On Fri, Apr 2, 2010 at 5:36 PM, Christopher Doty <
suomichris@gmail.com> wrote:
> So, I kind of distracted from this, but here is what I had come up with.
> Putting the language families first, and having alternate forms for
> different clusters, does work a lot better.
> Also, this is a bit weird, since six of these eight languages are
> Niger-Congo, so also isn't quite a random sample,
not too surprizing.
So, you want to create a prefix or series of prefixes for each of (how
many? more than a hundred?) families?
Is it really worth going to all that trouble? What's the advantage,
over just using bang-, ban-, baur-?
I don't think it would take 100, but I'm not sure of the specific number; in part, this is because our decisions about how to group things could change the number. As I mentioned yesterday, we could split Niger-Congo into Bantu and, basically, "other" Niger-Congo. Likewise with Chinese (which, linguistically, is either a worthless term, or a family and not a language). There are only five or six major ones on the list that I've been looking at, but if we keep to the highest level classifications possible, it shouldn't be too bad.
I think using the language families has any number of advantages. First, I was envisioning this as something that would be both useful to linguists and easily learnable. I think a list of alphabetized language names in Lojban would be really scary, and possibly worthless, if they all started exactly the same (granted, you could sort some other way, but still). An alphabetized list with family names at the beginning, though, would actually group languages together by family. Plus, I really just think learning words which are half the same as any other word for any other language is going to make learning language names REALLY difficult, and thus likely cause these to never be adopted.
> but it does illustrate, as
> do the examples from .xorxes., the idea of different initial elements. So,
> Niger-Congo might be {nirk} when the ISO is V initial, {nirko} when it is VV
> initial, etc....
If you do that, how do you distinguish code "aab" from code "oab"?
"nirko'abu" could be either. (At least if they both happen to be in
the same family.) And why would you want to change the vowel?
Er, right, I think I mistyped. {nirk} would probably be for any vowel-initial root (I haven't tried this out yet, but that is my sense). And I didn't change the vowel, I put it on the wrong line, I think: {nirko'abu} would be, in the system I proposed earlier, for language code abu, not aab. If we take {nirk} for any vowel-initial root, we'd have nirka'abV here.
> aaa -> nirka'a'a' (cf. banga'a'a)
> aab -> nirko'abu (cf. banga'abu)
> aba -> nirkaba or ... (cf. bangabu'a)
I could do "bangaba" too. The reason I prefer "bangabu'a"is that this
way every language name has four syllables. But if giving three
syllables to languages with VCV codes does not violate neutrality,
that can be done.
I don't think length violates neutrality--even if it does, we can't do much about it given the restrictions of Lojban phonotactics.
> abb -> ... bangabubu
> baa -> ... banbu'a'a
> bab -> babnicV or nirbabV (cf. banbu'abu)
"nirbabV" works with a nir- family code, but it wouldn't work with
other family codes.
> bba -> ... banbubu'a
> bbb -> tagbVbVbV (cf. banbububu)
tag- won't work for codes that begin with voiceless consonants.
Then tak- before a voiceless; or, language family names are carefully constructed so as to always end in a sonorant. Or some use their longer for here (tagnukVkVkV, say, for kkk). Or, the language family names are done in such a was to always have something that would make a nasty cluster.