Shanghainese

[Morrígan]

Over the last semester, I (and other undergrad students) had the opportunity to work with two graduate students from Shanghai to do an informant-study of their native language.

You can find a PDF of my paper here.

The research process was pretty cool, though I wish I had access to better recording gear, and had synchronized video and high-quality audio so I could do better phonetics work.

I was thinking about typing up what I had in a couple of posts, but I don't really feel like it right now. If anyone would like me to do so, I'll get around to it over the course of this next week or so.

So, anyone else work with Chinese languages like this?

[Soap]

Awesome. I still remember your PIE-based conlang (that was you, right?)

I notice a few typos in there ... is that due to the software, or are they actually in the paper? e.g. "not tone occurs in more tan one syllable type"

[Tengado]

Interesting! Shangahinese has \ps`\ as on onset? Wow.


In the section about Counter words, you say you have little clue as to their distribution. The ones you list as a,b,c,d,h all seem to have direct cognates with the usual Mandarin and Cantonese ones: a: k@, used for people and as a default for anything (it can replace the specific one), b, /t_hjau/ used for long thin things, not only dogs and fish (I too do not understand why dogs are classified as long and thin, and not along with other small animals), but also snakes, neckties, rivers, roads etc. c. pinyin zhi (tS followed by a syllabic voiced retroflex fricatve), used for small animas (spiders, birds, cats- cows and bigger animals have a different one), h, li - used for smal and round things, like grains of rice, pearls, teeth.
You labelled d as being "mountains." The mandrin cognate, zuo, is used for mountains, and also architectural objects: bridges, buildings etc. I can't spot obvious cognates for the other ones e /ps`i/ might be related to pian, used for thin flat things, but in Mandarin thsi isn't restricted to round ones.

In the evrbs ection, youmention past and future tense forms - are you sure the past tense l@ is really past tense? In mandarin the same morpheme is perfective aspect. And you mention that l@ ispost verbal whereas iO for future is preverbal - iO appears cognate to Mandarin /jau/, Cantonese /jiu/ - a verb meaning "will, want", which is why it is preverbal. This is a serial verb construction ratehr than a grammatical particle.

EDIT: if ps`i is related to /pjEn/, that makes sense: the s` would be from friction of the j - the j changing first into a palatal fricative presumably.


Interesting stuff!

[Morrígan]

Awesome. I still remember your PIE-based conlang (that was you, right?)

I notice a few typos in there ... is that due to the software, or are they actually in the paper? e.g. "not tone occurs in more tan one syllable type"

Woops. Those are probably in the paper. Unless this is actually the draft for proofreading; I thought it was the final, but I could be wrong.

@Tengado:

Yeah, I was pretty excited about [pɕ]. About the counters, there are two problems. We didn't have a huge number of quantified words, and only the informants new anything about Mandarin. I should have guessed that shape was a major factor, given that [lɪ] was used with "tooth" and "candy", and [pɕi] was used with assorted flat, round things - it probably is cognate to pian, since Mandarin codas in [n] correspond to 0 in Shanghainese.

You are right about the verbs. We hadn't gathered much data, and I don't think we tried looking at an aspect distinction. Also, you are correct about the serial verb construction; we were told that this was another verb, but I forgot. I spent most of the semester staring at spectrograms; I wish I had more time to look at morphology and syntax.

[Nortaneous]

You don't list /M\/ in the consonant chart, but it shows up in your examples (/loɰə/). Same with /B/ (/dɯβ̞ə/). Are they allophones of voiced consonants or something?

Tengado: /ɕ/ is /s\/. /s`/ is /ʂ/.

[Tengado]

I speak Mandarin and a little Cantonese, and the counter words in both tend to be assigned according to shape of the noun, and secondarily function (there's a counter for machines, one for things with handles). Chinese languages in general don't mark tense, only aspect.

I always forget which of s` s\ is which. s\ just looks so intuitively more retroflex than s` does

[Nortaneous]

with sufficient imagination, <`> looks like the retroflex hook, except in a different place

[Morrígan]

You don't list /M\/ in the consonant chart, but it shows up in your examples (/loɰə/). Same with /B/ (/dɯβ̞ə/). Are they allophones of voiced consonants or something?

Careful with your slashes. [ɰ] is an allophone of /ʔ/. I'm not sure where [β̞] comes from; it may just get inserted where /u/ is followed by a zero-initial. Of course, my entire analysis of there being a zero-initial could be wrong. The wikipedia article says that Shanghainese has /ɦ/ where I posit no initial.

[Ran]

Over the last semester, I (and other undergrad students) had the opportunity to work with two graduate students from Shanghai to do an informant-study of their native language.

You can find a PDF of my paper here.

The research process was pretty cool, though I wish I had access to better recording gear, and had synchronized video and high-quality audio so I could do better phonetics work.

I was thinking about typing up what I had in a couple of posts, but I don't really feel like it right now. If anyone would like me to do so, I'll get around to it over the course of this next week or so.

So, anyone else work with Chinese languages like this?

Wow, did you refer to any materials on Wu or Shanghainese before this?

[Morrígan]

Wow, did you refer to any materials on Wu or Shanghainese before this?

No, we were totally blind going in. We didn't know what language we would be studying, or who our informant(s) was/were going to be until the first session. Actually, the students we worked with had also been in other classes with us in the fall and ended up being our informants because the program's budget got cut to the point that we could not afford to pay informants. But they got course credit, so I guess that is pretty good.

I can't imagine how hard it must have been for some of the undergrads, who lack the knowledge of linguistic typology that even a novice conlanger is likely to have.

What ended up in the paper is somewhat abbreviated from the total of data I managed to collect, since I needed to turn out a final draft in time while working on three other major papers. It will probably be worth typing some of this stuff up in posts again.

There are some interesting contradictions between my data in what the Wikipedia article says, especially with respect to tone. The surface distribution of tones is mostly correct, except that the falling tone [52] does occur in syllables with voiced onsets: e.g. [mɔ̂] 'cat'. Thus, both types of unchecked syllable contrast falling and rising tones, and only in checked syllables is the contrast neutralized. Wikipedia indicates that this contrast is neutralized with voiced onsets, also.

[BloodMerchant]

I am a 'native' speaker of Shanghainese (my heritage language), so I couldn't really ignore this thread.

I do have some issues with the paper itself.

Mandarin codas in [n] are all allophonic in Shanghainese, just like in other Wu dialects. SO [n] coda could be [null], [nasal vowel coda] or [ɲ]. (Actually, any nasal coda in Middle Chinese is now allophonic in Shanghainese/ other Wu).

For many, [lɔtʰɯ] should be [lɔtɤ], since 多 is not aspirated in Shanghainese.

All 全部 should be [zibu], [ʑ̊øbu] or [dzøbu], not [sɛbu]. But that could be use of an alternate word, which would be 儕 [ze].

For 'I, me' should be extremely ambiguous because there are at least four different pronunciations.

They are either [ŋu], [ŋo], [ŋ], . But is a New Shanghainese pronunciation. [ŋ] is quite rare these days and is only heard in a few districts.

人 in Shanghainese is not pronounced as [nin], but as [ɲiɲ]. Its literary pronunciation is [zəɲ].

what should not be pronounced as [sāʔ], but rather [sa], since it does not have the entering tone which accompanies glottals. Any native speaker can distinguish between [saʔ] and [sa].

For bird, there are two different pronunciations.

[ɲiɔ] and [tiɔ]. (obvious literary/colloquial contrast)

Big is [dɤ] or [da]. (lit/colloq again)

Long is [zã].

Hair is [dɤɯfɐʔ].

I found [pɕi], [β], [ɰ] to be odd in Shanghainese. Because any type of assimilation is rare/seldom found in Shanghainese (where there usually one example that I know of), but is common in Fuzhou dialect (a Min dialect).

Shanghainese itself has quite a lot of pure vowels, though.

But that could be due to their accent, since Shanghainese is a dialect with multitude of accents, and my accent is closer to the Ningbo dialect.

There are other issues that I have, but I'll just read the rest of the paper.

(But I'm guessing that those who wrote the paper were basing Shanghainese sounds on English phonology)

Wu also lacks retroflex, with the exception of some dialects in Southern Zhejiang (Quzhou dialect), and possibly any other dialect that would be related to Quzhou dialect.

[Yiuel Raumbesrairc]

I'd like to have that done for Quebec French, how it could turn out.

[Morrígan]

Mandarin codas in [n] are all allophonic in Shanghainese, just like in other Wu dialects. SO [n] coda could be [null], [nasal vowel coda] or [ɲ]. (Actually, any nasal coda in Middle Chinese is now allophonic in Shanghainese/ other Wu

I saw no strong evidence for this. The informants had a difficult time distinguishing between English [ŋ] and [n], but in the data I analyzed, Mandarin [n] codas corresponded to a null coda in Shainghainese: 'mountain' and 'three' were both [sɛ̂]

For many, [lɔtʰɯ] should be [lɔtɤ], since 多 is not aspirated in Shanghainese.

This makes sense; looking at the spectrograms again, it would appear that I misidentified a plosive burst as aspiration. And the VOTs are different.

All 全部 should be [zibu], [ʑ̊øbu] or [dzøbu], not [sɛbu]. But that could be use of an alternate word, which would be 儕 [ze].

Aha! I always suspected that the onset was [z] but I couldn't tell for certain.

what should not be pronounced as [sāʔ], but rather [sa], since it does not have the entering tone which accompanies glottals. Any native speaker can distinguish between [saʔ] and [sa].

I don't doubt you on this, but the pitch readings strongly suggested that the pitch contour was [23], which is what I found for words which we were fairly certain has [ʔ] as a coda. It certainly wasn't [52], and it sure didn't sound like [35].

I found [pɕi], [β], [ɰ] to be odd in Shanghainese. Because any type of assimilation is rare/seldom found in Shanghainese (where there usually one example that I know of), but is common in Fuzhou dialect (a Min dialect).

[ɰ] is a sound which appears when a syllable with [ʔ] in the coda is followed by a syllable beginning with a vowel, like [ə] after adjectives. Or at least, this was our theory.

But I'm guessing that those who wrote the paper were basing Shanghainese sounds on English phonology

I hope you don't mean what I think you mean.

[Davoush]

I'm guessing [pɕi] is a result of strong palatalisation and aspiration occuring together. This is quite common in Mandarin too, and happens in other environments. I remember a non Mandarin speaker asking me why /pa ta/ sound like [pxa txa].

[BloodMerchant]

Mandarin codas in [n] are all allophonic in Shanghainese, just like in other Wu dialects. SO [n] coda could be [null], [nasal vowel coda] or [ɲ]. (Actually, any nasal coda in Middle Chinese is now allophonic in Shanghainese/ other Wu

I saw no strong evidence for this. The informants had a difficult time distinguishing between English [ŋ] and [n], but in the data I analyzed, Mandarin [n] codas corresponded to a null coda in Shanghainese: 'mountain' and 'three' were both [sɛ̂]

[n] codas can be any of the following:
You have been analyzing the -an rime of Mandarin, which is usually irregular in Shanghainese. Mandarin Rimes do not correspond and don't often fit into Shanghainese perfectly.

Using the -an rime as an example,

岩, 顔 - [ŋɛ]/[ɦi] (Mandarin yan) MC reconstruction generally states that they have both the same initial and nucleus, but different coda. (岩 ends in -m [ŋam] and 顔 ends in -n [ŋan] in MC)

三, 山, 散 - [sɛ] In that case, you would be correct. There are quite a lot of codas in Shanghainese that end in [ɛ] which correspond to -an (by extension, -m in MC).

For 難. 南 (Mandarin nan), it is 南 [nø] and 難 [nɛ]

看 - [kø] (Mandarin kan)

扇 - [sø] (Mandarin shan)

But now, looking at other [n] rime codas,

the -en ending for example.

Like I said, 人, 忍 are both pronounced as ren in Mandarin (I'm too lazy to put Mandarin tone), but correspond to [ɲiɲ] and [zəɲ]. Shanghainese does not have a [n] coda as in Suzhou dialect, but a [ɲ] coda.

-en endings in Mandarin usually end in -[əɲ] in Shanghainese. But a few exceptions make it end in -[a] in Shanghainese, such as 洒 (which has many pronunciations in Mandarin, including [sa], but it usually has one pronunciation in Shanghainese, which is [sa])
-ian endings in Mandarin ---> usually - in Shanghainese
-uan endings in Mandarin ----> usually -[ø] in Shanghainese; wan in Mandarin also corresponds to -[ø] as well.
-un endings in Mandarin -----> usually -[yɪɲ] or -[əɲ] in Shanghainese; but then -un endings in Mandarin is quite ambiguous.
-in endings in Mandarin -----> usually -[iɲ] in Shanghainese.

For fish, 魚 is typically pronounced as [ŋ] or [ɦŋ], but in certain words, it is pronounced as [ɦy].

(As an interesting note, even though 囝 [nø] {Mandarin jian} is stated to be pronounced the same as 囡 [nø] {Mandarin nan}, I personally think, according to the Guangyun {and Wenzhou dialect}, it should be [tɕi] rather than [nø]; since it has an intial value of 見 [tɕ] {MC [k]/[kʲ]} and a rime value of 仙 {MC [ian]/[ien]}.)

what should not be pronounced as [sāʔ], but rather [sa], since it does not have the entering tone which accompanies glottals. Any native speaker can distinguish between [saʔ] and [sa].

I don't doubt you on this, but the pitch readings strongly suggested that the pitch contour was [23], which is what I found for words which we were fairly certain has [ʔ] as a coda. It certainly wasn't [52], and it sure didn't sound like [35].

All -[ʔ] codas have an entering tone, again, which suggests that they used to have any of the Middle Chinese stops -[p], -[t] and -[k]. Shanghainese has 'lost' all those -[p], -[t] and -[k] stops and replaced with them with -[ʔ] while still retaining the entering tone. (Standard Mandarin has completely lost those stops, in comparison) Even some Mandarin dialects and Jin do that as well. There is no word with a -[ʔ] coda that does not have an entering tone. But when some of the words are put into a sentence, they undergo tone sandhi and would thus sound different.
There are two different entering tones: [5] and [23]. [5] being the Yin Ru and [23] being the Yang Ru, as according to the traditional Chinese tone classification.

I think you should be quite aware of the fact that there are three different types of pronunciation for modern Shanghainese (aside from local dialects in Shanghai). I think the one that you've might be listening to is the New pronunciation, which is relatively different from the more 'conservative' pronunciation.

I'm guessing [pɕi] is a result of strong palatalisation and aspiration occuring together. This is quite common in Mandarin too, and happens in other environments. I remember a non Mandarin speaker asking me why /pa ta/ sound like [pxa txa].

I think p should be more like [pʲ], [pɹ̝̊] or [pθ̠] however. In any case, [pɕi] could be possible, but I don't think there's any sibilation involved.

[Ran]

BloodMerchant, I think you're being way too harsh and nitpicky on a project that was basically done with zero information going in.

I mean, do you really think it's possible to start from scratch, use only raw data, and arrive at the complete corpus of our current knowledge of Shanghainese phonology within the space of one semester?

The stuff you're pointing out can probably be found in any reference book, but I don't think that's the point of this project.

[Morrígan]

The stuff you're pointing out can probably be found in any reference book, but I don't think that's the point of this project.

I'm fairly sure the point was to teach us how unbelievably frustrating and uncertain original field research can be.

I know how to analyze syntactic, morphological, and phonetic data, but all the data you work with in school is already transcribed (unless you are actually in a phonetics lab, which my program – we didn't even have a department – didn't have). Actually transcribing the data is so much harder than I even thought it would be. I knew I would have a clear bias because my brain wants to recover phonetic input as English phonemes, but this seemed to be a bigger problem in the vowels, than in consonants.

But anyway, I've read on Wikipedia, subsequently, that 'person' is [ɲɪɲ], which I was transcribing as [nɪŋ], even though I had correctly identified the sound in 'woman'.

As for the status of [pɕ], the spectrogram is pretty conclusive:


The first voiced region is [ɪ] from 'one', the 109ms of silence is [p] and the white noise of the putative [ɕ] speaks for itself.

Here is the power spectrum from [ɕ] in [ɕõ] 'chest'


Here is the power spectrum of [ɕ] in [pɕi]


They are not identical, but clearly very similar, and probably perceptually indistinguishable.

[Zortzi]

Okay. I think that the paper had a lot of analysis for such a small data set, but I think there could have been a little more questioning, because I think that if the data set was augmented a bit more, more patterns could surface.

My mom's relatives speak Suzhou and Ningbo Wu Chinese from pre-1949 times because they left China before the communists came. Perhaps I could bounce a few ideas and some data around?

[Morrígan]

Okay. I think that the paper had a lot of analysis for such a small data set, but I think there could have been a little more questioning, because I think that if the data set was augmented a bit more, more patterns could surface.

I couldn't agree more. Unfortunately, I did not have exclusive access to my informants and had to fight with nine other people to get data. My analysis of the vowel system is woefully inadequate, and I didn't even have time to look at tone sandhi effects.

There are a lot of Chinese students in Buffalo, so I'm hoping to eventually find more Wu speakers. I'm currently working with a Min speaker, so that will require my full attention.

[Zortzi]

Hmmm. Keep us posted!

But I'm curious. What kind of Min chinese is it?

[Morrígan]

Hmmm. Keep us posted!

But I'm curious. What kind of Min chinese is it?

Southern; I believe the speaker is from Tainan.