Or, in Jurchen,

<ša.nggiyan OX DAY> šanggiyan wihan inenggi

1. I started looking at Alexander M. Ščerbak's "Reconstrucing the Manchu-Tungusic Proto-language" (2012) tonight. It lists sample proto-forms divided into six semantic categories. I only have time to discuss the first, "Terms for Day, Night, Month, and Year" from a Jurchen/Manchu (J/M) perspective.

1a. *ineŋī 'day'

J/M and Oroqen are the only languages cited with -ŋg-. J/M have hardened intervocalic *-ŋ- to a prenasalized stop. This seems in line with the fortition of initial *ŋ- in Manchu gala 'hand' from Ščerbak's*ŋāla.

Jin Qizong reads Jurchen


as <nga.la> ngala.

However, the vocabulary of the Bureau of Interpreters has a Chinese transcription *xa la pointing to gala [ʁala] c. 1500.

If early written Jurchen lacked ŋ-fortition, then perhaps


should be transliterated <ngiyan ... DAY> ngiyan ... inengi [ŋʲan ... inəŋi].

1b. *dolbo 'night'

Why did Manchu dobori lose -l- if -l- was retained elsewhere in the same environment: e.g., in M golbon 'clothes rack'?

3.8.1:11: Ming Jurchen dialects may have retained -l-.


in the vocabulary of the Bureau of Translators was transcribed in Ming Mandarin as 多羅斡 *to lo wo which may have represented dolwo < *dolbo.

The corresponding form in the vocabulary of the Bureau of Interpreters was transcribed in Ming Mandarin as 多博力 *to po li which may have represented dolbori.

The Chinese transcriptions exemplify two approaches to a Jurchen coda -l absent in Chinese: insert a vowel after it or just ignore it.

1c. *bēga 'month' ~ 'season'

I'm surprised a *long vowel was reduced to mere palatalization in Jurchen

biya [pʲa] 'moon, month' (unchanged in Manchu except for the obsolescence of the character, of course).

1d. *anŋa 'year'

I suppose the palatality of the nasal of Jurchen

aniya [aɲa] (again, unchanged in Manchu)

was conditioned by *n which blocked *-ŋ-fortition: *-nŋ- > *-ɲŋ-*-jn- (as in Nanai ajŋani '') > *-nj- > [ɲ]. In the same volume, Janhunen (2012: 16) grouped Nanaic together with Jurchenic in Southern Tungusic. Might *-nŋ- > *-ɲŋ-*-jn- be a Southern Tungusic innovation?

Could a similar *-n- to *-j- shift have occurred in J/M 'gold'?

3.8.1:27: The earliest attested form of J/M 'gold' is Jurchen

<GOLD.un> alcun (or ancun?) 'gold' (originally spelled with a single character <GOLD>?)

Perhaps *alcun > *ancun > *aɲcun > *ajcun > *ajsyn > Manchu aisin [ajɕin]. But then why does have Manchu have an unrelated word alcu 'the concave side of a toy made from an animal's ankle bone' with the -lcu sequence that became -isi- in 'gold'? Absolute regularity would demand that alcu is either a loanword or originated from something other than *alcu: i.e., Manchu developed a new -lcu after the old one became -isi. Rozycki (1983: 27) identified alcu as a loanword in Tungusic from Mongolic, so a workaround to explain -lcu in terms of sound laws is unnecessary in that case. But what of, say, Manchu kalcun 'spirit' which has no Mongolic source? Why didn't it become †kaisin?

The fronting of the second vowel is a problem, as Manchu does have words with isu and aisu: e.g., gisun (not †gisin) 'word' and aisuri (not †aisiri) 'a kind of bird'. The 'missing link' form in Alchuka has a second vowel that is neither palatal nor labial: anʃïn.

The palatal c is also a problem, as Turkic and Mongolic have t, and the original vowel of the second syllable was not palatal, so this is not a case of *ti becoming ci (a change which didn't happen in Jurchen and wouldn't happen until Manchu).

In any case, Poppe's (1960: derivation of aisin from *alʲsin < *alʲtin < *altin as reported in Rozycki (1983: 24) doesn't look likely.

I wonder if the word was borrowed independently by Turkic, Mongolic or Tungusic from different varieties of some fourth type of language - perhaps Xiongnu or Rouran.

2. Today I found the Pyu phrase tiṁ priṅ·ḥ kdaṅ· 'LOC city ?' (27.6) which at first glance appears to have double case marking. kdaṅ· ooks like the second half of ṅit·ṁ kdaṅ· 'with, including'. But 'with in the city' makes no sense. Moreover, ṅit·ṁ kdaṅ· precedes nouns: e.g., ṅit·ṁ kdaṅ· saḥ 'with sons' (16.4A). Maybe kdaṅ· does not modify priṅ·ḥ 'city'. Maybe kdaṅ· even has nothing to do with ṅit·ṁ kdaṅ·.

3. I am not sure how to write kdaṅ· in phonological notation. I used to take it at face value as /k.daŋ/ with a period indicating a potential schwa. But lately I think it might be ambiguous.

One possible interpretation of Pyu preinitial-initial sequences (using velar-dental stop-a sequences as examples)

No schwa
Schwa (with lenition of following nonaspirates)
/kəta/ [kəda]
/kəda/ [kəða]

It is unclear if *schwa or some other minimal vowel contrasted with zero after preinitials. The above scenario assumes such a contrast existed but was not indicated in the script (except indirectly if a following consonant was lenited).

kt-type voiceless-voiceless sequences are absent from the 12th century Kubyaukgyi text, suggesting that /kt/ may have merged with /kət/.

The sequence ktha is hypothetical; the only instance of kth- in the entire corpus is kthor·ḥ '?' (27.6).

Aspirates are rare in Pyu. Aspiration after stops may not be phonemic: e.g., kthor·ḥ might be /ktorH/ rather than /ktʰorH/ or /kətʰorH/. (It cannot be /kətorH/ because an intervocalic /t/ would voice to [d], and the word would have been spelled †kdor·ṃḥ [kəðorH].)

/rH/ may have been voiceless [r̥] or [r] preceded by a vowel with phonation and/or a tone. THE DAY OF THE WHITE RAT

Or, in Jurchen,

<ša.nggiyan sin.ge DAY> šanggiyan singge inenggi

1. Japanese 蝦蛄 shako 'Oratosquilla oratoria' is a strange word. It is the only Japanese word I know of with sh- corresponding to standard Mandarin x-. It looks like a recent borrowing from Mandarin 蝦蛄 xiāgū, itself an interesting word for reasons I won't go into here. Yet shako ends in -o like a Sino-Japanese borrowing from Middle Chinese rather than -u, though I doubt Middle Chinese is relevant here. In short, the word seems as if it mixes borrowing patterns:

Middle Chinese
Hypothetical borrowing from Mandarin
Actual borrowing
ga, ka
xiā [ɕja˥]
[ku˥ ]

How was this word borrowed? When was it first attested? I presume it must have displaced a Japanese word since shako live in Japanese waters.

3.7.9:45: I should have read the Japanese Wikipedia article on shako before asking those questions. Going by what it says - I have no other references on hand - it seems the resemblance to Mandarin 蝦蛄 xiāgū is fortuitous.

The Edo period name for shako was shakunage because when boiled, it turned purple like a shakunage flower (Rhododendron subg. Hymenanthes). Shakunage is spelled as 石楠花 <ROCK CAMPHOR FLOWER> or 石南花 <ROCK SOUTH FLOWER>. I suspect that even though the spellings could be taken as meaningful, they are actually phonogram sequences. Shakunage then got shortened to shaku or shako, and the latter was then respelled as Chinese 蝦蛄 'mantis shrimp'.

If 石楠花 ~ 石南花 shakunage and 蝦蛄 shako are actually native Japanese words in sinographic disguise, their sha is in need of explanation since sha is normally only in loanwords. A major exception is 喋る shaberu 'to chat', a modern, common colloquial word whose origin is unknown to me.

A shift of -u to -o is unusual in Japanese. I can't think of any examples. Normally the vowel shift goes the other way around: -o > -u. So I wonder if shako is actually a more conservative form and if the association with shakunage was the product of later confusion. Shaku would then be a clipped form of shakunage or from shako with vowel raising. However, normally o-raising is not in final position, so that might favor the clipping hypothesis. I don't have the dialectological background needed to solve this problem.

3.7.6:41: Wiktionary lists Sino-Japanese Go-on readings ge and ku for 蝦 and 蛄, but my policy is to regard Sino-Japanese readings as hypothetical unless they occur in attested words. So many readings in dictionaries are generated on the basis of fanqie and knowledge of the general patterns of the two major strata of Sino-Japanese, Go-on and Ka-on. I don't know of any words in which 蝦 and 蛄 are read as ge and ku, so I only list ga, ka, and ko here on the basis of 蝦蟇 gama 'toad', 魚蝦 gyoka 'fish and shrimp', and 蟪蛄 keiko 'a kind of cicada'.

2. I didn't know about Eskayan, a constructed language of the Philippines, and its gigantic syllabary.

3. I also didn't know about Gustav Heldt's new translation of the Kojiki.

4. This would upset J. Marshall Unger (via Joanne Jacobs whom I haven't linked to in ages; emphasis mine):

There is no single way a brain becomes “rewired,” explains Wolf, a cognitive neuroscientist and director of UCLA’s Center for Dyslexia, Diverse Learners, and Social Justice. The process happens differently, depending on how we read. Readers of Chinese (an ideographic language) rewire differently from those who read Spanish (a logographic one).

It upsets me for three reasons.

First, languages can't be characterized by their writing systems.

Second, the Chinese script isn't "ideographic"; no writing system is. It isn't really "logographic" either, though that term is less wrong than "ideographic", as there is a partial correlation between Chinese words and Chinese characters.

Third, Spanish orthography is not "logographic"; Spanish is written in an alphabet, not a script with thousands of characters for words or morphemes. Strictly speaking, no writing system is logographic either - there are too many words in any language (Toki Pona aside) for the one-character-per-word principle to be viable.

5. John Candy died 25 years ago today. I didn't know he had Ukrainian ancestry, though I'm not surprised since Canada has "the world's third-largest Ukrainian population behind Ukraine itself and Russia."

Today I learned Canada has its own Ukrainian dialect. I was surprised to see cash register borrowed as a spelling-based кеш реґистер (?) kesh régyster rather than as a pronunciation-based кеш реджистер †kesh rédzhyster.

6. Seeing the word має <maje> 'has' in the Wikipedia article on Canadian Ukrainian made me check to see what other Cyrillic alphabets have є <je>. I forgot about Rusyn! And I didn't know about the letter's various usages over time and in Church Slavonic.

7. Via Viacheslav Zaytsev: Kychanov's (1970) decipherment and translation of "Гимн священным предкам тангутов" (Hymn to the Sacred Ancestors of the Tangut). I had heard of the text but didn't know about this study from almost fifty years ago! THE DAY OF THE YELLOW PIG

Or, in Jurchen,

< so.nggiyan PIG DAY> songgiyan uliyan inenggi

1. Not that it matters much, but when I tried to copy and paste 'pig' from the last day of the pig, I discovered that entry was missing from my index page! I've restored it; it'll eventually disappear after the entry preceding it does. As far as I know, I've never accidentally deleted an entry between entries like that before.

2. I'm stuck in Pirahã (P; getting tired of typing the tilde) mode now. It could be worse. I've only glanced at Daniel L. Everett's Don't Sleep, There Are Snakes: Life and Language in the Amazonian Jungle (2008). I'll read the whole thing eventually - I have two books to finish. I don't want to get too involved with Pirahã. But a glimpse of any language can offer data for future use, and so here I jot my notes on the little I've seen based on the sketch on pp. xi-xii in his book.

2a. Vowels

Here are the allophones:

/i/ [ɪ] ~ [ɛ] ~ [i]
/o/ [u] ~ [o]
/a/ [ɑ]

Here are the allophones in order of apparent frequency:

Phoneme \ frequency
[ɛ] [i]
[u] [o]

And here is a chart of the allophones:

[i] [u]
[ɪ] (gap 2)
(gap 1)
[ɛ] (gap 4)
(gap 3)

/i o/ constitute a class. They are the only vowels with allophonic variation and the only vowels that condition consonantal allophony (see below).

I don't know why o was chosen to symbolize the nonlow back vowel if its most frequent allophone is [u].

I am surprised there are no allophones [e] and [ʊ] (gaps 1-2). Are the P careful to avoid those vowels, or is Everett's description simplified?

The absence of [æ] and [ɔ] (gaps 3-4) may be motivated by the need to preserve 'buffer space' between /i o/ and /a/; such vowels combine characteristics of /i o/ and /a/.

2b. Consonants

Here are the allophones of the two most interesting consonants:

between /i/ and /o/
[ɺ͡ɺ̼] or [g]

/s/ palatalizes to [ʃ] before /i/.

The remaining consonants do not have allophony in Everett's introductory account: /ʔ h k t p/. (I won't go into the issue of whether [k] is really an allophone of [h].)

But later on p. 182, he talks about variation in 'head':

xapapaí ~ kapapaí ~ papapaí ~ xaxaxaí ~ kakakaí

The variation only affects voiceless labial /p/ and back consonants /ʔ h k/, not alveolar /t/ and not voiced /g b/. Vowels and tones (acute = high; unmarked = low) are stable.

The two voiced consonants can be regarded as front and back. (I almost wrote labial and nonlabial, but t'nonlabial' /g/ in fact has a linguolabial allophone [ɺ͡ɺ̼]; the subscript 'seagull' indicates linguolabiality.)

The nonlow vowels condition nonstop allophones of the voiced consonants: the lateral flap [ɺ͡ɺ̼] and the trill [ʙ]. I wanted to say continuant allophones, but Wikipedia says,

Whether laterals, taps/flaps, or trills are continuant is not conclusive.

Are /g b/ the results of a merger of a larger set of earlier voiced consonants?

- Were there originally three voiced consonants */g d b/?

- /g/ could be a merger of */g/ and */d/

- an earlier initial nasal allophone [ŋ] of /g/ could have merged with [n].

- [ɺ͡ɺ̼] could have originally been the /i o/-allophone of */d/

- this lateral flap allophone may in turn be a merger of an original *liquid and a lenited allophone of */d/; cf. how Korean intervocalic /r/ is a blend of the liquids *r and *l and lenited *t

- [g] could have originally been the /i o/-allophone of /g/

- Were nasals */ŋ n m/ originally distinct from stops */g d b/?

Looking at what little remains of P's extinct relatives may help to answer these questions. The initial consonant of Yahahí ~ Jahahí is intriguing, as P has nothing like it (anymore?).

Tangut Yinchuan font copyright © Prof. 景永时 Jing Yongshi
Tangut character image fonts by Mojikyo.org
Tangut radical and Khitan fonts by Andrew West
Jurchen font by Jason Glavy
All other content copyright © 2002-2019 Amritavision