« previous post | next post »

Jonathan Dushoff sent in this photograph of a sign in the Lukang (Lùgǎng 鹿港) public library in Taiwan (apologies for the reflection off the surface):

Jonathan says, "It's obvious how a computer would make that translation; not clear why a human (at the library!) didn't spot it."

The translation software (or somebody) made this mistranslation ("Invites the slipper") because of problems with polysemy, parsing, and homophony.  As a matter of fact, depending upon their frame of mind and level of familiarity with Chinese language and characters, even a human being may have to pause for a moment to correctly interpret the intended message.

The Chinese consists of three characters, each with bopomofo phonetic annotation along the right side):

qǐng 請 ("please; invite; request")
tuō 脫 ("take off; remove; shed; doff; escape; get away; come off")
xié 鞋 ("shoe")

Google Translate, Baidu Fanyi, and Bing Translator all render it perfectly as "Please take off your shoes."  Even iCIBA has "shoes off; please take off your shoes; take off your shoes; please take your shoes".

One begins to wonder how this mistake ("Invites the slipper") actually occurred.  Where did the "slipper" in the sign come from, if not from translation software (which doesn't seem to be the culprit in this case)?

The Mandarin word for "slipper" is tuōxié 拖鞋, which consists of two morphosyllables:

tuō 拖 ("drag; haul; tow; pull; draw; delay")
xié 鞋 ("shoe")

Thus, it would appear that the homophonous term tuōxié 拖 鞋 ("slipper") interfered with the processing of tuō xié 脫鞋 ("take off / remove shoes") and replaced it in the English translation.

I asked about two dozen native speakers of Mandarin if they thought that they pronounced tuōxié 拖鞋 ("slipper") and tuō xié 脫鞋 ("take off / remove shoes") exactly the same.  The results of my survey are rather astonishing.

Nearly all individuals who are highly literate in characters (humanists) and professional language teachers maintained that they pronounced tuōxié 拖鞋 ("slipper") and tuō xié 脫鞋 ("take off / remove shoes") in an identical fashion.  But there were two categories of native speakers who perceived a difference in their own pronunciation of the two expressions:  those who are highly qualified linguists and those who are not very literate in characters.  How can we make sense of this phenomenon?

I think that, when native speakers claim they are pronouncing these two expressions in exactly the same way, they are being unduly influenced by the characters, that they are indulging in what we may refer to as "reading pronunciation".  It's somewhat comparable to someone pronouncing "Wednesday" and "February" the way they are spelled instead of the way they are spoken in real life.

As for myself (although I am not a native speaker, I possess near native fluency in MSM), I have never felt that tuōxié 拖 鞋 ("slipper") and tuō xié 脫鞋 ("take off / remove shoes") were pronounced identically in actual speech.  Simply for innate, cognitive reasons, I'm certain that I make a slight pause between 脫 and 鞋 of 脫鞋 ("remove shoes"; VO), whereas there is no pause between the two syllables of the disyllabic noun 拖鞋 ("slipper") in actual speech.  For example, in these two sentences:

chuān tuōxié 穿拖鞋 ("wear slippers")
qǐng tuō xié 請脫鞋 ("please take off [your] shoes")

I will definitely insert a slight pause between the 脫 and 鞋 of 脫鞋 ("remove shoes"), but not between the two syllables of the noun 拖鞋 ("slippers").

When I asked my semi-literate or illiterate (in characters) friends who are native speakers of Mandarin why they thought tuōxié 拖鞋 and tuō xié 脫鞋 were not identical in pronunciation, most of them could not articulate any particular reason, but when I pressed them further, several of them said that it was due to the fact that tuō xié is a verb-object construction, whereas tuōxié is a noun.  Incidentally, I elicited their responses simply by wearing a pair of slippers and by taking off one of my shoes, and asking them to say what I was doing in each case, then asking them to tell me if they thought the word for "slipper" and the words for "take off" sounded exactly alike.

Now, when it comes to the linguists, we get much more sophisticated explanations, such as this one from Jiahong Yuan:

Attached is a recording I made. It contains two sentences:  shāngdiàn lǐ mài tuōxié 商店里卖拖鞋 ("in stores that are selling slippers"), and jìnmén yào tuō xié 进门要脱鞋 ("when you go inside you have to take off your shoes"). The words "tuoxie" are marked in the textgrid file. 拖鞋 and 脫鞋 are probably slightly different in my pronunciation, but the intuition is vague.

San Duanmu's proposal is that in Mandarin Chinese disyllabic words have a stress on the first syllable; compounds have a stress on the non-head word. So in 拖鞋 (a disyllabic word) the first syllable should be stronger, and in 脫鞋 (VO compound) the second syllable should be stronger (The phonology of Standard Chinese: pp.136). And the relationship between 鞋 and 拖鞋 is related to what San calls "elastic word length": see his papers 现代汉语词长弹性的量化研究 [A quantitative study of elastic word length in Modern Chinese] and "How many Chinese words have elastic length?".

Catherine, Yanyan and I did a study on the stress patterns of polysyllabic words in Mandarin, and we found that the first syllable of a disyllabic word is stronger: Catherine Lai, Yanyan Sui & Jiahong Yuan, "A Corpus Study of the Prosody of Polysyllabic Words in Mandarin Chinese", Speech Prosody 2010.

BTW, the drawing on the sign seems to suggest that patrons are encouraged to go barefoot in the library, which would be frowned upon in most American libraries.  On the other hand, I don't know how one might visually indicate that patrons are requested to enter the library in stocking feet.

[Thanks to Zhao Lu, Maiheng Dietrich, Grace Wu, Melvin Lee, Liwei Jiao, Rebecca Fu, Wei Shao, Ziwei He, Jiajia Wang, Andy Lee, and several informants who wish to remain anonymous]

Update –

From Karen (Kaiyan) Yang:

As for 拖鞋 and 脱鞋, in Putonghua I do pronounce them exactly alike. I asked some of my friends from Beijing and they said in Beijing Fangyan [VHM: Pekingese] it's also the same. But I think in real daily-life conversation, there are also cases when we emphasize on "拖" in 拖鞋, in order to distinguish it from other types of shoes, or emphasize on "鞋" in 脱鞋, in order to emphasize the mood to make the person take his shoes off. Despite this, I still think it's the exactly alike in Putonghua.

However, if you have interest on the situation in other topolects, such as mine, the Lower Yangtze Mandarin, the difference between the two is distinctive, for "脱" is pronounced out in entering tone in my topolect. There are also other examples of such cases when the pronunciations of two words show no difference in Putonghua, while differ a lot in topolect. Here's a table of some examples.

Lower Yangtze Mandarin

(my topolect)

全不 ʨʰũ pəʔ ʨʰyɛn pu
全部 ʨʰũ pʋɯ ʨʰyɛn pu
检察 ʨĩ tsʰɛʔ ʨiɛn tʂʰa
检查 ʨĩ tsʰa ʨiɛn tʂʰa
权力 ʨʰũ liʔ ʨʰyɛn li
权利 ʨʰũ lɿ ʨʰyɛn li
脱鞋 tʰʊʔ xɛ tʰuo ɕiɛ
拖鞋 tʰʊ xɛ tʰuo ɕiɛ
发钱 fɛʔ ʨʰĩ fa ʨʰiɛn
罚钱 fɛʔ ʨʰĩ fa ʨʰiɛn

Except for the last pair, the pronunciations of all the other pairs are exactly alike in Putonghua they while differ distinctly in my topolect. The point is that, "不" in "全不"·, "察" in "检察", "力" in "权力", "脱" in "脱鞋" are pronounced out in entering tone, which does not exist in Putonghua now.
The opposite case is the last pair. In Putonghua, 发 and 罚 differ in their tones, while in my topolect, they are both pronounced in entering tone, thus in the case of this pair, the pronunciations are exactly alike in my topolect, while they differ in Putonghua.

I don't know if cases in other topolects are similar or not, but I guess at least in topolects that have entering tones, the two pronunciations won't be exactly alike.

Sorry I wrote so many irrelevant things hahaha…. I don't know if you have interest or not, but I do feel proud that Lower Yangtze Mandarin, as well as other topolects, show more delicacy in pronunciation.



  1. Jonathan Dushoff said,

    July 22, 2014 @ 7:37 am

    Patrons seem equally welcome to browse the kids' section in socks or barefoot, as long as they remove their shoes.

    I'm the one who should apologize for the reflection; I would have had to take off my shoes to get a better angle :-).

  2. Jonathan Dushoff said,

    July 22, 2014 @ 8:02 am

    You could try asking people to say: 我要脫鞋 vs. 我要拖鞋. (Phonemically, these sentences are the same; one means "I want to take off (my) shoes", and the other "I want slippers".)

    I wonder if that would convince a lot of people that they do in fact say it differently. It's convincing to me, but I'm pretty far from being a native speaker.

  3. Lugubert said,

    July 22, 2014 @ 2:59 pm

    In my favourite Swedish library, people of all ages are asked to remove their shoes in the children’s department.

  4. Gene Buckley said,

    July 22, 2014 @ 3:32 pm

    A linguistics dissertation last year at Penn, by Yanyan Sui, also argues for subtle trochaic (initial) stress in MSM compounds. As mentioned in the post, this would account for the different pronunciations of the noun compound and V+O phrase.

  5. Rubrick said,

    July 22, 2014 @ 4:11 pm

    This post causes me to wonder tangentially what the state of the art is in text-to-speech for MSM (or other topolects, but presumably those would lag). A tonal language with character-based orthography would seem to present unique challenges, but perhaps there are some compensating advantages?

  6. Victor Mair said,

    July 22, 2014 @ 9:04 pm


    One of the hardest aspects of text-to-speech for Sinitic languages is the fact that many characters have more than one pronunciation.


    "The unpredictability of Chinese character formation and pronunciation"

    "Chinese characters with multiple pronunciations"


    The difficulty of learning Chinese characters is compounded by the fact that many of them are pronounced in more than one way, depending on the context. Among the 2,400 most common Chinese characters (which account for some 99 percent of the Chinese characters in most texts), one in five (20 percent) has more than one pronunciation. As bad as that is for those who need to learn to read Mandarin in Chinese characters, the situation is even worse, because the 500 most common Chinese characters, which comprise 80 percent of those used in most texts, are even more likely to have multiple readings.


    Here's a list of characters with multiple pronunciations:

    Many of them have three, four, or five different pronunciations. I think I once found a character that had eleven possible pronunciations and wrote them down on a paper that I taped to my wall, but the painters tore it down and threw it away when they were redoing my office.

  7. Victor Mair said,

    July 22, 2014 @ 9:44 pm

    From Jiajia Wang:

    I would say the same [that their pronunciations are identical]. However,I would stress on TUO xie if I would like to say pls TAKE OFF your shoes or you should wear a pair of TUOxie.

  8. Jonathan Dushoff said,

    July 23, 2014 @ 2:11 am

    Sorry, I missed the point that translation software doesn't seem to have caused the mistake. I should admit that I hadn't thought it through much when I said it was a computer mistake, but I still believe it.

    The entry systems I see native speakers use are adaptive. Thus, you could enter 脫, get the right character, then enter 鞋 (the last character in the phrase) and hit enter, without noticing that the computer had decided to change 脫 to 拖. In this case, the translation software would be correctly translating the wrong input, and I guess you could argue whether this was a computer mistake or an entry mistake.

    It's hard for me to see how a person could make this mistake without the help of a computer.

  9. Matt Anderson said,

    July 23, 2014 @ 10:04 am

    Was 敦 the character with 11 pronunciations? The Hànyǔ dà zìdiǎn lists 10 pronunciations for it (dūn, duī, tuán, diāo, dùn, dào, zhǔn, tūn, duì, and tún). The graph, when it appears in the "Wǔ shùn" 武順 section of the Yì Zhōu shū 逸周書, is also sometimes read as diàn 殿, which would bring the total up to 11 (I think this last one, though, is either a graphic error or a misunderstanding, not a real additional reading).

  10. Victor Mair said,

    July 23, 2014 @ 2:17 pm

    That's a good one, Matt! It shows that I wasn't hallucinating when I said that I had encountered at least one character that had eleven pronunciations in MSM. And dūn 敦 certainly means a lot to me, since it forms the first part of Dūnhuáng 敦煌, on the medieval manuscripts from which I spent over two decades of my life. So perhaps it really was dūn 敦 that I was thinking of and whose multiple pronunciations I had posted on my wall for many years till the painters destroyed that precious piece of paper (and many other treasures from my office!).

    Still, yesterday when I was trying to recall which character it might have been, I kept thinking that it was some character with the phonophore 豈, and that it had readings such as qi, kai, wei, yi, etc. (in various tones), one or more of which meant "mill", but I may be wrong about that. Perhaps I really was thinking of dūn 敦. Lord knows I would have had occasion to look it up many times, and I wrote several papers on the name Dūnhuáng 敦煌 in which I linked it to Ptolemy's Throana and ultimately to Indic draṃga (“frontier post”), and beyond that to Iranian *druvāna (“stronghold”).

    Here's an easily accessible reference: (see the second paragraph)

    See also Sino-Platonic Papers 16 (March, 1990) (mentioned several places therein).

    This one is harder to get, but it is much more complete (available in the Penn library system):

    Victor Mair, "Reflections on the Origins of the Modern Standard Mandarin Place-Name 'Dunhuang' — With an Added Note on the Identity of the Modern Uighur Place-Name 'Turpan'", in Li Zheng, et al., eds., Ji Xianlin Jiaoshou bashi huadan jinian lunwenji (Papers in Honour of Prof. Dr. Ji Xianlin on the Occasion of His 80th Birthday) (Nanchang: Jiangxi People's Press, 1991), vol. 2, pp. 901-954 (very long and detailed study).

  11. Matt Anderson said,

    July 23, 2014 @ 5:02 pm

    I did think 敦 seemed especially appropriate!

    But was the character in question 磑? When pronounced wèi, it means 'mill'.

    (I apologize in advance for the following being overly technical.)

    磑 only has 2 readings in most dictionaries (wèi and ái), and three in Hànyǔ dà zìdiǎn (wèi, ái, and gài), but the Kāngxī zìdiǎn lists 7 readings for it:

    《唐韻》《集韻》五對切《韻會》魚對切,x音x (wèi)
    《韻會》魚回切,音嵬 (wéi)
    《集韻》魚衣切,音沂 ()
    居希切,音機 ()
    《玉篇》公哀切,音該 (gāi)
    《集韻》魚開切,音皚 (ái)
    《字彙補》蒙破切,音磨 ()

    This doesn't include Hànyǔ dà zìdiǎn's gài (居代切), which it says is the same as x. Strangely, though, in a quick glance through my references, that's the only mention I can find of x having the reading gài—the Hànyǔ dà zìdiǎn itself only lists the reading and the Kāngxī zìdiǎn adds the reading ái (which I've already listed for 磑). The Guǎng yùn also includes the reading gāi (公哀切, again already listed under 磑), for whatever that's worth.

    So it might not be justified, but if you add gài and to the Kāngxī zìdiǎn's readings, you get 9 different pronunciations. According to the Kāngxī zìdiǎn, x additionally has the meaning 汔 (it just states "汔也" with no pronunciation given—it may be that the intended pronunciation for this meaning is likewise ).

    I'm on pretty shaky ground now, but adding that pronunciation would bring us to ten (wèi, wéi, yí, jī, gāi, ái, mò, gài, qí,and ). Who knows, maybe someone else has proposed another one?

    Of course, the words represented by three of the above modern pronunciations (wéi, gāi, and ái) only occur in reduplicated compounds, presumably used for their sound value alone. Wèi means 'mill' and supposedly means 'millstone', and , , gài, , and , assuming these readings really exist, are simply existing words normally written with a different character which can be written instead with this character. As far as I can tell, other than wèi, each of these readings, if you disregard occurrences in lexicological/phonological/graphological works, occurs only once in history (though it's certainly possible that some of them were used more often).

    With all this confusion, it seems reasonable to me that the editors of Hànyǔ dà cídiǎn, for example, seem to have just thrown up their hands and gone with only two readings, wèi for 'mill' and ái for, I think, all of the cases in which the graph is being used for its sound alone.

    A few characters broke this comment when I first tried to post it, so now they are all replaced by x's.

    In the fifth paragraph, "x音x" should read "竝音[酉+豈]", where [酉+豈] is a character pronounced wèi, and 竝 was originally an alternate form used to write the same word.

    The remaining x's, in the sixth paragraph (starting with "This doesn't include Hànyǔ dà zìdiǎn's gài…") and the following one, should all be [豈+幾], a character normally read as .

    All the problematic characters are part of Unicode's "CJK Unified Ideographs Extension B", so I guess that’s where the problem lies.

  12. Victor Mair said,

    July 23, 2014 @ 9:19 pm

    @Matt Anderson

    Thanks for all that good work!

    It really must have been 磑 that I was thinking of.

    Zhongwen da cidian #24952 gives the following pronunciations (using GR [they also have bopomofo]): wey, kae, wei, yi, ji, gai, air, moh, and gay, for a total of nine. Somebody (I don't think it was me) has written in by hand (I believe that it was written onto the camera ready plates) nái, and that would bring us up to ten. Although I don't know what the sanction for nái is, it doesn't occur on your list of ten, so that would bring us up to eleven. However, even without the nái, the nine in the Zhongwen da cidian list do not include your qì and qí, and your list does not include the kae of the Zhongwen da cidian. So, combining the two lists, we would have exactly eleven! That is what I must have had on my wall all those years. Mystery solved!

    Morohashi #24402 only gives a total of seven pronunciations for 磑.

  13. Victor Mair said,

    July 24, 2014 @ 8:06 am

    From Tingyu Liu, a trained and practicing anthropological field linguist:


    There will be a stress on the 脱 of 脱鞋 because it is a verb, and 鞋 will be a little shorter. No stress for 拖 of 拖鞋.



  14. Stephan Stiller said,

    July 24, 2014 @ 1:58 pm

    I think it is quite plausible that a human might not spot the mistake, the reason being precisely the homophonic nature of the two expressions in question. This is consistent with the idea that writing systems are primarily processed phonetically (by which we normally mean "phonemically": words are accessed via their phoneme sequences). In English you see this when people write "wierd", "per say", and "judge mental". Not making those mistakes requires a certain ability which not everyone has. Any non-phonemic orthography makes life difficult for part of the population, as the goal is nowadays general education, not just that of an elite.

    I do think that the distinction in pronunciation is phrasal, not lexical/phonemic, at least not yet. That is, VO-compounds (even if frequent and even if lexicalized) are probably treated as two words on a prosodic level. One can call this "stress", but if it's on the phrase-level, talking about "intonation" is in my opinion better.

    To answer a different question in this thread: Text-to-speech (TTS) is quite good if we convert from a text's pinyin representation. If we start from a text's representation in Chinese characters, we have the problem of polyphonetic characters (多音字; there is no standard English translation afaik, but I'd go with "polyphonetic") which Victor Mair discussed.

    If we do work with a pinyin representation, prosody in TTS will still be challenging (like for any language), but the syllabic output will be very clear. The reason is that there are relatively few phonemic (or phonetic, for that matter) effects/interactions of morphology across syllable boundaries in MSM.

  15. JS said,

    July 24, 2014 @ 2:01 pm

    @Prof. Mair,
    Though Tingyu Liu's account is exactly the opposite of San Duanmu's, who argues for stress on the first syllable of disyllables like 拖鞋 and on the O in V-O compounds like 脱鞋.
    We really need a well-designed experiment along the lines suggested by Jonathan Dushoff, where native speakers attempt to distinguish the two words when spoken in isolation or within ambiguous contexts. (You might think Duanmu would have tried something like this were he serious about the idea of Mandarin having "word stress.") I'm not quite prepared to promise to eat my 拖鞋 if this and similar pairs turn out to be readily distinguishable — but color me skeptical.

  16. Bathrobe said,

    July 28, 2014 @ 8:43 pm

    I would pronounce them the same. Of course if you wanted to distinguish them you could without too much effort, but in the normal course of things I suspect that the two aren't very clearly distinguished (if at all) in everyday speech.

  17. Victor Mair said,

    July 31, 2014 @ 5:24 am

    Once more on tuōxié vs. tuō xié — part 1

    In the lively discussion that followed the original post (o.p.) above, we have seen that there is a wide variety of opinion on the subject of whether tuōxié (N "slippers") and tuō xié (VO "remove shoes") are pronounced exactly alike. In general, the observation I made in the o.p., viz., that those who are highly literate in characters (esp. humanists) and professional Mandarin teachers of beginning to intermediate levels or those who are not yet possessed of advanced spoken fluency (I did not make this point in the o.p.) tend to say that they pronounce the two terms exactly alike, while those who are illiterate or partially literate in characters and those who are highly advanced in linguistic analytical skills tend to perceive a difference in the way the two terms are pronounced.

    Preface: Since this comment is multipartite and quite long, I will break it up into several sections.

    I. My own approach

    Because I always make an effort to enunciate clearly and not slur or gloss over my words, no matter what language I am speaking, I am especially attentive to elocution, and am always concerned that the person I'm talking to is not confused by what I say. In other words, I try as hard as possible not to be ambiguous in my speech. Consequently, I would never pronounce tuōxié (N "slippers") in exactly the same fashion as I do tuō xié (VO "remove shoes"). For tuōxié, rhythmically the syllables would last an equal amount of time and there would be no pause between them (to signal that they belong to a single lexical unit), but I would add a bit of stress to the first syllable to identify what kind of xié. For tuō xié, rhythmically the first syllable would be slightly longer than the second and there would definitely be a slight pause between the two syllables (to signal the verb and the object it is acting upon), while I would put slight stress on the second syllable to identify what it is that I am removing.

    If, after all of this, I think that there's still a chance that I might be misunderstood in a given context, I may adopt even further strategies for disambiguation, such as — for example — giving a slight hint of the underlying tone of what canonically would be specified as a neutral tone. Above all, I do not woodenly pronounce the syllables of a sentence as they would be stipulated in isolation according to a dictionary. Language is living, and it varies significantly according to sentence length or longer utterances. We all know how comical it can sound when someone pronounces English one syllable (or one word) at a time as though each of them was just picked out of a dictionary and mechanically rendered the same way in every sentence.

    When I first started developing these speech patterns, it was all very conscious and deliberate, but by now it has become second nature. I was weaned on Y. R. Chao, and he was very careful to make such phonological distinctions based on grammar and morphology. I was also fortunate to be married to one of the world's best Mandarin teachers (second year and higher levels) for forty-one years, and I followed her speech patterns implicitly.

    I follow the same precepts for all homophones and near homophones that belong to different grammatical categories (i.e., are different parts of speech), e.g., lǎoshi 老是 (ADV "always"), lǎoshi 老實 (SV [Stative Verb] "honest; frank"), lǎoshī 老師 (N "teacher"), etc. I should note, by the way, that the ABC Chinese-English dictionaries make the study of this kind of phenomenon much easier than it would be with other dictionaries, inasmuch as they are single-sort alphabetically ordered dictionaries that group homophones next to each other, regardless of the characters with which they are written.

  18. Victor Mair said,

    July 31, 2014 @ 5:31 am

    Once more on tuōxié vs. tuō xié — part 2

    In the following comments, I present the views of ten of the best speakers of Mandarin I know, whether they be native speakers or nonnative speakers who possess native fluency, and also those who are trained as phonologists and language teachers who are capable of analyzing their own speech patterns and the speech patterns of others. You will see that they all are sensitive to communicating clearly the differences between the two sentences. Although they may adopt different strategies to differentiate terms that have dissimilar grammatical properties and may describe their analyses differently, the main point is that they acknowledge that the pronunciation of the two terms is not identical and / or they feel the need to slightly restructure what they say to avoid ambiguity.

    Two other individuals whom I queried simply stated that they would pronounce the terms the same way, so there's nothing to discuss about their views. I seriously doubt that they do pronounce the terms exactly the same way, certainly not in different contexts, but it's not up to me to try to convince them otherwise than they already believe.

    All those whose views are presented in the following comments are responding to this question that I put to them (as suggested by Jonathan Dushoff, who also prompted the o.p.):


    Do you pronounce these two sentences *exactly* alike?

    wǒ yào tuōxié
    "I want slippers."


    wǒ yào tuō xié
    "I want to take off my shoes."

    If you pronounce them differently, please explain / describe how so (especially focusing on the tuōxié vs. tuō xié, of course).


    BTW, I was pleasantly surprised that Google Translate parsed the two sentences correctly.

    I should note that I (and at least one other respondent) would make a significant adjustment on the first two syllables to accommodate the distinctions in the two pairs of two syllables that follow, but since we are focusing on tuōxié vs. tuō xié in this post, I won't go into that here.

    [N.B.: Since these comments will largely be of interest to those who are at least somewhat familiar with Chinese characters, I will not adhere to my usual principle for Language Log posts of providing transcriptions and translations for all characters.]

  19. Victor Mair said,

    July 31, 2014 @ 5:46 am

    Once more on tuōxié vs. tuō xié — part 3

    II. San Duanmu

    This particular example is a bit complicated. In general, a compound should have the stress pattern SW (strong weak), and VO the pattern WS, where W is weaker than S but need not be 轻声. This is true in Shanghai and Chengdu, as well as most cases in English. However, in Beijing Chinese many disyllabic words are WS, such as 大学、酱油、etc., and apparently 拖鞋 as well. Some English NN are WS, too, such as 'room temperature' and 'apple pie'. In such cases, the stress difference becomes easily negligible or confuseable, in Beijing Chinese.

    The point is that if a disyllabic noun is SW in Beijing, where W need not be 轻声, then it should be clearly different from VO. There are such cases, though I don't have the data with me. A good topic for a student's term paper.

    By the way, YR Chao used the example 油井 vs. 有井,and he said the pronunciations are different.

    III. Jiahong Yuan

    It's an interesting question. I think my pronunciation of the two words would still be slightly different in this context: in "我要脫鞋", 鞋 is more stressed whereas in "我要拖鞋", "拖" is more stressed, like what Duanmu has proposed. If we put the two words in a context that the first syllable needs to be stressed/focused, I will probably pronounce them the same:

    vs. 你要托鞋,还是要皮鞋?

  20. Victor Mair said,

    July 31, 2014 @ 5:47 am

    Once more on tuōxié vs. tuō xié — part 4

    IV. Perry Link

    There's no doubt that I "think" the phrases differently, and so feel, subjectively, that there ought to be a slight difference. But whether there actually IS a difference I would want to leave to a mechanical device to determine. My subjective feeling is that for "I want slippers", the fourth tone on 要 would fall a bit farther and there would be a very slightly longer pause before tuo arrives.

    V. Tom Bartlett

    拖鞋 is a binomial noun while 脫鞋 is syntactically a verb-object unit.

    I never thought of this point, but I guess it’s an interesting one. I suppose, as suggested by the differing concatenization in the two romanized terms, corresponding to binomial noun and v-o, I would pause ever so slightly, perhaps often inaudibly (?), in the v-o 脫鞋 case.

  21. Victor Mair said,

    July 31, 2014 @ 5:48 am

    Once more on tuōxié vs. tuō xié — part 5

    VI. Brendan O'Kane

    I'm pretty sure I'd pronounce them differently: in "slippers," xie2 is less emphasized and has a shorter duration. That said, I'd probably avoid any ambiguity by saying "wo3 yao4 ba3 xie2 tuo1 le" — this sounds slightly more natural to me, though I can't say why.

    VII. Julie Wei

    I pronounce them pretty much the same. Repeating them consciously, however, I find for tuo xie ("take off shoes"), I stress 'xie" a little more, drag it out a little more. For tuoxie "slippers", I stress "tuo" a little more.

    VIII. Gloria Bien

    Gosh, Victor, if I still had an Ah-ma, I don't think she'd be able to tell which one I was saying.

    To a close friend, I might say 我要脱脱鞋, so as not to sound too rude and abrupt.

    A vendor might be quite shocked if I announced, 我要脱鞋, and would probably hear "我要拖鞋" no matter whether I pronounced it differently or not.

    Oh, wait–changing the subject–from "I" to "you," I must say when a friend asks "要不要拖鞋?/要不要脱鞋?I'm never too sure whether I'm asked whether I want a pair of slippers or whether a gentle suggestion is being made for me not to track up their floors.

  22. Victor Mair said,

    July 31, 2014 @ 5:48 am

    Once more on tuōxié vs. tuō xié — part 6

    IX. Mark Swofford

    I'd probably say 脫鞋 as an iamb and 拖鞋 as a trochee. But my Mandarin isn't fluent enough for that to count for anything in a survey.

    My wife says she wouldn't say them any differently (assuming one were indeed saying ""我要脫鞋" rather than what for her would be the more natural "我要脫鞋子"), unless "我要脫鞋" was in response to the question "Ni yao tuo shenme?" — in which case "xie" would be given more emphasis in the answer.

    X. Haitao Tang

    I would say the 1st slightly emphasizing xie and the 2nd slightly emphasizing tuo. However, the difference lies elsewhere. Probably people will say, Wo yao tuo xie le/ Wo yao yi shuang tuoxie. In this way the ambiguity is effectively removed.

  23. Victor Mair said,

    July 31, 2014 @ 5:53 am

    Once more on tuōxié vs. tuō xié — part 7

    XI. Andy Lee (late arrival)

    我要脫鞋 vs. 我要拖鞋 are really confusing because these characters pronounce the same correspondingly. And generally I would rely on the context to distinguish the meanings. And when the context does not help, I prefer to say 我要穿脱鞋 to differentiate.

    If I have to differentiate these two sentences with the exact words provided here, I’ll change the stress of the sentence to indicate the meaning. As for “我要脱鞋”,I would put stress on “鞋”, which is in fact emphasizing the object after the verb“脱”;for “我要拖鞋”, I would put stress on “拖”, and “鞋” on this occasion would be pronounced with “轻声(unstressed or light tone)”, which aims to emphasize the modifier of the object.

  24. Zhiqiang Li (Andy Lee) said,

    July 31, 2014 @ 8:42 am

    In fact, in order to avoid the misunderstanding, native speakers tend to add some characters or change the structure of the sentence to clarify the meaning. Take, for example, "我要脱鞋子" and “我要(双/穿)脱鞋”, or "我要把鞋脱了".

    And when it comes to the dialects of China, the misunderstanding would be less possible . For example, the 脱鞋 might be pronounced as "tuō haí" while "鞋“ in "拖鞋 will be pronounced ”hai'er“ by adding "er" to "haí" to indicate the difference in some region of Hubei province. And in some regions of Shan'xi province the term for ”拖鞋“ would be another phrase like ”洒鞋" pronounced as "sā hai". On this occasion there won't be misunderstanding.

  25. Victor Mair said,

    July 31, 2014 @ 10:53 am

    Once more on tuōxié vs. tuō xié — part 8

    XII. Christoph Harbsmeier

    This is indeed a very nice case to think about.

    It seems to me, as a very quick first reaction, irresponsibly off the cuff, that:

    1. Some native speakers regularly make the distinction, others don't.

    2. Some native speakers tend to make the distinction, others don't tend to.

    3. Some native speakers can make the distinction phonetically explicit when required in pragmatic context others cannot.

    4. Some native speakers think they make the distinction, other people think they do not.

    5. Some native speakers think they regularly make the distinction, others do not.

    6. Some native speakers think they can make the distinction phonetically explicit when required in pragmatic context, others do not.

    7. All the above varies quite radically with idiolect, subdialect, sociolect, and dialect, as well as very much with the pragmatic context and in particular the need or absence of a need to avoid misunderstanding in context.
    8. Probably, none of the groups of native speakers between which I am trying to distinguish in 1-6 above coincide completely with each other.

    9. No native speakers are reliable judges of all the distinctions that they do or do not make or tend to make on what occasions. One does make many distinctions that one is unaware of. (For example, few Danes are aware of all the vowel distinctions and allophones which they actually use quite proficiently in practice… I, on the other hand, am actually unable to hear some of those fine differences which they make so clearly… Such is life!)

    I think what one would need to do is to analyse (using something like the separate use of tuoxie and tuo xie in representative idiolects in different conversational and pragmatic contexts. One obviously does NOT want to ask native speakers how they make the distinction. One needs to make unsuspecting native speakers use these expressions in different contexts and analyse exactly what they tend to do in various contexts, and always in spontaneous non-demonstrative speech. Reading out texts would obviusly be of little use. Pragmatic contexts where one would need to make the distinction explicit could be useful to create at the end of the procedure, also.

  26. Victor Mair said,

    August 1, 2014 @ 10:17 am

    Once more on tuōxié vs. tuō xié — part 9

    XIII. From Yanyan Sui:

    My understanding is that the difference lies in whether the second syllable can be reduced. In a VO disyllabic phrase, the O syllable does not reduce, but in a compound, the second syllable can be reduced. This is because a VO phrase has two word stresses, but a compound has one word stress, and in the metrical nonhead position, the syllable is subject to reduction.

    For VO phrases, the O seems to have a higher level prominence projection, namely, O has more stress than V.

    I do not have quantitative evidence for stress at the phrasal level yet, it is only my observation that in VO phrases O seems to have a phrasal level prominence projection.

    The distinction between compounds and phrases is crucial to the discussion. Disyllabic compounds share the same metrical structure (syllabic trochee, by my assumption), and the syllable at the metrical nonhead is subject to reduction, although the extent of reduction may vary by compound structures and individual words, etc.

  27. Old Gobbo said,

    August 3, 2014 @ 10:03 am

    @Victor Mair
    I am afraid it is entirely irrelevant to the topic, but I have to tell someone, and with your suggestion that Danes can hear their language “accurately”, you are it.
    It may then be of interest to you that Tuborg, a very fine Danish brewer, is at present advertising a new bottled beer in France with the name “Skøl".

    This would be all very well, were it not that the Danish for “cheers” or “your good health” is actually “skål". (The two do not sound the same, and in fact “skøl" does not exist as a Danish word.)

    However I doubt if I will write to them to complain: a note in New Scientist (18jan14) offered the following warning -

    "The problem we face here is that if we point at anyone's mishandling of the English language, something is likely to go dreadfully wrong with our own use of it nearby. It occurs to us that this phenomenon should be called 'prosaic justice' "

  28. Victor Mair said,

    August 6, 2014 @ 6:42 am

    @Old Gobbo

    Very interesting. What you say about Tuborg's new "Skøl" beer in France reminds me of the pseudo-Danish Bronx ice cream brand name, Häagen-Dazs:


    Mattus invented the "Danish-sounding" "Häagen-Dazs" as a tribute to Denmark's exemplary treatment of its Jews during the Second World War,[4] and included an outline map of Denmark on early labels. The name, however, is not Danish, which has neither an umlaut nor a digraph zs, nor does it have any meaning in any language or etymology before its creation.[5] Mattus felt that Denmark was known for its dairy products and had a positive image in the U.S.[6] His daughter Doris Hurley reported in the 1999 PBS documentary An Ice Cream Show that her father sat at the kitchen table for hours saying nonsensical words until he came up with a combination he liked. The reason he chose this method was so that the name would be unique and original.[7]



    The person who made the comment about the Danes being able to hear their own language accurately is Christoph Harbsmeier, not me. item #9:


    "…few Danes are aware of all the vowel distinctions and allophones which they actually use quite proficiently in practice… I, on the other hand, am actually unable to hear some of those fine differences which they make so clearly… Such is life!"


    Christoph is German by birth, but received his higher education at Oxford and is married to a Dane.

  29. Victor Mair said,

    August 7, 2014 @ 4:44 pm

    From Cao Lin, a graduate student at Peking University; native of Shandong:

    I pronounced the two "wo yao tuo xie" sentences, but without finding out big difference. The problem is when "我 want 拖鞋", maybe I prefer to say "给我拖鞋". "我要拖鞋" isn't used frequently in my speech. I wonder whether this is idiolectal?

    This is a little confusing. It reminds me once my friend who studies linguistics asked me to pronounce some phrases about directions in Shandongese. There was a phrase "山脚下". However, according to my impression, we never use this phrase, instead, we say “山底下/山根儿里”. I still remember the anecdote you told in Beida's class, the one about "ga rou" instead of 买肉. I kind of dislike Mandarin now because people's speeches are limited so much by it. I haven't read lively and vivid fictions for a long time. Most young online writers can only use the same verbs and adjectives repeatedly, repeatedly and repeatedly.

RSS feed for comments on this post · TrackBack URI

Leave a Comment