[Lingtyp] wordhood: responses to Haspelmath

Sat Nov 11 18:52:16 UTC 2017

Dear all,

I think some of the comments here are taking too extreme a position (for
argument's sake?). Martin did not in his paper reject the idea that some
(or all) languages have their own sense of "words", even though in
different languages the criteria may be different. One of the clearest
examples (in my opinion at least) is vowel harmony in Turkish. But
unfortunately the rarity of vowel harmony makes that impractical for most
broad typological work.

So I think there are three possible positions:
1. Words can be defined consistently and effectively for all languages.
2. Words can be defined differently for different languages but can still
be considered valid at least for those individual languages, and
potentially in a comparative sense if we take into account how the criteria
for each language differ. (I guess that could potentially lead back to
position 1, although with an indirect definition.)
3. There are really no words at all, and that's just our perspective bias
(from writing spaces, from traditional grammar, whatever).

Some of the comments have argued for position 3, such as writing
grammatical descriptions without referencing words at all. But I don't see
why that needs to be the case. Why not consider position 2? Just like parts
of speech vary in different languages and it may not be possible to
consistently define "noun" or "verb" for all languages (of course that's up
for debate), individual grammars can still make reference to them in
language-internal terms (along with potentially unique word classes as
well).

A question then is whether "word" is a basic notion (say, part of UG), or
whether it is something emergent in some (or all) grammars, based on
varying criteria. Or whether it just isn't a valid concept at all in any
language.

Questions:
1) Can all grammars be written without referencing a concept like "word" at
all?
2) If we do reference "words", are we really talking about the same concept
in each language?
3) What criteria are used, and if they differ, is that because "words" are
the "same" across languages but manifested differently, or because "words"
are actually different (because of the different criteria), and then can we
still compare them in a relevant way?

I don't have an answer for those questions yet, though. But I'm enjoying
the discussion.

Daniel Ross
Ph.D. Candidate
University of Illinois

On Sat, Nov 11, 2017 at 10:38 AM, David Gil <gil at shh.mpg.de> wrote:

> Contrary to Östen (below), I am not quite ready to concede to Martin the
> impossibility of defining a comparative concept of "word" that will enable
> typologists to distinguish between isolating and polysynthetic languages
> (as well as various intermediate types).  I am currently working on a paper
> that will provide such a definition.  An extended abstract of the paper is
> attached here.
>
> David
>
>
>
> On 12/11/2017 02:23, Östen Dahl wrote:
>
> OK, we should forget about word boundaries in typology, but should we also
> do so when writing grammars? Could you write a grammar of a stereotypical
> polysynthetic grammar and make it look like an isolating one without using
> procrustean methods?
>
> (Didn’t Skalička have a rather idiosyncratic definition of polysynthesis?)
>
> Östen
>
>
>
> *Från:* Martin Haspelmath [mailto:haspelmath at shh.mpg.de
> <haspelmath at shh.mpg.de>]
> *Skickat:* den 11 november 2017 19:01
> *Till:* Östen Dahl <oesten at ling.su.se> <oesten at ling.su.se>
> *Kopia:* lingtyp at listserv.linguistlist.org
> *Ämne:* Re: SV: [Lingtyp] wordhood: responses to Haspelmath
>
>
>
> It's not crazy at all to say that isolating languages could be described
> as polysynthetic, and vice versa. (In fact, Skalička described Modern
> Chinese as polysynthetic in 1946.)
>
>
>
> The problem is that archetypes like isolating and polysynthetic are mostly
> stereotypes. They are not clearly defined, at least not without reference
> to a "word" concept (itself only based on intuition, i.e. stereotypes).
>
>
>
> Of course, morphosyntactic patterns are often more complex than simple
> strings of morphemes. But we don't really know in which ways these
> complexities cluster. Is it the case that languages with tense-person
> cumulation (to give just one example of a complexity) also tend to show
> case-number cumulation? Is it the case that languages with special
> bare-object constructions ("incorporation") tend to show phonological
> interactions between object and verb? We don't know yet, I think. By merely
> labeling languages according to a few archetypes, we won't find out.
>
>
>
> So yes, let's forget about word boundaries in typology until we have a
> very good way to draw them consistently (using the same criteria in all
> languages).
>
>
>
> Best,
>
> Martin
>
>
> Am 11.11.2017 um 18:40 schrieb Östen Dahl < <oesten at ling.su.se>
> oesten at ling.su.se>:
>
> Martin, I wonder if your views on these matters imply that a polysynthetic
> language could equally well be described as being an isolating one, and
> vice versa. That is, one should just forget about word boundaries and
> describe utterances as consisting of strings of morphemes. If you think
> this is not feasible, why?
>
> Best,
>
> Östen
>
>
>
>
>
> _______________________________________________
> Lingtyp mailing listLingtyp at listserv.linguistlist.orghttp://listserv.linguistlist.org/mailman/listinfo/lingtyp
>
>
> --
> David Gil
>
> Department of Linguistic and Cultural Evolution
> Max Planck Institute for the Science of Human History
> Kahlaische Strasse 10, 07745 Jena, Germany
>
> Email: gil at shh.mpg.de
> Office Phone (Germany): +49-3641686834 <+49%203641%20686834>
> Mobile Phone (Indonesia): +62-81281162816 <+62%20812-8116-2816>
>
>
> _______________________________________________
> Lingtyp mailing list
> Lingtyp at listserv.linguistlist.org
> http://listserv.linguistlist.org/mailman/listinfo/lingtyp
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/lingtyp/attachments/20171111/c145bd2d/attachment.htm>