Among all the oddities that exist between the Okinawan and Japanese data, the diminutive morpheme seems to be one of the most curious. In nearly all cases, the Okinawan morpheme/diminutive っくゎ kkwa, ぐゎー -gwaa and んぐゎ(ー) -ngwa(a) seem to correspond very closely to the Japanese equivalent こ【子】 ko, though there are some minor differences. It seems therefore simple and satisfactory to consider them cognates. But is this really the case? The following aims to investigate the idea and other proposals.
It seems easy to consider the two morphemes as cognates, especially since they are both only one syllable and begin with a velar plosive (k/g). It is likely that the variant -gwaa is the result of simple voicing while -ngwa(a) features both voicing and a reduced genitive particle. This would be directly comparable to the variant ご -go in Japanese, found in some words such as 女子 onago "woman; girl".
Nonetheless, this does not explain the vowel difference between the Okinawan morpheme kkwa and Japanese ko. If the Okinawan morpheme stemmed from the form /ko/, we would expect the modern reflex to be /ku/ due to vowel raising, but this is not the case. However, Thorpe (1983) originally proposed that the Ryukyuan form was suffixed with another diminutive, being *wa. Under this proposal, the form would have changed from /ko/ + /wa/ to /kwaa/.
I reject Thorpe's proposition under two premises. The first being that if such were the case, the form *koo or *kuu would be expected in Okinawan. I base this judgment on the fact that the mysterious morpheme *wa in Okinawan, which commonly changes the ending of a noun and creates doublets in the language (atabichi ~ atabichaa "frog"), may be related to the topic particle, whose usage differs from that of standard Japanese and may have been conflated and, when appended to a word ending in /o/ or /u/, becomes /oo/ or /uu/. In no other Okinawan word ending in -ku or -gu have I been able to find a variation with /kwaa/ and /gwaa/. The change /(k)owa~(k)uwa/ to /(k)oo~(k)uu/ is also supported by some other words, such as Japanese kowasu and Okinawan kuusun "to destroy".
The second is due to the initial preglottalization found in the Okinawan and Amami form kkwa. In Japonic phonology, a geminated or glottalized consonant often signals the reduction of a non-nasal and non-approximant consonant in the word. If the Okinawan form stems from /ko/ + /wa/, it does not explain why the preglottalization is present even when kkwa is used in isolation. This applies similarly to the idea that /kkwa/ stems from the Old Japanese form /kwo/.
Differences in usage
The form kkwa occurs in isolation and at the beginning of some words and can never surface after the reduced genitive morpheme -n. Its variants -gwaa and -ngwa(a), on the other hand, surface at the end of words. Thus, we can obtain kkwa "child", kkwa-muyaa "babysitter", in-gwaa "puppy", and っわぐゎー qwaa-gwaa "piglet". It is not possible to say *gwaa-muyaa "babysitter" or *in-kkwa "puppy". It appears that the form kkwa maintains more directly the semantic meaning of "child", whereas -gwaa functionally acts as a diminutive. The form -ngwa(a) appears to vary.
This differs from Japanese where the morpheme /ko/ can appear both at the beginning and end of a word, and can act as a diminutive in both case. Its variant /go/, however, can only appear at the end of a word and is more limited in usage. Comparatively, though, Japanese tends to prefer prefixing, especially with animals. So, for "puppy" we get 子犬 ko-inu, and for "piglet" we get 子豚 ko-buta.
I feel that this usage difference, along with the phonological discrepancies aforementioned, is pretty significant and merits the consideration of another candidate.
Relation to Japanese こら kora and ころ koro
A candidate that has not yet been put forward would be the Japanese morpheme こら【子等】 kora "child; children" and the diminutive or possible variant ころ koro. I suggest these based on the following:
(1) First, they are semantically equivalent, though kora arguably carries a plural or associative-plural suffix, which is not unlike the Japanese word 子供 ko-domo.
(2) Second, they can be reasoned on a phonological level, as the sequence /kora/~/kura/ sometimes becomes /kkwa/ in Okinawan. Compare Japanese 枕 makura to Okinawan makkwa "pillow". The collapse into one syllable and preglottalization also follows the rule that preglottalization or gemination may only occur in the case of a reduced non-nasal and non-approximant consonant. Further, when voiced, the sequence /gora/~/gura/ tends to become /ngwa/ in Okinawan, as in Japanese 櫓 yagura "(wooden) watchtower; high stage" and Okinawan やんぐゎ yangwa "tree house".
(3) Third, the morpheme koro (kora) and variants thereof is never used as a diminutive in word-initial position, always in word-final position. It is also historically attested in Japanese and remains common in some dialects:
犬児・狗子・狗 "puppy; dog" (also: enoko)
enkoro (Toyama, Yamanashi)
inkoro (Toyama, Chiba)
? ingwaa (Okinawan)
The form くろ kuro might also be a related variant; for example, in Kagawa, the word こどくろ kodokuro "children" corresponds to standard Japanese 子供 kodomo.
Given the information above, I would propose that the morpheme koro/kora is the most likely cognate to Okinawan's morpheme kkwa, as they tie in nicely on a semantic, phonological and syntactic level. I also reject Thorpe's original proposition that kkwa would be the result of /ko/ with a second suffixed diminutive, *wa, as this idea presents too many issues. However, I do believe the ending -aa, found in variation in words like atabichi ~ atabichaa "frog" or mayu ~ mayaa "cat", should be further investigated and its possible ties to the topic marker clarified.
Thorpe, Maner L. Ryūkyūan Language History. Unpublished doctoral dissertation. University of Southern California. May 1983.
As a slight addendum, it seems that the reconstructed form *kora was also proposed by Martine Irma Robbeets (2005) in her book Is Japanese Related to Korean, Tungusic, Mongolic and Turkic? (p. 254), while the form *kura was earlier proposed by Moriyo Shimabukuro (2003) in A reconstruction of the accentual history of the Japanese and Ryukyuan languages by comparison with 枕 makura and Okinawan makkwa "pillow".