Jörg Rhiemeier, On 06/04/2011 16:37:
> Hallo!
> On Tue, 5 Apr 2011 22:54:32 -0400, Jim Henry wrote:
>> On Tue, Apr 5, 2011 at 10:02 PM, And Rosta<[log in to unmask]>  wrote:
>>> Jim Henry, On 04/04/2011 21:20:
>>>> In general, for a conlang with one or more speakers to be interesting
>>>> for academic study, I would think you'd need a new descriptive
>>>> grammar, lexicon, etc. written *after* you have one or more fluent
>>> On the whole, yes -- if there's stuff in systematic usage of invented lgs
>>> that doesn't occur in natlangs, then that'd be of interest. But some
>>> linguistic theorizing considers matters of pure design -- of possible
> design
>>> solutions to the problem of how to achieve the functions lg must achieve
> --
>>> and to this theorizing, lg designs ought to be relevant, not usage, except
>>> maybe to the extent that usage proves human usability.
>> That last clause, "except....usability", seems to almost negate the
>> rest of the sentence -- if a language design isn't eventually proven
>> to be human-usable, by being learned and spoken by one or more (or
>> more rigorously, two or more) speakers, then it's not obvious to me
>> that any non-conlanger linguists would see such designs as being
>> relevant to their concerns.
> Nor do I see that, either.  It is not very difficult to draw up a
> weird conlang that does things deliberately differently than any
> human language.  That tells us nothing about the human language
> faculty, though.

No, but how is this observation relevant?

>> Some linguists may be more interested in *langue* than in *parole*,
>> but with natlangs, we only know the existence and nature of *langue*
>> by studying *parole*.
> Yep.  Grammars of natural languages are based on observations of the
> actual usage of the language (often on texts that are considered
> "authoritative" or "classical", but that is still "parole").

We clearly live in very different universes, you and me. I don't deny that there is a flourishing, and extremely tedious, body of work describing parole, replete with mind-numbing tables of frequencies and so forth. But there is a vastly larger, and vastly untedious, body of work that describes langue, not parole.

> there is a language, then there are descriptive or prescriptive
> grammars.  With conlangs, it is reversed: first there is a grammar,
> and then, if at all, there is a language.

This reads to me as mere argument-by-definition: you wish to define language as parole. (Admittedly, I wish to define language as langue.)

>>> In practice, I think
>>> this is the area where conlanging (more specifically, engelanging) has the
>>> greatest (i.e. least nugatory) scope for contributing to linguistics:
>>> specifically, in what ways, if any, is it possible to functionally improve
>>> on natlangs' design? To what extent are natural languages perfect?
>> Is a theoretical improvement in design interesting (for these
>> purposes) if it turns out that nobody is able to learn to speak said
>> engelang fluently?
> Well, it would at least indicate that that design improvement is
> impractical, and goes outside what the human language faculty can
> cope with.  We would discover something about the human language
> faculty, namely that there is something particular that it *cannot*
> do.

Aha! A point we agree on!