Re: THEORY: Storage Vs. Computation

At 7:24 pm -0600 14/6/99, Ed Heil wrote:
>Irregularity is a problem when a lot of language learning has to
>happen under less than ideal conditions.
Yep - for fairly obvious reasons.

>But surely it's not true that there is a unidirectional tendency
>towards loss of irregularity?  If that were so, we'd all have
>wondroudly regular languages by now.
Yes, indeed.  It's like all theories that posit some unidirectional
tendency in language - the theory would be fine if it weren't for all the
multitude of natlangs actually spoken!  As homines spaientes have been
using language for many millennia, unidirectionalism should've got us all
speaking the same by now - and we ain't.

>I suspect that on the contrary
>languages tend to lose irregularity and suppletion under stress (e.g.
>a language becomes a trade language and a lot of people have to learn
>as simple a form of it as they can get away with; Saxons get conquered
>by Frenchmen and the language finds a happy medium; soldiers spread
>Latin all over the empire and a lot of Gauls, Germans, and Dalmatians
>try to talk to each other in it; that sort of thing) -- but in spite
>of that they keep as much irregularity as they can get away with
In the case of the latter, they did not merely keep as much irregularity as
they could get away with, they added a whole lot more!

The big difference between Latin & the Romancelangs is the breakdown of the
case system in nouns & adjectives.  Romanian, Old French & Old Provencal
retained a two case system, the others have lost case entirely (the
pronouns, of course, retain some of the case apparatus just as English
pronouns do).   Yep - simplification, indeed.

But the verbs!  I started once to write a Prolog parser for French verbs,
but dropped the project - all those irregular present tenses were causing
problems.  (Of course, it could be done - I just didn't have the patience
:)

The Romancelangs have dropped such things as synthetic passives & 'deponent
verbs', but they've sure added a lot of irregularities to make up for that.
--------------------------------------------------------------------

At 9:39 pm -0600 14/6/99, Ed Heil wrote:
>No, I'd say that children learning language is an optimal
>circumstance, since their brains are primed for it and they can do it
>pretty much constantly (infants not having a whole lot else in the way
>of responsibilities in most societies).
Yep - it's clearly the optimal circumstance.  No other time in life is
language acquisition so easy.  My elder grandson picked up two languages -
French & English - with the same facility and chatters away in whichever
one he wants to with similar ease now.   This is quite normal in bilingual
situations and, I believe, it is not all unusual to find children growing
up with three first languages - happy children who are spared the efforts
needed when we have to acquire a second, third or whatever language latter
in life.


>But yes, it is indeed one of
>the things that sets limits on irregularity and suppletion in language
>-- in short, which makes calculation rather than storage necessary
>even though calculation is infinitely less efficient in *usage*.
I guess that's about right.

>Here's an analogy.  Two students take a physics test which requires
>them to know a lot of mathematical formulas.
[analogy snipped - but I think it was a very relevant one]

>So I'd say that language acquisition (especially under difficult
>circumstances) regularizes; language use irregularizes (or at least in
>language use there is no advantage in regularity since everything
>possible will be stored rather than computed).
...and I'd say that I pretty much go along with that.

Ray.