Polysyllabism in Sinitic and (phonemic) syllable stress
« previous post | next post »
AntC wrote:
To your recent point on the 'slippery, slithery' article …
There's a town on Taiwan's East coast 'Taimali' / 太麻里鄉. This name is from the indigenous Paiwan language [also here for the people]. [see wikip]
I naively pronounced it with stress on the first syllable. I was roundly corrected by the Taiwanese family I'm staying with for a Lunar New Year visit: that should be Tai(m)-'ali, with stress on second syllable.
I suspect they're modelling this on Tai-pei, Tai-chung, Tai-tung, etc. As you've repeatedly told us, the Tai- in Tai-wan is _not_ a Sinitic morpheme.
Wikip alleges Sinitic languages are syllable-timed; syllable stress is not morphemic — which would make sense _if_ Sinitic languages were monosyllabic. You beg to differ. So how does it go with stress within polysyllabic imported words?
Good observations and good question.
For a short answer, I would just say that, once a non-Sinitic word or morpheme gets swallowed up by the sinographic writing system, it also gets swallowed up by the phonology of the sinographic writing system. However, in spoken sinitic languages, it is possible that, as I have often demonstrated and put it, "intonation overrides tone", whether for sinitic or non-sinitic morphemes and lexemes. By "tone", we can also subsume other phonological and phonetic phenomena. See the "Selected readings" below.
Selected readings
- "How to call your relations" (1/27/25)
- "When intonation overrides tone, part 6" (5/9/21)
- "Language that exercises the brain; poetry and gradations of understanding" (1/7/24)
- "Surprising Transformations of a Beijing Street Name" (1/29/11)
- "Another Northeastern topolectal term without specified characters to write it" (7/22/20)
- "Russian Loans in Northeast and Northwest Mandarin: The Power of Script to Influence Pronunciation" (1/23/11)
- "Manchu loans in northeast Mandarin" (10/7/13)
- "Varieties of Mandarin" (10/25/17)
- "Triple topolectal reprimand" (5/29/16)
- "Our Taiwan" (11/19/13)
- "No character for the most frequent morpheme in Taiwanese" (12/10/13)
- "Taiwanese Morphemes in Search of Chinese Characters", by Robert L. Cheng (Zheng Liangwei), Journal of Chinese Linguistics, 6.2 (June, 1978), 306-314.
- "Sinitic languages without the Sinographic script" (3/5/19)
- "Dongbei Survival Guide: You can be funny and emotional in Dongbeihua, but be careful," by Ginger Huang, The World of Chinese
- "Northeastern Mandarin"
Chris Button said,
January 31, 2025 @ 8:20 am
Could you possibly clarify what you mean here?
AntC said,
January 31, 2025 @ 8:51 am
Could you possibly clarify
Hi Chris, I suspect my ear is not phonetically qualified enough to be reliable. (So I was hoping Prof Mair might be able to elicit a more reliable source.) My informants are primarily Putonghua speakers, but the older ones (and especially the matriarch who was most vehement in correcting me) are reasonably fluent in Hokkien, or at least claim to understand it.
I transcribed the (m) to show there's a vaguely nasal quality to the first vowel, but it is still open as in all the Tai- names in Taiwan. I've also noticed the family pronounces that Gorge near Hualien as more like Tai-roko.
It's as if (I'm folk-etymologising here) Tai- is a prefix for placenames-in-general, you put contrastive stress on whatever follows.
languagehat said,
January 31, 2025 @ 9:08 am
This has nothing to do with any "Tai- prefix" (for one thing, it's a different character); the Wikipedia article linked in the post says:
So it's a Chinese modification of a Japanese modification of an indigenous name.
languagehat said,
January 31, 2025 @ 9:11 am
On the other point, what you were hearing was not stress but tone. The second syllable has rising pitch (Tàimálǐ); this sounds to a stress-trained ear like a stressed syllable. What the Taiwanese family heard from you was not "stress on the first syllable" but an incorrect sequence of pitches.
AntC said,
January 31, 2025 @ 9:43 am
suspect my ear is not phonetically qualified enough to be reliable
Thank you @Hat for substantiating my suspicions. (I guess my informants know nothing of the name's history; nor of the similar history for Taroko.)
So my original question stands/the example is a distraction:
Is there syllable stress within polysyllabic imports? Are there regular rules? Can it be phonemic? Does it inherit from the source language? (supposing that has word-stress)
This is not a question about multi-syllabic compounds from pukka Sinitic morphemes.
Mark Liberman said,
January 31, 2025 @ 9:49 am
See also San Duanmu, "Syllable structure and stress." The handbook of Chinese linguistics (2014), as well as his other publications on related subjects.
Or for another perspective, Hana Třísková, "Is the glass half-full, or half-empty? The alternative concept of stress in Mandarin Chinese." Studies in Prosodic Grammar 2019.
Overall, there's a large linguistics literature on this subject.
Victor Mair said,
January 31, 2025 @ 11:13 am
汉语韵律语法丛书 A Series of Books on Chinese Prosodic Grammar
http://www.blcup.com/EnSeriesBook/index/1603
Includes a pertinent title by Duanmu San
9787561953884 汉语三音节韵律问题研究 Prosodic Studies of Chinese Trisyllables, Cui Sixing
汉语的双音化 The bisyllablization of Chinese, Zhuang Huibin, Zhao Pusong, Feng Shengli
9787561953853 汉语的弹性词 Elastic Words in Chinese, Qiu Jinping
9787561954201 声调、语调与句末语气词研究 Tones, Intonation and Sentence-Final Particles
9787561944110 汉语的韵律词 Chinese Prosodic Words, Pei Yulai
9787561944134 音步和重音 Foot and Stress, Duanmu San
9787561944127 汉语韵律语法问答 Questions and Answers on Chinese Prosodic Grammar
9787561943533 汉语合偶双音词 The Combined Disyllabic Words in Chinese, Wang Yongna
9787561943625 汉语的最小词 The Minimal Words in Chinese 洪爽 著 Hong Shuang
9787561943526 汉语嵌偶单音词 The Embedded Monosyllabic Words in Chinese, Huang Mei
9787561943502 汉语的韵律形态 Prosodic Form in Chinese, Wang Lijuan
9787561944103 汉语的句法词 Chinese Syntactic Words, Zhuang Huibin
9787561935330 新汉语水平口语考试 HSKK（高级）应试指南（含1MP3）刘芳 编著
Courtesy of Antonio L. Banderas
Chris Button said,
January 31, 2025 @ 1:10 pm
@ Ant C
My suspicion then is that the matriarch is pronouncing "m" as [ʋ] in accordance with Paiwanese "Tjavualji". You are then associating the [ʋ] sound with a "vaguely nasal quality" to the vowel preceding it.
Incidentally, there's a long history of /m/ ~ /ʋ/ interchange in Sinitic. It's something I've noted on LLog in the past regarding characters like 卯, whose m- pronunciations seem to come from ʁ → β → ʋ → m. Conversely, Old Chinese m- → ʋ- in Early Middle Chinese in most environments before -u-.
Jonathan Smith said,
January 31, 2025 @ 9:13 pm
We can dodge some of the issues haunting discussions of "lexical stress" in Mandarin by considering just fully-toned (i.e. no "neutral tone") two(or more)-syllable words… and OK monomorphemes to disentangle from problems with "compounding"/"lexicalization"/"the word".
The notion that you could retain the phonological form of such baseline items — zhīzhū 'spider', fǎngfú 'seemingly', Tàimálǐ, etc., etc. — but shift the position of something called "stress" to produce a contrasting cromulent word makes no sense and can't even begin to be explained to a linguistically naive Mandarin speaker. I.e. (I think all would agree?) the idea of zhīZHŪ ≠ ZHĪzhū or something is plain wacky. Since that's the case, Mandarin can't have "lexical stress" in anything like the sense in which the term applies to say English. The intonational properties of these words and others (beyond tone) fall out of more general stuff — trochaic rhythm (a Mandarin thing and one which over time has yielded lexicalized "neutral tone" words), declining FO (a human language thing), contrastive emphasis (ditto), etc.
San Duanmu said,
February 1, 2025 @ 11:08 am
Thanks to Victor for informing me of this discussion. It is common that place names (and personal names) tend to be conservative, even though the language has changed elsewhere, that is, the same orthography is pronounced differently by people from elsewhere. The BBC is fully aware of this. In its pronunciation guide, there is an advice: "as a matter of courtesy, the bearer of a name or title should be referred to by the pronunciation which he himself prefers; and that place names should be pronounced as they are locally…A name is usually a matter of vital moment to those closely and often emotively concerned with it, and unfavourable reaction to a mispronunciation…is immediate."
Zev Handel said,
February 1, 2025 @ 7:17 pm
Interesting discussion.
Seems to me that we can't draw any meaningful conclusions without hearing recordings of AntC's pronunciation and of the Taiwanese family's. Otherwise we are just speculating based on our prior assumptions about the prosodic effects at play.
AntC said,
February 2, 2025 @ 3:47 am
@Zev, what I was hoping to elicit was Prof Mair's contacts in Taiwan confirming the canonical Taiwanese pronunciation/stress pattern amongst L1 Sinitic speakers. This would not necessarily be the indigenous Paiwan pronunciation, nor necessarily what you'd guess from the Chinese character spellings — noting these have varied historically.
My pronunciation is certainly no evidence of anything. There are reasonable explanations above for what I thought I was hearing.
Victor Mair said,
February 2, 2025 @ 9:48 pm
From South Coblin:
I believe the primary point of what Jonathan Smith is saying is that “Mandarin can't have 'lexical stress' in anything like the sense in which the term applies to say English.”
Mandarin does of course have stress, but it is used purely for emphasis, not as a morphological marker. Phonetically, it is produced by lengthening a syllable and increasing its sonic volume. The tone contour of the stressed syllable is preserved, but is attenuated by the lengthening. For example, one can say 我說的是“老人”， 不是 “小人‘. “What I said was an “
old
person” not a “
petty
person”. Where the bold forms (in Chinese) and the underlined ones (in the translation) are stressed. But stress in Mandarin cannot be used morphologically as in the contrasting English forms “That is just a rèpeat of what John said.” vs. “Please repeàt what John said.” I suspect this is what Smith is getting at.
I have never heard of the Taiwanese town they are talking about in your log. I asked Jing [South's wife], and she hasn’t either. She refuses to try to pronounce it on the basis of its Chinese written form. Since she has never heard it said, she won’t say it. (That’s what makes her a good informant! She only says what she knows. She doesn’t try to make up things to please the inquirer!) If you would like to hear the name pronounced by a Taiwanese speaker who presumably does know it, you can do so at this site:
I have no idea what kind of Mandarin the “matriarch” in these discussions speaks. I would want to hear what she says with my own ears before offering an opinion on her rendition of the word. I don’t know how accurate the writer’s account is or how good his ear is. Guessing would be a waste of time.
Chris Button said,
February 2, 2025 @ 10:58 pm
I think it is important to distinguish between a stressed syllable and an accented syllable. A stressed syllable can attract an accent.
So, the verb proJECT has stress on the second syllable, where it can attract an accent. The noun PROJect has stress on the first syllable, where it can attract an accent.
In standard Mandarin, an unstressed syllable can lose its ability to carry tone (i.e., accent).
Chris Button said,
February 2, 2025 @ 11:14 pm
Instead of "i.e., accent", I should have said "and therefore accent"
Jonathan Smith said,
February 3, 2025 @ 9:51 am
Yes, what South said —
"lexical stress" like English = the position of stress in multisyllabic words is variable and needs to be regarded as part of a word's phonology; two words could differ only in terms of this position (above for examples)
nothing like this is found in Chinese languages AFAIK. if we like, we could regard (some of?) them as having a kind of "fixed stress" — e.g., Mandarin 2-syllable nouns may be regarded as X-x trochees, with unstressed/destressed final syllables susceptible to being "neutral tone"ed. Whereas many Wu and Min languages seem to have "fixed stress" of the opposite kind i.e. […x-]x-X iambs, with non-final unstressed syllables historically affected by the so-called "sandhi" tone changes i.e. to varying degrees neutralized.
Actually one could argue (I have…) that emergence of lexical *third* tone on the second syllable of certain (e.g.) Beijing Mandarin words is an alternative or advanced kind of neutralization, mirroring certain Min-type changes in that the "neutralized" syllable rejoins the regular tonal paradigm
Chris Button said,
February 3, 2025 @ 5:20 pm
A distinction such as dōngxī "east-west" vs dōngxi "stuff" is a stress distinction. The use of neutral tone in the second syllable of dōngxi may be compared to the use of schwa in an unstressed English syllable.
Presumably when intonational tone (i.e. accent) kicks in, the intonation nucleus can only occur on a stressed (i.e. tone bearing / non-neutral tone) syllable, who tone it then distorts? So, "fundamental" can only bear accent on "fund" and/or "ment" in the same way that dōngxi can only bear it on dōng?
Chris Button said,
February 3, 2025 @ 5:21 pm
"who tone it then distorts" should say "whose tone it then distorts". Or better still "whose lexical tone it then distorts".
Jonathan Smith said,
February 3, 2025 @ 6:05 pm
If one wants to go there, the point WRT e.g. dōngxi 'thing' ≠ dōng-xī 'east-west' is that you can't describe these words (or any similar Mandarin pair) as O-o ≠ o-O, as could be the case for an e.g. English minimally contrastive pair. That much is clear. Instead one might say that the lexicalization of the former — which began as exactly the same dvanda-type compound which the latter remains, with 'east-n-west' > 'here-n-there'~'this-n-that' (cf. recent LL post on such words) — has seen it pounded more deeply into the pervasive trochee-mold. So the 2nd syllable of the word 'thing' is of late (at least in Beijing etc.) lexically specified as "light" -o, in which respect it contrasts with all the other *three* syllables on display.
Chris Button said,
February 3, 2025 @ 7:51 pm
I think the relationship between dōngxī vs dōngxi and 'project vs pro'ject is the same stress-based relationship. It is immaterial whether the difference surfaces as differences in tone or vowel. The stressed syllables may then attract the accent or intonational (non-lexical) tone, but they do not need to be accented–that comes down to the speaker's intonation.
Jonathan Smith said,
February 3, 2025 @ 8:44 pm
To return to the topic of the post, Question was is there [variable] stress within polysyllabic imported [or maybe also native] [Mandarin] words (such that e.g. "Tàimálǐ" is really to be specified as either 'Tàimálǐ [what AntC felt theythemselves said] or Tài'málǐ [what AntC felt interlocutor said] or Tàimá'lǐ [other option])? Chris Button's Answer: yes (I guess?). Actual answer: no, variable stress is not a thing in Mandarin. Bringing in such a device means 1000 times more tools than there is work to do, and fails to account for the sounds from the mouths, and is counter to native intuition, and is wrong.
Chris Button said,
February 3, 2025 @ 9:03 pm
Project (n) vs project (v) is not variable stress. There is a historical reason why the noun is stressed on the first syllable and the verb on the second syllable. It holds across many other pairs and is lexcially encoded for the part of speech. A speaker cannot vary it.
I don't actually think any of this discussion is pertinent to the question at hand. I wasn't really sure why you brought the topic up at all in the first place, but it's an interesting topic nonetheless. So, I figured I would add some thoughts.
As for my answer to the question at hand, it had absolutely nothing to do with stress or accent (which are not the same thing). Instead I guessed that it might be more to do with the effect of the articulation of the "m" causing a perceived shift in how the syllables were being heard:
~flow said,
March 8, 2025 @ 6:02 am
Having read the discussion I now have two (crazy?) ideas in my mind:
(1) Languages like English can be described as tonal. Like e.g. Mandarin English has tones but only a single 'full' one that contrasts with a neutral (default) tone (not sure whether you'd need a second or maybe a third 'full' tone for secondary stress which I believe is largely following rules in English once you know where the primary stress is?)
At any rate we can start out with a model that has a single tone contrasting with a neutral one with the stipulation that a 'word' must have a single full tone; it naturally follows that the *position* of that full tone then becomes a way to distinguish words (symbolically, 'project' has either pattern 1-0 or 0-1, that is either the first or the second syllable is stressed). This then is, in essence, no different from a Mandarin word with tones 2-4 as compared to one with the same basic syllables but tonal pattern 4-2. It's just that Mandarin has more than one way (namely four ways) to put stress on syllables and there's no restriction on the number of any tone within a word. (There's sandhi e.g. for the third tone but even if you follow the often-given advice of pronouncing a 3-3 word as if it were a 2-3 it's still a surface phenomenon, not a lexical restriction.)
(2) Mandarin does have a stress-like thing; people maybe can't imagine "the idea of zhīZHŪ ≠ ZHĪzhū" (Jonathan Smith) but they regularly know and use the distinction between dōngxī "east-west" and dōngxi "stuff" (Chris Button). In the above notation, the latter is pattern 1-1 contrasting with pattern 1-0. There are certain restrictions on where and how many syllables can have the neutral ('zero') tone, and a word (well, actually most words) may have no neutral tone at all.
This way of describing the phenomena we observe in English and Mandarin would put both languages onto the same playing board, as it were, it's just the rule books that differ, not unlike chess and checkers—still very different games but played on the same board. It's hard to describe chess as a set of differences to figure skating, but that's easily done when comparing chess to checkers, which might lead to some insights.
Provided the idea is not totally nuts.