<div dir="ltr">Hi all,<div><br></div><div>With regard to the concern about losing information when abstracting, I think that this is mainly a problem if indeed all typologists did was to say whether a language does or doesn't have a particular category. While it's true that this is a problem in some frameworks (language X has some but not all of the properties associated with account Y of phenomenon Z, therefore X doesn't really "have" Z but rather "pseudo-Z"), this doesn't have to be a problem for typology.</div><div><br></div><div>That's because given a broad definition of a phenomenon, one can treat properties of grams/constructions as variables to be typologized over rather than as criteria for belonging to a category. For example, one could (1) say that since a particular construction has to be limited to noun roots, Warembori doesn't "have" incorporation. Or one could say (2) that the nature of the incorporand is not a criterion for whether a given construction is or isn't incorporation, but rather a variable that can have different values. Balthasar Bickel and his colleagues have called this "multivariate analysis," and have applied it to a few domains (for example, clause linkage). Of course, this is also basically what other typologists have always done, e.g., Joan Bybee and her colleagues in their pioneering Evolution of Grammar, where TAM categories were defined explicitly and rather broadly, and then particular functional and formal properties of grams were typologized over, thereby giving typology tremendous insights into the diachronic typology of TAM.<br><br>In this way, individual categories in particular languages don't lose their particularity (Bickel argues for this explicitly in his paper on clause-linkage), and typologists have lots of material to work with, with lots of space for new insights about language. To go back to the matter at hand, I think Martin's suggestion of a unified notion of genifiers is a salutary one (and this is orthogonal to the actual name chosen), but it's possible that a more bottom-up approach to typologizing over the properties associated with individual grams in individual languages might lead to a non-arbitrary clustering of properties, thereby alleviating the concern over the arbitrariness of where one puts the cut-off point.<br><br></div><div>Best,<br></div><div>Eitan<br><br></div><div><br></div><div><br></div><div><br></div></div><div class="gmail_extra"><br clear="all"><div><div class="gmail_signature" data-smartmail="gmail_signature"><div dir="ltr">Eitan Grossman<div>Lecturer, Department of Linguistics/School of Language Sciences<br></div><div>Hebrew University of Jerusalem</div><div>Tel: +972 2 588 3809</div><div>Fax: +972 2 588 1224</div></div></div></div>
<br><div class="gmail_quote">On Wed, Mar 22, 2017 at 10:18 PM, Martin Haspelmath <span dir="ltr"><<a href="mailto:haspelmath@shh.mpg.de" target="_blank">haspelmath@shh.mpg.de</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div text="#000000" bgcolor="#FFFFFF">
I fully agree with Mark on this:<span class=""><br>
<br>
<div class="m_-6060647318619433019moz-cite-prefix">On 22.03.17 00:25, Mark W. Post wrote:<br>
</div>
<blockquote type="cite">
<p>It seems to me that what we're really talking about here is the
same thing that we usually talk about, which is that there are
no cross-linguistically watertight categories, but we want to do
typology anyway, so what do we do? We can select a semantic
parameter (a "comparative concept") in terms of which categories
may be similar across languages, but they will differ in other
respects. If we focus on those other respects, we can end up
with a different typology. It may be that the real difficulty
here is that our traditional category-labels, and the categories
they are designed to capture, are multi-dimensional.<br>
</p>
</blockquote>
<br></span>
Genifiers may of course cumulate with other functions, e.g.
referential specification (as in Mark's Yi examples), or
definiteness (as in Spanish definite articles <i>el/la</i>), or
number. These forms are thus simultaneouly articles and/or number
markers, but from the perspective of genification, they are
genifiers.<br>
<br>
I also agree with Randy:<span class=""><br>
<br>
On 22.03.17 03:42, Randy LaPolla wrote:<br>
<blockquote type="cite">
<p>Whenever we make a higher abstraction we are moving one more
step away from the facts of the languages. The terms “gender”
and “noun classifier” are already abstractions across a range of
different phenomena, and so there is some loss of information
about the diversity of forms when we use such terms, and if we
then make a categorial merger of these two forms, as suggested,
we then lose even more information.</p>
</blockquote>
<br></span>
Of course we "lose" information when we generalize across languages.
But one needs to understand that typological categorization is very
different from description. It may be that "there is the danger that
this usage filters back into descriptions of languages", but this is
so only if describers think that describing a language means putting
pre-established labels on them. They should be warned against this,
by emphasizing that comparative concepts are meant for comparison,
and can be used for description only if there is no significant
cross-linguistic variation in the relevant domain (e.g. "1st
person", or "fricative").<br>
<br>
To get back to my original question: Does anyone know a reason why
one shouldn't define "gender" as a comparative concept in the
following way:?<br>
<br>
A <b>gender system</b> (= a system of gender markers) is a system
of genifiers which includes no more than 20 genifiers and which is
not restricted to numeral modifiers.<br>
<br>
I'm not very comfortable with a definition that makes reference to
an arbitrary number, but I'm even less comfortable with a situation
where we have no definition for "gender" at all, because this makes
much of the earlier literature unreadable. It seems to me that
Corbett (1991) is about gender systems roughly in the above sense.<br>
<br>
Best,<br>
Martin<span class=""><br>
<pre class="m_-6060647318619433019moz-signature" cols="72">--
Martin Haspelmath (<a class="m_-6060647318619433019moz-txt-link-abbreviated" href="mailto:haspelmath@shh.mpg.de" target="_blank">haspelmath@shh.mpg.de</a>)
Max Planck Institute for the Science of Human History
Kahlaische Strasse 10
D-07745 Jena
&
Leipzig University
IPF 141199
Nikolaistrasse 6-10
D-04109 Leipzig
</pre>
</span></div>
<br>______________________________<wbr>_________________<br>
Lingtyp mailing list<br>
<a href="mailto:Lingtyp@listserv.linguistlist.org">Lingtyp@listserv.linguistlist.<wbr>org</a><br>
<a href="http://listserv.linguistlist.org/mailman/listinfo/lingtyp" rel="noreferrer" target="_blank">http://listserv.linguistlist.<wbr>org/mailman/listinfo/lingtyp</a><br>
<br></blockquote></div><br></div>