Pronunciation: 出租车司机 (first tones in a row)

Discussion in '中文+方言 (Chinese)' started by Sibutlasi, Aug 23, 2013.

  1. Sibutlasi

    Sibutlasi Senior Member

    Hello :)

    According to one of the tone adjustment ('tone sandhi') rules stated in various grammars/textbooks of Mandarin, two consecutive first tones T1+T1 are realized as T4+T1. Since Mandarin phrases may contain sequences of 2, 3, 4, 5…, perhaps even more, consecutive syllables with T1, and tone adjustment rules elsewhere apply across word boundaries, a primary tonal sequence like T1+T1+T1+T1+T1, as in “chu1 zu1 che1 si1 ji1” (= "taxi-driver"), may yield very different results depending on how the T1+T1>T4+T1 rule applies.

    For example, if the rule applied recursively irrespective of constituent structure, the outcome should be T4+T4+T4+T4+T1. If, on the contrary, the rule applied ‘cyclically’, bottom up, first to the smaller constituents [chu1 zu1] and [si1 ji1], then to the intermediate constituent [[chu1 zu1] che1], and finally to the whole noun phrase (or compound) [[[chu1 zu1] che1] [si1 ji1]], the result should probably be [[[T4+T1]+T1] [+T4+T1]]. There are still other possibilities, but such details are not clarified in the grammars/textbooks I have seen.

    When I made the Text-to-Speech Site Pal machine pronounce “chu1 zu1 che1 si1 ji1” for me, eight different native Mandarin voices respectively produced tonal sequences very approximately corresponding to the following ‘melodies’:

    1. D + C sharp + C + B + B flat
    2. C sharp + C + B + A + B flat
    3. C + B + B flat + A flat + A
    4. G sharp + G sharp + A + B flat + F sharp
    5. A flat + A + B flat + B + F sharp
    6. B + B flat + A + A + B flat
    7. A + A + B + F sharp + F sharp
    8. B flat + B flat + B flat + A + C

    Obviously, since I was not using special equipment, I cannot claim absolute accuracy, but I did listen carefully and searched for the pitch of each syllable on the piano keyboard several times and I can be reasonably sure that the tonal sequences and intervals used by those eight speakers correspond fairly closely to the melodies just transcribed.

    As you will observe, there are significant differences between 1, 2, 3 or 6, on one side, and 5, 7 or 8, for example (and Cantonese speakers at Text-to-Speech used still other pitch sequences I have not transcribed).

    My question, then, is this: How exactly does the T1+T1 > T4+T1 rule apply? And, in this particular case, is there a preferred tonal rendition of “chu1 zu1 che1 si1 ji1”, and how can the tonal-melodic differences above be explained?

    Thank you in advance.
  2. stellari Senior Member

    Mandarin Chinese
    I would pronounce all five characters with exactly the same frequency.

    I think the 11->41 rule you mentioned applies only when the first character is "一", and obviously does not apply in the phrase 出租车司机.

    In Mandarin, a tone is not defined by its absolute frequency: different speakers have different pitches for T1. Even for the same speaker, he may use different pitches at different occasions. In TTS engines with recorded sounds, each character is recorded separately and therefore a fluctuation in frequency is expected; Also when the sounds of single characters are merged into words and even longer audio signal, additional frequency components may be introduced. TTS is still very far away from producing natural speeches indistinguishable from those produced by human. As such, it cannot be trusted as a source of speech analysis.
    Last edited: Aug 23, 2013
  3. xiaolijie

    xiaolijie Senior Member

    English (UK)
    That is anchronistic, stellari! Although machines may be completely wrong, they are at least consistent and consistency means reliability. In the same way, we should start using Google translation for analysis of grammar, and all the unnecessary grammatical complications that have bothered us so far should therefore be gone!
  4. Sibutlasi

    Sibutlasi Senior Member

    Thank you for your attention stellari.

    The grammars/textbooks I referred to, however, do not say anything about the T1+T1>T4+T1 tone sandhi rule being restricted to sequences starting with "yi1", although they all cite "yi1" as a particularly frequent example (just as they cite the "bù > bú" case when illustrating the rule T4+T4>T2+T4); otherwise, I would not have posed my question in the first place.

    The first sentence of your third paragraph is, nevertheless, correct (as it is for all other tone languages I am aware of, too). Absolute pitch is irrelevant here. What matters is the tone/pitch intervals each speaker uses, whether they start with "chu1" at D6, D5, C4#, C5#, etc. In my 'melodies' above I omitted the fact that they occur at different octaves (4th to 6th), depending on e.g., whether the speakers are male or female and their apparent age, among other factors (e.g. speed rate).

    I was also inclined to take T-T-S data with a pinch of salt, for the reasons you mention, but I have so far found that nearly all of the pronunciations I obtained from T-T-S did correspond rather well to data subsequently elicited from native speakers, Pleco software, online pronouncing dictionaries, and other sources. As a case in point, the tone interval sequence the Google Translator generated for "chu1 zu1 che1 si1 ji1" was exactly that of T-T-S Voice 1. above = D + C sharp + C + B + B flat.

    Thank you for your personal data, anyway. Assuming you have a good musical ear, that will add a ninth (and so far unattested!) possibility to my current inventory of tonal realizations for "chu1 zu1 che1 si1 ji". I will call it the 'zero choice', since it negates the general validity of the tone adjustment rule T1+T1>T4+T1 and the existence of any tone adjustment in such a case.

    Yet, I find it hard to believe that the clear syntactic and semantic functions and constituency groupings among the characters in "chu1 zu1 che1 si1 ji1" should have no tone adjustment effect whatsoever on the pitch of its five syllables. In theory, that should not happen, so I am curious about what other native speakers may have to say about the 'music' of such isotonous series.

    Thank you!
    Last edited by a moderator: Oct 21, 2014
  5. yuechu Senior Member

    Canada, English
    Hello Sibutlasi,

    If I understand correctly, you are talking about having listened to and transcribing the tones of 8 different synthesized 'native' Mandarin voices? (as I know that these programs/demos often have many voices per language and can sound quite real)
    I haven't heard that program, but you may want to try listening to native speakers instead since, as you say, machines cannot always be 100% trusted in terms of accuracy.

    First of all, as stellari pointed out, T1+T1 --> T4+T1 is not a general rule in Mandarin and only for 一 (to my knowledge). If anything, T1 + T1 may become T1 + T0/5(轻声) sometimes, but that's the only tone change I know of there (for standard Mandarin).

    I searched for some videos of "出租车司机" and found the following two pronunciations which you can listen to:
    1) This one is on CCTV and is very standard/official.
    Example 1: CCTV-法治在线-20110611-广东揭阳杀害出租车司机案-1/2 (t=2m11s)
    As stellari mentioned, it is the usual "monopitch" pronunciation. I think this is the most usual and would recommend it as well, especially for a beginner. But if I am not mistaken, I think it is often pronounced chu1zu1che1 si1ji1 with "chu1zu1che1" being higher and "si1ji1" dropping to a lower level (ex: 4 semitones) in 北方话/Northern dialect. I'm not sure about 南方。。
    The "che1", I think, also may drop slightly at the end in anticipation of "si1ji1". As you may have noticed, tones in Chinese are all relative since everyone speaks at different pitch levels. Also, pitch "drops" or "falls" are also different from person to person.

    Example 2: 【飞碟说】出租车司机的爱恨情仇
    It should be the first word you hear in the link (t=18s). It is not a good example (and probably shouldn't be here) because it isn't a typical 普通话 pronunciation. I think it is best to stick with the 'textbook' T1 T1 T1 T1 T1 pronunciation above.
    I think it often has little to no tone adjustment but that it can, especially in informal speaking. The rhythm marks the largest difference.
    I've noticed that the more standard/formal Chinese is, the less clearly polysyllabic words词 are differentiated through tone when speaking. Tones tend to follow more predictable patterns. (this is just my observation, as I have nothing to back it up.. ) I think the same thing happens in a lot of languages though (that the prosody tends to follow more predictable/regular patterns). Or perhaps I am just thinking of the news....

    I hope the first video's "chu1zu1che1 si1ji1" will help you!

    I was originally thinking that some of the final notes (F sharp, F sharp, C) went up at the end. They all go down, right? (I think I read it wrong)

    For a more precise analysis, I would suggest you use a sound analysis program/graphing the sounds so that you can see the changes in pitch visually. Even tones which are supposed to be level can sometimes change (rise or drop) in speech (especially informal speech) in anticipation of other syllables. (once again, just my observation) I don't think this noticeably happens very much though...

    Also, I was trying to play your 'melodies' of 出租车司机 on the piano (since my mental 'sight reading' has become a bit rusty over the years) and... well, it's hard to compare to speech, I find. Once again, I think listening to actual sound files of native speakers will prove most helpful.

    EDIT: Ah, now I know why you are comparing the pitches to notes on the piano--the "robot" voices do indeed give different pitches similar to what you would hear on a musical instrument. I just listened to Google Translate's voice and, although comprehensible, it does not sound like a native standard Mandarin speaker to me...
    Last edited by a moderator: Oct 21, 2014
  6. stellari Senior Member

    Mandarin Chinese
    Seems to me your textbook(s) may have given you the wrong impression that T1+T1->T4+T1 is general rule. Most textbooks and online tutorials I have seen list this tone change as a special rule exclusively for 一.

    I listened to the speech produced by Google translator. It is intelligible, but sounds rather weird, funny, and machine-like. The pronunciation of 出租 is still okay, but when it comes to 车司机, the tone dropped too much that it sounds a casette tape gets twisted. I also checked SitePal. For this particular phrase, only 'Lily' would almost pass for a natural human speech, and her tone has the lowest fluctuation among all eight.

    I am not saying that tone fluctuation is forbidden here. For example, I might sometimes pronounce 司机 with a slightly lower pitch (at most a semitone) than 出租车 to indicate that it is a separate word. However, this is not tone sandhi as it is not mandatory.

    Like I said, TTS produces intelligible speech, but if you really want to know how real people would pronounce those sounds, you should trust native speakers more than computers.
  7. YangMuye

    YangMuye Senior Member

    You must have misunderstood Google translation and have to apologize to her, xiaolijie. ;)

    She was NOT taught a few "right", "consistent" grammar/phonology rules, rather, she learnt from REAL Chinese, e.g. through reading newspapers, novels, watching TVs and of course learning "textbooks"(corpus). :D
    We teach her how to learn rather than what we have learnt, and let herself develop her own grammar.

    All problems are simplified into a statistics or probability problem, the answer depends on: whether she has seen the answer or not; how familiar she is with the pattern or similar paterns(combination of sound or words).
    Her language ability improves over time.

    So she must NOT be consistent and reliable, just like we natives. ;)
    Last edited: Aug 24, 2013
  8. Sibutlasi

    Sibutlasi Senior Member

    Hello baosheng.

    Thank you very much for your answer. I've listened to the videos you pointed me to, and, indeed, particularly in the first one, I notice very little deviation from the flat 'textbook' pronunciation. Unfortunately, I am not yet at the stage at which I could benefit from listening to videos like that and would never have thought of looking for one in which "chu1 zu1 che1 si1 ji1" was likely to be used. The chances of hitting on one would have been negligible. Also, I have no access to native Mandarin speakers here, so I must trust online dictionaries, recordings accompanying textbooks, the Google Translator, Text-to-Speech-like sites, etc. As to the (apparently) fake-rule T1+T1 > T4+T1, it is just another of many pseudo-rules (or inaccurately stated rules) I have already discovered (thanks to stellari and you, in this case).

    Hello baosheng.

    Thanks again. In general, pitch values gradually descend on the whole, but melodies 2, 3, 6 and 8 rise in pitch on "ji1", which I found surprising. Since, on general linguistic grounds, modifiers tend to be more prominent than their modifieds (and complements more prominent than their heads, etc.), I expected "chu1 zu1" to be pitch-wise more prominent than "che1", "chu1 zu1 che1", in its turn, more prominent than "si1 ji1", "chu1" more prominent that "zu1", and "si1" more prominent than "ji1". The intonation of 1., hence, seemed to me fairly in tune with general principles (although I would have expected sharper pitch differences between modifier and modified); the other intonations I transcribed in my original post all contained features I did not expect, particularly the end-rise on "ji1" in several of them. As to the 'mono-pitch' rendition that textbooks offer and now both stellari and yourself broadly endorse, it seemed to me prima facie implausible, since it entails that intonation is not made use of at all when it comes to signalling the difference between modifiers and modifieds (heads). Other languages, even those with fairly rigid syntax (e.g., English, German,...), do not 'waste' the obvious support intonation may offer in that respect, why should Mandarin do? That's approximately the way I reasoned when I decided to open this thread.
    Now, in retrospect, I realize that Chinese tones are lexically fixed, which possibly leaves no leeway to manipulate pitch for the expression of syntactic-semantic function, and, of course, I know that a rigid syntax alone suffices to signal the functional+semantic status of constituents if necessary, but in that case only the mono-pitch intonation 11111 should occur, don´t you think? The variation the TTS voice generators did show (if faithful to the way various real native speakers might intone that sequence, as I assumed), must then be 'free variation', i.e., modulation left entirely to the taste (or intonational needs) of individual speakers. Of course that may well be true, but I found, and still find, that conclusion aesthetically disappointing, :).

    Nevertheless, I do not think it necessary or practical (at this stage, at least) to use phonetics lab equipment to find out what native speakers do in full detail. As I had occasion to tell our moderator just in case my questions suggested otherwise, I am not doing research on Mandarin phonology (or any other aspect of the language, for that matter). I am just trying to teach myself Mandarin in a rational way, since in my little corner of the world I have no access to reliable teachers. Since resources like the Google Translator or the Text-to-Speech site are not as trustworthy as I thought, I will from now on look for elementary level, but genuine, recordings by native speakers, as you recommend, but I have not got there yet, :).

    Kind regards

    Indeed, so it seems. Textbooks and even standard reference grammars in English are full of inaccurately stated rules, so I'm no longer surprised to learn that this is just another.
    Of course I had noticed the shortcomings of the Google Translator (at all levels), but until now I had not tried to get from it anything but isolated words and the results seemed approximately comparable to those I could obtain from Pleco, the Oxford Talking Chinese Dictionary and other pronunciation dictionaries. It seems reliability decreases as soon as the input is a phrase or a sentence. Same for TTS, then? Anyway, I'll adopt 'Lily' as my favourite TTS informer.

    Thank you!
    Last edited by a moderator: Oct 21, 2014
  9. Skatinginbc

    Skatinginbc Senior Member

    Mandarin 國語
    I wonder if that sequence came from a speaker who mistook 死鸡 "dead chicken" for 司机 "driver". I would have difficulties understanding his/her speech, in other words.
  10. SuperXW

    SuperXW Senior Member

    Despite the complicated, professional question thread which I couldn't finish reading... x.x I really can't think of any case I would say or hear chu1zu1che1si1ji1 being pronounced in various tones... (as long as the speaker is speaking standard Mandarin.) For me, it's 11111 without question.

    In general, people raise the pitch when they want to stress a certain word or character, but I don't see any part in 出租車司機 worthy to get a higher pitch than others...
    Last edited: Aug 26, 2013
  11. Sibutlasi

    Sibutlasi Senior Member

    Thank you, SuperXW, for your categorical judgment!

    Despite what theories demand and theoreticians would like to hear, the native speaker 'is always right'. If you systematically intone "chu1zu1che1si1ji1" as a 'flat' monopitch 11111 sequence, then surely I, as a learner, had better try to get used to doing the same myself, particularly since other native speakers have manifested the same opinion. I'm already trying, although my performance remains rather unnatural :)!

    Nevertheless, "chu1 zu1 che1 si1 ji1" is a fairly complex nominal 'compound', it has various internal constituents with different syntactic and semantic functions, and, in principle, according to what you say (= "In general, people raise the pitch when they want to stress a certain word...") such differences should make certain syllables (e.g., "CHU1ZU1") get more stress than others (e.g. "che1"). Why? because "chu1 zu1 che1" is a special kind of "che1", and it is important to stress what makes that kind of vehicle special, i.e., that it is a vehicle for hire.

    When you say that you do not see "any part worthy of a higher pitch than others", is it that by 'stress' you mean just 'contrastive stress'? (cf. A: Did you speak to your sister at the funeral? B: "I SAW her, but did not actually SPEAK to her", where verbs exceptionally get extra stress (> higher pitch) because they are in explicit contrast). Now, I wonder: Would the first two words of "chu1 zu1 che1" be more strongly stressed and higher in pitch than "che1" if you wanted to convince a stingy wife that you got home by public bus ("gong1gong2 qi4 che1"), rather than in an (expensive) taxi? Also, if you wanted to say something like "I'm a BUS driver, not a mere TAXI driver", would "gong1gong2 qi4 che1" and "chu1zu1 che1" get more stress and higher pitch that the two tokens of "si1ji1" following each of them?

    Thanks for your attention.

    Last edited by a moderator: Oct 21, 2014
  12. SuperXW

    SuperXW Senior Member

    You are right on this. There IS a case where we'd like to highlight chu1zu1 in this compound. I can think of the following scenario:
    A: He's a driver? So he can drive the truck away?
    B: He's only a CAB driver. He doesn't have the licence for driving a truck.

    However, I think Northern Chinese are more used to alter the tones according to their emotions, while Taiwanese Mandarin speakers' voices are more "standardized".

    Also, don't forget Chinese has a list of "mood particles" and special structures to help express the feelings.
    For "I SAW her, but did not actually SPEAK to her." Sometimes we do stress the words SAW and SPEAK, sometimes we don't, because each tone are supposed to have a relatively fixed pitch range. We often add some particles to express the feeling (Words like "but" "actually" also have the function.):
    For example, 我看是看见她了,只不过没和她说话而已
    Last edited by a moderator: Oct 21, 2014
  13. Skatinginbc

    Skatinginbc Senior Member

    Mandarin 國語
    I think the two "tones" you mentioned perhaps refer to different notions: one concerning "intonation" cross word boundaries (pitch variation that is NOT used to distinguish words), the other concerning "tone" within the word boundary (pitch variation used to distinguish words). In standard Mandarin, every word in 出租车司机 should be pronounced with the "first tone" (i.e., "flat tone"--pitch remains steadily flat within each word). It is fixed, no tone sandhi, no stress induced tonal changes. Any deviation from the flat tone will bring about either a sense of regional accent (e.g., a reduced tone for 机, perhaps someone from the north?) or a loss of intelligibility (e.g., si3ji1 "dead chicken" for si1ji "driver").

    If we start to consider paralinguistic factors like "expressing emotion", "emphasis", "contrast", etc, we are entering the territory of "intonation", which is not just about pitch contour across words. It involves other prosodic features as well, such as loudness, tempo, rhythm (or syllable length), and so on. For instance, one may scream to a taxi driver "Hey, chu zu che si jiiiiiii", where 机 receives a higher pitch and prolonged duration. There are too many possibilities that can give reasons to emphasize any segment of the phrase. I think what is more useful for a beginner is to learn "normal" (unmarked) speech. 车 is the head of the noun phrase 出租车. 司机 is the head of the noun phrase 出租车司机. 司机 is a person, not a machine 机. What interesting about 出租车司机 is that two heads (i.e., 车 and 司) meet each other right at the word boundary (出租车 + 司机). If the Site Pal the OP referred to is this website ( search "出租车司机"), I think five out of the eight speakers place some sort of emphasis on 车 and 司. For instance, the speaker "Liang" raises the pitch slightly for 车 and gives a little pause (technically not a pause, but a prolonged duration) to separate the two words: 出租车 and 司机. The speaker Linlin and the speaker Yaling raise the pitch for 车 so dramatically that I would consider their intonations "unnatural". The speaker Lisheng raises the pitch for 车 just so lightly in anticipation of the climax 司 (i.e., anticipatory assimilation in pitch). The speaker Mei-Ling did likewise. Two (Lily and Ting-Ting) of the rest of the speakers keep 车 level with 出租. The speaker Hui is an exception. She treats 出租车司机 as one single unit and gives 出 the highest pitch.
  14. Sibutlasi

    Sibutlasi Senior Member

    Thanks, SuperXW #15. The example you cite would indeed be another case of 'contrastive stress', but no special 'emotions' need be involved. The process works automatically, so to speak. The other example I cited (SAW vs. SPEAK) was just for clarification in case my terminology was unfamiliar. For present purposes I'm interested only in the Chinese counterpart of "taxi-driver" and in how the syntax-semantics of "chu1zu1 che1 si1ji1" could affect the 'music', the tones, of its various syllables.


    Thank you very much, Skatinginbc. I entirely agree with the distinction you make in your first paragraph: only the 'unmarked' (default, non-contrastive,...) intonation of "chu1zu1 che1 si1ji1" is relevant to my question as originally intended. Since tone is used to distinguish lexical items, obviously, speakers cannot in general replace T1 with, say, T2 (unless there happens to be a 'lexical gap' just at that point, in which case no misunderstanding could arise anyway). Although I was led by a bad grammar to (wrongly) assume that the rule T1+T1-> T4+T1 applied in general, not just to "yi1", that much was clear to me from the start. That is why I brought in the 'music' (i.e., the specific pitch associated with each token of T1 on the corresponding lexemes within that compound) in my question: the way the various voices provided by the TTS engine at Site Pal realized the 11111 sequence suggested A LOT of 'allotonic' variation (= very significant pitch differences that might still be perceived by native speakers as 'irrelevant' - parallel to 'allophonic' variation within a 'phoneme'). Since the compound? "[[[chu1zu]1che1]] [si1ji1]]" is realized by the TTS 'voices' completely out of context, such absolute pitch differences could not be attributed to contrast, focus, or any other discourse or contextually-determined factor. If they occur (and they are not wrong), they must be 'allotonic' variation of the first type, within the fixed tone each lexical item has. Of course, we learners do not know how much pitch variation is allowed before native speakers start understanding something else.

    [By the way: thanks for letting me know that the 'head' of "si1ji1" is "si1", not "ji1", but that is a different issue].

    Last edited by a moderator: Oct 21, 2014
  15. SuperXW

    SuperXW Senior Member

    I knew I have problems on terminologies, just I was not sure how to express my exact idea in English. Thank you for pointing them out. :)
    Last edited by a moderator: Oct 21, 2014
  16. Skatinginbc

    Skatinginbc Senior Member

    Mandarin 國語
    No. That's not true. There is little allotonic variation. For instance, the Speaker Hui at Site Pal seems to pronounce 出租车司机 with a "G, F, D#, D, D" melody. If you pay attention to each syllable, you shall realize that there is little pitch variation within each syllable. It is like if you play "G, F, D#, D, D" on a piano, there will be no pitch change within each note. Every key on a regular piano is what Chinese call a "flat tone" because once you press a key to make a sound (no matter it is A, B#, C, or whatever), the pitch of that sound will stay the same until the end. Your transcribing each syllable into a music note in your original post reflects the fact that in your mind each syllable can be represented by a music note. And that fact should have spoken volume to you that every syllable represents a "flat tone" that has a constant, unvarying pitch like a music note or a key on the piano.
  17. tarlou Senior Member

    I'm actually not sure what is going on in this long thread :(. Just to clarify, the 1st tone in standard Chinese does have pitch requirement, and a flat tone with a wrong pitch can sound like a different tone that does not exist in standard Chinese. In other dialects this is important, for example, 天津话 has two falling tones (阴平 and 去声). There are also many dialects/languages with several different flat tones in both north and south.
    For the sound by Hui at Pal, I can't hear significant pitch variations even between different syllables. Only 司 is slightly off, but Skatinginbc marks it with the same pitch as 机. I'm bad at music and can't figure out the exact pitch at all. Maybe the extent of variation matters here, and slight lower or higher flat will only be considered as an intonation/stress/emotion/inaccuracy.
    Last edited by a moderator: Oct 21, 2014
  18. Skatinginbc

    Skatinginbc Senior Member

    Mandarin 國語
    Hi, Tarlou,
    Your post made me reread Post #18. And I just realized that the OP seems to be interested in the relative pitch distance between syllables, rather than the intra-syllablic pitch contour I discussed in Post #20, which focused on OP's choice of word "allotonic" and therefore took the wrong direction.
    The conclusion of your post seems to be: Not a lot of variation is permitted with regard to the relative pitch distance between two adjacent first-tone syllables. And I agree with you. To my ear, the permissible interval between two adjacent first-tone syllables cannot exceed a "major second" (two semitones); otherwise, it would trigger a sense of "abnormality" (accent, stress, or inaccuracy). The speaker LinLin at SitePal has a jump of "minor third" (three semitones) from 租 to 车 and I found it "unnatural" (as if speaking with a strange regional accent).
    司 of the speaker Hui sounds slightly off perhaps due to our cognitive need to either separate 出租车 and 司机 or at least keep them at the same level. As I said in a previous post, Hui is the only speaker that pronounces 出租车司机 with a falling intonation.
    Last edited: Aug 28, 2013
  19. Sibutlasi

    Sibutlasi Senior Member

    I think you have misunderstood me: Of course within each syllable, as intoned by each speaker, pitch changes are small, and difficult to quantify without proper equipment. What I meant was that if you compare the absolute pitch with which each of the eight Mandarin speakers at the Site Pal TTS realizes T1 on the same syllable of "chu1zu1che1si1ji1" there is considerable variation, and as a consequence the intervals between any two syllables often do not coincide, although certain regularities are detectable, for example there is a clear descending pattern that several speakers use. [By the way, in my post I did not list the 'melodies' in the order the speakers are listed in the TTS window. I tried to order them a bit to emphasize the existence of several groups of intonation patterns. That may cause discrepancy if somebody replicates my little 'experiment' and compares my 'melodies' with their own. I would have to repeat the process in order to restore the correct pairing of melody-speaker, but it probably is no longer worthwhile the effort].
    Last edited by a moderator: Oct 21, 2014
  20. Skatinginbc

    Skatinginbc Senior Member

    Mandarin 國語
    Didn't I just say so in post #22? :)
    I don't see such pattern. Can you please name those speakers so we can judge it ourselves?
    Last edited: Aug 29, 2013
  21. Sibutlasi

    Sibutlasi Senior Member

    Let me quote you in one of your earlier posts: "For instance, the Speaker Hui at Site Pal seems to pronounce 出租车司机 with a "G, F, D#, D, D" melody." Don´t you call that a descending pattern? The Google Translator offers another. Other speakers do produce descending patterns on "chu1 zu1 che1", one suddenly has a sharp rise on "che1", others have a rise on "ji1", etc. If you have listened to the Mandarin speakers at Site Pal's free TTS and you do not notice any significant differences in intonation between the way they realize the 11111 sequence, we are wasting our time. Let's forget it.
  22. Skatinginbc

    Skatinginbc Senior Member

    Mandarin 國語
    Sibutlasi, the reason I am so concerned about your "descending pattern" is that other learners of Chinese may read this thread and learn from it. I am afraid they would take the "descending pattern" as the norm.
    What I disagree with is your claim that there is a "clear descending pattern that several speakers use". I said in #16 that Hui is an exception. She treats 出租车司机 as one single unit and gives 出 the highest pitch (and the pitch keeps dropping from there). I said again in #24 that she is the only speaker that pronounces 出租车司机 with a falling intonation (a steadily falling intonation till the end).
    I thought we were discussing only those eight speakers at SitePal. You can pick only samples that agree with your hypothesized pattern from one billion of Chinese speakers and get quite a few number, but the result obtained from the pick-and-choose method is often not representative of the norm.
    So the so-called "descending pattern" applies to only two or three syllables, rather than to the whole phrase 出租车司机? In that case, the term "descending pattern" can be quite misleading, can't it?
    That's not true. I did notice significant differences in intonation among those Site Pal speakers. And I have described my observation in Post #16.

Share This Page