Tones and pitch accent in Shanghainese.

[ala]

钱乃荣 《上海语言发展史》:

“两个平升调阴去和阳去只是由于声母的清浊而形成开头的高低，两个短促调阴入和阳入的高低也是由声母的清浊原因造成，也是两个平升调，只是阴入调高而促，听不出升调了。只有一个阴平降调与其他4个调形成明显对立。实际上上海话如今的5个调本质上只有一降一平升两个纯属调形性质的音位对立，平升的4个调的差异都是声母和韵尾的辅音造成的。

第44页

声调向重音化倾向进化。上海话的声调从8个合并成5个，实际上只余下一个降调（阴平）和一个平升调，变得十分简单。这使得上海人读声调时，自由变体可以相当宽泛，如降调读成“53”“51”“552”都不影响听感，平升调读成平降升调也不会影响理解。语音随着词汇语法词双音节连调成为主流以后，上海话在吴语中最快进化到“延伸式”连调，后字都失去了独立的声调而弱化粘着，重又向屈折语变化。前字有声调音位的作用，除此以外，只有一高一低或一低一高，上海话语流中的语音词读音已像日语的读法。目前，上海话语的语流中，相对稳定的音位有两类，一类是声母，一类是前字声调，这两类为首的音位对上海话语音正起着重要的稳定作用。值得注意的是，在青年中，有的常用词读成前字都是44，最后一字为低升调的读法，如：睏觉kuəɲ44 kɔ13，一点点ʔiɪʔ4 ti44 ti13，做勿来tsu44 vɐʔ4 lE13，规规矩矩kuE44 kuE44 ʨy44 ʨy13，这种读法有缓慢发展趋势，这是上海话向重音化语言转变的前兆。”

第74-75页

上海文化通史 第三篇 语言:

”一、8个单字调到5个单字调

这5个单字声调分别为“1.52;5.334;6.113;7.5;8.12”。实际上，这5个单字调中5与6，7与8都是由声母清浊对立而形成的，7、8与5、6的对立是由韵母的促声（入声）和舒声差别形成的，如果这些对立因素都归为声、韵母中辅音的对立，那么真正的声调对立只有阴平的降调和其他的平升调两个了。可以预言，上海话的声调离消亡已为期不远。

二、复杂型连读调到延伸型连读调

上海城区方言在150年里从“复杂型”跳过第三阶段的“简单型”（如无锡、江阴）进入第四阶段的“延伸型”。这是一种最简化的连读变调类型，5个连读调的调型和5个单字调的调型相同，即前字的声调覆盖后字音节，后字都失去调子共用前调。阴入开头的连读调又已并入阴去。上海话的连读调的简化已使上海话离重音语言不远了。

值得注意的是，一种新的类似重音形式的连读调开始在一些常用词中形成。语音词的前几个音节不论原来是什么声调，都读成无声调差别的高平调，只在最后一个音节上用一个低平升调，以划分语音词界限。如“交关”读“44+113”，“困觉”读“44+113”，“一点点”读“44+44+113”，“半个钟头”读“44+44+44+113”。”

第252-255页

[cuthbert]

,maybe

[ala]

Tones and pitch accent patterns in Shanghainese. The tone of each individual character in Shanghainese do not matter; instead what is critical is the accent pattern of the entire word.

The high pitch accented syllable ("falling" toneme) is indicated with an acute accent. A word that starts with the "falling" toneme has a high initial pitch (in the diagram below indicated by a blue line above the syllable), the second and subsequent syllables however all experience a drop to a lower pitch (indicated by the blue line below the syllable). A word that starts with the "regular" toneme has a low initial pitch, a sudden rise in pitch in the second syllable, and then a sustained dip in pitch for the third and subsequent syllables (as seen in the Neubanyou example below).

Hence, the 本 in 日本，蚀本，本事 is pronounced differently according to the word. Likewise the 友 and 朋 are also pronounced differently: 朋友，男朋友，友好

Below is another method of visualizing the pitch accent patterns. On the left is the pitch accent patterns and on the right are the Shanghainese examples. Each pattern exhibits increasing syllable length.

[shibo77]

从声调到语调.

Neat!

How common is this "44+44+44+113" sandhi? Does it happen to all sentences?

It immediately reminded me of Mandarin's 上声, "35+35+35+214".

Do you think this process is happening to Mandarin's 上声?

Mandarin's 上声 mostly exists only as 单字调 as well.

是否会纳入1A,1B,3A呢?

I think for practical purposes, excluding 单字调. Mandarin could be

55 -- 51 --

35 -- -- --

Yielding three tones:

55 平

35 升

51 降

上声 is a bit difficult/awkward to produce, I think.

Also, on the note of tone pitch, how would one describe the contours of Japanese's High Low?

(H)igh

(L)ow

Would it be similar to Shanghainese's?

34 Flat pitch

53 Down pitch

Thanks in advance!

-Shibo

[Altair]

Also, on the note of tone pitch, how would one describe the contours of Japanese's High Low?

(H)igh

(L)ow

Would it be similar to Shanghainese's?

34 Flat pitch

53 Down pitch

From the way, I have always heard Japanese described, I do not think that the five point scale would be appropriate. As far as I am aware, absolute pitch plays no more role in Japanese than it does in English. I think this is why no one talks about Japanese or English “tones.”

To be accurate, the Japanese pitch accent is indeed realized as a sudden drop in pitch, but the pitch heights are always relative. There is just a high and a low. Similarly, the initial rise in pitch has no contrastive function in the language and so a 3-5 rise would never contrast with a 1-3 rise. The relative pitches will be determined by the overall intonation of the sentence.

As for the relative distance of the high and low pitches, I think that a drop of two “levels” is probably the normal minimum. This will, however, vary according to the overall sentence intonation. This intonation will almost invariably impose several distinct pitch levels matched to each phrase. After each potential hiatus, one has the option of establishing a new overall pitch height for the sentence.

By the way, I also think that the five point scale can be misleading in discussing different varieties of Chinese. In my case, I actually went through a period when I found it easier to distinguish and reproduce the 6 to 7 Cantonese tones than to do the same with the 4 to 5 Mandarin tones. This was partly because of experiences I have had with other tonal languages and partly because of the descriptions of Mandarin I read. Because of these two experiences, I had thought that the differences in the Mandarin tones was primarily one of pitch and so was having a great deal of difficulty matching this to what my Mandarin tapes produced. In other words, a “musical” approach to the tones seemed to work with pretty well with Cantonese, but very poorly with Mandarin. Several of my books did indeed say not to “sing” Mandarin, but rather to speak it as if I “meant” it, but they gave no real description of how a beginner should actually do this.

The problem for me was that Cantonese has little differences in stress and can be spoken with something approximating absolute pitch distinctions. Mandarin, however, contrasts pitch contours much more than pitch heights. Differential stress is also a very prominent feature, although very, very few grammar books give any clue about this. In fact, most grammar books and reference materials seem to give incorrect or outdated linguistic information about Mandarin stress.

In citation speech and in stressed positions, I also believe that Mandarin tones are distinguished by relative volume changes over the length of the syllable. Each of the four main tones has unique volume characteristics that seem to be quite important to respect.

Lastly, I remember reading in one specialist article that analysis of recorded Mandarin showed quite a number of surprises about tones in connected speech. For instance, some rising tones were actually realized as falling pitches, or vice versa.

[ala]

Also' date=' on the note of tone pitch, how would one describe the contours of Japanese's High Low?

Would it be similar to Shanghainese's?

34 Flat pitch

53 Down pitch [/quote']

I think you have misunderstood how Shanghainese accent High and Low works. In fact Japanese and Shanghainese have similar accent systems, being that neither have clear tonal contour or position during conversation. The flat tone (34) and down tone (53) doesn't really exist in conversation in Shanghainese. Shanghainese pitches are only consistently measurable on a 5 point scale for the pronunciation of individual characters. The contour also exists only on the readings of individual and isolated characters. The only time there is a "down" pitch in conversation is when emphasizing or cursing.

In conversation, there is only a relative differentiation of high and low (and most Shanghainese are not aware of this until told). High being around 5 to 2, and low being around 1 to 4. The actual pitch varies according to the word, sentence, mood, gender, age, etc. There is a lot of flexibility in pitch pronunciation in Shanghainese (developed from the diversity of the population). If you get the pitch wrong, you have an accent, but you will likely be understood. A very approximate 5 point scale of Kónkonchitsuo 公共汽车 (bus) is 55-33-33-32, for Shinshidde 信息台 (info desk) is 33-55-32, for Shazide 写字台 (desk) is 33-44-32, for Detzi 台子 (table) is 11-34, for Ciotzi 饺子 (dumping) is 33-55. But it is much easier and systematic to just say H-L-L-L; L-H-L; L-H-L; L-H; and L-H since in conversation that's what they are: relative high and low, the precision of the 5-pt scale is unrealistic. The reason why the same 子 is at different pitches is because the first syllable in 台子 (de) is voiced, the jump in pitch for 子 thus starts from a lower starting point.

A Shanghainese can accurately read an unfamiliar romanized writing with only the high pitch on the first syllable of the word marked (if the first syllable is a low pitch, then no marking need to be made at all). Everything else is automatic. This is quite similar to the Japanese haSHI and HAshi phenomenon.

The flat and falling toneme distinction in characters is important because it allows one to predict the pattern (HLLLL.. or LHLL.. or LHHH, etc), but it does not reflect the actual pitch (which will always be relative and dependent on the actual word and sentence). And to address Altair's comment, in Shanghainese there is no contrast between a 12 and 34, a 12 would occur because of voicing (which naturally is lower, as in the case of Japanese as well), and there are never more than one fall and rise in a Shanghainese sandhi phrase.

[shibo77]

Very interesting!

I screwed up on the Japanese pitches, sorry!

But I don't know the meaning of the English words.

What is a toneme?

In Chinese.

声调tone

语调intonation

These describe 语调intonation only:

调核tone nucleus?

调冠tone onset?

调头tone initial?

调尾tone final?

Also, I think most textbooks stress that Chinese tones are not "musical" tones. The musical approach was abandoned early last century. Musical tones do not change but Chinese tones are relative.

I don't fully support the 5 point scale, but I still think that it is probably the best at the moment.

This one shows Mandarin's four pitch tones.

T1,T2... are the citation tones.

T1 55

T2 35

T3 214

T4 51

And when involved with tone sandhi and 语调intonation, it becomes relative and the volume goes up and down:

The five point scale marks off the important parts of the contour, which have to exist to be considered of that tone. If otherwise, then tone sandhi have occured. In this way, the five point scale is pretty adequate to describe tones and tone sandhi.

But for relative pitch and intonation, it is not the correct way. I was then wondering how Shanghainese have relative pitches marked by the five point scale:

34 Flat pitch

53 Down pitch

I am confused with 5 to 2 for High pitch, 4 to 1 for Low pitch. If this is possible, then it would be possible with Japanese haSHI/HAshi.

Japanese is at the relative pitch phase, a little different from English.

English is just intonation, and usually only 口气语调(implicating intonation?).

When analysing from 声调tone to 语调intonation, the tones can still be distinguished but only the relative pitch is changing, hence the arrows moving up and down.

In Mandarin there are 3 high tones, 55, 35, 51, because they start high. When in a sentence, they are marked (H)igh pitch.

In Mandarin there is 1 low tone, 214, because it starts low. When in a sentence, it is marked (L)ow pitch.

55+55+55+55+55+55+55+55+55

张中斌今天修收音机。

35+35+35+35+35+35+35+35+35

吴国华明年回阳澄湖。

51+51+51+51+51+51+51+51+51

赵树庆半夜上教育部。

The low tone becomes a little messy and has tone sandhi, but the five point scale can still help you find the proper tones for the individual syllables in the sentence:

214+214+214+214+214+214+214+214+214 changes to ????

李小宝九点写讲演稿。

Obviously the 35+35+35+214 proposed by Chao Yuanren is incorrect.

It's more like (45+55+21)+(35+212)+(21)+(24+45+21). The pitch would be (HHL)(HL)(L)(HHL)

(....) <<circles off words.

Talking of relative pitch, Mandarin have relative pitch just as Shanghainese or Japanese.

讲演稿 Jiang-yan-gao HHL perhaps 升升降 24+45+21.

one can't just mark these 45, 212, 24, 21 as new tones. Some sat 21 is called "partial third". They are merely representations from High and Low pitches.

How the third tone character is pronounced in a sentence is determined more by the relative pitch of the previous syllables(HHL) than merely what citation tone the previous syllable is(35+35+35+214).

In this way, Mandarin's 上声 is not much different than the relative pitch of Shanghainese, when it only exists (as 214) as 单字调character tonal pronunciation. Otherwise in daily speech, the tone 上声 doesn't really have a standing in 声调tones, but more in 语调intonations.

-Shibo

[taibeihong]

what happens to monosyllabic words? for example, pronouns like ngu and nong in a sentence like "ngu shotach nong", do the pronouns keep their low rising intonation, or do they change somehow (but without forming a sandhi phrase with shotach)?

And what about at the beginning of verbs? when verbs are preceded by rach or vach, does their intonation pattern depend on those particles? would shotach (LH) become LHL in vashotach? what happens with "prefixes" like "raerach"? do they enter into a sandhi phrase with the verb?

And at the end of verbs, when particles for aspect and tense are added, do these form a sandhi phrase with the verb? or are they sepparate?

[ala]

what happens to monosyllabic words? for example, pronouns like ngu and nong in a sentence like "ngu shotach nong", do the pronouns keep their low rising intonation, or do they change somehow (but without forming a sandhi phrase with shotach)?

They form a general sentence sandhi phrase, but separate from shotach. There is a very small rising intonation (sometimes none) in ngu. Citation tones are never really pronounced unless for emphasis.

And what about at the beginning of verbs? when verbs are preceded by rach or vach, does their intonation pattern depend on those particles? would shotach (LH) become LHL in vashotach? what happens with "prefixes" like "raerach"? do they enter into a sandhi phrase with the verb?

Yes, vach is always attached in the same sandhi phrase as the core or auxillary verb (whichever comes first). Always, you never see vach by itself. raerach is separate from the core verb (so 勿勒了白相 = vaeraerach basshan). Attachments like -rach, -raehe, etc are always part of the verb sandhi phrase if they follow the verb (Youshici basshanraehe). This almost gives verbs a conjugative quality.

And at the end of verbs, when particles for aspect and tense are added, do these form a sandhi phrase with the verb? or are they sepparate?

All agglutinative complements (到、好、脱、脱了、光、了、得、牢、上去、下去、etc) are part of same sandhi phrase. All basic postpositions (like 马路浪) are part of the same sandhi phrase as the noun (上、浪、浪向、下、里、里向、中). But more complex locatives such as 当中、旁边、后头、前头 are separate. All prepositions are separate.

Most of the partitioning occurs pretty naturally (they kind of make sense). I'll try to compile a more orderly set of rules sometime. One partition that is kind of interesting is names. The surname and the first character of the given name are pronounced as one sandhi phrase; and the last character is separate.

[SassybutSweet]

I have done various searches to try and find some way to learn shanghinese on the internet, But to no avail!! I'am learning Mandarin on the internet with good results!!! I can't seem to find a good progam on the internt for Shanghinese, is there one available?! If so can anyone tell me where?! I noticed some good research on this page, and I was wondering where you got your resource from!!That will definitily help me in my studies of Shanghinese!!! Thank You!!!

[Altair]

What is a toneme?

A toneme has the same relationship to a tone as a phoneme has to a sound or a morpheme has to a word. :Toneme," "phoneme," and "morpheme" are theoretical constructs that are realized in true speech as tones, sounds, and words. Usage of the word "toneme" implies that an underlying tone may be expressed in true speech in various ways, including perhaps by complete omission.

I think that current linguistic theory posits that tones are not features of the syllables they are associated with, but rather free standing components of language. A toneme may be spread over more than one syllable or be transferred from one syllable to another. A syllable may drop out, but leave its toneme to affect the syllables that remain. There are instances of this sort of thing in both Cantonese and Mandarin, but I think it is more frequent in Cantonese.

In Chinese.

声调tone

语调intonation

These describe 语调intonation only:

调核tone nucleus?

调冠tone onset?

调头tone initial?

调尾tone final?

The English words sound great. I am not sure what they mean, but can make some educated guesses.

Also, I think most textbooks stress that Chinese tones are not "musical" tones. The musical approach was abandoned early last century. Musical tones do not change but Chinese tones are relative.

I think this is only a subset of the issue I was discussing. Let me try to clarify.

As far as I know, no language in the world makes use of absolute musical pitches; however, there are many where pitch relative to one's personal vocal range is meaningful. As I understand it, Cantonese has many tones that are contrasted exclusively by pitch, since they have exactly the same contours. From what I understand, in Cantonese, a listener can distinguish three different pitch heights of a syllable like "si" (e.g., 诗, 试, and 是) purely by estimating where the pitch lies within the speakers vocal range and without having to hear any other syllables from the speaker to give a relative pitch.

In learning a bit of Cantonese, I was able to use this pitch stability to learn to reproduce and recognize tones. Many of the tone sequences corresponded to discrete musical intervals that I could hear, such as fifths, fourths, and minor thirds. Of course, Cantonese speech is not the same as singing, because there is some variation in pitches and because all the intervals are substantially flatted (i.e., slightly out of musical tune).

I think that Mandarin has the same underlying pitch stability. I read an article about a year ago, perhaps in the New York Times, that described a study of absolute pitch. In the study, "Chinese" speakers recorded specific words at substantial time intervals (with perhaps days, weeks, or months in between recordings). The speakers consistently used the same pitch for the individual words. English speakers did not do the same. The study proposed that this might account for the apparent fact that Chinese speakers are more likely to have a sense of absolute musical pitch than English speakers.

For all these reasons, I think that the 5-point scale works with Cantonese and Mandarin. My experience is that it also works with some African languages that have consistent pitches. I do, however, have some doubts about a language like Hausa (spoken across Northern Nigeria), which is reported to have significant downdrift. I do not speak any Hausa, but I get the impression that it does not maintain descreet pitch levels and that high tones can end up being lower than low tones and that low tones can end up being higher than high tones in a given sentence, because relative pitch height is much more important than absolute pitch height.

The problem with Mandarin is that the picture is clouded by a number of factors that do not exist in Cantonese. Mandarin has significant tone sandhi, variable stress, and only one level tone that can easily establish a pitch level. Also, all Mandarin tones differ in contour and in the distribution of sound volume. All these factors serve to diminish the importance of pitch. Because of this, I find that approaching Mandarin solely in terms of pitch yields seriously deficient results, as is shown in the tone graphs in the post above.

Japanese has a completely different pitch structure. The pitches involved in a Japanese sentence are mostly determined at the phrase level and are due to two different phenomena.

First, Japanese is best analyzed at the level of moras (or morae) rather than at the level of syllables. If the first mora of a Japanese phrase is not "accented" it has a relatively low pitch. The next mora is pronounced at a relatively high pitch, and all subsequent moras in the phrase are pronounced more or less at the same pitch. There is, of course, some downdrift. When a new phrase is begun, even within the same sentence, an optional juncture can be recognized and the pitch will drop once more to begin the next phrase in the same way as the first. The result of all this is that a Japanese sentence, even without any accented words, will generally have a marked two-level pitch structure.

The second feature of Japanese that determines pitch is the presence of word accents. A word may optionally have one, and only one, accented mora. Surprisingly, the accent of an accented mora has an effect on that mora only if it is the first mora in a phrase, otherwise the pitch of the mora itself is never affected. In fact, there is no phonetic way to tell from the mora itself that it is "accented." What can happen is that the following syllable will have a lowered pitch, which will be carried on until the end of the phrase, even across word boundaries. This lowering of pitch always takes place within a word, but takes place across word boundaries only in the case of certain particles. Even though Japanese has many words that are "accented" on the last mora, this accent is often not actually realized phonetically in any way. When words consist of only one mora, there is never a contrast in accent or pitch when the words are cited in isolation. The accent can only manifest itself in specific situations, as I have described.

Because of all these factors, I think it would be very misleading to think of Japanese words as having tones or to use the 5-point scale in describing them.

From what Ala has said, it seems that Shanghainese is moving towards an accentual system that has a surface similarity to the Japanese one. The similarity is that it is easiest to verify whether or not a syllable in a word has a high pitch and then to assign automatic pitches to all other syllables. Thinking of each syllable as having a distinctive pitch or tone contour is apparently misleading.

Obviously the 35+35+35+214 proposed by Chao Yuanren is incorrect.

It's more like (45+55+21)+(35+212)+(21)+(24+45+21). The pitch would be (HHL)(HL)(L)(HHL)

Thanks for posting the pitch graphs, they seem to verify certain hypotheses about Mandarin tone sandhi that I have read. When I get a chance, I will post what I have learned on a more suitable forum to verify whether it matches up with what native speakers feel. I am particular interested in the issue of tone sandhi covering third tones, since this is a major problem for non-native speakers trying to read material out loud like "李小宝九点写讲演稿。" I agree that Chao's theory does not seem to fit the facts, even though it is parroted in just about every grammar book I have read.

[ala]

Talking of relative pitch' date=' Mandarin have relative pitch just as Shanghainese or Japanese.

讲演稿 Jiang-yan-gao HHL perhaps 升升降 24+45+21. [/quote']

24+45+21 would be LHL on a high-low scale.

one can't just mark these 45' date=' 212, 24, 21 as new tones. Some sat 21 is called "partial third". They are merely representations from High and Low pitches.

How the third tone character is pronounced in a sentence is determined more by the relative pitch of the previous syllables(HHL) than merely what citation tone the previous syllable is(35+35+35+214).

In this way, Mandarin's 上声 is not much different than the relative pitch of Shanghainese, when it only exists (as 214) as 单字调character tonal pronunciation. Otherwise in daily speech, the tone 上声 doesn't really have a standing in 声调tones, but more in 语调intonations. [/quote']

No, this is still not quite the same as Shanghainese tones. In Shanghainese NONE OF THE CITATION TONES MATTER (ZERO!) except for the very first syllable of a sandhi phrase (each phrase is usually a lexical item).

In the Mandarin example; 李小宝 / 九点 / 写 / 讲演稿。 (45+55+21)+(35+212)+(21)+(24+45+21), if I added a 1st, 2nd, or 4th tone character into any of the above word, the tone of that character will be very apparent, and it will alter subsequent tones. In Shanghainese, if I added any character (regardless of citation tone) after the words above (and treated it as part of the same word), the citation tone of that character is ignored completely and will follow the general sandhi pattern (albeit some differences in voicing and Rusheng) of that word (pattern initiated by the first syllable). Thus subsequent syllables in a Shanghainese word do not follow the citation tones, NOR are they at all dependent on the citation tones. This is far from the case in Mandarin, which has tone sandhi but is still dependent on the tone grouping of the characters. The Shanghai dialect is unique in this (it is also unique within the Wu dialects).

上海话的连续语音变化已经脱离与原调相关的连续语音变化规则。 不管原调，一律按某种规律连续变调。

[shibo77]

Thanks! I understand now. Did you make the website: zanhei.com ? It's looking good!

-Shibo

[ala]

Thanks shibo. 最近很忙，可能要到明年才好完成。 中文版打算现在就写起来。

[Altair]

Ala, I have now had a chance to peruse the websites mentioned below and consider more of what you have posted on Shanghainese. What do you think of this Wikipedia site (http://en.wikipedia.org/wiki/Shanghainese) and this other site called Conversational Shanghai Chinese (http://www.linguasinica.com/Shanghai/default.html)? What dialect of Shanghainese do they describe? The same one you have been talking about? Although I have trouble with the phonetic fonts on both sites, I was able to follow the explanations of the phonology quite clearly.

By the way, Shibo, you should especially like the discussion of tone Sandhi on the Conversational Shanghai Chinese site, since it uses the 1-5 numbering.

I now have a couple of questions and comments about some of the earlier discussions. First, I think that rather than speaking only about voiced and unvoiced consonants, I think it would be better to talk also about the concept of murmur. Here are two sites that describe the phenomenon: Sound patterns in Human Language: phonation and Phonation.

I think it is important to mention “murmur” because according to the Shanghainese sites I refer to above, “murmur” characterizes the entire syllable, and not just the initial consonant. Also, initial “m,” “n,” “ng,” and “l” seem always to be voiced, but when they begin low-toned syllables, they also have murmur. The descriptions of murmur also make even more understandable why the consonants with this feature in Middle Chinese tended to generate lower tone variants in the later dialects of Chinese.

One little tidbit I found interesting was the Shanghainese pronunciation of日本人 (“shebbennin). The makes the English pronuncation of “Japan” much more understandable than one would guess from Mandarin, Cantonese ([yutt boon], or Japanese itself.

If I am correct in my understanding so far, I am confused about what this site (http://www.linguasinica.com/Shanghai/html/lesson01.htm) implies is a three-way distinction between `fi (“to fly”), “fi (“fat”), and similar words beginning with “v.” Are there both “f’s” and “v’s” with murmur? A sound like “t” can have a three-way contrast between unvoiced unaspirated, unvoiced aspirated, and murmured, but I cannot figure out how to apply this to “f.” I have the same question about `si (“west”), ”shi (“neat”), and any word beginning with a “z” sound. Can anyone explain?

Ala, I also think I misunderstood the scope of the sandhi phenomenon in Shanghainese compared with Japanese pitch accentuation, even though I think your explanations were rather clear. In Japanese, pitch levels tend to cover more than just lexical items, usually extending to entire phrases or even short sentences. It seems that Shanghainese can have four or five word sentence with three or four tone levels represented. This practically never occurs in Japanese. In Japanese, a short sentence will often have only one pitch phrase and maybe two at the most. In other words, the audible domain of Shanghainese tone sandhi appears to correspond to lexical units; whereas in Japanese, the domain tends to corresponds phrases.

For those who have read this far, who have no idea of what I am talking about, but who speak some French, I can make an analogy to French intonation patterns. French phrases tend to begin with a low pitch and then rise to a constant level. Each time a new phrase is begun, the pitch can optionally back down before rising again in a repeating pattern until the end of the sentence. Japanese intonation phrases are divided in the same way, except that a drop in pitch will commonly occur within a phrase under certain conditions that are hard to summarize briefly. Basically, the drop in pitch has to do with the lexical pitch of a word, the place of the pitch within the word, and the structure of the phrase within which the word occurs. Unlike French, there is little drop in pitch simply because a sentence comes to an end. [/url]

[ala]

Ala, I have now had a chance to peruse the websites mentioned below and consider more of what you have posted on Shanghainese. What do you think of this Wikipedia site (http://en.wikipedia.org/wiki/Shanghainese) and this other site called Conversational Shanghai Chinese (http://www.linguasinica.com/Shanghai/default.html[/url'])? What dialect of Shanghainese do they describe? The same one you have been talking about? Although I have trouble with the phonetic fonts on both sites, I was able to follow the explanations of the phonology quite clearly.

linguasinica site is pretty in-depth, although the word selection and pronunciation is rather outdated. The romanization is also very difficult to understand (it tries too hard to fit Mandarin romanization).

Just some examples:

请问 is romanized as cinven by linguasinica. But we actually pronounce it as cinmen (chinmen).

Question particle, 呢 is romanized as ni by linguasinica. Clearly no one under the age of 65 would pronounce 呢 as ni, we pronounce it instead as a very clear ne /nE/ (which is from Ningbo-hua, "ni" is from Suzhou-hua).

linguasinica uses a pronunciation that is a cross between Old and Middle Shanghainese (老中派).

The wikipedia site is a jigsaw puzzle, it has Old form of Shanghainese in its phonology, and it is very wrong (especially in the diphthongs). The tone sandhi description however is Middle and New, although the radical newer patterns are not included. No one in city proper speaks Shanghainese with six tones today, maybe in the villages nearby.

Old, Middle, New all have the same consonants (In the New form, there's some merging of /ng/ with either /h^/ or /n/). The consonant is a very strong stabilizing element in Shanghainese.

I think it is important to mention “murmur” because according to the Shanghainese sites I refer to above, “murmur” characterizes the entire syllable, and not just the initial consonant. Also, initial “m,” “n,” “ng,” and “l” seem always to be voiced, but when they begin low-toned syllables, they also have murmur. The descriptions of murmur also make even more understandable why the consonants with this feature in Middle Chinese tended to generate lower tone variants in the later dialects of Chinese.

Yes. The murmer corresponds to the traditional class of 浊音 characters, which is roughly translated as "voiced." And it is also why nasals and approximants can be "voiced" or can also lack that "voiced" quality.

One little tidbit I found interesting was the Shanghainese pronunciation of日本人 (“shebbennin). The makes the English pronuncation of “Japan” much more understandable than one would guess from Mandarin, Cantonese ([yutt boon], or Japanese itself.

Shanghainese for 日本 sounds very similar to French Japon. But supposedly they got it from a rare Japanese dialect. 日本 would be pronounced Nippen [gniI?p@n] in Shanghainese if the term didn't go through Mandarin first. Two readings for 日 in Shanghainese are [z@?] and [gniI?].

Are there both “f’s” and “v’s” with murmur? A sound like “t” can have a three-way contrast between unvoiced unaspirated, unvoiced aspirated, and murmured, but I cannot figure out how to apply this to “f.”

There is no three way distinction in f/v. However in 老派, the two distinctions are [f] and [beta]. Today (under 65) the two distinctions are clearly: 飞 [fi] and 肥 [vi].

Regarding s, sh, and z. Perhaps linguasinica is talking about , [z], [dz]. I can say everyone since 中派 has lost all trace of [dz] distinction (治、迟). But after checking out the site, I don't think that's what linguasinica meant:

sheqnidi syzeffen hoku' date=' finish classes at twelve forty,

sheqnidi syzerrfen ciq jonvai. eat lunch at twelve forty-five[/quote']

Several problems. First of all, we see 十 used in two cases per sentence. But the first one is spelled sh (12), the second one is spelled with a z (40, 45). So sinicalingua is clearly taking into account what the linguist YR Chao said about Northern Wu dialects: 清音浊流. The first 十 is sh, the second 十 is z. In fact, sinicalingua's romanization scheme is very similar to the one devised by YR Chao himself. 清音浊流 description has actually been rather disproved lately, instead some studies show that Northern Wu dialects have voicing throughout the sentence, though the voicing is indeed weaker at the beginning of a sentence.

but I cannot figure out how to apply this to “f.” I have the same question about `si (“west”), ”shi (“neat”), and any word beginning with a “z” sound. Can anyone explain?

Hope that answered the predicament of the "three-way differentiation" for f/v and s/z....

The second problem is that, the 十 in 四十 40 should be voiceless in Shanghainese (十 in 四十五 45 is voiced). Again, linguasinica is very outdated (by at least 3/4 of a century).

In other words, the audible domain of Shanghainese tone sandhi appears to correspond to lexical units; whereas in Japanese, the domain tends to corresponds phrases.

Yes, very much based on lexical units, although postpositionals (locatives), complements, indirect object pronouns are part of the same utterance. Smaller than Japanese phrases for sure, but somewhat similar to the HAshi and haSHI example.

[Altair]

Ala, thanks for your explanation. I think I understand now.

I think I also have a better appreciation for you preference for word spacing. I think I can imagine how reading Shainghainese in characters could be irritating, if you are constantly trying to figure out what pattern of tone sandhi to give your readings without having to read to far ahead.