Writing Shanghainese, part 2

« previous post | next post »

No one in this Douban thread (so far) can identify the script in the image below:

At first, I was going to post this as "Unknown script", with these thoughts:

I'm inclined to think that it's not a hoax.  Too many regular recurrences. Looks like a phonetic script.  And it's gotta be East Asian because of the orientation, format, overall appearance, etc.  Kinda 'Phags-pa-like or Hangulish — block assemblage of "letters".

But then Jichang Lulu reminded me that this had already come up on Language Log in comments to this post: "Writing Shanghainese" (5/25/16).

The very first comment to that post was by Frédéric Grosshans:

Readers of this post might be interested by a 2012 blog post by David Helliwell on some books they have in the Bodelian Library at Oxford. These books were written in the 1850’s by protestant missionaries in Shanghainese, using an original phonetic writing system. According to one of the few comments, the few pages scanned on the blog post show some phonetic differences with modern Shanghainese.

To which, in the second comment, I replied:

Thank you very much for this extremely interesting and valuable information. It is great to know about David Helliwell's excellent blog, and I am particularly pleased to learn about the creation of a phonetic writing system for Shanghainese already in the 1850's. This complements well what we already knew about the gradual, general trend toward phoneticization of Chinese writing during the last century and more, adding powerful new evidence and depth to our findings.

Jichang Lulu at one point had pretty much taught himself how to read "New Phonetic Character", but there were a few graphemes that he never figured out.  I suspect that, within a day or two, he might be able to tell us a bit more about how the script works.  You will note that Helliwell has already provided transcriptions and annotations in his descriptions of the books pictured in his blog post.  Texts written in the New Phonetic Character hold great promise for telling us about 19th c. Shanghainese.

[h.t. Joel Martinsen]


  1. Victor Mair said,

    July 22, 2016 @ 5:42 pm

    From Jichang Lulu (within an hour or two, not a day or two):

    It's the title page of a Shanghainese Gospel of Luke, printed in Shanghai in 1859. Likely the translation David Helliwell mentions towards the end of this post.

    The translation is by Cleveland Keith and the transcription by A.B. Cabaniss of the American Southern Baptist Mission. The script itself (called New Phonetic Character) was created by another Baptist missionary, Tarleton Perry Crawford. The original version (presumably the one seen here) was created for Shanghainese, but Crawford later enriched the script to fit the sounds of Mandarin and other topolects. There are more links and details in Helliwell's post above.

    As for the page in the Douban thread, this is my reading, with a rough-and-ready transliteration of the New Phonetic Character:

    耶穌降世一千八十九年 上海土白
    yá-sú kong-sz ih-ts'ien pah-páh m-zah-kyeu-nyeN zong-hé t'ú-báh
    AD 1859 Shanghai local vernacular

    lú-ká-dseN fuh-yung-sû
    Gospel of Luke

    YaN-fúng kyeu-nyeN mung-Yá k'uh

    The last line refers gives the year by the era system (Xianfeng 9) and possibly mentions the printer.

    The transliteration is based on the one given by a Shanghainese textbook published by Protestant missionaries, and provided in Helliwell's post. In it, g/ds/d/b/z are, or were at some point, (slack) voiced; k'/ts'/t'/p' are aspirated; a final -h is a glottal stop; a final -N indicates nasalisation of the previous vowel; -eu is a back vowel (modern Shanghai [ɤ]); acutes probably show a tonal contrast, but it's not obvious how their use matches the Shanghai tonal (or accentual) system. I'm not sure how to interpret some of the symbols.

    As I mentioned in my comment to Helliwell's post, the New Phonetic Character shows distinctions modern Shanghai has lost. It's not used entirely consistently though, so I'm not sure how faithfully it represents mid-19C Shanghainese. One thing to remember is that many of these works were produced by non-native speakers (the missionaries) who hadn't spent a decade in Shanghai when the phonetic script works began to be published.

  2. Bathrobe said,

    July 22, 2016 @ 11:07 pm

    九年 appears twice but the markings (tone markings?) are different at each occurrence. In the first line there are two dots, one at the top lefthand corner of 九, one at the right in between the two words. In the third line, the first dot remains the same, but the dot before 年 is missing. Instead, there appears to be an acute accent (´) below 年, which is not found in the first line. Why would they be different?

  3. Jichang Lulu said,

    July 23, 2016 @ 9:07 am


    Indeed. This is what I meant when I talked about inconsistencies and stuff I'm not sure how to interpret.

    I think the acutes are actually commas, but used as 'spaces', as word boundary markers. If you look at the little 'corpus' of New Phonetic Character texts in Helliwell's post, you'll see title pages in Phonetic Character written without any commas at all, while poetry (NPC, hanzi), while prose has both commas as separators and <。> as punctuation.

    As for the dots to the left and right of syllables, they seem to correspond approximately to the rising (shang, B) and departing (qu, C) tone categories, mainly reflected in the third and fourth tones of Standard Mandarin. I think that was the intended meaning of the symbols, based on this dual script table for tone C and the rest of the mini-corpus for tone B.

    But there are two problems with this. One is, as you noticed, that the tone marks aren't used terribly consistently, namely the rule for the left dot appears to be 'mark a shang tone, or something, if you feel like it'. Perhaps having more dual hanzi-NPC texts would help pin down the precise intended meaning of the dots and decide which cases are simply errors.

    The second problem is that modern Shanghai doesn't have a shang-qu contrast. Shang and qu-tone syllables with voiced initials (阳上, 阳去) have merged with the voiced ping ('level', A, MSM tone 1) tone (阳平), while voiceless shang and qu (阴上, 阴去) have merged with each other but remain separate from voiceless ping (阴平). This means in particular that for many initials, Standard Mandarin will show a shang-qu contrast lost in modern Shanghai, shockingly enough buy-sell: 买卖不分.

    Perhaps someone knowledgeable in Wu historical phonology could advise on the likelihood of Shanghainese distinguishing 阳上 from 阳去, or 阴上 from 阴去, ca. 1850. If it was in the process of losing that distinction, that would explain its inconsistent marking by the missionaries. If it had already lost it, the fact they tried to mark it at all could be explained by their exposure to different varieties of Chinese, and that of their informants.

    The elephant in the room is, of course, Shanghai tone sandhi. If you count the tones I described above and add two 'entering' (ru) tones, with voiced and voiceless initials (阴入, 阳入), you get five tones for modern Shanghai. But those are only distinguished for monosyllables in citation form ('citation' itself being a mode apt to elicit artificial distinctions). The melody of a longer phrase is essentially a function of only its length and the tone of the first syllable. A transcription such as NPC faces the choice between marking tone before and after sandhi, i.e. between syllabically marking citation tones and ignoring sandhi (as in pinyin or Jyutping) and 'accentuate' a whole phrase by marking only the first syllable (as in Simmons' Shanghai Romanisation). And of course I don't know to what extent the (relatively simple) rules of modern Shanghai tonal sandhi applied 150 years ago. In view of variation within Wu not far from Shanghai, sandhi could have worked quite differently back then.

    I might be wrong, but I think the Protestant missionaries in Shanghai tried to do it pre-sandhi. That's suggested by a cursory examination of the running prose in our mini-corpus, and is only to be expected if we think the Romanisation systems (e.g. for Mandarin) the Shanghai Baptists might have been exposed to. Tellingly, they mark a final glottal stop as -h, just like POJ ('Church' romanisation for Minnan), developed by the English Congregationalist Medhurst in the 1830s. Minnan topolects can have rather hardcore tone sandhi, but POJ ignores it and marks tones syllabically, pre-sandhi. Another source of inconsistencies in NPC texts could be (imperfectly understood) Shanghai sandhi colliding with attempts to write all morphemes in citation form.

    Part of what I see as inconsistencies could be just my misunderstanding of the intended logic, but NPC texts do unequivocally show inconsistencies elsewhere (e.g. hesitations in the final for 十 'ten' (Modern Shanghai (Simmons) zeq)). And nothing explains the apparent departing-tone marking for 年 (NPC nyeN, Modern Shanghai ni) you signalled.

RSS feed for comments on this post