Nouns exhibiting vowel fronting

Question

As touched upon in another thread, there are several nouns that exhibit a kind of vowel shift in older forms, where the ending vowel is fronted when the noun is used on its own to become /i/ or /e/, compared to unfronted vowel forms /u/ or /o/ or /a/ when the noun is used in a compound. Examples include:

• 天: あま vs. あめ (also for 雨)
  • 天照{あまてらす}
  • 雨合羽{あまがっぱ}
• 上: うわ vs. うえ
  • 上着{うわぎ}
• 金: かな vs. かね
  • 金槌{かなづち}
  • 金屋{かなや}
• 神: かむ vs. かみ
  • 神上{かむあ}がる
  • 神所{かむどころ}
• 黄: く vs. き
  • 黄金{くがね}
• 口: くつ vs. くち
  • 轡{くつわ} from 口{くつ} + 輪{わ}
• 木: こ vs. き
  • 木漏{こも}れ日{び}
• 声: こわ vs. こえ
  • 声色{こわいろ}
• 手: た vs. て
  • 袂{たもと} from 手{た} + 本{もと}
  • 手折{たお}る
• 月: つく vs. つき
  • 月読{つくよみ}
• 火: ほ vs. ひ
  • 炎{ほのお} from 火{ほ}の穂{ほ}
  • 火中{ほなか}
• 目: ま vs. め
  • 瞬{まばた}き from 目{ま} + 叩{はた}く
  • 瞼{まぶた} from 目{ま} + 蓋{ふた}

Now for the questions.

1. Does anyone have a list of all nouns known to exhibit this kind of vowel shift?
   • Did all nouns in ancient Japanese or proto-Japanese exhibit this vowel shift?
   • If it were only some nouns, were these nouns categorizable as any clear class of nouns? As an example of a noun class, there are inseparable nouns in Polynesian languages, which generally cover things like body parts and spiritually important terms, much like many of the Japanese vowel-shift nouns that I'm aware of.
2. Is there any clear membership in 甲類 or 乙類 for these nouns, as compared to similar nouns that don't exhibit any vowel shift? One example is 上 kami₁ with the 甲類 み and that doesn't have any kamu form, vs. 神 kami₂ with the 乙類 み and that does have a kamu form.
3. What research is there into this phenomenon? Are there any specific titles or authors that cover this?
   • One theory I've read about (possibly in Shibatani's The Languages of Japan, which I've since misplaced) is that these nouns, when used in standalone contexts, were appended with the now-obsolete Old Japanese い, an emphatic nominalizing particle. Over time, this fused with the preceding vowel to produce vowel fronting. As evidence for this, the term カムイ appears in Ainu as a possible borrowing from pre-Old Japanese, before any such sound fusion, clearly manifesting a distinct む and a distinct い sound.
   • Another theory that I've only come up with on my own is that this might be somehow related to verb conjugations, where the 連用形{れんようけい} always ends in either /i/ or /e/. Verb stems, when used as nouns, always use the 連用形, at least in modern Japanese. Perhaps this is a reflection of some phonetic constraint or requirement in an ancient stage of the language, that is also reflected in these standalone noun forms?

Answer 1

Straight from jawiki on カム/神:

"カムヤマトイワレヒコ、カムアタツヒメなどの複合語で「神」が「カム」となっていることから、「神」は古くは「カム」かそれに近い音だったことが推定される。大野晋や森重敏などは、ï の古い形として *ui と *oi を推定しており、これによれば kamï は古くは *kamui となる。"

Answer 2

大和言葉（和語）の複合語造語法に則ります。音韻交替です。

The following changes may happen when forming compound words:

'e' turns into 'a' （雨宿り・金沢・船旅）
'i' turns into 'u' （くつわ・かむなぎ）
'i' turns into 'o' （木陰、木霊、最寄り）
'o' turns into 'a' （白髪・たなごころ）

If the first syllable of the second word begins with an unvoiced consonant, it often becomes voiced. 連濁です。

'ts' turns into 'tz'（横綱・手綱）
'f' turns into 'b' （まぶた）
't' turns into 'd' （酒樽）etc.

This is like the English...

"tooth" -> "teeth"
"foot" -> "feet"
"goose" -> "geese" 
"cow" -> "kye" would be an older English example)

...except that it is used in compound word formation rather than pluralization.

These words that undergo this vowel-shift are very often compound nouns of Japanese origin （複合和語名詞）. Although this also happens in the 動詞 example of 「くる」⇒「こない」.

The class that these words belong to could be called 複合和語, though I don't believe that term is in general use.