Re: "Theory informs practice" - OK?

On Tue, 11 Nov 2008 14:19:23 -0800, David J. Peterson wrote:
>The problem, when it comes to natural languages, is when
>theory constrains creation unnaturally.  I've seen this happen
>dozens of times.  Take a construction like this:
I initially thought you were trying to take the definition of
"natural(istic) language" wider than it should. I'm sure we can agree that
the folloing is a necessary condition:
 1) The language must be usable by humans
I'd add a 2nd one, however.
 2) It must be possible for the language to come about by linguistic change
because otherwise everything we've created here would automatically fall
under "natural", no matter how engelangy.

Condition 2) has some interesting consequences, when we take it in place of
attestedness. One I can think of is that not every feature is as likely as
others to evolve, and a feature with a truly zero probability seems hard to
imagine (outside of physiological impossibilities in phonetics). So does
this mean naturalness of a grammar is not a boolean, but a probability
variable? Could we say that Ubyx was less naturalistic than Uzbek? How much
approximately?

(It would most likely also do good for linguistics in general to keep this
condition in mind when developing any theory... look at what evolution did
for biology, why shouldn't it do the same for linguistics?)

Then you write this part:

>As long as there's some sort of historical explanation for how
>this happened (and I can think of a couple), there's no reason
>that a human can't use this language, even though the T element
>is buried in an NP.
So it seems we agree, in principle. However... just because we can think of
it, it does not necessarily imply that a diachronic process should be called
natural. For example, labialization of velars, develarization of
labiovelars, dentalization of labiovelars (Greek), velarization of labials
(was it Iroquois?), velarization of dentals (Hawaiian), labialization of
interdentals (some Athabask branches) and linguolabialization of palatalized
labials are all attested processes; but does that mean we could then take a
small step forward and claim a POA chain-shift of p > t > k > p or k > t > p
> k as natural? I'd be raising an eyebrow if someone ask'd if their
diachronics make sense and they had a line like that.

Or worse yet, how about a chainshift to k > l > m > n > o > p > q? I'm sure
you could here, too, argue for the phonetic pathways of each change, or even
how "pwn" represents the first stages of an orthographically-motivated
version of such a change, but seriously now, is it really natural?

But it might be possible to even argue the position that _any_ regular sound
change should be taken as "natural". This would be a convenient point to
draw the line, since here, decidedly unnatural alternativs are not hard to
think of. For the sake of fun: how about a change that dispenses with
synonyms by meshing the phonemes of the words together in some pre-decided
fashion? Or a change that codes every word to binary, then flips every nth
bit if the word has a number of phonemes divisable by n?

Furthermore, when conworlding comes into picture, there's also the option
that linguistic change works inherently differently - Frathful Dispersion,
Basque monks, etc - tho any sensible definition of a "human" to speak the
language probably implies processes such as coining slang terminology to be
possible.

Then again, this could all be a red herring: is the history of a language
still a part of the language, or is the naturalness of a language distinct
from the naturalness of its history?


>This, of course, all has to do with languages that attempt to
>look natural. When it comes engineered languages, visual
>languages, logical languages, artistic non-natural languages,
>philosophical languages, etc., it seems to me that one has to
>create one's own theory.
Agreed again in principle, tho the distinction seems circular when no
reference to real natlangs is made: universal grammar (whatever that turns
out to truly contain, so intentionally no caps) is the sum of rules that
apply to all natural languages, and a language is natural(istic) if it
abides to universal grammar...? Going with a theory is surely useful, but it
need not imply that the outcome isn't natural.


>Linguistics is the study of *natural*
>languages, so when it comes to what one can or cannot create,
>linguistics doesn't have much to say.  Certainly, one can take
>claims that are made and test them, but unless they're particularly
>grandiose claims or foundational assumptions, testing them
>won't be that interesting, it seems, since linguistics attempts to
>explain what is, and not what can be imagined.
>
>-David
I could imagine taking this the other way around: designing a conlang to
conflict with a theory, then seeing if it can be taught to (and persist as
spoken by) linguistically uneducated speakers.

John Vertical

Reply