Re: Merian H-4: Grammar and Phonology.

En réponse à "Thomas R. Wier" <trwier@...>:

>
> Unfortunately, the Classical Arabic system is beside the point.
> The job of a typologist is to infer from many different points of
> data from various sources (geographical, familial, etc.) how
> languages across the world tend to behave.  Classical Arabic
> is one data point, which is fine, but IIRC statisticians believe
> that you need to have at least *30* data points to form a normal
> curve, and typically linguistic typologists look at far many more
> data points than that (usually over 150, depending on the amount
> of data that has been published).
>
Yeah, but then compare what's comparable. We were talking about a language,
Merian, which has two series of stops, voiceless vs. voiced (you can see that
because it has /t/ vs. /d/ and /k/ vs. /g/). And even if in general voiceless
stops are more frequent than voiced stops, in the particular case of languages
which oppose a voiceless series and a voiced series, when they have a gap in
the bilabial position, it's always the voiceless /p/ which lacks, never the /b/
(actually, there is no known example of a language, which has voiceless vs.
voiced stops, which has /p/ and lacks /b/, while there is at least one - and I
believe more - language which has /b/ but lacks /p/, in a system where it
opposes voiceless and voiced stops. In fact, in PIE there is a reconstructed
*p, but no *b - except in one dubious root -, and this fact alone was
considered unnatural enough to bring the whole glottalic theory - because in an
ejective series, it's normal to lack the bilabial articulation. The presence of
reconstructed voiced aspirates without voiceless aspirates was not a good
reason enough for this theory, since in this case the theory doesn't explain
well the evolution of those phonemes).

>
> But I wasn't talking about Classical Arabic's obstruent
> system, but rather obstruent systems in general.  There is
> a robust amount of evidence to support my claim.  I suggest you
> look through the UCLA Phonological Segment Inventory database
> if you believe that I am in error. (I do not have ready access
> to this right now; otherwise, I would quote it.)
>
But your evidence proves nothing. Of course voiceless stops are more frequent
than voiced stops in general, that's to say when you take all languages of the
world. But typology shows only trends, and when you take a particular group of
languages (whether affiliated or not, but sharing the same features), those
trends may very well be reversed. I was reacting to your proposition that if
Merian had to have only one bilabial stop, it had to be /p/, because voiceless
stops are more frequent than voiced stops, and thus to make the language
naturalistic this had to be. Unfortunately, your proposal, though based on
typological reasons, is flawed, since it led to a system (a voiceless series of
stops with a /p/ vs. a voiced series of stops lacking a /b/) which is not
attested anywhere, while the system I proposed (a voiceless series of stops
without /p/ vs. a voiced series of stops with /b/) is attested at least once
(and I'm sure more), and thus, for the little we can know, more naturalistic
than the one you propose (the fact that it's not attested anywhere doesn't mean
it's not humanly possible, but it does indicate that it's at least less
frequent than what I proposed). There may be only one example, but that's
compared to no example at all of the system you proposed.

>
> No, I would deny that, because one data point is not sufficient
> to prove typological patterns -- in fact, is completely incompatible
> with the notion of "typology", since typology ipso nomine implies
> more than one point of data. Again, I would refer you to the UCLA
> Phonological Segment inventory database, which says that front
> rounded vowels and back unrounded vowels are statistically speaking
> about equally common in languages of the world.
Again, you're absolutely not contradicting me, since you're absolutely missing
the point. Whether back unrounded vowels are more or less frequent than front
rounded ones in general is besides the point. The point was that you cannot
find any language that has front rounded vowels without their unrounded
counterparts (Do you know any language with /y/ but no /i/? Or with /2/ but
no /e/? I don't personnally), while you can very well find languages with back
unrounded vowels lacking their rounded counterparts (Japanese with /M/ but
no /u/ is an example, but I know three other examples, namely Jaqaru - spoken
in Peru - Ñungubuyu adn Alawa - spoken in Australia - which all three have the
following vowel inventory: /a/ /i/ /M/, that's to say no rounded vowel at all!)

>
> > (Japanese is an example with its /M/, but
> > across Asia this is a common phenomenon).
>
> But if we're aiming about the typology of a human language --
> which was our topic of discussion, not an Asian language
Well, by saying this, you seem to imply that Asian languages are not human
languages. I think it's not what you meant :))) .

-- this
> information matters only if you put it into its larger context.
>
But you then may miss the point. Information matters if you put it in a larger
context, but if the context is too large information disappears, completely
diluted. Typology says what features are "average", that's to say most
frequent. But it's not reason enough to dismiss everything that's not average
as non humanly possible language. Especially when doing that you dismiss
perfectly attested systems in favor of systems that have never been attested as
far as we know. You were talking about phoneme frequency in general, forgetting
that everything has to be thought as systems in language. Though voiceless
stops are far more frequent than voiced ones, in systems which oppose them,
lacking /p/ but having /b/ has been attested, while having /p/ but lacking /b/
(which you proposed as more naturalistic) never has. In the same way, even if
in general front rounded vowels are as frequent as back unrounded ones, vowel
systems which have front rounded vowels without their unrounded counterparts
are unattested, while vowel systems which have back unrounded vowels without
their rounded counterparts are. In short, those systems are more frequent than
the former.

In short, you were talking about frequencies of mere phonemes, while it was
more to the point and meaningful to talk about frequencies of phonological
*systems*.

Christophe.

http://rainbow.conlang.free.fr

Take your life as a movie: do not let anybody else play the leading role.

Reply