On Friday night, I looked for every instance of the Old Korean dative-locative particles in the hyangga corpus. I've listed these attestations in chronological order below:

Spelling Poem Author King at time of composition
(前)乃 *(tsyən)-ʌy or *(tsen)-ɯy 願往生歌 廣德 文武王 (Shilla; 661-681)
?衣希 *ɯyhɯy
*hɯy 獻花歌 佚名老人 聖德王 (Shilla; 702-737)
*la 祭亡妹歌 月明師 景德王 (Shilla; 742-765)
也中 *laKɯy, 惡希 *axɯy 讚耆婆郎歌 忠談師
良中 *laKɯy,  羅 *la 禱千手觀音歌 希明
*la 處容歌 處容 憲康王 (Shilla; 876-886)
*ɯy 禮敬諸佛歌 均如大師 (923-973) (Koryo Dynasty;
exact dates of composition unknown)
良衣 *laɯy, 惡中 *akɯy 稱讚如來歌
*la 廣修供養歌
*la 隨喜功德歌
惡之 *akɯy, 阿希 *ahɯy 請轉法輪歌
*aKɯy, (夜)未 *(pam)-ɯy 請佛住世歌
惡中 *akɯy 普皆迴向歌

All of the pre-均如 Kyunyŏ attestations are in the 三國遺事 Samguk yusa that was compiled centuries later. The 均如傳 Kyunyŏjŏn (1075) containing all ten of Kyunyŏ's hyangga appeared over a century after his death. Hence the extant spellings may or may not match the originals.

I have revised my reconstructions of the Old Korean particles. My reasoning for my new reconstructions follows.

(前)乃 *(tsyən)-ʌy: cf. Sino-Korean 前 tsyən 'front', 乃 nʌy. The phonogram 乃 represents the final consonant of 'front' and the locative particle. *ə and *ʌ are in different Middle Korean vowel classes but coexist here, implying that Old Korean did not yet have vowel harmony.

But perhaps 前 'front' was *tsen  with a nonhigh vowel in the same harmonic class as *ʌ.

Sino-Korean readings generally date from the 8th century, so perhaps an earlier Late Old Chinese reading *nəyʔ influenced the choice of 乃. LOC 乃 *nəyʔ was a phonogram for Old Japanese *nə (since there was no LOC *nə) and may be a phonogram for OK *n.ɯy (since there was no LOC *nɨy or *nɯy). Maybe there is no need to reconstruct an *-ʌy locative. All other evidence below points to an *-ɯy locative.

衣希 *ɯyhɯy: *ɯy 'genitive' + *hɯy 'locative'? Kim Wan-jin (1980: 116) rejects the locative interpretation of these graphs.

*ɯy: cf. Sino-Korean ɯy, Late Old Chinese *ʔɨəy; is this *ɯy < *ɣɯy < *xɯy < *kɯy, or is it the genitive reused as a locative? If my earlier reconstructions with *-ʌy were correct, this version of the postposition would have been transcribed with a graph read as ʌy in Sino-Korean and *ʔəy in Late Old Chinese: e.g., 愛.

*hɯy: cf. Sino-Korean hɯy, Late Old Chinese *xɨəy. If my earlier reconstructions with *-ʌy were correct, this version of the postposition would have been transcribed with a graph read as hʌy in Sino-Korean and *xəyʔ in Late Old Chinese:: e.g., 海.

*la: alternates with 也 *la; cf. Sino-Korean rang, the use of this graph for Old Japanese ra (an idiosyncratic usage that spread from the Korean peninsula?) and its idu reading a (< *la?) An example of an OK *l that later disappeared? 良 ~ 惡 alternations suggest synchronic allomorphy (*l after some nouns, zero initial after others) or 良 could be an archaism reflecting a lost *l- whereas 惡 could be an innovation reflecting the zero initial after *l- was lost.

*la: alternates with 良 *la; cf. Old Chinese *ljajʔ

*Kɯy: initial consonant could be *k or *h; *-ɯy based on 希, 衣; cf. idu reading aɯy for 中

惡希 *axɯy: cf. Sino-Korean ak, hɯy; is -k + h- an attempt to write *x?; cf. the romanization of Cyrillic х as <kh>; maybe OK h was simply *[x], the voiceless counterpart of *ɣ, and *[x] didn't back to h until was lost by the end of the 16th century. It's also possible that was *[ɦ], the voiced counterpart of h.

Before end of 16th century *x (or *h?; hangul ㅎ) (or *ɦ?; hangul ㅇ)
End of 16th century h zero

The exact place of articulation may not matter because at no point was there a phonemic distinction between velar and glottal fricatives in Old or Middle Korean. By convention, the Middle Korean back fricatives are romanized as h and ɣ (G in Yale romanization).

*la: cf. 良 *la ~ 也 *la; Sino-Korean ra

良衣 *laɯy: could be *laɣɯy? No Sino-Korean readings with ɣ-.

惡中 *akɯy: cf. Sino-Korean ak

惡之 *akɯy: 之 is 'genitive'; cf. Middle Korean ɯy 'genitive', 希, 衣 ending in *-ɯy

I wrote,

The spelling 惡之 <ak.GENITIVE> suggests that OK *akʌy may in turn be a compound of *ak 'locative' (?) plus *ʌy 'genitive'.

Now I reject this analysis in favor of *(l)a + *kɯy. If the morpheme boundary were between *k and (my earlier *ʌ), I would expect to see the monosyllabic *l-variant spelled as *lak, not *la.

阿希 *ahɯy: cf. Sino-Korean a

(夜)未 *(pam)-ɯy: 夜 is a logogram for 'night', a word ending in *-m and presumably cognate to Middle Korean pam 'id.'; 未 Late Old Chinese *mɨyh represents the final *-m of 'night' and the locative suffix *ɯy

One might expect to see different stages of lenition of the medial consonant across the centuries: e.g .,

*akɯy > *axɯy > *aɣɯy > *aɯy

But in fact the spellings, taken at face value, indicate no such sequence. The earliest spelling with a consonant 希 has *h, whereas the latest spellings which might be from the same man mix *k, *h, and even or what may have been zero: 惡中 ~ 惡之, 阿希, 良衣.

Here are two possible explanations:

1. The 惡-spellings are archaic and reflect a pronunciation with *k that was already extinct by the 8th century. The 希-spellings reflect the *h that was current. By Kyunyŏ's time, even *h might have lenited to or zero, resulting in 良衣 *laɯy.

2. The spellings reflect synchronic variation: a spectrum ranging from *akɯy to *aɯy and even just as *ɯy existed as late as Kyunyŏ's time.

My reconstructions conflict with some post-OK idu readings attested in hangul (discrepancies in bold):

良中: ahʌy, ahay, aəy, əɯy as well as aɯy

The -h- is more archaic and dates before Kyunyŏ's time.

The vowel variation reflects different harmonizations of OK *ahɯy:

ahʌy, ahay, aəy: lower second vowel to harmonize with the first

aəy is not harmonic according to Middle Korean rules, but ə is still more like a than *a

əɯy: raise first vowel to harmonize with the second

I think *lakɯy is the most archaic form since

apheresis (*l-loss) is more likely than prothesis (*l-addition)

to dissimilation goes against a trend toward vowel harmony between OK and Middle Korean.

Tangut fonts by Mojikyo.org
Tangut radical and Khitan fonts by Andrew West
Jurchen font by Jason Glavy
All other content copyright © 2002-2012 Amritavision