A Remarkable Phonetic Correspondence in the Chinese Numerals in Korean and Japanese

Both Korean and Japanese have two sets of numerals (words for numbers) for the numbers from 1 to 10: a native set and a set borrowed from forms of Early Middle Chinese (2nd to 12th century). The native sets are unrelated to each other but those borrowed from Chinese of course are, and the relationship is fairly obvious:

Korean

Japanese

1

il

ichi

2

i

ni

3

sam

san

4

sa

shi

5

o

go

6

[r]yug

roku

7

chil

shichi

8

phal

hachi

9

ku

kyu, ku

10

shib

ju

Especially 2, 3, 4, 5, and 9 are almost or completely the same, when we take into account that initial `n-‘ followed by an i-sound tends to disappear in Korean. Japanese words cannot end in a consonant except `-n’, and if such a word is imported into Japanese it is extended with a vowel, usually -u or -i. So the -u in the word for 6, `roku’, does not belong to the word en can be ignored, which makes it a good match for `[r]yug’.

The words for 1, 7, and 8 look similar too when we remember that initial `p-‘ in Japanese turns into `h-‘. But there is something funny here with the final consonants: where Korean has `-l’ Japanese has `-chi’, or actually `-ch’, since the `-i’ is just stuffing. With three occurrences in such a small sample this can hardly be coincidence. Now this is an unusual correspondence, and requires a sound in the original language that can appear as `l’ in one language and as `ch’ (as in `church’) in another.

When we pull in the numerals from Middle Chinese and from three present-day Chinese languages from the Wikipedia the mystery clears somewhat but new questions pop up:

Korean

Japanese

Mandarin

Cantonese

Meixian

Mid.Chin.

1

il

ichi

yaat

yit

ʔyit

Kor. -l ~ Jap. -chi ~ MC -t

2

i

ni, ji

èr

yee

nyi

nyiy

3

sam

san

sān

saam

sam

sam

4

sa

shi

sei

si

shi

5

o

go

ng

ng

ngu

6

[r]yug

roku

liù

loogh

liuk

lyuwk

r = l

7

chil

shichi

tsaat

tshit

tshit

Kor. -l ~ Jap. -chi ~ MC -t

8

phal

hachi

baat

pat

peat

Kor. -l ~ Jap. -chi ~ MC -t

9

ku

kyu, ku

jiǔ

gau

kiu

kyuw

10

shib

ju

shí

sap

sĕp

dzyip

We see first of all that the Korean and Japanese numerals are very similar to Meixian and Middle Chinese, and less to Cantonese. Mandarin is even more distant, in spite of the fact that Beijing, where Mandarin originates, is much closer geographically to Korea and Japan than the Guangdong region which is where Cantonese and Meixian are spoken now. Apparently Mandarin and Cantonese changed much more over time than Meixian, Korean and Japanese did, at least as far as numerals are concerned.

Next we see that our Korean `-l’ / Japanese `-ch’ correspond to a `-t’ in Middle Chinese, and again the correspondence is systematic. And it is striking that in all three cases a long `aa’ precedes the `-t’ in Cantonese. Still the path from Middle Chinese `-t’ to Korean `-l’ is not obvious.

A fair lot is known about Middle Chinese pronunciation, but most of it concerns initial consonants and vowels; much less is known about final consonants. But the Wikipedia page “Historical Chinese phonology”, section “From Old Chinese to Early Middle Chinese / Medials and finals” mentions that some Old Chinese vowels ended in `r’, and that in Early Middle Chinese this `r’ turned a following `t’ into a retroflex `t’, somewhat like the last consonant in the American pronunciation of `Burt’.

This retroflex `t’ seems a good candidate for the origin of our -l/-ch/-t trio. Its retroflex nature would make it easy for it to turn into a Korean `l’; its sound is less clear and more hissing than a normal `t’, so it could easily turn into -ch[i] rather than into -ts[u] in Japanese; and it can of course easily lose its retroflexness and become a normal `t’ in present-day Meixian and Cantonese. And the disappeared `r’ may help explain the long `aa’ in Cantonese.