The fourth sort of sinograph is 假借 jiajie, translated by Djamouri (2006: 10) as 'false loan'. Jiajie characters represent a word by using an already existing character for another (nearly) homophonous word. These rebuses are 'false' in the sense that they are not WYSIWIG (what you see is what you get): a drawing of A can represent a homophone B whose referent may look nothing like A (if it looks like anything at all). The examples he gives are

a stick-figure drawing of a child

originally for Old Chinese tsə' 'child'

also used for Old Chinese slə' 'snake (calendrical term)'

(The mismatch of initials [ts- : sl-] is difficult to explain, and may indicate that the reconstructions are wrong - maybe 'child' had st-? In Pulleyblank's [1991: 52] eccentric reconstruction, there is no mismatch: 'child' and 'snake' both had velar initials [ky- and G-].)

a drawing of one person following another

originally for Old Chinese pi' 'compare', pi's, Npi's 'follow'

also used for Old Chinese pi', pi's 'ancestress'

A simple English example would be a drawing of an eye for the pronoun I. That case - and many actual cases in Chinese (e.g., a drawing [於] of a crow [Old Chinese 'a] for the OC preposition 'a 'in') - demonstrate how the jiajie technique is useful for writing words that can be difficult to visualize.

Did tangraphy have a jiajie component? One probable example is

TT1730 ALSO (Gong tsyï 1.30)

which also appears as a semantic component meaning 'small' (cf. the shapes of Chn 小 'small', 少 'few') in other tangraphs: e.g.,

TT1496 SMALL (Gong zyi 1.11)

TT1174 FINE (Gong syiy 1.36)

TT0147 FINE-SILK (Gong tsyi 2.10)

TT5522 FINE (Gong tshyïy 1.42)

TT2597 NEEDLE (Gong Ga 1.17)

compound of METAL + TT5522 FINE

TT4063 FINE-DUST (Gong dzu 2.1)

compound of EARTH + TT5522 FINE

TT1949 INFANT? (Gong ze 1.8)

TT3584 SMALL (Gong tsəy 1.40)

looks like PERSON + SMALL/ALSO

TT4015 (a [place?] name) (Gong sew 2.38)

compound of EARTH + SMALL/ALSO

sounds like Chn 小 'small', 少 'few'

TT5700 SMALL-BIRD (Gong lew 2.38)

compound of WAIST + SMALL/ALSO

(WAIST may be used here and elsewhere as jiajie for BIRD)

TT5798 CALF (Gong mə̣ 1.68)

compound of an element resembling Chn 羊 'sheep' + SMALL/ALSO

TT3951 LIGHTLY-SIT? (Gong tswe 1.12)

compound of 2/3 of SIT + SMALL/ALSO

Note that it is impossible to interpret SMALL/ALSO as 'small' or 'also' in all the tangraphs it appears in*.

What seems to have happened is that

- Chn 小 'small' and/or 少 'few' inspired the shape for the tangraphic component SMALL:

- and the Tangut word for 'also' (tsyï 1.30) which sounded like 'small' (tsəy 1.40) was written as SMALL

- while the word 'small' (tsəy 1.40) was written with a different tangraph

with PERSON added to the left. (Why?**)

However, nothing in Tangraphic Sea confirms a jiajie origin for this tangraph or any other. TS analyzes ALSO as

all of TT1496 SMALL (Gong zyi 1.11) minus (not plus)

left of TT1496 SMALL (Gong zyi 1.11)

Next: Category V: Shapes and sounds.

*SMALL/ALSO does not appear to be a semantic element in the following tangraphs (which do not sound like ALSO tsyï 1.30):

TT0282 SHAMAN (Gong wyï 1.69)

TT0500 DARE-EAT-DRINK-MUCH (Gong kyï 2.61)

compound of MOUTH + SMALL/ALSO

looks like it should mean something like 'eat/drink very little'

TT2570 SWORD (Gong byïr 1.86)

compound of METAL + SMALL/ALSO

TT3365 SORE/WOUND (Gong lyi 2.9)

looks like PERSON (did this word only apply to people?) + 二 (function unknown) + SMALL/ALSO

TT4541 DITCH (Gong khow 1.54)

with a mystery top component ユ also found in WINTER (looks like ユ + COLD)

TT5514 SHAKE (Gong Gu 1.4)

That is not an exhaustive list.

TT1376 BIG-THUMB (Gong myaa 1.23)

looks like NOT + SMALL/ALSO, but doesn't simply mean 'big'. Nothing in the tangraph suggests 'finger'. Why not write this as BIG + FINGER?

The function of SMALL/ALSO is uncertain in:

TT1076 (a surname) (Gong tya 1.20)

written with ONE on the left in Grinstead (1972: 81)

TT3801 (a kind of insect - a small one?) (Gong kyow 2.48)

looks like INSECT + SMALL/ALSO

**Tangraphic Sea analyzes

TT3584 SMALL (Gong tsəy 1.40)

as a cryptosemantic compound of

right of TT5070 FEW (Gong zyïïr 1.92)

all of TT1730 ALSO (Gong tsyï 1.30)

though SMALL could be misinterpreted as SMALL-PERSON or as a compound of PERSON from some tangraph other than FEW plus SMALL. I wonder if SMALL was intended to represent a Tangut B word 'small' consisting of a prefix homophonous with 'person' + a root 'small'. ÉCDUeR 3: THE SIX SORTS OF SINOGRAPHS: 會意 HUIYI

The third sort of sinograph is 會意 huiyi, translated by Djamouri (2006: 10) as 'combining ideas'. Huiyi characters represent a word "[t]hrough the juxtaposition of two figurative components evoking iconically the object or the event to which the designated word refers". The example he gives is

for Old Chinese ris 'cut; sharp; profit'

a combination of

禾 (for Old Chinese gway 'ear of millet') and

刂 (for Old Chinese taw 'knife').

Note that 利 ris does not sound like either 禾 gway or 刂 taw - or a combination of the two (e.g., gwaw). The pronunciation of a huiyi character is not predictable on the basis of the pronunciations of its parts. Moreover, there is no way to visually distinguish between a huiyi character (a semantic compound) and a phonetic-semantic compound (PSC): e.g., 利 looks like it could be a PSC of 禾 (phonetic) + 刂 (semantic) pronounced something like 禾 gway, but it's not. 利 does not stand out among PSCs like


which consist of various phonetic elements plus the semantic element 刂 'knife'. One simply has to memorize that 利 is an exception to the strong tendency for compound sinographs to be PSC. The vast majority of sinographs are PSCs.

Conversely, the majority of tangraphs are thought to be semantic compounds: e.g.,


TT4106 RIDE (Gong dzeey 2.34) analzyed as

TT4105 PERSON (Gong dzywo 2.44) atop

TT5233 HORSE (Gong ryiry 1.74)

This alone would make tangraphy the world's only semantocentric writing system. Even sinography, usually assumed to be 'pictographic', is actually phonocentric, like all other writing systems - except tangraphy? But tangraphy is even more unusual: its huiyi characters consist of parts of characters thought to invoke wholes. Such partial invocation is the exception in Chinese but is apparently the rule in tangraphy. I don't think it is possible for a successful writing system like tangraphy to be semantocentric - or, worse yet, cryptosemantocentric (i.e., based on partly hidden semantic elements). Could readers realistically 'fill in the blanks' when confronted with tangraphs consisting of part of graph A and part of graph B: e.g., how could they know that the left side of

TT4105 PERSON (Gong dzywo 2.44)

was from the right side of

TT3769 HEART (Gong nyiiy 2.39)

rather than from one of the 19 other tangraphs with that component (which never occurs by itself) on the left or the 8 other tangraphs with that component on the right: e.g.,

TT4095 FILIAL-PIETY (Gong 1.31) or

TT5344 FEAR (Gong kyïy 1.42)

analyzed as right of FEAR (Gong lhyi [rhyme unknown]) + left of HEART

but the right of FEAR lhyi looks like

ONE-OF-TWO (Gong dzyiy 2.33)

and how is anyone supposed to know the right side of FEAR kyïy 1.42 is from HEART as opposed to PERSON? TT5344 could be misinterepreted as ONE-OF-TWO-PEOPLE (e.g., a spouse - half of a couple).

Granted, neither FILIAL-PIETY nor FEAR evoke PERSON, but HEART doesn't suggest PERSON to me either - after all, humans aren't the only creatures with hearts. One could also argue that 禾 'ear of millet' doesn't suggest 利 'cut; sharp; profit' either, but at least graphs of the 利 type are the minority in Chinese, whereas such graphs are supposed to be the majority in Tangut. Could tangraphy really have been so successful if its characters were based on oblique semantic associations of ambiguous parts of other characters? My answer is no. (Longtime readers already know what I think tangraphy really is; newer readers can look forward to an explanation in a future post.)

Next: Category IV: False loans. ÉCDUeR 2: THE SIX SORTS OF SINOGRAPHS: 指事 ZHISHI

The second sort of sinograph is 指事 zhishi, translated by Djamouri (2006: 10) as 'indicating an event'. Zhishi characters represent a word "[t]hrough iconic 'evocation' of the event or the object of experience". The examples he gives are

上 下

'up' and 'down'

One points up and the other is a near-mirror image pointing down.

Tangut has no zhishi characters, though it does have zhishi elements for 'top' and 'bottom' - actually the same shape in with different locations and orientations:

Examples of the element TOP:

TT0188 TOP-OF-HEAD (Gong chyiw 2.4)

TT0207 STAR (Gong gyịy 1.61)


TT0209 WEAR-ON-TOP (Gong chiey 2.31)


TT0238 HEAD (Gong lyụ 2.52)

related to Old Chinese 首 hlu' 'head'?

TT0240 BRAIN (Gong no 2.42)

TT0243 BRAIN (Gong no 2.42)

both related to Old Chinese 腦 nu' 'brain'?

Examples of the element BOTTOM:

TT0392 SEED (Gong sywï 1.3)

seeds are planted in the ground (bottom)?

TT1686 SEED (Gong lywi 1.11)

GRAIN element added to left; cf. Chn 禾 'ear of millet'

TT3515 BED (Gong jywi 2.9)

where one lies down?

TT3749 SIT (Gong dzuu 2.5)

a calque of Chn 坐 'sit', with two PERSON tangraphs instead of Chn 人 and the element 'bottom' instead of Chn 土 'earth'; related to Old Chinese 坐 dzoy' 'sit'?

TT4024 EARTH (Gong thyi 2.10)

borrowed from Chn 地 'earth'; the native word for 'earth', lyị̈ 2.61 (related to Old Chinese 地 rlays 'earth'?), is written without the 'bottom' element)

TT4413 BEDDING (Gong lyoo 2.46)

TT4782 ?BOTTOM (Gong wyi 2.9)

TT2996 SOLES-OF-SHOES (Gong thwo 2.42)

TT5271 FALL (Gong kiew 1.44)

TT5390 LEVEL (Gong kạ 2.56)

looks like EVEN/EQUAL ka 1.17 (phonetic/semantic) + BOTTOM

i.e., on the same plane / level (ground)?

TOP and BOTTOM are ironically not in

TT3619 TOP/HEAD (Gong Gu 1.4)

TT5446 DOWN/BOTTOM (Gong chhyïy 1.42)

used to refer to the tops and bottoms of tangraphs in the analyses in Tangraphic Sea.

Why not?

Conversely, why is TOP in

TT0083 DRAGON (Gong we 1.8)

the TOP animal? (but the bottom components don't comprise an independent tangraph)

TT0183 OX (calendrical term) (Gong myuu 2.6)

TT0190 OX (Gong byu 1.2)

one might think there couldn't be two TOP animals, but there are several!

TT0195 SEE (Gong lyiy 2.33)

is TOP supposed to be like a turned version 罒 of the 目 'eye' at the top of Chn 見 'see'?

TT0202 BROTHERS (Gong byu 2.2)

analyzed as MOTHER (see below) + PERCEPTION (Gong byu 1.2; phonetic)

TT0203 RICH (Gong lo; rhyme unknown)

no analysis known

people at the TOP are RICH?

TT0225 TURQUOISE (Gong khu 1.1)

analyzed as GOLD (see below) + WATER zyïïr 2.85

TT0226 GOLD (Gong kiẹ 1.66)

not TOP + METAL (METAL is not in this tangraph at all)

the bottom is RECOGNIZE (Gong tsywar 1.82, said to have the right of TRUTH [Gong Giey 1.34]): RECOGNIZEd as being the TOP (metal)?

analyzed as top of RICH + bottom of TRUTH (Gong Giey 1.

TT0241 SUMMER (Gong jywiy 2.32)

TT0245 MOTHER (Gong mya 1.20)

And why is BOTTOM in

TT3340 COMPREHEND (Gong dar 1.80)

looks like HOLY + WAIST + BOTTOM

to comprehend is to get to the bottom of something?

TT3730 (a surname) (Gong mey 2.7)

looks like pho 1.49 (used to transcribe Chn-filtered pron. of Skt pha) + BOTTOM

TT3854 MASTER (Gong wyi 2.9)

looks like PERSON + BOTTOM rather than PERSON + TOP)

TT5231 ?SADDLE BLANKET (Gong lyu 1.3)

looks like ORNAMENT tshyịy 2.54 (< HORSE + PERSON?)

TT5762 SPIRIT (Gong myiry 2.68)

looks like WAIST + BOTTOM

written in Nevsky (1960 II.319) and Grinstead (1972: 72) as WAIST + [WAIST + BOTTOM]; there is nothing bottom-ish about Nevsky's Chinese translations 通 'penetrate' or 神 'spirit'

The answer for TT3854 MASTER wyi 2.9 is obvious if one looks at the reading wyi 2.9 of TT4782 ?BOTTOM above. Nonetheless, BOTTOM is an ironic choice for an element in MASTER.

Next: Category III: Combining ideas. ÉCDUeR 1: THE SIX SORTS OF SINOGRAPHS: 象形 XIANGXING

One of the highlights of my academic career was my invitation - expenses paid! - to Paris for a 2000 conference on sinography (Chinese writing). Articles based on papers presented there (including mine) have been reprinted in Écriture chinoise: données, usages et représentations (Françoise Bottéro and Redouane Djamouri, eds., 2006; hereafter ÉCDUeR). I got my complimentary copy in the mail on Saturday. Anyone who enjoys this blog - at least when I talk about Chinese characters - would enjoy this book. From the back cover:

This volume gathers together the points of view of different linguists on the Chinese script. They all stress the fact of the primacy of the phonetic relation between graphic signs and words of the language they represent. Regardless of whether we examine the earliest attestations of Chinese characters, or look at the principles that govern their formation and development, reconstruct the archaic phonological forms, try to understand Chinese traditional notions, or describe contemporary usages of writing in China, phonetic primacy is verified in every case. In taking into account the indispensable nature of this phonetic primacy, the alternating perspectives of vision and reading are thus dealt with in a thorough manner.

The body of the volume kicks off with Redouane Djamouri's "The development of the writing system in Early China: between phonographic necessity and semiographic efficiency". This is a short and effective introduction to sinography going all the way back to its origins.

A week ago, I quoted Steve Farmer's law:

No ancient literate civilizations are known - not even those that wrote extensively on perishable materials - that did not also leave long texts behind on durable materials.

This applies to 商 Shang sinography from the 13th-11th c. BC. According to Djamouri (emphasis mine),

Today we have at our disposal more than 150,000 inscribed fragments of tortoise shells and ox scapulae. The number of written signs carved on each fragment varies from one to more than two hundred signs and the number of distinguishable characters throughout the whole corpus is about 4,500 to 5,000. (p. 8)

Compare that high figure of 200+ with the figures of 17 or 26 for the longest example of the Indus script.

These early sinographs are not to be confused with the

geometrical or broadly figurative marks ... dating from the seventh millennium BC until the Shang period ... found on pottery, stones, bones and even tortoise shells ... inscribed in isolation or without any discernible interdependency between them. They cannot, as such, be considered as graphic signs, nor can they be considered as the forerunners of the Chinese written signs attested in the Shang inscriptions.

Many speculations have been made on the eventual affiliation between these marks (or some of them) with the Early Chinese writing signs used in the Shang inscriptions. But none of them seems really convincing for they consist of purely analogical links based on formal [i.e., visual] similarities with no possible functional interpretation based on linguistic criteria. (pp. 8-9)

Sinographs have been traditionally classified into six categories, the 六書 liushu 'six writings'. Djamouri describes five of them, ignoring the problematic 轉注 zhuanzhu 'turn-pour' category of extended uses of characters, and adds a nontraditional sixth category. All seven categories already existed during the Shang period:

The first is 象形 xiangxing 'imitating the external form': drawings of what the sinograph represents: e.g.,

'person' (a two-legged stick figure).

Compare Tangut

TT3344 PERSON (Gong dzywo 2.44)

which looks like a more elaborate version of 人. But the more frequent tangraph for 'person' is

TT4105 PERSON (Gong dzywo 2.44)

which is not a xiangxing graph. Its right side is a drawing of a person, but what about its left-hand and center elements? The Tangraphic Sea analyzes TT4105 PERSON as

right of TT3769 HEART (Gong nyiiy 2.39) +

all of TT3344 PERSON (Gong dzywo 2.44)

and HEART does not appear to be a drawing of a heart. If anything, it is like a partial mirror image of TT4105 PERSON (and it too contains TT3344 PERSON, albeit on the left instead of on the right).

TT3344 is unusual in tangraphy because no other tangraphs belong to the xiangxing category. Grinstead (1972: 53) goes so far as to say tangraphy has "no pictures at all". (However, some tangraph components are pictorial in origin.)

Compare Djamouri's three other examples of xiangxing sinographs with their nonpictorial tangraphic counterparts:

'child' (a stick figure) :

TT1131 CHILD (Gong gyi 2.10, Tib. transcription Hgyi [Nevsky 1960 II.504])

'eye' (turned 90 degrees in a modern font) :

TT1850 EYE (Gong mey 1.33)

'horse' (灬 are its four legs) :

TT5233 HORSE (Gong ryiry 1.74)

whose partial similarity to Chn 馬 'horse' may not be coincidental. Even if the left side of TT5233 is derived from the Chinese xiangxing graph 馬 'horse', TT5233 is still not a xiangxing graph since its right side is presumably not a drawing of a part of a horse.

Next: Category II: Indicating an event. WHAT'S IN IT?

Writing about 之 Old Chinese 'it' made me wonder about a question I've left unanswered for years: why is it that the word for 'in it' is

焉 < 於 + ?

Old Chinese 'an < Old Chinese 'a 'in' + ?

rather than something like 'at < 'a + tə?

(Note that there is no graphic relationship whatsoever between 焉 'in it', 於 'in', and 之 'it'. If Chinese writing were 'ideographic', one would expect 'in it' to be a combination of the characters for 'in' and 'it'. But 焉 OC 'an 'in it' is a drawing of a bird called an 'an, and 於 OC 'a is a drawing of a crow [OC 'a, also written 烏].)

Here are two more words like 'in it':

然 < 如 + ?

OC nan 'be thus' < OC na 'like' + ?

(instead of expected nat; 之 OC can also mean 'this')

云 < 曰 + ?

OC wən 'say it' < OC wat 'say' + ?

(note the vowel change in addition to -n instead of -t)

Why do these words end in -n instead of -t (< 'it; this')? Today I wondered if the -n was a remnant of a pre-OC case form like tən preserved in reduced forms (-ən, -n) in contractions (cf. the initial consonant loss in 'em < them):

'a 'in' + -ən 'it' = 'an 'to it'

na 'like' + -ən 'this' = nan 'be thus'

wat 'say' + -ən 'it' = wən 'say it'

Neither 'it' nor this hypothetical tən look anything like 厥 OC kot (third person possessive) or 其 OC (third person possessive; postdates kot). Maybe there was a pre-OC third person paradigm like




? (OC does not have a 3rd person subj. pron.; if pre-OC had one, maybe it was kot-like in pre-OC)




kotə >

Oblique (required by convention after 'in', 'like', 'say'?)

kotən > tən > -ən

I have no explanation for 其 OC gə. I'd like to regard it as an unstressed form of 厥 OC kot, but there is no reason for its initial to shift from k- to g-.

Members of pronominal paradigms don't have to sound alike: e.g., in English:

1st sg.: I (vowel-initial) <> me, my (m-initial)

1st pl.: we (w-initial) <> us, our (vowel-initial)

This split between vowel and non-vowel initial forms is also found in the Sanskrit cognates of those pronouns:

1st sg.: aham 'I' <> maam 'me', mama 'my'

1st pl: vayam 'we' <> asmaan 'us', asmaakam 'our'

The vowel-initial 1st pl. forms are from Proto-Indo-European n- forms: e.g.,

PIE ns > Proto-Germanic uns > Eng us

(the -n- is still in Ger uns 'us', Dutch ons 'us'; initial n- is in Latin nos and in Romance: e.g., Fr nous)

PIE nos > Skt nas (short equivalent of asmaan 'us', asmabhyam 'to us', asmaakam 'our')

As far as I know, no one has tried to claim that all the 1st sg. and/or 1st pl. IE pronouns share a single root. Similarly, there may be no need to assume that OC kot (3rd p. poss.) and OC tə(n) (3rd p. obj.) shared a single root: maybe pre-OC had a mixed k/t paradigm just as IE has a mixed vowel/m paradigm for 1st sg.

How do such mixed, suppletive paradigms arise? Wouldn't such paradigms (e.g., go / went) be harder to learn than more regular paradigms (e.g., a hypothetical go / goed)? Yes, though the high frequency of suppletive items makes them a bit easier to learn than they otherwise might be. Anglophone children hear go and went so many times that they can't help but memorize the pair, whereas they (or anyone) would have a much harder time learning a suppletive paradigm for a rare word. That still doesn't answer the question of why the complication of suppletion arises in the first place.

One might guess that language contact might lead to suppletive paradigms: e.g., people would create a new paradigm out of bits and pieces from different languages. But I can't think of any attested example of such a 'Frankensteinian' paradigm. There is no evidence that leads us (or at least me) to think that the IE first person pronouns are a mixture of forms from two or more pre-IE languages. Go and went are both equally English (and IE) in origin; neither is a Norse or Old French import. And even if the mixture hypothesis were correct, that still begs the question: why mix at all? WHITE LADLE (PART 2)

Why is the Mandarin suffix -de (indicating possession, among other things) written as

a combination of

白 + 勺

bai 'white' + shao 'ladle'?

In Part 1, I mentioned a variant

with 日 ri (< Old Chinese nit) 'sun' instead of 白 bai (< OC Nprak) 'white'. Both were semantic elements in 的/旳 OC tlewk (< t-lew-k) 'bright, brilliant', a member of a word-family sharing a core root l-w 'bright'*. 勺 OC Ntlawk 'ladle', OC tlawk 'to ladle' was a phonetic element.

的 OC tlewk also meant 'mark in a target' (because such marks were bright points?; or was this an unrelated homophone?)

In Mandarin, 的 'target' is pronounced di and is used to write the unrelated possessive suffix -de (pron. [tə]) which has survived phonetically unchanged from Old Chinese tə. (The scope of functions of [tə] have, however, changed between OC and Mandarin.)

One may wonder how the possessive suffix was written in OC: it was a drawing of a foot (之, a variant of 止 OC tə' 'stop', another drawing of a foot). 之 stood for two or more unrelated homophones:

1. OC 'go' (hence the drawing of a foot)

2. OC 'this'

3. OC (third person object pronoun)

4. OC (genitive)

3 and 4 could be extended uses of 2, but I doubt there is any connection between 1 and 2-4.

In Mandarin, 之 is pronounced zhi (the expected descendant of OC tə), not de. Why? My guess is that OC underwent normal phonetic changes when stressed: OC > Middle Chinese chï > Md zhi, whereas unstressed OC remained intact as Md de [tə]. Since Md -de [tə] no longer sounded like 之 Md zhi (< OC tə), it was written with a different, similar-sounding character 的 for an unrelated word Md di 'target'. -de also has other rebus spellings to graphically distinguish its different uses:

得 (orig. for de 'get') as a verbal/adjectival suffix preceding a complement

地 (orig. for di 'earth') as an adverbial suffix '-ly'

底 (orig for di 'bottom') as a noun suffix (see Xinhua zidian 1971 ed., p. 78) or an adjectival suffix (see Lin Yutang). (Obsolete?)

*Here is a maximal list of Old Chinese l-w 'bright' words. Most are written with the semantic elements 'fire', 'sun', and 'white'; others are written with homophonous sinographs containing irrelevant semantic elements: e.g., OC rlewk 'brilliant' was written with 濯 (semantic element 氵 'water'), originally for the unrelated word OC rlewk 'wash' (< 'make bright' if one wanted to force a connection?).

Glosses are from Karlgren 1957: 287-?; only relevant glosses are listed.

灼 OC tlewk 'burn; brilliant; illuminate; brightly, clearly'

爚 OC lawk 'shine'

濯 OC rlewk 'brilliant'

曜 OC lews 'shine, brightness'

燿 OC lews 'shine, gleam'

耀 OC lews 'shine'

爍 OC ngəhlawk 'shine'

The following words may be reconstructed without l and may not actually belong to this family.

鶴 OC Nklawk 'glistening white' (extended use of Nklawk 'crane', or was it the other way around?)

(for 'bright' > 'white'; cf. Proto-Indo-European leuk 'light, brightness', source of Greek leukos 'light, bright; white')

爝 OC Nstlawk 'torch'

燋 OC rstlawk 'torch'

(日+高) OC klaw' 'white, brilliant'

縞 OC klaw' 'white silk' (extended use of above word)

熇 OC 'hlawk, 'hluk (< '-s-l-w-k?) 'blaze; flame'

翯 OC 'hrluk (< '-s-r-l-w-k?), Nklawk 'rich white colors (of birds)'

昭 OC tlaw 'bright; illustrious; enlighten', tlaw' 'glorious'

照 OC tlaws 'shine, shine on, enlighten'

炤 OC tlaws 'shine on; brilliant, manifest, visible'

尞 OC Cəlews 'burnt-offering'

燎 OC Cəlew, Cəlews, Cəlews 'burnt offering; torch; flame, burn; brilliant'

(C = unknown consonant)

皦 OC klew' 'bright; distinct'

髐 OC khlew 'bleached, white (bones)' 

曉 OC 'hlew' 'make clear, understand' (< '-s-l-?)

燒 OC hlew 'burn' (< s-l-?; 'make bright'?)

晈 OC klew' 'bright'

皎 OC klew' 'bright'

These words are both semantically and phonologically questionable:

鑠 OC ngəhlawk 'beautiful, fine' (< 'bright'?)

紹 OC Ntlaw 'beautiful' (< 'bright'?)

(沃 atop 金) OC 'lawk 'silver, silvered' (< 'bright'?; cf. the 'white' words above)

朝 OC rtlaw 'morning' (< 'bright time'?)

僚 OC Cəlew, Cəlew' 'fine, good, lovely' (< 'bright'?)

EG Pulleyblank has proposed a genetic relationship between Sino-Tibetan and Indo-European. Although I do not agree with this hypothesis, this root l-w (sometimes with -e-, -k: lewk) does resemble Proto-Indo-European leuk 'light, brightness', the source of light and many other words.

There is a lot of morphology in the above words that is beyond my understanding. I have no idea what the various prefixes and suffixes are doing, or what the significance of the e ~ a ~ zero vowel alternation is. (I also suspect that there were alternations in vowel length, since I think that Old Chinese short vowels may have been one source of the 'rising' tones of Middle Chinese [and may have occurred in OC precursors of MC 'departing' tone syllables as well].)