12.7.21.23:59: 羅杰瑞 JERRY NORMAN'S "THREE MIN ETYMOLOGIES" (1984) REVISITED

I wish to honor the late Jerry Norman by posting a few notes about one of his articles which is available for free at Persée.

Norman reconstructed Proto-Min (PM) 'house' as *tšhio^C which he regarded as a cognate of dictionary Middle Chinese (MC) 戌 *ɕuəh (my reconstruction) 'to garrison the frontier' and proposed that 'garrison' became 'house' (p. 179):

This sort of change typically occurs in frontier regions where the first carriers of a new language are military colonists; the first settlements are in fact forts or stockades, manned by soldiers, sometimes accompanied by their families. As the region loses its frontier character and is integrated into the larger linguistic community, it is very easy for these miltiary terms to lose their military sense and take on purely civilian meanings. This is precisely what I think happened in early Fujian [i.e., where Proto-Min developed]. After the defeat of the non-Han [i.e., non-Chinese] Yue tribes in the Han dynasty, the region became open to Chinese settlement. The first settlers were undoubtedly military colonists who dwelt in garrison forts, śju-* [= my *ɕuəh]. Once Fujian lost its frontier aspect and was integrated into the greater Chinese-speaking ecumene, this military term lost its specifically military character and became the ordinary word for 'house'.

Schuessler (2009: 151) listed Cantonese ³³tsʰy^CD (sic)** and Taishan tsʰui²¹ 'house' as additional cognates. One might assume that the shift of 'garrison' to 'house' occurred in both PM and Proto-Yue (PY)***, but as far as I know, the Cantonese word for 'house' is 屋(企) uk (khei), attested in Chalmers (1859: 169) as uuk(-k‘iˊ)****.

There is, however, a Cantonese 處 tsʰy 'place'. Was/is this word used for 'house', and could the PM and PY (or at least Taishan) words for 'house' be from Middle Chinese 處 *tɕɨəh 'place'?

If I am wrong and Norman is right, why do the PM, Cantonese, and Taishan words for 'house' have an affricate absent from dictionary MC 戌 *ɕuəh?

One possibility is that the southern affricates incorporated prefixes that were either lost or never added in the north: e.g.,

*C-š- > PM *tšh-

If there was ever a prefix in the southern version of 戌, it did not result in a PM 'softened' initial *-tšh-.

Another possibility is that Old Chinese (OC) nonemphatic *hn- developed differently in southern Chinese languages. (Underlining indicates emphatics. Vowel height is correlated with emphasis.)

Northern OC *hn- and *hl-

Southern OC *hn- and *hl-

Nasals are like both obstruents and sonorants. In the south, the OC nonemphatic nasal stop *hn- could have initially lost its nasality while remaining a stop (unlike *hl- which was never a stop) before palatalizing into an affricate.

Norman listed four examples of the correspondence PM *tšh- : dictionary MC *ɕ- other than 戌: 深鼠試手. Baxter and Sagart (2011) reconstructed 鼠 and 手 with *hn- in OC. OC 鼠 *hnaʔ < ?*snaʔ 'rat' may be cognate to Tangut

2nɔ̣ 2nẹ < *snroH-sneH 'rat' (a reduplication of a pre-Tangut root *snVH?)

Although they reconstructed 深 with *hl-, I initially didn't see why it couldn't have had *hn-. Schuessler (2009: 366) reconstructed 深 with *hn-, though he also wrote that

The initial consonant in this series [with phonetic 穼] is very uncertain; it could be n or l or something more complex.

On closer examination*****, I think that series is indeed lateral, so 深 does present a problem for my hypothesis.

試 is definitely in a lateral phonetic series containing MC *d- (e.g., 代 MC *dəjh < OC *ləks). Could l ~ n confusion which is common in modern Chinese languages have occurred at an earlier stage: e.g., was 試 *hnəks rather than *hləks in pre-Proto-Min? 

It remains to be seen whether *hn- should be reconstructed in more words that have the PM *tšh- : MC *ɕ- correspondence.

Norman reconstructed two PM words for 'eye':

*mhək (cognate to 'eye' in other Chinese languages)

*m(h)ət ~ *m(h)it (a borrowing from a substratum Austroasiatic language; cf. Vietnamese mắt and Shorto 2006's Proto-Mon-Khmer *mat 'id.')

I wondered if *m(h)it could be from

Scenario 1: 目 pre-PM ?*mik < OC ?*miwk, cognate to Written Tibetan (WT) mig 'eye'

cf. OC *tsit < *tsik 'joint'; the final velar remains in WT tshigs 'joint' and the glide of Tangut 1tseʳw < *rʌ-tsik is a remnant of an earlier *-k

Scenario 1a: 7.22.20:38: PM *mhək could be from pre-PM *s-mik with vowel lowering (triggered by harmony with a presyllabic *ə?) preceding the shift of *-ik to *-it:

Pre-Proto-Min root	√*mik 'eye'
Variation	*s(ə?)-mik	*(s-)mik
Vowel lowering	*s(ə?)-mək
*-ik > *-it		*(s-)mit
*s-reduction	*hmək	*m(h)it
Vowel lowering		*m(h)ət

Scenario 1b: 7.22.20:57: Norman (1984: 183) noted that

This sort of vocalic alternation between forms in *ə and *i is very common in Min.

Perhaps this alternation is not due to sporadic vowel lowering but is due to emphatic/nonemphatic variation: e.g.,

目 'eye'

Proto-Sino-Tibetan	*miwk
Old Chinese (lowering/centralizing of vowel unexplained; no evidence for prefix at this level)	*məwk
Pre-Proto-Min: prefix (added in PPM or retained in PPM from OC and lost elsewhere?)	*sʌ-mək	*sʌ-mək	*(s)mək
Emphatic harmony; emphasis blocks raising of *ə	*sʌ-mək	*sʌ-mək	*(s)mit
Proto-Min: initial simplification: *s(ʌ)m- > mh-	*mhək	*mhət	*m(h)it

來 'come'

Proto-Sino-Tibetan	*rə
Old Chinese (*mʌ-prefix potentially implied by use of 來 as phonetic in  麥 *mʌ-rək 'wheat')	*m-rə
Pre-Proto-Min	*mʌ-rə	*(m)rə
Emphatic harmony; *ə bends in different directions depending on whether the preceding consonant is emphatic or not	*mʌ-rəɨ	*(m)rɨə
Proto-Min: *r- > *l-; *ɨ > *i	*ləi	*li

Scenario 2: pre-PM ?*mik < OC ?*m-liwk, a prefixed cognate of OC 覿 *liwk 'to see' (which Sagart 1999: 154 viewed as a cognate of his OC 目 *mr-liwk 'eye'; he regarded *m- as an "agentive" prefix)

But on second thought, Norman's Austroasiatic derivation is more likely if there are no other cases of PM *-ət ~ *-it corresponding to OC *-iwk.

3. The Min word for 'to run aground'

Norman reconstructed PM *khâi^C 'to run aground' and regarded the Fuzhou reflexes of PM *-âi as the result of dialect mixture:

Is there evidence for similar mixture in other aspects of Fuzhou?

The graph 艐 for 'run aground' had three readings in dictionary Middle Chinese:

*tsoŋ 'to adhere to sand and cease to go'

*kah 'to adhere to sand and cease to go'

*kaih 'to arrive'

Norman pointed out that only the first reading matches its phonetic MC 㚇 *tsoŋ and proposed that

perhaps kai- [= my *kaih] and khâ- [= my *khah] should be considered "kun" readings, that is, reading associated with the character purely on semantic grounds******.

But where did these k-readings and PM *khâi^C 'to run aground' come from? MC 艐 *kah < OC *kals and PM *khâi^C vaguely resemble Khmer

កើល kaəl < *kəəl 'to run aground'

I am hesitate to declare PM *khâi^C and MC *kah to be from Austroasiatic because the word is unattested in Old Khmer and I cannot find any cognates within Mon-Khmer, so it cannot be reconstructed on the Proto-Mon-Khmer level. It could be a Khmer innovation that never existed in what is now southern China. Moreover, the vowels do not match. I would expect an MC cognate of Khmer *kəəl to be *kɨi < *kəi < *kəl.

*7.22.1:12: The hyphen indicates an MC 'departing tone', not that śju- 'garrison' is a prefix.

**7.22.14:26: I would expect Cantonese ³³tsʰy or tsʰy^C with only one kind of tone marking. Tone D is not possible for Cantonese open syllables.

***7.22.16:59: Cantonese and Taishan share no common ancestor below the Proto-Yue level according to JM Campbell's classification as presented on Wikipedia.

****7.22.15:43: I do not know the origin of the second syllable of Cantonese 屋企 uk khei 'house'.  企 means 'to stand on tiptoe' which has nothing to do with 'house', so I think it is purely phonetic in 屋企. Chalmers' romanization k‘iˊ for 企 implies that -i broke to -ei after velars sometime between 1859 and the 20th century. 

*****7.22.16:13: The phonetic 穼 is doubly strange.

First, it and its variant  𥥿 have a dot on top absent from its derivatives 深琛探.

Second, the Guangyun MC dictionary listed two fanqie for it:

所今切 = *ʂim < OC ?*ksləm

杜覽切 = *damʔ < OC *lamʔ

Neither matches the MC reading *ɕim that is my equivalent of Karlgren's (1957: 176) *śi̯əm and Schuessler's *śj̯əm^A (2009: 366).

Although MC initial *d- points to OC *l-, MC initial *ʂ-  is usually from *sr- rather than laterals or nasals. Like Pulleyblank (1991: 67), I think *ks- is a possible OC soruce for MC *ʂ-, so my *ksl- is an attempt to account for an anomalous retroflex initial in a lateral series.

******7.22.22:20: Norman's "kun" is the Japanese term for native Japanese readings of Chinese characters which are usually translation equivalents of their Sino-Japanese readings borrowed from Chinese.

'to adhere to sand and cease to go' and 'to arrive' do not have much semantic overlap. I suspect a two-stage process:

1. MC 㚇 *tsoŋ to adhere to sand and cease to go' used to write unrelated synonym *kah 'id.'

2. MC 㚇 *kah 'to adhere to sand and cease to go' used to write unrelated near-homophone *kaih 'to arrive'

It's remotely possible that the two meanings could be related in one direction or the other:

'to arrive' > 'to arrive badly' > 'to get stuck' > 'to get stuck in sand' = 'to adhere to sand and cease to go'

'to adhere to sand and cease to go' > 'to get stuck' > 'to come to get stuck' > 'to come to a place' = 'to arrive'

*kaih 'to arrive' is phonologically more archaic than *kah 'to adhere to sand and cease to go', as the former preserves the OC *-i lost in the latter. Could this mean that 'to arrive' is the older of the two meanings? If so, then there cannot be any connection with Khmer 'to run aground' unless the semantics went full circle (!):

I doubt that 'to arrive' is the older meaning and that Chinese and Khmer independently (!) changed it to 'run aground':

But then if I still wanted these words to be related, I would have to claim that 'to arrive' has an innovative meaning coupled with archaic *-i-retention:

PM *khâi^C with both the old meaning and the old *-i would then be like a 'missing link' between an OC *kais 'to run aground' (unattested!) and MC *kah 'to adhere to sand and cease to go'. (PM *khâi^C is not literally a link since PM is not ancestral to MC; it is a sister, not a mother.)

It is simpler to regard the Chinese and Khmer words as unrelated soundalikes.