Irregularity in human languages (was Re: irregular conlangs)

Quoth R. Nierse:
> Ed:
> > Yep.  That's more or less what I was saying.  Irregularity is a
> > problem for learning, but not at all a problem for use -- because
> > regularity is an advantage for learning, but not at all an advantage
> > for use.
> IMO most language users are persons that have acquired the language as a
> kid. Children don't learn a language (they absorb it) so they don't make a
> distinction between regular and irregular forms. Words are just words and
> children use them when they need them. At a later stage children discover
> the regularities and start making all irregular verbs, adjectives etc.
> regular. After that they notice that the language community uses some
> irregular forms instead of regular and those are remembered. Can't be too
> much, because we can't rememeber all irregular forms, so the irregular ones
> that are used most will be remembered.
I have to disagree with that last sentence.  I'm _sure_ that it's not a
memory limitation that restricts the irregular verbs.  You make a good
point though, about children first learning all verbs as lexemes, then
learning the rules (and trying to apply them indiscriminately), and
finally going back to the correct irregular forms where applicable.

Hypothesis: the restricting factor on which verbs are irregular are
"those verbs which are used often around children, and which children
themselves use often".  Thus I wouldn't expect, for instance, "compute"
or "automate", whereas I would expect "fall", "run", "sneak", "jump",
"dive", etc.

Test: Hit Google for a list of english irregular verbs.  Found
        http://www.englishclub.net/reference/irregular/verbs.htm
and     http://www.sk.com.br/sk-irrve.html .

Results: Out of a few hundred irregular verbs on the two pages, I judge
the following to be "non-kid" verbs:
        bid broadcast cost lend spend understand arise bear befall
        beget behold breed cast fling forbid overcome overtake slay
        sweep undergo understand uphold tread

Conclusion: I adjudge my hypothesis basically correct, with a few
caveats (to explain away the counterexamples :).

Group 1: bear befall beget behold bid(bade) fling slay tread uphold
I'd say that these words are all out of _common_ use; they still enjoy
play as archaic words and in formulaic phrases ("uphold the law"), but
they don't get enough use to be reanalysed as weak verbs.

Group 2: bid(bid) broadcast cast cost lend spend
These words are only pseudo-irregular; they would obey the -ed/-ed
paradigm except for phonological concerns.

Group 3: understand arise forbid overcome overtake undergo understand
These words are all compounds (opaque, but compounds nonetheless) which
follow the rule of their base verb.

So the only two I can't explain (that I saw) were "breed" and "sweep".
I suppose "sweep" could maybe be a little-kid verb.  I further suppose
that "breed" might be a little-kid verb among the people that use it
(that is, *I* didn't use it as a kid, but then I don't use it now; but
someone who uses it today, e.g. on a farm, may well have grown up using
it).  Or maybe they're just counterexamples... what's a rule without
exceptions, anyway? :)

This has a lot of relevance (if true) for the people doing human
conlangs---it's not *just* the frequency of a verb, but also its
likelihood of use around kids, that makes it a good candidate for
irregularity from the normal paradigms.  Maybe other people can support
my claim with reference to other languages?

--
-=-Don Blaheta-=-=-dpb@cs.brown.edu-=-=-<http://www.cs.brown.edu/~dpb/>-=-
"Retraction: The 'Greek Special' is a huge 18 inch pizza and not a huge 18
inch penis, as described in an ad. Blondie's Pizza would like to apologize
for any confusion Friday's ad may have caused."
                        --correction printed in The Daily Californian