As all of the people reading have probably noticed, I tend to write a lot about Indo-European. I try to fit in some of my other interests every now and then, but since I study Indo-European linguistics in university, I'm far more informed on that subject than the others.
But today I will actually write about something completely different, namely Japanese, and its earlier forms. Much like Indo-European Japanese actually has a form of reduplication in the verb. In fact Japanese has multiple forms of reduplications, also in nouns. What is remarkable about these reduplication, is that the reduplicated element will gain voicing. Allow me to give several examples.
First of all the nominal reduplications.
人人 ひとびと hitobito 'people' (note that earlier, /h/ was probably /p/ or /ɸ/).
島島 しまじま simajima ' islands'
所所 ところどころ tokorodokoro ' here and there' (tokoro on itself means 'place')
度度 たびたび tabitabi ' often' (tabi no itself means ' time' as in how many times something happened. Notice that the second element isn't voiced yet, due to there being a voiced element in the word tabi already. 2x voiced in one element is disallowed. Sometimes this even seems to extend to the whole compound).
While it's not uncommon to reduplicate to form plurals, for example see Indonesian, it is rather weird the second element is voiced. Intervocalic voicing might be something you'd have in mind, but the above example already show it's perfectly possible to have intervocalic voiceless consonants.
In fact originally these voiced consonants where prenasalised and due to this prenasalisation the following consonant became voiced. But why would you prenasalise the consonant of a plural formation? It's difficult to make sense out of. This prenasalised element which I'll call *N is there though, and also in a very different kind of reduplication, namely Verbal reduplication.
Verbal reduplication isn't a complete reduplication though, not even a complete reduplication of the stem, it is solely a reduplication of the first syllable.
So a verb like CVC-u would become CVNCVC-u.
One of the more, especially to anime watching crowd, famous reduplicated nouns is:
続く つづく tsuzuku ' to continue' with archaic spelling of 'zu' as ' du'. This verb is used to indicate 'to be continued. This is a reduplication of 付く つく tsuku ' to adjoin, to be attached'.
There's countless of these kinds of verbs for example:
止まる とどまる todomaru ' to remain'
止まる とまる tomaru ' to stop' (note that when written in kanji, there's no difference in how it is written).
Similar to Indo-european reduplication Japanese reduplication seems to give a sense of 'itterativity' or 'perfectivity' . Though in neither Japanese or Indo-European this could be called a fixed semantic meaning of the construction. There's also countless Japanese words that still have their reduplicated counterpart with little to no semantic change, or sometimes only the reduplicated word is found. For example in
縮む ちぢむ chijimu 'to become small' (once again archaic spelling of ' ji' as 'di'), But there is no verb ' chimu'.
But my point of this article besides illustrating this interesting phenomenon is that there is a certain problem with this construction from a historical point of view. Lets take tsuzuku as an example.
In Proto-Japanese the phonemic form would have been */tuNtuku/ which is derrived from */tuku/ (I'm skipping the problem with the vowels 1. because I don't know the exact proto-japanese vowels of these word and 2. because it's not relevant here). What exactly possesed the Proto-Japanese speakers to reduplicated this word while inserting a prenasalisation phoneme? Why not just **/tutuku/ Seems perfectly normal and not ambiguous, the extra */N/ actually seems to be over doing it.
Language generally finds a most ideal solution whenever possible. And even when it would create ambiguities language generally doesn't care. This extra */N/ is not ideal but superfluous, unless it had some kind of function. Maybe the same kind of function as it has in plural-reduplication. It's somehow a marking that says ' Hello! I'm tying two elements together!'. Odd little thing.
I've written about it before, since it also shows up in other compounds, generally tatpuruṣa compounds, display the insertion of the */N/, for those forms it could be explained as a simplification of particles like no and ni (genitive, and locative, respectively).
But in dvandva compounds, this */N/ is absolutely prohibited to appear, resulting in lovely pairs like these:
山川 やまがわ yamagawa 'Mountain stream'
山川 やまかわ yamakawa 'mountains and streams'.
This prohibition in dvandva compounds feels slightly odd though, because, if anything reduplicated plurals are most like dvandva compounds, compounding with themselves.
And then there's the sometimes incredibly stubborn */N/ that doesn't show up when you really do expect him to, like in the word for Tokyo
東京 とうきょう toukyou ' Eastern Captical'. This is a typical tatpuruṣa compound why isn't it **tougyou?
This phenomenon is far from being explained. And sometimes the distribution of prenasalised consonants actually does seem to work much like how Altaic languages distribute their voiced consonants. What happened here? Were there maybe prenasalised, voiced and voiceless consonants where voiced and prenasalised consonants merged historically into prenasalised? Is that even possible?
Many questions, few answers. Throw about your ideas, think about it, or just enjoy this post. :-)