In my last entry, I wrote about three types of nom characters for Vietnamese mắm 'salting':

1. m-phonetic characters: e.g., 𩻐 = 魚 'fish' + right of 鎫 m- 'head ornament for a horse' (Sino-Vietnamese reading unknown but presumably similar to its nom reading mâm)

2. c-phonetic characters: e.g., 鹵 'salt' + 禁 cấm 'to forbid'

3. b-phonetic characters: e.g., 酉 'liquor' + 稟 bẩm 'to receive from above'

The third type of mắm-character must have been devised at a stage when 'salting' had an initial closer to the initial of 稟 (i.e., stage 1 or 2 below):



b [ɓ] m [m]

The first type of mắm-character must date from stage 2 or 3.

The second type of mắm-character continues to baffle me. If I didn't know anything about Vietnamese or Chinese, I might propose a solution involving a labiovelar, but labiovelars did not exist in earlier Vietnamese, and禁 never had a labiovelar or a velar-labial cluster *kw- in Chinese. Did 'salting' once have a cluster *kɓ- in Vietnamese? There is no support for *k- in other Vietic languages.

I looked for other cases of c-phonetics for syllables with *ɓ- and other labial initials in the Nom Foundation's Kiều index and only found a single example: biếng khuây 'unforgettable' was written as 更亏 in line 246 of the 1872 version of Kiều. 更 is normally read as canh 'watch of the night' and cánh 'more'. Khuây is 'forget', and I doubt 更 has semantic relevance in 更亏: why write 'unforgettable' as 'watch forget' or 'more forget'? 更更 canh cánh 'obsessed' appears earlier in the line, so I wonder if 更 for biếng later in the line is an accidental substitute for the b-phonetic character that appeared in earlier editions. FORBIDDEN SALT

Last night I mentioned two examples of phonetics representing Vietnamese syllables with different onsets in the nom script. Here's a third.

As Vietnamese cuisine becomes more popular, more Americans are becoming familiar with nước mắm 'fish sauce'. Nước is literally 'water' and mắm is 'salting'. I do not know of any Sino-Vietnamese reading like mắm. The only similar Middle Chinese syllable was 鋄/鎫 *muamˀ 'head ornament for a horse'. I cannot find a Sino-Vietnamese (SV) reading for that rare character; in theory it should have been *vãm or, if it was borrowed earlier, *muộm. 鎫 was used as a phonetic symbol for the native Vietnamese word mâm 'tray', so its SV reading must have contained the consonant sequence m-m. Variations of its right side were used as a phonetic in nom characters for mâm 'tray' and mắm 'salting': e.g., 𩻐 mắm (with 魚 'fish' instead of 金 'metal' on the left side). (nomfoundation.org also lists a similar character with the codepoint U+29DE0 which may be a typo for U+29ED0, the codepoint for 𩻐. U+29DE0 is for a different character 𩷠 from a source in Taiwan. I cannot find the other 𩻐-like nom character in Unicode.)

There are two other types of characters for mắm which aren't in Unicode yet, so I have to describe them in terms of their semantic and phonetic components:

variations of 鹵 'salt' + 禁 cấm 'to forbid' (the latter is also a phonetic loan for the native Vietnamese word bấm 'to press')

酉 'liquor' + variations of 稟 bẩm 'to receive from above' (more on 稟 here)

Why was mắm written with a b-

Why was mắm written with a b-phonetic? Were the latter two types of characters devised when mắm still had an initial implosive *ɓ-? (Many other Vietic languages still have b- in this word: e.g. Sơn La Muong bam³. Is their b- implosive?) And why was bấm 'to press' written with a c-phonetic 禁?

8.16.1:32: I suspect Proto-Vietic *ɓamʔ 'salting' (as reconstructed in the SEAlan

phonetic? Were the latter two types of characters devised when mắm still had an initial implosive *ɓ-? (Many other Vietic languages still have b- in this word: e.g. Sơn La Muong bam³. Is their b- implosive?) And why was bấm 'to press' written with a c-phonetic 禁?

8.16.1:32: I suspect Proto-Vietic *ɓamʔ 'salting' (as reconstructed in the SEAlang Mon-Khmer Languages Project database) is a Vietic innovation. I have not found any potential true cognates in other langauges in that database. Halang măm 'salt fish' and Mnong măm 'salted fish' are probably Vietnamese loans in those Bahnaric languages, and Bolyu mjaːm¹³ 'salt' may be a lookalike; its -j- matches nothing in Vietic. BIT-TẢI-R ROOF

In my last entry, I couldn't explain why 宰 was read as tể instead of tải in Vietnamese. Today I checked various nom dictionaries and found the reading tải in Vũ Văn Kính's Bảng tra chữ nôm sau thể kỷ XVII (18, 19, 20) (Table for Finding Nom Characters after the 17th Century (18, 19, 20)). Unfortunately the book did not provide a context for tải, so I don't know if that syllable was a now-extinct Sino-Vietnamese reading or (part of) a native word. I also don't know if that reading predates the 18th century. My guess is that the taboo substitution occurred in the 18th century (hence the inclusion of the original reading tải in Vũ's book), and that most works only include the later altered reading tể and its spinoffs tẻ and tỉa. (It would be unusual for an -ai character to be used to write -e and -ia syllables, so I assume the latter two readings postdate tể.)

The Nom Foundation's Kiều index lists yet another reading in line 2873 of the 1870 version: tề. However, page 206 of its romanized text of that version has the usual reading tể.

I just realized that although characters could be used as nom phonetic symbols and components without regard for tone (e.g., 宰 tể for tề in Kiều?), all taboo deformations I have seen retained tones along with onsets*. Final consonants could be slightly changed: e.g., hoàng [hwaːŋ] became huỳnh [hwiɲ]. The hierarchies of 'loyalty' for nom phonetics and taboo deformation were slightly different:

nom phonetics: vowel quality > onsets, codas, tones**

taboo deformation: onsets, tones > codas > vowel quality

Nom phonetics were generally used for syllables with similar vowels: e.g., 宰 tể could not represent a syllable like tổ even though it had the same onset, tone, and zero coda. However, native Vietnamese words had more onsets and onset-coda sequences than Sino-Vietnamese, so there was more freedom to use phonetics to represent syllables with different onsets or codas: e.g.,

la as a phonetic with semantic 出 xuất 'to go out' in 𠚢 ra 'to go out' (there are no r-syllables in Sino-Vietnamese)

n as a phonetic with semantic 口 khẩu 'mouth' in 𠵘 mồm 'mouth' (there is no syllable môm with any tone in Sino-Vietnamese)

I'll look at another example tomorrow. Note how tone is disregarded in the latter case. 羅 la can represent là 'to be' with a different tone.

*The spellings of initial onsets could change because of quốc ngữ spelling conventions: e.g., kiểu [kiəw] became cảo [kaːw].

**8.15.1:30: Vietnamese tones historically had 3 x 2 categories. Each tone name exemplifies its tone.

 voice quality *plain *creaky *breathy
*voiceless initial > *upper register ngang sắc hỏi
*voiced initial > *lower register huyền nặng ngã

The reconstructed category names no longer necessarily describe the modern tones: e.g., huyền is breathy, ngã is not breathy and is higher than hỏi, etc.

There seems to be a hierarchy of tonal 'loyalty' in nom:

1. Retention of original tone in phonetic symbol/component.

2. Use of phonetic for syllable with opposite-*register tone: e.g., 羅 la for là above.

3. Use of nonplain tone phonetic for syllable with any other *nonplain tone: e.g., 禮 lễ for lấy, lạy, and rẻ in Kiều. (Lễ 'ceremony' and lạy 'to bow' are in fact the same Chinese word borrowed into Vietnamese during two different periods.)

4. Use of phonetic for syllable with any tone: e.g., 永 vĩnh for the *plain tone syllable vành as well as *nonplain vắng and vạnh in Kiều. (Is there an example of a phonetic used for syllables with all six tones?)

The most 'loyal' phonetics have readings ending in stops. Stop-final syllables in Vietnamese can only have *creaky tones: e.g., 越 việt also represented the *creaky-tone syllables vượt, vết, and vớt, but could not represent *noncreaky tone syllables like *vVt, *vV̀t, *vV̉t, or *vṼt which were impossible in Vietnamese. BIT-TỂ-R ROOF

I regret not paying attention to Vietnamese until I started reading Bernhard Karlgren's books after my first semester of graduate school over twenty years ago. I remember flipping through a Vietnamese-English dictionary and being astounded by all the words I could recognize because they were Chinese borrowings. (Of course the native words were totally alien to me, as I had never studied an Austroasiatic language before, much less one that was closely related to Vietnamese like a variety of Muong. I didn't even know what Muong was!) I soon learned the sound correspondences between Sino-Vietnamese and what I was more familiar with (Mandarin, Cantonese, Sino-Japanese, and Sino-Korean). Since then I've committed many Sino-Vietnamese readings to memory and can guess still others using those correspondences. When I look at Vietnamese, I can usually 'see' the characters for Chinese loans. However, there are 'blind spots': i.e., exceptional readings.

One such reading that I can't explain is tể instead of the regular reading *tải for 宰 'minister'. (The title refers to the components of the character: 宀 miên 'roof' and 辛 tân 'bitter'. Why those add up to 'minister' is a topic for another time.) 宰 belongs to the Middle Chinese 海 *-əj (> later *-aj) rhyme category which usually corresponds to three rhymes in Sino-Vietnamese:

Old borrowings: -ơi [əːj]

Later borrowings with nonlabial initials: -ai [aj]

Later borrowings with labial initials: -ôi [oj]

tể is the only instance of [e] in this category. It is probably not an archaism from Old Chinese since Middle Chinese *-əj goes back to *-ə, not *-e. I do not know of any cases of *-ai becoming in Vietnamese.

宰 must have been read with a front vowel when it was used as a nom phonetic symbol to write the unrelated native Vietnamese words lẻ tẻ 'scattered' and tỉa 'to trim'.

Although there are modern Chinese languages in which this rhyme has become e-like, they are geographically distant from Vietnamese with the sole excpetion of only one variety of Pinghua (Guilin Yanshan Zhuyuan which has tse with an irregular tone). I doubt tể is a borrowing from Guilin which is 400 miles from Hanoi.

Is tể the last survivor of a long-dead trend of monophongization in earlier Vietnamese or the source dialect of Sino-Vietnamese? I doubt it.

8.14.1:42: Could the sui generis reading tể be the product of taboo deformation? But would 宰 have been used in a name? Tể is not in this long list of deformed readings that I just found. (The original readings are in the "Âm chính" 'main sound' columns; the altered readings are in the "Âm trại" 'mispronounced sound' columns.)

I tried looking for tể in de Rhodes' dictionary to see if tể existed in the 17th century. However, I couldn't find it or my theoretical regular *tải with any of the meanings of tể.  THOUGHT-BEARING HAPPY PROGRESS?

Thai names contain many Indic elements, so they should be transparent to me. However, they often contain surprises. For instance, last night I encountered the name

จินตหรา สุขพัฒน์ <cinthrā sukhbaḍhn˟> [tɕintaraː sukʰapʰat] (?) 'Chintara Sukapatana'

which looks like it should be from an Indic *cinta-harā sukha-baḍhana-. However, only [sukʰa] < Sanskrit/Pali sukha- 'happy' is straightforward. The remaining three components puzzle me:

- I would expect the final long vowel of Sanskrit/Pali cintā 'thought' to remain intact in compounds; this same shortening is also in regular Indic loanwords in both Thai and Khmer (so perhaps the shortening is of Khmer origin)

- apparently <hrā> is pronounced as if it were a monosyllabic native Thai word [raː] rather than the expected [haraː] from Sanskrit/Pali harā 'bearing' (f.). There is a Thai word หรา <hrā> [raː], but I don't think it is part of this name because it is an adverb 'boldly', not an adjective which should follow [tɕinta] 'thought'.

- although I'm accustomed to Pali vaḍḍhana- 'increase' (< Sanskrit √vṛdh) becoming the regular Thai word พัฒนา <baḍhnā> [pʰattʰanaː] 'progress', I didn't expect it to be clipped to [pʰat]. (์ <˟> indicates silent characters.) Was *[sukʰapʰattʰan] too long? Is the feminine ending [aː] always absent from Thai surnames?

Chintara's birth name is

จิตติมาฆ์ <cittimāgh˟> [tɕittimaː]

which has mysteries of its own. [tɕitti] is from Sanskrit/Pali citti-, a variant of cintā 'thought', but what is [maː] from Sanskrit māgha- 'name of a constellation' (> 'third lunar month' in Thai and Khmer) doing, and why was its final consonant dropped? Compare มาฆ์ <māgh˟> [maː] with เมฆ <megh> [mek] < Sanskrit/Pali megha- 'cloud' whose final <gh> [k] is not silent.

Has the phenomenon of dropping perfectly pronounceable segments in Indic loans in Thai been studied? (Some dropping is required to make Indic loans fit the constraints of Thai phonology: e.g., จันทร์ <candr˟> 'moon' is [tɕan] because Thai does not permit final consonant clusters.) C-RUTSUBO

Li (2008: 721)  listed the first syllable of

4538 5544 2ko' 1riuʳ 'crucible'

as a borrowing from the second syllable of Chinese 坩堝 *kã ko 'crucible'.

I am skeptical of a connection between the two words for the following reasons.

First, I do not know of any cases of the rhyme of 堝 *ko borrowed as Tangut rhyme 54 -o'. I use the symbol ' to indicate that rhyme 54 was similar to rhyme 51 -o yet different in some unknown way. I only know of a single case of rhyme 54 transcribing Chinese *-o:

5388 2bo' for Chinese 摩 *mbo (Gong 2002: 436)

Chinese *-o was normally transcribed with rhyme 51 -o (see Gong 2002: 456 for many examples).

I originally was going to write that I thought it was unlikely that 坩堝 *kã ko would be cut in half by the Tangut, but in fact 堝 *ko is attested as an independent word in the Song Dynasty. I suspect it is a specialized use of 鍋 *ko 'cooking pot' written with a radical 土 to match 坩 which is attested as an independent word in the Tang Dynasty.

So my third objection is now my second:  Even if 4538 is a borrowing from Chinese, what is 5130? No homophone of 5130 is an adjective that would make sense as a modifier of 4538. Here are Li's (2008) glosses for the other tangraphs pronounced 1riuʳ:

0968 'all'

1403 'complain'

2147 'sweep'

2324 'sigh'

2542 'gadfly'

2543 'hate'

2812 'cherish'

3491 'bright star'

3493 'firefly'

3737 'frivolous'

4364 'wooden framework'

4437 'auspicious'

4713 'world'

5130 'subdue'

I think Tangut 2ko' 1riuʳ 'crucible' is an indivisible disyllabic word that is a coincidental soundalike of 堝 *ko.

As for the title, Nishida (1986: 43) translated Tangut 2ko' 1riuʳ 'crucible' as Japanese rutsubo. No native Japanese word can begin with r-, so the word must be of at least partly foreign origin. I think it might be a compound of Middle Chinese 爐 *lo (borrowed into Japanese as *ro which then became ru after raising) 'stove' and the native word tsubo 'pot'. FROZEN WHITE WATER IN THE BLACKSMITH'S CRUCIBLE

Andrew West pointed out that 4053 1ʔwọ 'ice' from my recent entries occurs three times in The Ode on Monthly Pleasures: e.g., in the 'common language' line 3B of the section on the second month. The 'ritual language' line 3A is slightly longer:

2.3A 1 2 3 4 5 6 7

Li Fanwen number 0804 4051 3052 1659 5441 4538 5544
Reading 2diə 1kiʳw 1nioʳ' 2lew 2swi 2ko' 1riuʳ
Gloss PERF cold water white blacksmith crucible

The 'common language' line only has a single direct match:

2.3B 1 2 3 4 5 6 7

Li Fanwen number 1490 4053 1572 0185 1452 3956 -
Reading 1tsʊʳ 1ʔwọ 1phɤõ 2nwiə 1nia 1dʐi -
Gloss winter ice white spring PERF melt -

Nishida (1986: 43) translated those lines as topic-comment sequences:

A. 'The cold, white water - a crucible of materials'

B. 'Spring melts the white water of winter'

He proposed four parallel pairs:

A1-2 'cold' : B1 'winter'

A3-4 'white water' : B2-3 'white ice'

A5 'materials' : B4 'spring'

A6-7 'crucible' : B5-6 'melts' (my 'melted')

A1 0804 2diə is a perfective prefix originally indicating motion toward the speaker. Perhaps its combination with A2 4051 1kiʳw 'cold' could be translated as 'frozen' (i.e., finished becoming cold). A 'common language' perfective prefix is not what I would expect in the 'ritual language' if the latter was an unrelated substratum language.

As far as I know, A2 4051 is only in dictionaries and this ode. If it is a 'ritual language' word, it demonstrates that 'common language' affixes can be attached to 'ritual language' vocabulary.

A3 3052 1nioʳ' is a 'common language' word for 'water' that is not very common. It is the Tangut name of the Chinese trigram ☵ for water. If the 'ritual language' were a low-prestige substratum language, I would not expect its words to be used to refer to concepts from a high-status culture. I should look into the names of the seven other trigrams; none match the common words for their concepts.

Similarly, A4 1659 2lew is a 'common language' word for 'white' that is not very common.

I do not know why Nishida translated A5 5441 2swi as 'materials'. A note in Homophones text D equates 2swi with 'iron artisan' (i.e., 'blacksmith'). 5441 is also a verb 'to (s)melt' (Kychanov and Arakawa 2006: 542), so a blacksmith was a 'smelter'. 5441 can also mean 'mother-in-law' (Li 2008: 858), but I assume the character is used for two different unrelated words. Unfortunately, I do not see this term for 'mother-in-law' in Jacques 2012 which mentions the term

3986 4893 1niə 1vɨə

A6-7 4538 5544 2ko' 1riuʳ 'crucible' is a indivisible disyllabic word which is not in the 'common language' to the best of my knowledge. Neither half can stand alone. Li (2008: 721) regarded the first syllable as a loan from the second syllable of 坩堝, but this is problematic for reasons I'll go into in my next entry.

A5-7 'blacksmith's crucible' makes little sense as a gloss for 'spring melted'. It is an odd metaphor for spring, as a crucible is much hotter. Nishida's 'crucible of materials' is even more puzzling.

I would translate the B line as a topic-comment sequence:

'As for the white ice of winter, spring melted it.'

B6 1452 1nia is a perfective prefix originally indicating downward motion, so B6-7 1452 3956 1nia 1dʐi 'down-melt' (i.e., 'melted') is reminiscent of English melt down, though they are not equivalents: the former could be translated as 'melted down' but not 'melt(s) down' and the latter has an extended meaning '(emotionally) collapse'.

Tangut fonts by Mojikyo.org
Tangut radical and Khitan fonts by Andrew West
Jurchen font by Jason Glavy
All other content copyright © 2002-2014 Amritavision