Theiling Online    Sitemap    Conlang Mailing List HQ   

Re: THEORY: irregular conlangs

From:Ed Heil <edheil@...>
Date:Thursday, September 30, 1999, 18:51
I think at one time we had a discussion about "storage vs.
computation" in the brain.  I read in Kosslyn & Koenig's popular work
on neuroscience, _Wet Mind_, that the brain's storage-and-retrieval
mechanisms are so much more efficient than its computation mechanisms
that where possible behavior patterns will always be stored, rather
than computed anew for every use.

In linguistics terms, this means that it is more effective to store
"past of walk = walked" and look it up each time than it is to
calculate anew each time, "stem of walk = walk.  -ed is past morpheme.
 put them together to get the past and we get walked."  Even though
the latter would involve less use of memory.  Kosslyn and Koenig say
that for practical purposes we must consider the storage capacity of
the human memory to be unlimited, and retrieval to be cheap and easy.

This is often the opposite of the situation in computer science,
where computation is fast and effective and storage is expensive and
limited, and retrieval is difficult and time-consuming.

Notice what this situation means, though.  It means that in terms of
language use, it is just as effective to have an irregular form as to
have a regular form -- they're both going to be stored and looked up.
Sure the regular form is easier in terms of computation, but
computation is irrelevant for any frequently used language element.

Now, it's more difficult to *learn* an irregular form than to *learn*
a regular form, but it is absolutely no easier to *use* a regular one
than an irregular one.

In fact, in terms of information transfer, it is probably easier to
use an irregular one because it may be more distinctive.  It's harder
to mistake "went" for "go" than it is "goed" for "go."  (Note that
non-distinctive irregular forms, like "lie/lay/lain", get mixed up,
but virtually nobody mixes up "go" and "went.")

In situations where language learning takes place under especially
poor conditions (i.e. a lot of people have to learn the language as an
adult, or it's not a primary language), you're likely to lose
irregular forms.  In situations where language learning takes place in
childhood in a community where that language is primary, it's likely
to keep them: witness Navajo, where as far as I can tell there are no
regular verb forms whatsoever.

So if your conlang is designed to be *learned easily* then it's a
good idea to avoid irregularity, but if it's supposed to be
*naturalistic* and the language is a primary language of a community,
with a lot of first language learners, then by all means, let it have

Boxcars are pulling an Ed of sorts out of town.

Daniel Andreasson wrote:

> Hejsan allihopa. (Hello.) > > In Sociolinguistics class today we had a discussion > about irregular forms in languages. You know, strong > verbs and adjectives, etc. "Go, went, gone". > "Good, better, best". Obviously it is the most > frequent verbs and adjectives that are irregular. > The question is why. We got the explanation that > because they are so frequent, you can produce it > faster if it is a lexeme rather than a conjugation. > > That is, "went" is a whole other word than "go". > You need the word "went" and then you pick it out > from your brain immediately. If the past form of > 'go' would have been 'goed', you would have had to > first think of 'go' and then put the suffix '-ed' > to it. Therefore 'went' is faster and more > economic. IMHO this seems reasonable. (Although I > guess I should have to ask my Psycholinguistics > teacher first.) > > It seems to me that there should be some correlation > between frequency and irregularity. There should be > a border somewhere, where a word isn't so frequent > anymore that it becomes irregular. > > Anyway. My two questions. What do you guys think > of this? And do you do this in your conlangs? > AFAIK, in most languages the copula verb is > irregular, but most conlangs seem to be very regular. > Am I right or wrong? I know many of you (as I once) > want an extremely logical language, one that you > have to invent because there aren't any logical > natlangs. But those of you who persue a natlangy > touch of your conlang, how far do you go in your > irregularities? > > Daniel Andreasson > > >