Re: Redundancy + Ambiguity = What? (+ another question)

On Tue, Sep 2, 2008 at 8:37 AM, Veoler <veoler@...> wrote:
> Redundancy + Ambiguity = What?
>
> There have been some talking about redundancy, and sometimes about the
> naturalness of ambiguity in a naturalistic conlang. But that got me thinking,
> shouldn't they have "opposite effects", so to say?
They're orthogonal; you might can increase or decrease either without
perhaps directly affecting the other, as long as you allow other
variables to vary in response (e.g., average morpheme length).


> Notice how it have ambiguity, since all words are polysemous. But the
> language does also have redundancy, since it only uses four of nine possible
> syllables. Is this a realistic picture of a natlang?
I would guess that that degree of redundancy is unlikely to
occur in a natlang.

> Imagine now language B, a hypothetical(?) engelang:
> Notice how both the polysemy and the redundancy are removed.
>
> Now, do the lack of redundancy makes B inferior to A, in actual use? They
> both have nine concepts on nine possible words. The possible mishearing in
> language B is compensated with the lack of polysemy.
In this case I think it mostly is, because the alternate words
a word could be turned into by a mishearing are mostly in
other distributional categories from the intended word.

> I want my engelang to have the redundancy of a natural language, but,
> considering the complete lack of polysemy (on morpheme level), do I need it?
To some extent, yes.


> Another question is: do regularity increase the need for redundancy?
I'm not sure.


> I want my engelang to have the same degree of redundancy as the average
> natlang, and be fit for real use, but exactly how much redundancy is that?
> There was talk about not having any minimal pairs in the language a while
> back, but is that degree of redundancy really needed?
I would guess that "no minimal pairs at all" is more redundancy
than most languages need; but "no minimal pairs within the same
distributional category" is perhaps a reasonable design criterion
for an engelang.  "No minimal pairs" makes sense for a language
designed for a noisy environment.

--
Jim Henry
http://www.pobox.com/~jimhenry/