haspelmath at EVA.MPG.DE
Mon Aug 18 13:35:38 UTC 2003
Dan Everett wrote:
>> My conclusion from this state of affairs is that the term "iteratives"
>> should not be used at all -- once a term has been widely and prominently
>> used in two different senses, it is usually better to give it up and use
>> alternatives (e.g. just "pluractional" and "repetitive").
> I don't think I agree with Martin on this suggestion. Terms are always
> going to be used and misused in various ways and to abandon a term
> because it has become confusing will simply increase confusion
> exponentially. I suggest that terms be defined when they are used. If
> one wishes to use the term 'iterative' for an action which occurs only
> once, that is fine, so long as the linguist defines the term clearly
> at the outset of his/her study. If there is a contrast in a single
> language between multiply repeated events vs. simple, single repeats,
> then different terms should be used and defined.
> Much the same ambiguity is found already in IPA symbols. A [p] in
> Piraha, for example, is not a [p] in Portuguese. They are similar, but
> not identical. I wouldn't be helping my fellow linguists if I failed
> to note this distinction. But I would also be remiss if I invented
> another symbol, since the two [p]s both fall under the broad intention
> of the IPA symbol, [p]. (They are both voiceless bilabial stops, but
> in Piraha the lips are flatter.)
> All of our terminology is simply a set of approximations, always in
> need of fine-tuning on a case-by-case basis. The failure to recognize
> this and to believe that a single term can be used in a more or less
> constant sense across languages is one of my pet peeves with
> typological research in general.
Dan makes two different points here: One is that one should define ones
terms in order to avoid confusion; the other one is that every language
has different categories but one should still use the same terms for
roughly similar categories.
But the first point is not in contrast to my earlier point: Of course
it's fine if one defines one's terms, but if one is not content with
just being understood, if one has the additional goal of improving
linguistic terminology (so that at one point in the future, we'll have
terms that we all understand in roughly the same way), then one should
proceed as I suggested: simply not use a term that has been widely and
prominently used in two different senses (because it will be very
difficult to get other linguists to agree precisely which of the two
senses should be abandoned).
And of course I agree with the second point: We shouldn't use new terms
for each new language, but we should be aware that terms for
language-particular categories are language-particular (e.g. by
capitalizing them, as in the widely known Comrie-Bybee-Croft
convention). I think typologists are more aware of this than
non-typologists, so I'm not sure that Dan's remark about "typological
research in general" is fair. I think that people who just work on a
particular language are much moire likely to look for "the passive" or
"the dative" or "the [p]" etc. in this language than people who look at
lots of languages simultaneously.
In practice, this means that the Lezgian category should be referred to
as "Repetitive" (as I did in my grammar), the Zoque category should be
referred to as "Repetitive" (as Jan Terje Faarlund is doing in his
grammar), although there are no doubt differences between these two
categories. And neither category should be called "iterative" or
"Iterative", because with this term, we do not even approximately know
what it is supposed to mean.
Martin Haspelmath (haspelmath at eva.mpg.de)
Max-Planck-Institut fuer evolutionaere Anthropologie, Deutscher Platz 6
Tel. (MPI) +49-341-3550 307, (priv.) +49-341-980 1616
More information about the Lingtyp