is in Triplicane. Its name is from Sanskrit

paartha 'son of Pṛthaa' (a matronymic with vṛddhi*)

saarathi 'charioteer'

saa- < sa- < *sm- 'with'

rath- 'chariot'

-i (derivational suffix)

i.e., 'one who is with a chariot' - its driver

Can you think of English cognates for *sm- 'with' and rath- 'chariot'? Answers here and here.

(3.29.0:05: German cognates include zusammen 'together' and Rad 'wheel'.)

The Tamil for 'Parthasarathy Temple' is

பார்த்தசாரதி கோயில்

p aa r Ø t ta c aa ra t i / ee k aa y i l Ø

Tamil has no aspirates, so Skt -th- [tʰ] corresponds to -tt- [tt] (why geminate?) and -t- [ð].

Tamil c is [s] in medial position.

The combination of ee and aa around ka signify the completely different vowel oo, so

ee k aa y i l Ø = kooyil

Similar combinations exist in other Indic scripts: e.g., Devanagari को ko [koo] is क ka + ा aa with an े e [ee] atop the ा aa..

*3.29.0:48: In this particular case, a stem with a vṛddhi vowel followed by the derivational suffix -a indicates 'having to do something with ...':

pṛthaa 'Pṛthaa' > paarth-a 'son of Pṛthaa'

(-aa is feminine and -a here is masculine)

-aar- is, strictly speaking, not a vowel. But one can think of the vṛddhi 'vowels' as aa (which has been shortened in later pronunciation except before r) plus a root vowel:

manas- 'mind' > maanas-a (not maaanas-a) 'mental'

mitra 'friend' > *maaitr-a > maitr-a 'friendly'

puruṣa 'person; man'' > paauruṣ-a > pauruṣ-a 'manly'

is a syllabic consonant that behaves like a vowel. It loses its syllabicity if preceded by a vowel: aa + ; = aar, not aaṛ.

If a stem already ends in -a(a), this -a is replaced by the homophonous suffix -a (e.g., in paarth-a, maitr-a and pauruṣ-a above).

The derivational suffix -i is preceded by a stem with a vṛddhi vowel. Hence

saratha 'with-chariot' + -ya = saarath-i

The -a of a stem is lost before -i. The combination of vṛddhi plus -i usually signifies a patronymic, but saarathi is 'charioteer', not 'son of Saratha'. IS THE BACHELORS' PARADISE IN A SACRED LILY TANK?

Page 179 of The Man Who Knew Infinity has a double-layered etymology for the 'bachelors' paradise' of India:

Triplicane, a corruption of [Tamil] Tiru Alli Keni, meaning 'sacred lily tank'.

At first, I was puzzled by the -p-. I would have expected its Anglicized form to be something like Truealleycane or Trolleycane. Then I saw that Wikipedia listed the Tamil form as Tiruvallikeni with -v- which is like a -b- and thus sort of like a -p-. This -v- was presumably

inserted between these vowels [u and a] in order to 'smooth the transition' from one vowel to another.

- Harold F. Schiffman, A Reference Grammar of Spoken Tamil, p. 20

-v- is in the Tamil spelling at Wikipedia:


t i ru va l Ø li ee k ṇ i

I explain the use of Ø here.

ee is written before க ka in Tamil, even though it is pronounced after k. A NEW NAME FOR A NEW VILLAGE

Reading The Man Who Knew Infinity led me to find the Wikipedia article on Ramanujan's native language:

Tamil is also spoken by significant minorities in Malaysia, Mauritius, Vietnam, Réunion as well as emigrant communities around the world.

I had never heard of Tamil speakers in Vietnam. Thinking that they might be French-speaking, I then looked up the former French colony of Pondicherry, only to discover that it was now called Puducherry 'new village:


pu tu ch Ø ee ch ri = puðuchcheeri

Tamil ு u fuses with preceding p and t in writing:

pa + ு u = பு pu

ta + ு u = து tu

If a Tamil character containing inherent a is combined with a vowel symbol, the original a is lost, so pa + u is pu, not pau.

Like Korean ㄷ t, Tamil t becomes [d] between vowels, so putu is [pudu].

(3.28.0:30: According to p. 14 of Harold F. Schiffman's A Reference Grammar of Spoken Tamil, Tamil t becomes [ð] between vowels. I got [d] from this Wikipedia chart which leaves out the fricatives mentioned in the text above it.)

The ch- of cheeri 'village' is doubled before pudu 'new' (because it's preceded by a short vowel? - see section 2.1 of Nagaraja, "Gemination of stops in Tamil").

I explain the use of Ø here.

ee is written before ச cha, even though it is pronounced after it: சே ee ch = chee.

The fusion of ர ra and ி i into ரி ri is straightforward.

How did the French name Pondicherry end up with an -n- and -i- absent from Tamil? REALM OF THE STUPID

or मन्दराज्य mandaraajya is an etymythology for Madras:

manda 'stupid'

raajya 'realm; kingdom'

raaj 'to rule; king' (cognate to regal, reign, and the rea- of realm)

-ya (derivational suffix)n

Another etymythology is Madre de Dios. I found both on p. 82 of The Man Who Knew Infinity. Wikipedia has yet another explanation for Madras.

Madras is now called



The Tamil spelling could be transliterated as

e ch n Ø ai n

The vowel symbols ெ e and ை ai must be written to the left of a consonant (symbolized by a dotted circle) even though they are pronounced after consonants. Hence செ e ch = che and னை ai n = nai.

I use Ø to symbolize the ் puḷḷi (Skt viraama), a superscript dot that indicates the absence of a vowel after a consonant. Hence ன் n Ø = n. If the puḷḷi is removed, an a is pronounced after the consonant: ன na. अनन्त्यज्ञनरः ANANTYAJÑANARAḤ

is my Sanskrit translation of the title of the book I've been reading:

anantya 'infinity'

an- 'not' (cognate to in-, un-)

ant- 'end' (cognate to end)

-ya (derivational suffix)

jña 'know' (cognate to know)

naraḥ 'man' (cognate to andro-)

David Boxenhorn recommended it to me. Although it deals with mathematics rather than linguistics, I can empathize with Ramanujan (p. 58):

"For him, it wasn't what his equation stood for that mattered, but the equation itself, as pattern and form. And his pleasure lay not in finding in it a numerical answer, but from turning it upside down and inside out, seeing in it new possibilities, playing with it as the poet does words and images, the artist color and line, the philosopher ideas."

I am more interested in sound patterns than in the meanings sounds convey. I see languages as phonetic sculptures to be disassembled and reassembled. A dead language like Tangut is like a shattered sculpture waiting centuries to be restored. CLAUSON (1964) AND NISHIDA (1976)

To prove I haven't entirely forgotten about Tangut, I should post any nontrivial Tangut material that I find online from now on.

Last week, I found PDFs of two articles:

Sir Gerard Clauson's (1964) "The Future of Tangut (Hsi-Hsia) Studies" is now more like a window into the past. Tangutology has advanced considerably over the past 45 years. Nonetheless many questions Clauson raised still have no definitive answers.

The one question that really bugs me is: what is the raison d'être of the Mixed Categories volume of the Tangraphic Sea? Why were all tangraphs with voiced affricate initial readings plus a seemingly random mix of other tangraphs placed in a separate volume organized by initial class rather than by rhyme as in the other two volumes? On p. 67, Clauson suggested that MC could have contained tangraphs with initial consonant clusters whereas the other TS volumes (his "Sea of Characters") could have contained tangraphs with single consonant onsets, but I am not aware of any evidence for complex onsets in MC.

Nishida Tatsuo's (1976) "Hsihsia, Tosu and Lolo-Burmese Languages" contains many proposals for cognates and sound correspondences. I'd like to reexamine these proposals using my reconstruction instead of his: e.g., his first correspondence:

Tangut (Nishida) Tangut (this site) Tosu Written Burmese Akha Nyi-Lolo Proto-Tibeto-Burman
-i (R10)
-iɦ (R11)
-ɨi (R10)
-i (R11)
-iɨ (R31)
-i -a -a *-a

(NB: Nishida's R = 'rising tone' whereas my R = 'rhyme': e.g., his L10 = my R10 1.10 and his R28 = my R31 2.28.)

R10 and R11 occur in nearly complementary distribution before different types of initials. There is no reason for Nishida's to occur only after velars and alveolars. I think the distinction between the two rhymes involved different medials (-ɨ- : -ø-?, -ɰ- : -j-?, -ø- : -j-?) conditioned by the preceding onset. Hence I would reconstruct a single *-i as the source of R10 and R11 in pre-Tangut.

The -i- in my -iɨ R31 is also conditioned by the preceding onset and can be deleted in pre-Tangut.

Since I would rather not have earlier *-a split into *-i and *-ɨ at random in Tangut, I wonder if the palatal and nonpalatal rhymes reflect earlier low front and back vowels:

*-æ > *-i > -ɨi R10, -i R11 (depending on initial)

*-ɑ > *-ɨ > -iɨ R31

Preinitials and/or presyllables could also have been conditioning factors: e.g.,

*C1Cxa > Cɨi R10

*C1Cya > Ci R11

*C2Cya > Ciɨ R31


*CiCxa > Cɨi R10

*CiCya > Ci R11

*CɨCya > Ciɨ R31

(The subscripts indicate different unspecified consonants.) WHICH WIFE OF A CHINESE LEADER SPOKE WITH A GEORGIA ACCENT?

I had no idea such a woman existed until Friday.

I did, however, know about the wife of a Chinese leader (a.k.a. Николай Владимирович Елизаров*) who spoke Russian (and presumably also Belarusian). Only one of their children "could converse in Russian with their mother".

*Were these names chosen completely at random? They don't sound anything like his real name in any Chinese language and his father was not a Vladimir. 天理敎 TIANLIJIAO/CHOLLIGYO, 眞光 ZHENGUANG/(CINGWANG?)

I used to be interested in the 'New Religions' of Japan and I wish I knew more about their counterparts in Korea and Taiwan. This list of officially recognized religions in Taiwan includes two Japanese New Religions, Tenrikyo and Mahikari - presumably known as Tianlijiao and Zhenguang (the Mandarin readings of the Chinese characters for their names). I wonder what draws non-Japanese (including Thai and Brazilians [Japanese-Brazilians?]) to Japanese New Religions.

Tenrikyo came to Korea in 1893 and now 'chOlligyo' has 276,000 adherents in South Korea. Here's a photo of chOlligyo headquarters in Seoul. Can one be a patriotic Korean while believing that God took a Japanese woman as his living shrine, that God lives in the Jiba ('Place') in Japan, and that the founder's nonstandard 19th century Japanese dialect is the sacred language?

Mahikari isn't in this list of Japanese New Religions in Korea, but it would presumably be known as Cingwang (the Korean pronunciation of the Chinese characters for its name).

ADDENDUM: Some Japanese New Religions incorporate Judeo-Christian elements and some are allegedly anti-Semitic. A SINO-BURMESE SWADESH LIST

How different is Taiwanese from Mandarin? See for yourself at Wiktionary's "Sino-Tibetan Swadesh lists" which should really be called a Sino-Burmese Swadesh list since Burmese is the only non-Chinese language in it.

Many cognates are obscured by changes: e.g.,

'I': Mandarin wo < *ŋo < *ŋaʔ < *ŋajʔ : Burmese ŋa 'I'

cf. Written Tibetan ŋa, Tangut ŋa

'one': Taiwanese tsit < *?tit < *?tik : Burmese tiʔ < Written Burmese tac < *tik

(The Mandarin cognate is zhi 'single' < Old Chinese *tek)

cf. Written Tibetan gcig < *k-tik, Tangut lew < *Cʌ-tek or *Cʌ-tik

'two': Mandarin er < *r < *ʐi < *ɲih < *ni(t)s : Burmese hniʔ < Written Burmese hnac < ?*s-nik

cf. Written Tibetan gɲis, Tangut nɨɨ < *nɨɨH < ?*-s

'three: Mandarin san < *s(l)əm : Burmese θoʊn < Written Burmese suṃḥ < ?*sum-s

cf. Written Tibetan gsum, Tangut sọ < ?*ssom < ?*C-som

'four': Mandarin si < *sli(t)s : Burmese le < Written Burmese leḥ < Old Burmese liy ?[lɨjh]

cf. Written Tibetan bʑi < *?p-lji (see Guillaume Jacques 2004), Tangut lɨɨʳ < ?*r-lɨɨ

The above list assumes that 'I' and the numerals were not Old Chinese loanwords into Proto-Tibeto-Burman (or if one prefers, early non-Sinitic Sino-Tibetan languages).

Tangut fonts by Mojikyo.org
Tangut radical font by Andrew West
All other content copyright © 2002-2009 Amritavision