The morpheme -ne (根), and its allomorphic variants -na and -ni, appears in a huge amount of native vocabulary in the Japonic languages and dialects. But even though it's so widespread, it has somehow lost its value and fallen out of productive usage. The real question is, however, what exactly was its purpose? And further to that, what conditions brought about its allomorphic variants?
To answer this, we need to compile some data.
- ine "rice plant" (cf. yone "husked rice") :: i "soft rush (plant)"
- hane "wing, feather" :: ha (ibid.) :: wa (counter for birds and rabbits)
- kitsune "fox" :: kitsu (ibid.) :: kitsu "good fortune"
- hone "bone" :: ?ha "teeth"
- fune "boat" (cf. Kyushu hone) :: ho "sail"
- mine "(mountain) peak"
- mune "breast"
- mune "pillar"
- une "rib"
- kane "money" :: kin "gold"
- sane "core" (in Kyushu, variant of "tane")
- tane "seed, source"
- sune "lower leg"
- one "(mountain) ridge" :: o "(mountain) slope"
- yane "roof" :: ya "house"
- kine "mallet" :: ki "tree"
- hana "flower" :: ha "leaf" :: ho "head of a plant" (? Mandarin: huā, Cantonese fà "flower")
- suna "sand" :: su "sandbank"
- kamina "hermit crab" :: kami "crab"
- kona "flour, powder" :: ko (ibid.)
- shina "item"
- tana "shelf"
- tsuna "rope; hawser" (cf. tsunagu "to tie") :: tsu "harbour, ferry"
- hina "doll" :: hi "princess" :: hi "female servant"
- wana "snare" :: wana "loop" :: wa "circle, loop"
- yuna "woman who works in bathhouse" :: yu "hot water"
- kani "crab" (cf. Kagoshima gane, Koshikijima gaane, and Shuri gaani; Korean ge)
- kuni "country, homeland" :: gun "district, country" :: ku "ward, district"
- tani "valley" :: ta "field"
eni "fate, relationship" :: e (ibid.)
- yani "resin" :: ya "melting"
- wani "shark; crocodile" :: ?ha "teeth"
- uni "sea urchin"
- dani "tick" :: ?dani "horse's load" :: da "packhorse; worthless"
yuni "boil" :: yu "hot water"
The list above is non-exhaustive, contains some conjectures and may feature some terms and relations that do not fit. For instance, we find the terms yuna "bathhouse woman" and yuni "boil", both which relate to yu "hot water" (and by extension, "bathhouse"). The morpheme -na in yuna might be derived by association with the term onna "woman", but it's not clear. And the morpheme -ni in yuni appears to be connected to the verb 煮る niru "to boil", disqualifying it. Other such terms could equally be the product of separate morphemes or derived through nominalization of historically irregular verbs ending in -nu, -nuru, -neru or -niru. One example of this is mane "mimicry" versus maneru "to mimic".
In any case, some terms ending in -ne observe a shift to -na in compounds. For example, ine "rice (plant)" shifts in the term inada "rice field" and inago "rice grasshopper". The morpheme -ne is also sometimes spelled with the Kanji for 根 "root", but it's unclear if a relation exists, or if this is just phonetic spelling. Similarly, the form -ni is sometimes spelled with the Kanji 丹 "red" or 胆 "bile", though these are phonetically spelled. The reason we can also stipulate a related connection with the variant -ni is due to the discrepency between regional forms, as in the word for "crab", where the Southern Japonic data points to the historical form *gane.
So what does it mean?
Unfortunately, the data is very inconclusive. In some cases, there's no real significant shift in meaning. In others, we can note that the shorter form has a larger semantic extension, while the suffixed form seems to be more limited. In this case, it could be possible to consider it some form of diminutive. However, this is not always the case.
Given the fact that almost all the words above are bimoraic (bisyllabic), I'd like to further propose the idea that the suffix is also used to prevent the monomoraicity (monosyllabicity) of these words. Similar phenomenons are observed in other languages, an example being Cree which appends a trailing epethentic vowel when an otherwise monosyllabic word is used on its own without other morphemes.
For words ending in -ni, it's possible they may be derived from the nominal form of the verb 似る niru "to resemble".
Relation to Korean?
Korean has the diminutive suffix 이 -i which can be used after names ending in consonants. In the vocative case (i.e. when calling out to someone), the suffix becomes 아 -ah for names ending in a consonant, or 야 -yah for those ending in a vowel.
The interesting thing about the non-vocative form 이 -i is that it only appears after a consonant. Considering Japanese rules only permit -n in syllable final position, it's quite possible that, assuming any relation, the diminutive was reanalyzed as -ni (here, assuming the origin would have been -i in Japanese, though -e and -a are possible candidates).
And, while the diminutive is only used with proper names in standard Korean, it seems to appear with regular nouns in peripheral dialects. For example, Japanese 蟹 kani (Kyushu gane) corresponds to standard Korean 게 ge (pronounced /ke:/ in the Gangneung dialect ), which in turn corresponds to the Jeju dialectal form 깅이 ging-i. Standard Korean 새 sae "bird" (pronounced /sɛ:/ in Gangneung) also corresponds to the Jeju form 생이 saeng-i. The reason the Jeju data is so important here is that it further reveals a final -ng where standard Korean has none, so a parallel with Japanese is not implausible. Numerous mainland Korean dialects also exhibit a final off-glide [j] at the end of both the words for 'crab' and 'bird', so perhaps standard Korean reduced the ending -ng-i to -i, which may have further coalesced in the word for "crab": *gang-i > *ga-i > ge (or *geng-i > *ge-i > ge).