14.12.03.23:45: WERE TANGUT RETROFLEX INITIALS PRONOUNCED WITH LIP-ROUNDING?

Tangut Grade III rhymes are normally preceded by 'vigilant' initial consonants. Although those consonants may seem to have nothing in common at first glance, I presented a case for their similarity here. Tonight I will add to that case by drawing parallels between Tangut and modern European languages. Flemming (2013: 55) wrote,

Non-anterior coronal fricatives are often produced with lip-rounding, for example post-alveolar fricatives in English and French and retroflex fricatives in Polish are all produced in this way (Ladefoged and Maddieson 1996: 148, Puppel, Nawrocka-Fisiak, and Krassowska 1977: 157). Again, there is no articulatory basis for this relationship. From an auditory point of view, however, we can see that rounding serves to enhance the distinctiveness of these sounds.

I now suspect that Tangut retroflex shibilants (Class VII tʂ-, tʂh-, dʐ-, ʂ- and Class IX ʐ-) were pronounced with lip-rounding like Polish retroflex cz dż sz rz/ż [tʂ dʐ ʂ ʐ]. This lip-rounding was sufficient for them to behave like Class II v- and Class IX l- which may have been partway between [ɫ] and [w] (cf. Polish ł [w] < *ɫ). The -w- of the irregular Tibetan transcription zhwa (as well as zha) for

4184 1ʂɨa 'to appear; a transcription character for Sanskrit śa and ca'

may reflect that lip-rounding. A medial -w- was phonemic after Tangut labialized retroflex shibilants: e.g.,

0372 1ʂwɨa 'to hold'

was distinct from 4184 1ʂɨa above, just as English schwa is distinct from Shah.

Although English [ɹ] may be pronounced with lip-rounding, I cannot tell if Tangut Class IX r- had such rounding because it only appeared in rhymes in which the Grade III/IV distinction was neutralized. If Tangut r- was retroflex, I would expect it to behave like the other retroflexes and have lip-rounding. But if Tangut r- was an alveolar tap or flap [ɾ], then it wouldn't have lip-rounding.

14.12.01.23:18: EDWIN G. PULLEYBLANK (1922-2013)

Edwin G. Pulleyblank, perhaps my greatest influence, passed away last year. I found the task of writing about his impact on me to be so daunting that I kept putting off this entry. But I thought it would be appropriate to finally type something after my entry on Roy Andrew Miller, another man who changed my life. Short and late is bad, but nonexistent is worse.

My first exposure to Chinese historical phonology was in my second year as a Japanese major at Berkeley. My Classical Chinese class used George Kennedy's ZH Guide: An Introduction to Sinology as a textbook. Kennedy used simplified forms of Bernhard Karlgren's reconstructions of Middle Chinese with Gwoyeu Romatzyh-influenced tonal spelling to romanize Chinese characters in the book: e.g., ZI HOJ for 辭海 'word sea' (the ZH in the title). In a Japanese language history class, I used the Yunjing rhyme tables - presumably with help from Karlgren's reconstruction - to try to figure out the eight 'vowels' of Old Japanese. I can't remember what I concluded. It was almost certainly wrong: i.e., not much like what I wrote in my PhD dissertation eight years later.

Then I went to Japan in the summer of 1991 and picked up The Gakken Great Sino-Japanese Dictionary with Tōdō Akiyasu's Old and Middle Chinese reconstructions which were similar to Karlgren's.  Tōdō's Middle Chinese reconstruction was the first that I became familiar with. It was easy to transition from it to Karlgren's own reconstruction in his Analytic Dictionary of Chinese and Sino-Japanese which I bought in the fall of 1992.

Two characteristics of Karlgrenian* reconstructions are

- -y- or -y-like medial consonants** in over half the syllables: e.g., in syllables such as kya, chyi, pyu, etc.

- a large number of vowels

Karlgren's reconstruction had fifteen, including six kinds of a. Ramsey (1987: 131) described his vowel inventory as "exceedingly complex" and regarded "this overabundance of vowels [as] the least satisfying part of Karlgren's reconstructions."

Apparently no attested language has that much -y-. But I didn't know that at the time. How could the great Karlgren be wrong?

So when I first glimpsed Pulleyblank's Middle Chinese at a bookstore shortly before graduating Berkeley, I was shocked by his reconstruction. It was so unlike Karlgren's. It couldn't be right ... could it? And I wasn't going to pay $40 to give it a chance.

After returning to Hawaii and enrolling as a graduate student at the University of Hawaii, I spent much of 1993 borrowing every book of Karlgren's I could find in the library. I internalized Karlgren's Middle Chinese reconstruction and his arguments for it. (It is to Karlgren's credit that he could explain himself in a way that a beginner could understand.) I was a true believer.

In the fall, Leon Serafim lent me his copy of Middle Chinese. Now that I could read it at my leisure for free, I began to see the error of my ways.

I was disturbed by how Karlgren's -y- often corresponded to nothing in Sino-Vietnamese, Sino-Korean, and Sino-Japanese, even though those languages could have consistently reflected it - if it had really existed.

Moreover, I had begun my study of Chinese-like Southeast Asian languages, and neither they nor living Chinese languages had vowel inventories like Karlgren's.

Pulleyblank's reconstruction had looked so odd to me at first sight because it dealt with both these issues. Pulleyblank had reconstructed far fewer -y- and a smaller, more realistic vowel system.

Daniel Bryant provided the background of Pulleyblank's revolutionary reconstruction:

As EGP [Edwin G. Pulleyblank] himself has explained, it became clear to him, both from problems encountered in the use of Chinese transcriptions of foreign words in the writing of Background [The Background of the Rebellion of An Lu-shan] and in his attempts to bring historical phonology to bear on grammatical questions in Classical Chinese, that the reconstructions by Bernhard Karlgren of the premodern phonology of Chinese simply could not be made to agree with a broader range of evidence than Karlgren had used. Confronted with the monolith of Karlgren’s system and his reputation as the virtual creator and certainly the greatest master of modern studies of the historical phonology of Chinese, a lesser scholar might well have either suppressed the difficult evidence or, even more likely, have looked for a less problematical field in which to make a career. Nothing is more characteristic of EGP than his resolve to get to the bottom of the problem, whatever difficulties might arise in the form either of enormous toil or of injured sensibilities - Karlgren, for all his greatness as a philologist, regarded the reconstruction of premodern Chinese pronunciation as an accomplished task and, not incidentally, as one of his own accomplishing. The full force of his derision awaited anyone so incautious as to challenge the validity of his system, and all the more so in the case of EGP, whom the evidence soon compelled to move beyond the stage of simply amending Karlgren’s system in a few details to that of setting out to rethink the entire structure.

Pulleyblank made me "rethink the entire structure", but Karlgrenian reconstructions continued to be popular for half a century. I wish Pulleyblank had seen his reconstruction eclipse its Karlgrenian competitors during his lifetime. Alas, that was not to be.

I was never Pulleyblank's student, but I did see how his work did not receive the embrace that it deserved, and I learned the value of standing up for oneself and not giving in. Pulleyblank did his own thing by gathering unexpected evidence and challenging old assumptions. If only more were so bold.

I tried to follow Pulleyblank's example when reconstructing Tangut. My Tangut sound system has been de-Karlgrenized compared to its competitors. I trimmed excess -y- (= -i- in my notation) and simplified the huge vowel system as much as I could. All of my Tangut vowels are variants on a mere six vowels: u, i, a, ə, e, and o. I've gone outside the Sino-Tibetan box, looking to Korean and Khmer for inspiration when positing sound changes between Tangut and pre-Tangut. I don't see Tangut as a sui generis language; I see it as just another human language that might have been less exotic than people think.

All scholars should emulate Pulleyblank and, in Bryant's words, "think clearly about the effects of ignoring relevant material or of imposing arbitrary limits on the scope of one’s sources." Evidence is everywhere, and we mustn't hesitate to leave our comfort zones to find it - even if it forces us to abandon our initial conclusions or even reject the consensus. Unlike Karlgren who barely changed his Chinese reconstruction over the years, Pulleyblank constantly refined his reconstruction, and I myself consider my Tangut reconstruction to be a work in progress, subject to revision upon the discovery of new data. Stand your ground even when your peers reject you, but don't stand still for long. As this song I heard today put it, keep on moving! Pulleyblank never stopped. Nor will I. May his restless, indomitable spirit live on in us.

*12.2.4:33: I use the term 'Karlgrenian' to refer to Karlgren-like reconstructions such as Tōdō's as well as Karlgren's own reconstructions.

**I use y as a lay-friendly substitute for Karlgren's -i̯- and others' -j-.