12.4.28.23:48: MUL-LING OVER KOREAN L-STEMS

Last night I could have mentioned three more verbs that could be confused with the mut-verbs:

물-다  mul-ta 'bite'

물-다  mul-ta 'pay'

물-다  mūl-ta 'go bad, spoil, turn sour'

(The hangul spelling of all three verbs is identical despite differences in vowel length. Not all speakers distinguish long and short vowels.)

물-어  mur-ŏ 'bite/pay/spoil/ask'

물-어  mūr-ŏ 'spoil'

묻-어  mud-ŏ 'bury'

What would be a good test of confusion between the two? I Googled

사체(死體)-를 물-어

sachhe-rŭl mur-ŏ

'bury a corpse'

with the wrong infinitive and got the legitimate

사체(死體)-를 물-어-뜯-

sachhe-rŭl mur-ŏ-ttŭt-

'bite a corpse' (lit. '... bite-INF-tear apart')

since Google does not take Korean word spacing into account.

I had a hard time with Korean verbs with l/r as a student. Apparently even native Koreans do. Today I found this passage in Martin (1992: 238; I have changed his romanization to match mine and added emphasis and hyphens):

The l-doubling vowel stem has a shape which ends in vowel + rŭ-. When the infinitive (-ŏ/-a) or the past tense (-ŏss-/-ass-) is attached, the vowel ŭ drops, as expected [the vowel sequence ŭ-ŏ is absent from Korean verb conjugation], and the remaining l geminates [= doubles] - as not expected:

purŭ- > pull-ŏ 'calls'

morŭ- > mōll-a 'does not know' (the long ō in the infinitive [of 'does not know'] and forms derived from it is an irregularity).

Many Koreans regularize these verbs by doubling the l everywhere; they treat the stems as pullŭ-, mollŭ-, etc.

This error seems much more common than confusing the conjugation of 'bury' and 'ask':

-ㄴ다 -nda is a present ending for vowel-final stems. The equivalent ending for consonant-final stems is -는다 -nŭnda which was in two examples from last night:

묻-는다  mun-nŭnda 'buries' (present; -t > -n before -n-)

묻-는다  mun-nŭnda 'asks' (present; -t > -n before -n-)

Going back to where I started from, -l stems like 물-  mul- lose their -l before certain endings and become vowel-final stems: e.g.,

문다 mu-nda (not *물-는다 *mul-lŭnda** with -l) 'bites'

문다 mu-nda (not *물-는다 *mul-lŭnda with -l) 'pays'

문다 mū-nda (not *물-는다 *mūl-lŭnda with -l) 'spoils'

*The Korean 'infinitive' is not like a European infinitive. It can even be the sole finite verb in a 半말 panmal 'half-speech' style sentence (23:57: hence Martin's translations with '-s': 'calls'). So I am not comfortable with the term even though some linguists use it. I would rather not argue that Korean verbs have an infinitive and a homophonous finite form.

*Korean n- becomes l- after -l, though this is not reflected in hangul spelling (in < >):

물-는다 <mur-nŭnta> mul-nŭnda > mul-lŭnda

There is only one hangul letter ㄹ <r> for [r] and [l].

12.4.27.23:59: BURYING QUESTIONS, ASKING CORPSES

Two Korean verbs have identical forms in dictionaries:

묻-다  mut-ta 'bury'

묻-다  mut-ta 'ask'

They remain identical as long as consonant-initial suffixes are added: e.g.,

묻-고 mut-ko 'bury, and ...'

묻-고 mut-ko 'ask, and ...'

묻-는다  mun-nŭnda 'bury' (present; -t > -n before -n-)

묻-는다  mun-nŭnda 'ask' (present; -t > -n before -n-)

However, the -t of 'ask' (but not 'bury') becomes -r before vowel-initial suffixes: e.g.,

묻-었다 mud-ŏtta 'buried' (-t- > -d- between vowels)

물-었다 mur-ŏtta 'asked'

묻-으면 mud-ŭmyŏn 'if ... bury'

물-으면 mur-ŭmyŏn 'if ... ask'

(All vowel-initial suffixes are written with the zero consonant letter ㅇ shaped with a zero.)

Given the substantial overlap between the paradigms of 'bury' and 'ask', I wondered if Koreans mix them up. I Googled the following phrases with correct and incorrect conjugation:

Given the extreme rarity of errors, Korean learners have no excuse not to conjugate these two verbs correctly.

However, I was surprised to see 31,300 results for *듣-었다 *tŭd-ŏtta, an error for 들-었다 tŭr-ŏtta 'heard', the past form of 듣-다 tŭt-ta 'hear' which conjugates like 묻-다 mut-ta 'ask'. Why is the wrong form of 'heard' much more common than the wrong form of 'asked'? (I presume 'hear' and 'ask' have similar frequencies as common verbs of speech; one hears and asks more than one buries, so I would expect 'bury' to have a lower frequency.) There is no homophonous verb*듣-다 *tŭt-ta with -d before vowel-initial stems, so 듣-었다 *tŭd-ŏtta 'heard' must be by analogy to 듣-다 tŭt-ta and other -t-final forms of 'hear'. (The shift of t to d between vowels in *tŭd-ŏtta is automatic.)

In theory, Koreans could reanalyze -t stems as -n stems due to forms like 묻-는다  mun-nŭnda, but I can't find any examples of

*사체(死體)-를 문-었다 *sachhe-rŭl mun-ŏtta 'buried a corpse'

*질문(質問)-을 문-었다 *chilmun-ŭl mun-ŏtta 'asked a question'

with -n instead of the correct stem-final consonants. Perhaps this is because Korean verbs with nasal-final stems are rare*; there are not enough of them to serve as analogical models for conjugating the far more common verbs with -t stems.

*4.28.1:12: If Alexander Vovin (2010) is correct, this rarity is the result of a sound change:

- Old *-nt stems became -t stems:

*munt- > mut- 'bury'

There are no longer any -nt stems in Middle Korean or modern Korean.

- Old *-t stems became -r stems before vowels:

*mut- > mur- 'ask' (but still mut- before consonants)

- Old *-n stems almost always became vowel-final stems:

*on- > o- 'come'

but -n- remains in the imperative on-ŏra! 'come!'

but all other imperatives of modern vowel-final stems lack -n-: e.g.,

po-ara! 'see!',  not *pon-ara! 'see!'

Exception 1: an- 'embrace' might be an archaism that retained its -n

Exception 2: shin- 'put on one's feet' might have retained -n to continue to resemble its source, the noun shin 'shoe'.

These are the only two 'pure' modern -n stems. The others listed in Martin (1992: 364) are abbreviations of stems ending in vowels or clusters:

non- ~ nonŭ- 'distribute'

mun- ~ munŭ- 'demolish'

kkŭn- ~ kkŭnh- 'cut'

12.4.26.23:59: PROTHESIS IN PRONOUNS: SWEDISH AND SLAVIC

Here's yet another Wikipedia passage that had me thinking, Is this the real deal?

[Swedish] ni ['you' (pl.)] is derived from an older pronoun I*, "ye", for which verbs were always conjugated with the ending -en. I became ni when this conjugation was dropped; thus the n was moved from the end of the verb to the beginning of the pronoun.

I VERB-en > n-i VERB-?

But Swedish also has a verb-subject construction in which the -en would directly precede I:

VERB-en I > VERB n-i

Was this second order extremely common in older Swedish?

This change has a parallel in Old Norse.

Slavic languages also have a prothetic n- in pronouns. These selected forms give you some idea of the variation within Slavic:

Polish and Russian n-forms are used before prepositions. The locative always has n- since it is always preceded by prepositions. One  might then conclude that the n- is an old final consonant of one or more prepositions that came to be associated with the following pronoun. Yes, I guessed right for once! Bacz (2009: 168) wrote (emphasis mine):

According to the historical grammars of Polish (e.g. Kuraszkiewicz 1972: 130-31), the pronominal third person n'- [i.e., palatalized n-] forms replaced the original suppletive j-forms in the declensional paradigm of the pronouns on, ona, ono 'he, she, it'  when -n, the final consonant of the prototypical Slavic prepositions *vъn (modern w) 'in' and *sъn (modern z) 'with' shifted and mechanically attached itself to the locative and the instrumental j-forms of the following pronouns, respectively. The shift is illustrated by the examples in (4) taken from Doroszewski and Wieczorkiewicz (1972: 92).

(4) Forms before the shift : Forms after the shift

*vъn-jemь-LOC. 'in him' : w nim-LOC.'in him'

*sъn-jimь-INST. 'with him' : z nim-INST. 'with him'

The initial n' of the prepositional pronominal forms in the locative and the instrumental after the prepositions w 'in' and s/z 'with' has, with time, generalized to the other prepositions used with these cases [which did not end in -n] (such as przy nim-L 'next to him', po nim-L 'after him', etc.) and to the other prepositional cases: genitive, dative and accusative (o niego-G 'to him', ku niemu-D 'toward him', przez niego-A 'because of him').

4.27.0:11: Bacz introduced me to Polish preposition-pronoun contractions: e.g., weń < w niego 'in him'. In English, contractions may be considered nonliterary, but in Polish, these contractions are literary, which explains why I haven't seen them in grammars for learners.

*Danish still has I 'you' (pl.) without n-. Norwegian once had I but now has dere which was "only slowly breaking its way into literary language" in the early 20th century (Groth 1914: 71). Note that this capitalized I is a second person plural unlike its first person singular English homograph I. I assume northern Germanic I is cognate to

English ye, you

Dutch jij ~ je (coincidentally a homograph of French je 'I') 'you' (sg.) and jullie 'you' (pl.; < jij + lui 'people')

German ihr 'you' (pl.)

and even Sanskrit yūyam (nom.), yuṣma- (oblique stem) 'you' (pl.)

Icelandic has þið 'you' (pl.; originally dual; cf. Old Norse þit). The old plural is þér (same as Old Norse), presumably cognate to Norwegian dere and homophonous with the dative singular of Icelandic þú 'thou'.

The þ- of the Old Norse pronouns (and by extension Icelandic and the d- of Norwegian dere) is from a verb ending (emphasis mine):

The nominative forms are often suffixed to the verb, e.g. mæli-k 'I speak', má-k-at 'I cannot' (-at 'not', frequent in poetry). Similarly heyrðu and skaltu < skalt þú. Such occurrences with the dual and plural forms of the second person pronoun led to re-analysis on the part of the speakers: skuluð ér > skuluðér was subsequently interpreted as skulu þér. Hence the alternate forms þit and þér [for original dual it and plural ér] and the frequent use of the 3rd person plural [with þ-] in place of the 2nd person.

Old Norse ér 'you' looks like a cognate of German ihr 'you' (pl.).

12.4.25.23:59: DEFINITE ARTICLES IN COLOGNIAN: WHAT'S THÉ DIFFERENCE?

Last Tuesday, I returned to blogging after a long hiatus because I was driven to look into the dubious Wikipedia claim that there were Hmong in the Tangut Empire. And last Saturday I found an error in the description of Gan tones in the English Wikipedia (now corrected!). So as much as I love Wikipedia, I don't believe everything I read on it. There are errors there (and here too - sigh). Tonight I saw these two passages and initially wondered what they were describing (emphasis mine):

English Wikipedia: Colognian has two distinct sets of definite articles indicating focus and uniqueness

German Wikipedia: Die bestimmten Artikel des Kölschen haben jeweils zwei Ausprägungen, eine betonte und eine unbetonte, wovon die betonte Variante mit dem entsprechenden Demonstrativpronomen zusammenfällt. Sie wird vor allem benutzt, wenn auf einen bestimmten unter mehreren möglichen oder einen bereits bekannten Gegenstand Bezug genommen wird. „Es dat et Enkelche?“ Ist das Ihr Enkelchen / eines Ihrer Enkelchen? Aber: „Es et dat Enkelche?“ Ist es dieses Enkelchen?

'The definite article of Colognian has two forms, one stressed and one unstressed; the stressed variant coincides with the corresponding demonstrative. It [the unstressed variant?] is used especially when referring to a certain one among several possiblities or one already known object of reference. „Es dat et Enkelche?“ Is that your grandson/one of your grandsons? But: „Es et dat Enkelche?“ Is it this grandson?

I was confused at first because I took "distinct" to mean 'segmentally distinct' which seemed to clash with stressed vs. 'unstressed'. I was erroneously assuming that a difference in segments could not be accompanied with differences in stress.* Wikipedia didn't seem to be wrong - but I was! D'oh! Then a light clicked on ... the right one, I hope:

Es dat et Enkelche?

Is that the (unstressed) grandson (= your grandson or one of your grandsons)?

Es et dat Enkelche?

Is it the (stressed) grandson?

Et ('the'; neuter nominative singular) looks like it should be a shorter, unstressed derivative of the demonstrative dat (cf. Dutch het 'the' (neuter) and dat 'that') referring to a certain known object of reference (e.g., your grandson). Its stressed counterpart is dat 'the' (cf. standard German das 'the' (neuter) with -s < -t) which is identical to the demonstrative dat.

4.26.1:00: My attempt at a table of Colognian definite articles:

I got dä and de from this standard German-Colognian translator site and from browsing the Colognian Wikipedia (which oddly does not seem to have an article on Colognian). Ah, I went to the site of the Akademie för uns Kölsche Sproch 'Academy for Our Colognian Language' and used its German-Colognian online dictionary to find the other nonneuter pronouns der and die which happen to look like standard German. I was hoping for more exotic forms like d-less ä and e which would be to dä and die what d-less et is to dat.

Note the asymmetry: stressed m. 'the' is like Dutch 'the', but the other stressed 'the' are like German 'the' and Dutch 'that'; conversedly, unstressed m. 'the' is like German 'the', but the other unstressed 'the' are like Dutch 'the'. I would have predicted that German-like der was the stressed form, but it's actually the unstressed form even though it's longer than unstressed dä.

*4.26.1:10: Stressed and unstressed forms can be segmentally quite different: e.g.,

English a can be stressed [ej] or unstressed [ə] (though this is not indicated in spelling).

The dative singular of the Polish second person pronoun (i.e., 'to thee') can be stressed tobie [tɔ́bʲe] or unstressed ci [tɕi] (< earlier ti?).

How could I have forgotten that?

12.4.24.23:54: SOURCE OF THE SUN?

I looked up the tonal term (and 'sun' among other things) 陽 yang (as in yin-yang) in Schuessler's ABC Etymological Dictionary of Old Chinese (1997: 558) which linked it to Siamese ปลั่ง <plaŋ1> 'shiny', following Unger (in Hao-ku 33 (1986), a journal I've never seen. Let's suppose the Siamese word was borrowed from Chinese. Perhaps the earliest reconstructible form of the word in Chinese was

*pɯ-laŋ

I reconstruct a generic high vowel (symbolized as *ɯ) in the presyllabic prefix to condition the partial raising of the root vowel:

*pɯ-laŋ > *pɯ-lɨaŋ > *lɨaŋ > *jɨaŋ > modern Mandarin yang

*l lenited to *j before higher vowels, whereas it became 'emphatic' (pharyngealized; indicated with underlining) before lower vowels and hardened: e.g., in

餳 *laŋ > *laŋ > *daŋ > modern Mandarin tang 'sweets'

written with the same phonetic 昜 *laŋ plus semantic 𩙿'eat'

is 糖 Md tang < *daŋ 'sugar' cognate?

which might have lost a high vowel presyllable still in

𩛿 ~ 餳 *sɯ-laŋ > *sɯ-lɨaŋ > *slɨaŋ  > *zlɨaŋ > *zɨaŋ > modern Mandarin xing 'dried sweets'

Note the variation in phonetics: 易 as well as 昜 with an extra stroke*.

None of the 'sweets' words have early attestations.

Possible cognates:

昌 *kɯ-s-laŋ > *khlaŋ or *kɯ-t-laŋ > tɕhɨaŋ > Md chang 'bright'

*s-laŋ-ʔ > *hlaŋ-ʔ with a glottal stop suffix might be the source of Vietnamese láng 'shiny'

炳/昺/邴 *T-pɯ-laŋ-ʔ > *rplɨaŋʔ > *prɨaŋʔ > *pɨaŋʔ > Md bing 'bright'

Old Chinese is often reconstructed with medial *-r- all over the place. This is odd. Could a lot of these *-r- be from earlier *t- or even *l-preinitials? I use *T- to represent an uncertain coronal preinitial (*t-, *l-, *r-).

*Could a 日 drawing of the sun have been a phonetic *lVK in 易 *lek 'to change' as well as 昜 *laŋ 'south side'?

4.25.0:05?: No, going by Grammata serica recensa, early forms of 日 don't resemble what became 日 in the modern form of 易.

I also thought 日 could be a *lVK phonetic in 昌 (see above), but once again the early forms don't match.

12.4.23.23:59: DARK AND HEAVY YET LIGHT

The Wikipedia article on Cantonese phonology referred to the yin and yang tone categories as "dark" and "light". That choice of terms made me realize that the Chinese "light flat" (yangping) tone corresponded to the 玄 huyền 'dark' tone category in Vietnamese. And just now I noticed that the Chinese "light rising/entering" (yangshang/ru) tones correspond to the nặng 'heavy' tone category in Vietnamese. (But note that the "light" of Chinese tones refers to  brightness, not  weight.)

Vietnamese tonal terminology has no terms corresponding to yin and yang: each category has its own distinctive name containing the tone it represents:

(The table above is only for tones in native Vietnamese words. The relationship between Chinese tones and tones in Chinese borrowings in Vietnamese is very complex and beyond the scope of this entry.)

Speaking of huyền, I have recently been puzzled by the Vietnamese term huyền sử ca which I have seen translated as 'epic'. Contexts indicate that huyền sử without ca means 'myth'. I assume this word is Sino-Vietnamese

玄 huyền 'dark' (and by extension, 'mysterious')

史 sử 'history'

歌 ca 'song'

but have not been able to verify that etymology. If I am correct, its internal structure is Chinese: modifier-modified rather than Vietnamese modified-modifier. I can't find 玄史 in any Chinese dictionary, though Google lists 8,170 results. 玄史歌 has zero results in Google; perhaps it was coined in Vietnam, even if 玄史 was borrowed from Chinese.

Next: Source of the Sun?

12.4.22.23:51: DEPARTING THE DEPARTING TONE: MORE TONES IN GAN DIALECTS

The eight tone categories of Middle Chinese were

The terms are not necessarily descriptive: e.g., there is no guarantee that 'level' category tones were level in all Middle Chinese dialects (languages)? And it is difficult to imagine what contours were associated with the 'departing' and 'entering' tones. It is safest to think of the terms as arbitrary, though they may have been partly meaningful for the (standard?) dialects to which they were first applied. (The yin/yang and upper/lower terminology postdates the four-way categorization devised before the great tone split.)

In 南昌 Nanchang Gan, yangshang syllables developed different tones depending on their initials:

- sonorant initial > shang

e.g., 馬 'horse': Middle Chinese *mæ + yangshang = Nanchang ma + shang

- obstruent initial > yangshang

e.g., 夏 'summer': Middle Chinese *ɣæ + yangshang = Nanchang ha + yangqu

Thus the table above could be rewritten for Nanchang Gan as

But other Gan dialects had different merger patterns.

The Chinese Wikipedia sums up* the different descendants of the qu tones in Gan dialects. I have partly reorganized its categories below:

A. Dialects with only one qu tone category resulting from a yinqu-yangqu merger: 景德鎮 Jingdezhen, etc.

B. Dialects retaining the yinqu and yangqu categories

B1. Dialects with a single yinqu category: Nanchang, etc.

B2. Dialects with a split yinqu category:

The distinctive treatment of tones in syllables with aspirated voiceless initials is reminiscent of Thai and Lao:

B1a. 湖口 Hukou, etc.:

*unaspirated voiceless initial + yinqu > yinqu 1

*aspirated voiceless initial + yinqu > yinqu 2

B1b. 遂川 Suichuan which lacks yangqu:

*unaspirated voiceless initial + yinqu and voiced initial + yangqu > yinqu 1

4.23.1:25: This implies that voiced initials were unaspirated at the time of this change, even though Gan is known for having aspirated reflexes of voiced initials. Does Suichuan have aspirated reflexes of voiced initials: e.g.,

*b- + yangqu > *b- + yinqu 1 > ph- + yinqu 1?

*aspirated voiceless initial + yinqu > yinqu 2

How many other Chinese varieties have this kind of initial-based split of a yin tone?

C. Dialects with only one qu tone category not resulting from a yinqu-yangqu merger:

C1. Yinqu as sole surviving qu tone; yangqu > yinping: 都昌 Duchang, etc.

C2. Yangqu as sole surviving qu tone

C2a. Yinqu > yangping: 新喻 Xinyu, etc.

C2b. Yinqu > shang: 石城 Shicheng (and others?)

C2c. Yinqu split: Tacheng (a subdialect of Nanchang?), etc.

unaspirated initial + yinqu > yangping

aspirated initial + yinqu > shang

D. Dialect(s?) with no qu tone categories:

In 寕岡 Ninggang, yinqu has become yangping and yangqu has become yinping.

*The Chinese Wikipedia's section on Gan tones only distinguishes between three out of the four types of Middle Chinese initials:

清 clear: unaspirated obstruents

次清 second clear: aspirated obstruents

濁 muddy: voiced obstruents

It is not clear from this section alone whether 次濁 second muddy initials (voiced sonorants) conditioned the same tones as voiced obstruents. For simplicity, I will use the yin-yang terminology even though it's possible that yangqu tones in syllables with voiced sonorants may have developed differently from yangqu tones in syllables with voiced obstruents.