show this version as well, but in the box labelled "The Unicode Standard" is the image from the Unicode standard. It has the Chinese variant of the character.
HTH
Jim
Richard Warmington said,
February 3, 2022 @ 11:54 pm
You can see the difference here:
https://en.wiktionary.org/wiki/%E9%BC%BB#Alternative_forms
Fanolian said,
February 4, 2022 @ 12:06 am
Adding to Akito's and Jim Breen's comments:
There are three 鼻s in the first paragraph. In the HTML source code, the first 鼻 ("nose" (bí 鼻)) has specified lang="zh-CN" and is shown with a zh-CN font of your choice.
The later two do not have their languages specified separately and thus they are always shown with the same font[1]. It defeats the purpose of the sentence.
[1]: The font can be a TW/KR/HK one depending on your OS locale and browser settings.
other one spoon said,
February 4, 2022 @ 12:22 am
A Chinese teacher once told me that the English "beezer" (British slang for "nose") came from Mandarin bízi (鼻子). Is that possible?
Wolfgang Behr said,
February 4, 2022 @ 4:38 am
Re: Victor's "side note about tóng 瞳 ("pupil") that has long fascinated me, viz., the character is composed of the semantophore for "eye" (mù 目) and the phonophore tóng 童 ("pupil", [i.e. "student"])." —
It would be interesting to know how widespread the semantic motivation for 'pupil' constructed via 'little person', i.e. 'reflection of oneself in the eye of the interlocutor' really is. As pointed out by Erwin Reifler (on whom see https://en.wikipedia.org/wiki/Erwin_Reifler), apart from English pupil (< Old French < pupille < Latin pūpilla 'little girl, orphan' or < pūpillus, diminutive < pūpus 'little boy, child'), this semantic link not only occurs in Chinese (whence Japanese), but also in Hebrew and Greek: κόρη ‘young girl, daughter’ ( 'pupil of the eye' (where -u in miu 'dwarf' is a diminutive suffix attached to the common word for 'man') and in mig.gi rgyal.mo 'queen in the eye'; or in Thai luuk.F dtaa.M dam.M 'pupil of the eye' (where lluk.F ลูก means 'child, offspring, cub' etc.). Are there any examples in other, areally unrelated language families? This sure would be a nice candidate for Mattis List's super useful CONCEPTICON (https://concepticon.clld.org/), which already has a concept set for EYEBALL (#3897).
Notice that the other Late Classical Chinese word for 'pupil' lú 矑 'pearl of the eye', first attested in Yáng Xióng 揚雄 (53 BC—18 CE), works on a different semantic basis: one of the central meanings of the phonophor lú 盧 is simply 'black'. Xíng Gōngwǎn 邢公畹 long ago argued (MZYW 1989.1) that this word is the source of dtaa.M ตา 'eyeball' in the Thai word cited above, since Thai *ta-.M regularly < *tra- and lú 盧 < OC *C.ra-. The Modern Thai word would then be a combination of both motivation pathways, it seems: 'child+black+eyeball'.
Ross Presser said,
February 4, 2022 @ 8:55 am
@Richard Warmington: Thanks for the page pointer which does indeed show the difference properly in a browser. However, the HTML source actually still uses the exact same Unicode code point; the rendering difference is done by selecting a different font using a CSS class.
<span class="Hani" lang="zh">鼻</span>
<span class="Jpan" lang="ja">鼻</span>
So do not assume you can copy and paste into your own writing without carefully checking the font used after you paste!
Neil Kubler said,
February 4, 2022 @ 8:59 am
That the morphemes for "pupil (student)" and "pupil (part of the eye)" are homonyms in both English and in Chinese (童 and 瞳, both tóng), as Victor and Wolfgang Behr have noted, is interesting and may have a semantic basis — but we should keep in mind this could also be due to coincidence. Another example of the same phenomenon, almost certainly due to coincidence, are the morphemes for "swallow (bird)" and "swallow (down one's throat"), which are homonyms in English and also in Chinese (燕 and 嚥, both yàn).
Jonathan Smith said,
February 4, 2022 @ 9:13 am
Also cf. tong2ren2 瞳仁 ~ 瞳人 'pupil of eye' where 'kernel' and 'person' are further old homophobes… IDK
Jonathan Smith said,
February 4, 2022 @ 9:14 am
Ak :D, homophones*
Rodger C said,
February 4, 2022 @ 11:02 am
Reminds me of the ad I saw in PMLA ca. 2000 for a position teaching "contemporary anglophobe literature."
TonyK said,
February 4, 2022 @ 11:33 am
The Hungarian word for "iris" is "szivárványhártya", which literally means "rainbowmembrane".
tsts said,
February 4, 2022 @ 1:42 pm
@other one spoon: Not an expert, but it seems unlikely to me that beezer is derived from Mandarin. There are webpages claiming this, but with no sources provided. The term is documented in English going back over a hundred years, which to me makes Mandarin not that likely, and in Cantonese it would be pronounced "bay ze" so not that close (in case it came via HK).
Beyond "nose", it also means "head", "smart fellow", and "excellent". One source suggests a portmanteau of beak and sneezer. But my guess would be the Random House dictionary suggestion that it might derive from Spanish "cabeza".
Terpomo said,
February 4, 2022 @ 3:38 pm
@TonyK
Probably the Mandarin and Hungarian words are both calqued from "iris" which is, after all, Latin for "rainbow".
JK said,
February 4, 2022 @ 3:45 pm
I just submitted properly formatted examples illustrating glyph differences, but I see in the resulting code that Language Log strips my language tags away, so I'll try to post something better. Until then, I hope this comment will prevent the first one (awaiting moderation) from showing up.
JK said,
February 4, 2022 @ 4:07 pm
Adding to Fanolian’s comment and others' comments:
In HTML, the author/publisher can use language tags for their text as shown in the following examples. What those tags will and won’t achieve on the recipient’s side will depend entirely on how the recipient’s browser (or whatever other software they use the text with) is configured, and on what fonts are available on the recipient’s system. Unfortunately, it is impossible for me to post examples, because this blog's software strips away all my language tags. But you can copy my examples, paste them into a simple text editor, save them with the filename extension .html and then open that file in your browser or other software. Doing so should illustrate what VHM meant to show you only if you have at least one font installed whose glyph shapes follow Chinese convention and another one following Japanese convention and your browser (or other software) is configured to use each of those fonts where appropriate. All of which is obviously beyond my control, but you can easily configure your browser or install fonts, it’s not rocket science, so statements like “I use XY browser (or XY operating system), hence the characters display this or that way” don’t make much sense.
To specify the language (Sinitic* and Japanese, respectively) but not the region, you can use the following HTML code for the Chinese characters:
To specify the region but not the language:
Your software will almost certainly render these as looking alike; I’m not aware of any current software that would (or can easily be configured to) display a difference here, but I think it’s entirely possible.
In addition to primary and extended language subtags and region subtags, there are also script subtags (for example,
Latn), variant subtags (for example,fonipafor "International Phonetic Alphabet") and extension subtags. For example,specifies "Mandarin Chinese" + "Latin" + "Pinyin romanization". Information from language tags can be useful not only to help the software choose an appropriate font (or help select the appropriate glyph from fonts such as the multi-language or multi-region editions of “Noto Serif CJK” whose characters have a default shape but also include the three other ones), but also for screen readers for visually impaired people, where the screen reader, in order to pronounce words correctly, would need to know which language or dialect and which orthography or romanization scheme it is dealing with. Further reading: "Language tags in HTML and XML" (English · Russian · Japanese) and "Choosing a Language Tag".
Leaving the realm of language tags, one can of course directly specify one or several particular fonts in order of preference (without using a language tag to specify language or region). If you’ve got the “Noto Serif CJK {SC/TC/JP/KR}” fonts installed (they come pre-installed with some distros), here’s an example:
This should tell browsers to render the text using the font “Noto Serif CJK TC”. It defaults to glyph shapes that follow Taiwanese convention, unless you also use a
langtag specifying that the text is, for example, Japanese. In the SC, TC, and KR editions, the default shape for this particular character looks alike in version 1.001; only the JP edition has a different default shape for it. But I think specifying particular fonts is not the proper way to go about these things.↑ According to the language subtag registry,
zhis the macrolanguage “Chinese” and currently includes “Min Dong Chinese” (for which the individual code iscdo), “Jinyu Chinese” (cjy), “Mandarin Chinese” (cmn), “Northern Ping Chinese”/“Northern Pinghua” (cnp), “Pu-Xian Chinese” (cpx), “Southern Ping Chinese”/“Southern Pinghua” (csp), “Huizhou Chinese” (czh), “Min Zhong Chinese” (czo), “Gan Chinese” (gan), “Hakka Chinese” (hak), “Xiang Chinese” (hsn), “Literary Chinese” (lzh), “Min Bei Chinese” (mnp), “Min Nan Chinese” (nan), “Wu Chinese” (wuu), and “Yue Chinese”/“Cantonese” (yue).JK said,
February 4, 2022 @ 5:19 pm
(if this is too long or too off topic, please delete.)
Clarifying and adding to Fanolian’s and others' comments:
How to format text when posting something (not in the LL comments, but somewhere where you have full control over the code):
In HTML, the author/publisher can use
langtags for their text as shown in the following examples. What those tags will and won’t achieve on the recipient’s side will depend entirely on how the recipient’s browser (or whatever other software they use the text with) is configured, and on what fonts are available on the recipient’s system. I cannot post finished examples here, because this blog's software strips my submissions of itslangtags. But you can copy the following half-baked examples, paste them into a simple text editor, save them with the filename extension.htmland then open that file in your browser or other software. Doing so should illustrate what VHM meant to show you only if you have at least one font installed whose glyph shapes follow Chinese convention and another one following Japanese convention and your browser (or other software) is configured to use each of those fonts where appropriate. All of which is obviously beyond my control, but you can easily configure your browser or install fonts, it’s not rocket science, so statements like “I use XY browser (or XY operating system), hence the characters display this or that way” don’t make much sense.To specify the language (Sinitic* and Japanese, respectively) but not the region, you can use the following HTML code for the Chinese characters:
(Again, the difference will not show in this comment; you'll have to copy and paste it, save it as an HTML file, and open that in a browser.)
To specify the region but not the language:
Your software will almost certainly render these as looking alike; I’m not aware of any current software that would (or can easily be configured to) display a difference here, but I think it’s entirely possible.
In addition to primary and extended language subtags and region subtags, there are also script subtags (for example,
Latn), variant subtags (for example,fonipafor "International Phonetic Alphabet") and extension subtags. For example,specifies "Mandarin Chinese" + "Latin" + "Pinyin romanization". Information from language tags can be useful not only to help the software choose an appropriate font (or help select the appropriate glyph from fonts such as the multi-language or multi-region editions of “Noto Serif CJK” whose characters have a default shape but also include the three other ones), but also for screen readers for visually impaired people, where the screen reader, in order to pronounce words correctly, would need to know which language or dialect and which orthography or romanization scheme it is dealing with. Further reading: "Language tags in HTML and XML" (English · Russian · Japanese) and "Choosing a Language Tag".
Leaving the realm of language tags, one can of course directly specify one or several particular fonts in order of preference (without using a language tag to specify language or region). If you’ve got the “Noto Serif CJK {SC/TC/JP/KR}” fonts installed (they come pre-installed with some distros), here’s an example:
This should tell browsers to render the text using those fonts. For example, the TC edition defaults to glyph shapes that follow Taiwanese convention, unless you add a
langtag specifying that the language is Japanese. In the SC, TC, and KR editions, the default shape for this particular character looks alike in version 1.001; only the JP edition has a different default shape for it. But I think specifying particular fonts is not the proper way to go about these things.↑ According to the language subtag registry,
zhis the macrolanguage “Chinese” and currently includes “Min Dong Chinese” (for which the individual code iscdo), “Jinyu Chinese” (cjy), “Mandarin Chinese” (cmn), “Northern Ping Chinese”/“Northern Pinghua” (cnp), “Pu-Xian Chinese” (cpx), “Southern Ping Chinese”/“Southern Pinghua” (csp), “Huizhou Chinese” (czh), “Min Zhong Chinese” (czo), “Gan Chinese” (gan), “Hakka Chinese” (hak), “Xiang Chinese” (hsn), “Literary Chinese” (lzh), “Min Bei Chinese” (mnp), “Min Nan Chinese” (nan), “Wu Chinese” (wuu), and “Yue Chinese”/“Cantonese” (yue).JK said,
February 4, 2022 @ 6:01 pm
To specify the language (Chinese and Japanese, respectively) but not the region, you can use the following HTML code for the Chinese characters:
The code
zhis from the language subtag registry. It is for “Chinese” and covers “Mandarin Chinese” (cmn), “Yue Chinese”/“Cantonese” (yue), “Literary Chinese” (lzh) and 13 more. The glyph difference VHM was getting at will not show in this comment, so you'll have to copy and paste the code into a text editor, save it as an HTML file, have at least one font installed whose glyph shapes follow Chinese convention, another one following Japanese convention, make sure your browser (or other software) is configured to use each of those fonts where appropriate, and open your HTML file in the browser. It’s not rocket science, so statements like “I use X browser (or Y operating system), hence the characters display this or that way” don’t make much sense. I cannot post finished examples because this blog's software stripslangtags from comments.To specify the region but not the language:
Your software will almost certainly render these as looking alike; I’m not aware of any current software that would (or can easily be configured to) display a difference here, but I think it’s entirely possible.
In addition to primary and extended language subtags and region subtags, there are also script subtags (for example,
Latn), variant subtags (for example,fonipafor "International Phonetic Alphabet") and extension subtags. For example,specifies "Mandarin Chinese" + "Latin" + "Pinyin romanization".Information from language tags can be useful not only to help the software choose an appropriate font (or help select the appropriate glyph from fonts such as the multi-language or multi-region editions of “Noto Serif CJK” whose characters have a default shape but also include the three other ones), but also for screen readers for visually impaired people, where the screen reader, in order to pronounce words correctly, would need to know which language or dialect and which orthography or romanization scheme it is dealing with. Further reading: "Language tags in HTML and XML" (English · Russian · Japanese) and "Choosing a Language Tag".
Instead of using language tags, one can specify particular fonts to use. If you’ve got the “Noto Serif CJK {SC/TC/JP/KR}” fonts installed (they come pre-installed with some distros), here’s an example:
This should tell browsers to render the text using those fonts. For example, the TC edition defaults to glyph shapes that follow Taiwanese convention, unless you add a
langtag specifying that the language is Japanese. In the SC, TC, and KR editions, the default shape for this particular character looks alike in version 1.001; only the JP edition has a different default shape for it. But I think specifying particular fonts is not the proper way to go about these things.Richard Warmington said,
February 4, 2022 @ 7:54 pm
@Neil Kubler
Re: the morphemes for "swallow (bird)" and "swallow (down one's throat") … are [coincidentally] homonyms in English and also in Chinese (燕 and 嚥, both yàn).
Yes, and furthermore, 燕 is an old variant of 宴 (also pronounced yàn), which means "feast", i.e. an event where one swallows a lot of food and drink.
Alex said,
February 5, 2022 @ 12:44 am
During the spring festival week once again talk and commentary how Chinese will be an alternative global language. I guess nose sneeze cough etc aren't too important LOL
Chris Button said,
February 5, 2022 @ 9:18 am
This goes back to Pulleyblank 1973. It is quite persuasively argued and also notes the variant Jiyun EMC reading sit for 四 EMC siʰ "four".
The s~t interchange has support elsewhere too. See here for loanword examples: https://languagelog.ldc.upenn.edu/nll/?p=53219#comment-1591017
The connection of 童 with 瞳 is noted by Wang Li in his 1982 同源字典.
Also, note how some earlier inscriptional forms of 童 show an eye being pierced.
Fair point, although the inscriptional forms of 童 might also bolster an etymological connection. It's worth looking at the rest of the word family to see where it really sits (compare Wolfgang Behr's comment about 矑 and 盧).
Incidentally, 燕 and 嚥 recently came up here: https://languagelog.ldc.upenn.edu/nll/?p=53219#comment-1591181
I follow Schuessler in rejecting the idea that that 自 and 鼻 are phonetically related.
There's also some similarity between the pronunciation of 己, which I reconstruct with *x- ( https://languagelog.ldc.upenn.edu/nll/?p=53265#comment-1591189 ), and 自. I think the idea of a possible connection/confusion between the two is worth investigating.
Chris Button said,
February 6, 2022 @ 6:38 am
Regardless of any later confusion there, the etymological connection of 自 "from, self" seems to be with 次 "subsequent" and related words. Compare how Persian az "from" and signature come from the same IE səkw- root as subsequent. The use of 自 being simply a graphic borrowing.
Chris Button said,
February 6, 2022 @ 7:18 am
Looking at Takashima's 2020 article "Paleography, Historical Phonology, and Historical Lexicology: 'Kneeling Women with Their Wrists Crossed' and 'Slaves with Their Eyes Gouged'" (also discussed here: https://languagelog.ldc.upenn.edu/nll/?p=52312#comment-1589034 regarding 女 as a slave), it seems the eye piercing in 童 with its sense of "(ignorant servant) boy" fits well with his discussion of 民 as "ignorant populace" where the eye piercing is clearly depicted.
Chris Button said,
February 6, 2022 @ 7:23 am
And of course we have the actual depiction of an eye 臣 with the meaning of "vassal, servant …"
Diana Bloom said,
February 6, 2022 @ 12:30 pm
Could the meaning of "nose" be related to the phrases "lose/save face"?
Wolfgang Behr said,
February 7, 2022 @ 5:47 am
(VHM: reposting Wolfgang's note, which has been truncated by our system)
Re: Victor's "side note about tóng 瞳 ("pupil") that has long fascinated me,
viz., the character is composed of the semantophore for "eye" (mù 目) and
the phonophore tóng 童 ("pupil", [i.e. "student"])." —
It would be interesting to know how widespread the semantic motivation for 'pupil' constructed via 'little person', i.e. 'reflection of oneself in the eye of the interlocutor' really is. As pointed out by Erwin Reifler (on whom see here), apart from English pupil (< Old French < pupille < Latin pūpilla 'little girl, orphan' or < pūpillus, diminutive < pūpus 'little boy, child'), this semantic link not only occurs in Chinese (whence Japanese), but also in Hebrew and Greek: κόρη ‘young girl, daughter’ (< IE *ḱerh3-, 'to grow'), also 'pupil of the eye' <<. (in: Étude sur l'Étymologie des Caractères Chinois, Bulletin de l'Université de l'Aurore (1944) 1-36). The same metonymy appears in Tibetan mig-gi miu 'little man of the eye' --> 'pupil of the eye' >> (where -u in miu 'dwarf' is a diminutive suffix attached to 'man') and in mig.gi rgyal.mo 'queen in the eye', or in Thai luuk.F dtaa.M dam.M 'pupil of the eye' (where lluk.F ลูก means 'child, offspring, cub' etc.). Are there any examples in other, areally unrelated language families? This sure would be a nice candidate for Mattis List's super useful CONCEPTICON, which already hasa concept set for EYEBALL (#3897).
Notice that the other Late Classical word for 'pupil' in Chinese lú 矑 'pearl of the eye', first attested in Yáng Xióng 揚雄 (53 BC—18 CE), works on a different semantic basis: one of the central meanings of the phonophore lú 盧 is simply 'black'. Xíng Gōngwǎn 邢公畹 long ago argued (MZYW 1989.1) that this word is the source of dtaa.M ตา 'eyeball' in the Thai word above, since Thai *ta-.M regularly < *tra- and lú 盧 < OC *C.ra-. The Modern Thai word would then be a combination of both motivation pathways, it seems: 'child+black+eyeball'.