is it undergeneralization?

Suzanne E Kemmer kemmer at RUF.RICE.EDU
Wed Apr 9 04:32:22 UTC 1997

If we're looking for terminology to describe
this sort of phenomenon, I think it's exactly
parallel to the "repartition" that Breal described for
lexical splits in diachrony--differentiations in the
history of etymologically the same word. An example
going on right now is a pronunciation difference
in _virtually_ depending on whether it means
'practically', vs. 'pertaining to virtual, electronic
connections' (as in "She laughed, virtually").
The older meaning allows more phonological elision.

With _bees_ this differentiation happens to be at the
constructional level, and it is in the
ontogenetic development of the grammar rather
than the larger-scale change of the lg.
in a population over time. It's like the
case of _gots_ for 'has' which is endemic in child
language (and bigger and bigger kids seem
to use it).

I'm uncomfortable with 'undergeneralization'
for this kind of differentiation, since that
term sounds like restricting the usage of a morpheme
to a range smaller than what it has in the adult lg.
There are plenty of examples of the latter in child
language, but it seems different from what is going on here.
(E.g. using a word for a type to designate
an instance: _kuh_ to mean only the child's juice

'Overgeneralization' is appropriate
or not depending on the perspective you take.
>>From the point of view of the occurrence
of -s, it's putting regular
inflection in a place it doesn't belong,
on the strength of the entrenched regular pattern.
The child's schema for -s is then more general than in
the adult language.

Yet at the same time, because of
the child's different understanding
of the predicate _be_ (compared to the
adult language) in these 'act' contexts,
it's not just a matter of replacing an irregular
verb with its regular correlate,
like saying _goed_ for _went_.

Actually, this whole question makes
me wonder what kids are actually doing
when 'replacing' irregulars with
regulars  (which I understand they
do only a portion of the time--they
do keep producing irregulars the
whole period of overgeneralization).

We don't know what kinds of semantic
distinctions they might be making
until we look, and my assumptions about
language learning leave the possibility
open for such distinctions, rather
than precluding them.


More information about the Funknet mailing list