<div dir="ltr"><div><div><div>Dear all,<br><br></div>re the 3+ examples provided as a counterexamples for my very tentative generalization. I accept them, but consider them as such to different extents.<br><br></div><div><b>David suggested that Roon</b> is a counterexample. In his example, a vast number of nouns may refer, roughly, either to an object or to the substance produced from this object (or a part of this object). I think this is indeed a good counterexample to my claim, in being an example of a system that is, to a considerable extent if not overwhelmingly, reference-based and flexible. I would not take the counter-objection suggested by Johanna (that this is a productive unmarked derivational process, or conversion) because I have a feeling that this would be a way to wave off all counterexamples, which would make my generalization totally non-falsifiable. </div><div>I think that I need more details on this system - David, is there any reference? - to see how exactly the system works. Obviously, not all nouns may be re-conceptualized as substance in an equally easy way, at least under usual circumstances (but note David's missionary example), so it is interesting how overwhelming flexibility is in this system.<br><br></div><div><b>Doris suggested Maasai</b> as a similar situation. I would argue that it is not fully similar, though a counterexample in its own way. In her paper, she describes how Maasai nouns shift between masculine and feminine classes depending on whether the speaker construes them as small or big or wants to convey a pejorative meaning. After reading about this data, I would probably separate size-effect from evaluative effect; and would certainly distinguish between flexibility and reference-based-ness. Both - size-based and attitude-based gender shifts (or switches, as I think Sasha Aikhenvald calls them) lead to flexibility, but not both are equally reference-based in the strict sense. Evalutative effects are less (or not) related to the property of the referent but convey speaker attitude. (In this sense, David's example provides what seems to be a truly reference-based effect to a much greater extent. You could argue even here that speaker's stance is involved, i.e. the same tree may be considered as an object or a substance, but this, again, would probably make my claim non-falsifiable). I also think that size-based effects are more reference-based than pejorative/politeness effects - though size-effects also include a great deal of speaker's subjectivity, more than in the object-substance example. <br><br>I think the distinction between flexibility of gender (wider) and reference- vs discourse/attitude-based systems is an important one. The former is a more general and more formal notion, the latter are functional subtypes of the former, what a flexible system may convey (probably a scale rather than a true taxonomy).<br><br></div><div>Also, I could argue that speaker attitude (at least evaluation - good vs. bad) is still compatible with my original functional idea that you need to have a strong anchor for agreement, because speaker attitude holds through an utterance; but this would be another way of being an anchor, and certainly modifies my original claim - may be rightly. So in any sense this is a counterexample to the tentative generalization and an adjustment for the functional explanation. If there are more examples like that, I would now modify the original claim in the sense that gender systems (defined as involving agreement) tend to be rigid but may become flexible - but flexible more on discourse/indexical basis than on reference-basis. David's system remains a counterexample.<br></div><div><br>Finally, <b>Katharina suggested Movima</b> as a counterexample. In Movima, nominal categorization is expressed by articles, and category assignment is flexible. This I would probably not accept as counter-example, for the following reason. To me, Movima is not a clear case of gender system (contra Martin above), because it is not a clear case of agreement. Typical agreement involves different lexical items being 'hosts' to agreement markers across different constituents. Articles are commonly considered grammatical elements, and described as periphrastic nominal markers. Under this view, articles can not be considered targets of agreement because, being grammatical, they belong to the same grammatical word that is, or which contains, their supposed controller. If one adopts this view on articles, there is no agreement involved, because nominal categorization is expressed on the noun itself (periphrastically), not "repeated" on other lexical items/constituents to indicate belonging to the same syntactic unit (to use Walter's words).<br></div><br></div>Surely, this depends on your view on articles and agreement. Other views on articles may lead to consider Movima (and probably other languages) as a counterexample to my tentative generalization. But I have a feeling that many, like myself, would not consider what is usually called an article to be target of what is usually called agreement. On the other hand, articles often have recent and transparent sources of grammaticalization (from demonstratives etc.) which would fit my definition of agreement; and where such demonstratives show flexible gender, they would be better counterexamples. As I do not expect that what happens to Movima articles happened to them _because_ they ceased to be demonstratives and started being articles, so these counterexamples are probably out there in the big world.<br><div><br><div>Michael Daniel</div><br><br></div><div class="gmail_extra"><br><div class="gmail_quote">2017-03-31 22:22 GMT+02:00 Martin Haspelmath <span dir="ltr"><<a href="mailto:haspelmath@shh.mpg.de" target="_blank">haspelmath@shh.mpg.de</a>></span>:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">The Movima system is surely a gender system, not only according to my definition, but also according to Corbett's (1991) definition.<br>
<br>
It is gender according to my definition because it does not rely on a notion of asymmetric "agreement". Recall that I said that a genifier system is a system of markers which occur on nominal modifiers, on predicates or on anaphoric pronouns, and each of which *reflects or contributes* a property other than person or number. (And a gender system is a genifier system that is not restricted to numeral modifiers and has at most 20 members.)<br>
<br>
It is gender according to Corbett's older definition because it is used on anaphoric pronouns (where it seems to work much like English he/she/it, which Corbett regards as gender forms).<br>
<br>
(The same goes for Roon, the Austronesian language mentioned by David Gil.)<br>
<br>
Misha Daniel proposes to base the definition of gender on "agreement", but he provides no definition of agreement. Stereotypically (and "canonically"), agreement reflects properties of a noun, but what do we do if a marker contributes information to a noun, and if this happens commonly? The notion of "agreement" does not seem to help here. Moreover, people often seem to think that numeral classifiers are evidently not agreement markers, but as Walter Bisang pointed out, Thai-style numeral classifiers are not that different from stereotypical agreement, and neither are Kilivila classifiers (as noted by Grev Corbett and Sebastian Fedden).<br>
<br>
Best,<br>
Martin<div><div class="m_451588021156008535h5"><br>
<br>
On 31.03.17 11:16, <a href="mailto:khaude@uni-koeln.de" target="_blank">khaude@uni-koeln.de</a> wrote:<br>
</div></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div><div class="m_451588021156008535h5">
Dear colleagues,<br>
<br>
In Movima (isolate, South-West Amazon), referential elements (i.e. determiners/articles and pronouns) differentiate between human male, human female, non-human, and plural referents. Inanimate ("neuter") referential elements, when used with a human noun, indicate non-specific or derogatory reference. So for sex-neutral human nouns like *dichiye* 'child', we get:<br>
<br>
m. kus dichiye 'the/a boy'<br>
f. kinos dichiye 'the/a girl'<br>
pl. kis dichiye '(the) children'<br>
n. kos dichiye 'some child (nonspecific); that child (derogatory)'<br>
<br>
With nouns that imply sex, the alternatives are of course limited - e.g. *itilakwa* 'man' cannot be combined with a 'fem.' referential element. Still, they can be combined with a neutral element (if referring to an animal or with the above-mentioned effects for humans: kos itilakwa = 'the male (animal); some man; that man - derogatory'), which shows that the choice of the referential element does not depend on the noun, but on the referent.<br>
<br>
So this is not agreement, and not even gender (although I labelled it gender in Haude 2006). I am wondering now whether this system shouldn't, in fact, rather be considered a case of classification ...<br>
<br>
Best,<br>
Katharina<br>
<br>
Haude, Katharina. 2006. A Grammar of Movima. Doctoral diss., Radboud University Nijmegen. <a href="http://webdoc.ubn.ru.nl/mono/h/haude_k/gramofmo.pdf" rel="noreferrer" target="_blank">http://webdoc.ubn.ru.nl/mono/h<wbr>/haude_k/gramofmo.pdf</a><br>
<br>
<br>
Quoting Johanna NICHOLS <<a href="mailto:johanna@berkeley.edu" target="_blank">johanna@berkeley.edu</a>>:<br>
<br>
</div></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div><div class="m_451588021156008535h5">
I'd be inclined to call that a kind of derivation (or conversion), highly<br>
productive for some languages and lexically restricted for others. Much<br>
like what produces Spanish manzana (F) 'apple (fruit)' vs. manzano (M)<br>
'apple (tree)', except more widespread in the lexicon.<br>
<br>
Johanna<br>
<br>
On Thu, Mar 30, 2017 at 9:39 AM, <<a href="mailto:dlpayne@uoregon.edu" target="_blank">dlpayne@uoregon.edu</a>> wrote:<br>
<br>
</div></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div><div class="m_451588021156008535h5">
Another language of essentially the “referentially-based” type that David<br>
Gil describes is Maa (Maasai):<br>
<br>
Payne, Doris L. 1998. Maasai gender in typological perspective. *Studies<br>
in African Linguistics* 27.159-175.<br>
<br>
<br>
- Doris Payne<br>
<br>
<br>
*From:* David Gil <<a href="mailto:gil@shh.mpg.de" target="_blank">gil@shh.mpg.de</a>><br>
*Sent:* Thursday, March 30, 2017 9:32 AM<br>
*To:* <a href="mailto:lingtyp@listserv.linguistlist.org" target="_blank">lingtyp@listserv.linguistlist.<wbr>org</a><br>
<br>
Dear all,<br>
<br>
Misha writes:<br>
> I am unaware of any language that has a reference-based<br>
> (overwhelmingly non-rigid) noun categorization device used for what is<br>
> an obvious typological instance of agreement. If you are aware of such<br>
> example, it would be interesting to know more about it. On the other<br>
> hand, Walter's message indicates that there are consistently rigid<br>
> systems that are not used for agreement - his Thai example.<br>
As a possible example of a "language that has a reference-based<br>
(overwhelmingly non-rigid) noun categorization device used for what is<br>
an obvious typological instance of agreement", I offer Roon, a language<br>
of the South Halmahera West New Guinea subgroup of Austronesian, spoken<br>
off the northern coast of western New Guinea. (I believe the facts in<br>
closely related Biak, for which there are two recent reference grammars,<br>
are similar in the relevant respects.)<br>
<br>
Roon has two genders, animate (encompassing humans, animals and plants)<br>
and inanimate. (Gender is only marked in 3rd person, leading me to<br>
wonder whether it can be collapsed with person into a single<br>
higher-level feature, though Grev Corbett has tried to talk me out of<br>
that!) Agreement is straightforward: verbs agree with their subjects,<br>
and most nominal attributes agree with their head nouns, with respect to<br>
person, number and also gender (animate/inanimate). The agreement<br>
markers are prefixal, sometimes involving metathesis with the first<br>
segment of the root of the host agreement target.<br>
<br>
What's less clear is whether this is a *reference-based*<br>
noun-categorization device. Against its characterization as such are a<br>
*very* small set of lexical items, including for example the words for<br></div></div>
'coffee' and 'money' that are grammatically animate even though they<div><div class="m_451588021156008535h5"><br>
are semantically inanimate. On the other hand, many, perhaps most<br>
words, are flexible, and can belong to either class (animate or<br>
inanimate) with largely predictable semantic effects. Most commonly, the<br>
animate form refers to the animal/plant as a whole, while the inanimate<br>
form refers to either part of the animal/plant, or a substance or<br>
foodstuff made out of it. For example, one word means 'pig' when<br>
animate and 'pork' when inanimate, similarly another word means 'coconut<br>
tree' when animate and 'coconut (fruit)' when inanimate. (I must admit<br>
I haven't checked what happens with human referents, but given that they<br>
were still cooking the occasional missionary in the mid-1800s, I suspect<br>
I know the answer.)<br>
<br>
So does this count as "reference-based"?<br>
<br>
David<br>
<br>
-- <br>
David Gil<br>
<br>
Department of Linguistic and Cultural Evolution<br>
Max Planck Institute for the Science of Human History<br>
Kahlaische Strasse 10, 07745 Jena, Germany<br>
<br>
Email: <a href="mailto:gil@shh.mpg.de" target="_blank">gil@shh.mpg.de</a><br>
Office Phone (Germany): <a href="tel:%2B49-3641686834" value="+493641686834" target="_blank">+49-3641686834</a> <+49%203641%20686834><br>
Mobile Phone (Indonesia): <a href="tel:%2B62-81281162816" value="+6281281162816" target="_blank">+62-81281162816</a> <+62%20812-8116-2816><br>
<br>
______________________________<wbr>_________________<br>
Lingtyp mailing list<br>
<a href="mailto:Lingtyp@listserv.linguistlist.org" target="_blank">Lingtyp@listserv.linguistlist.<wbr>org</a><br>
<a href="http://listserv.linguistlist.org/mailman/listinfo/lingtyp" rel="noreferrer" target="_blank">http://listserv.linguistlist.o<wbr>rg/mailman/listinfo/lingtyp</a><br>
<br>
<br>
______________________________<wbr>_________________<br>
Lingtyp mailing list<br>
<a href="mailto:Lingtyp@listserv.linguistlist.org" target="_blank">Lingtyp@listserv.linguistlist.<wbr>org</a><br>
<a href="http://listserv.linguistlist.org/mailman/listinfo/lingtyp" rel="noreferrer" target="_blank">http://listserv.linguistlist.o<wbr>rg/mailman/listinfo/lingtyp</a><br>
<br>
<br>
</div></div></blockquote></blockquote><div><div class="m_451588021156008535h5">
<br>
<br>
Quoting Johanna NICHOLS <<a href="mailto:johanna@berkeley.edu" target="_blank">johanna@berkeley.edu</a>>:<br>
<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
I'd be inclined to call that a kind of derivation (or conversion), highly<br>
productive for some languages and lexically restricted for others. Much<br>
like what produces Spanish manzana (F) 'apple (fruit)' vs. manzano (M)<br>
'apple (tree)', except more widespread in the lexicon.<br>
<br>
Johanna<br>
<br>
On Thu, Mar 30, 2017 at 9:39 AM, <<a href="mailto:dlpayne@uoregon.edu" target="_blank">dlpayne@uoregon.edu</a>> wrote:<br>
<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
Another language of essentially the “referentially-based” type that David<br>
Gil describes is Maa (Maasai):<br>
<br>
Payne, Doris L. 1998. Maasai gender in typological perspective. *Studies<br>
in African Linguistics* 27.159-175.<br>
<br>
<br>
- Doris Payne<br>
<br>
<br>
*From:* David Gil <<a href="mailto:gil@shh.mpg.de" target="_blank">gil@shh.mpg.de</a>><br>
*Sent:* Thursday, March 30, 2017 9:32 AM<br>
*To:* <a href="mailto:lingtyp@listserv.linguistlist.org" target="_blank">lingtyp@listserv.linguistlist.<wbr>org</a><br>
<br>
Dear all,<br>
<br>
Misha writes:<br>
> I am unaware of any language that has a reference-based<br>
> (overwhelmingly non-rigid) noun categorization device used for what is<br>
> an obvious typological instance of agreement. If you are aware of such<br>
> example, it would be interesting to know more about it. On the other<br>
> hand, Walter's message indicates that there are consistently rigid<br>
> systems that are not used for agreement - his Thai example.<br>
As a possible example of a "language that has a reference-based<br>
(overwhelmingly non-rigid) noun categorization device used for what is<br>
an obvious typological instance of agreement", I offer Roon, a language<br>
of the South Halmahera West New Guinea subgroup of Austronesian, spoken<br>
off the northern coast of western New Guinea. (I believe the facts in<br>
closely related Biak, for which there are two recent reference grammars,<br>
are similar in the relevant respects.)<br>
<br>
Roon has two genders, animate (encompassing humans, animals and plants)<br>
and inanimate. (Gender is only marked in 3rd person, leading me to<br>
wonder whether it can be collapsed with person into a single<br>
higher-level feature, though Grev Corbett has tried to talk me out of<br>
that!) Agreement is straightforward: verbs agree with their subjects,<br>
and most nominal attributes agree with their head nouns, with respect to<br>
person, number and also gender (animate/inanimate). The agreement<br>
markers are prefixal, sometimes involving metathesis with the first<br>
segment of the root of the host agreement target.<br>
<br>
What's less clear is whether this is a *reference-based*<br>
noun-categorization device. Against its characterization as such are a<br>
*very* small set of lexical items, including for example the words for<br>
'coffee' and 'money' that are grammatically inanimate even though they<br>
are semantically inanimate. On the other hand, many, perhaps most<br>
words, are flexible, and can belong to either class (animate or<br>
inanimate) with largely predictable semantic effects. Most commonly, the<br>
animate form refers to the animal/plant as a whole, while the inanimate<br>
form refers to either part of the animal/plant, or a substance or<br>
foodstuff made out of it. For example, one word means 'pig' when<br>
animate and 'pork' when inanimate, similarly another word means 'coconut<br>
tree' when animate and 'coconut (fruit)' when inanimate. (I must admit<br>
I haven't checked what happens with human referents, but given that they<br>
were still cooking the occasional missionary in the mid-1800s, I suspect<br>
I know the answer.)<br>
<br>
So does this count as "reference-based"?<br>
<br>
David<br>
<br>
-- <br>
David Gil<br>
<br>
Department of Linguistic and Cultural Evolution<br>
Max Planck Institute for the Science of Human History<br>
Kahlaische Strasse 10, 07745 Jena, Germany<br>
<br>
Email: <a href="mailto:gil@shh.mpg.de" target="_blank">gil@shh.mpg.de</a><br>
Office Phone (Germany): <a href="tel:%2B49-3641686834" value="+493641686834" target="_blank">+49-3641686834</a> <+49%203641%20686834><br>
Mobile Phone (Indonesia): <a href="tel:%2B62-81281162816" value="+6281281162816" target="_blank">+62-81281162816</a> <+62%20812-8116-2816><br>
<br>
______________________________<wbr>_________________<br>
Lingtyp mailing list<br>
<a href="mailto:Lingtyp@listserv.linguistlist.org" target="_blank">Lingtyp@listserv.linguistlist.<wbr>org</a><br>
<a href="http://listserv.linguistlist.org/mailman/listinfo/lingtyp" rel="noreferrer" target="_blank">http://listserv.linguistlist.o<wbr>rg/mailman/listinfo/lingtyp</a><br>
<br>
<br>
______________________________<wbr>_________________<br>
Lingtyp mailing list<br>
<a href="mailto:Lingtyp@listserv.linguistlist.org" target="_blank">Lingtyp@listserv.linguistlist.<wbr>org</a><br>
<a href="http://listserv.linguistlist.org/mailman/listinfo/lingtyp" rel="noreferrer" target="_blank">http://listserv.linguistlist.o<wbr>rg/mailman/listinfo/lingtyp</a><br>
<br>
<br>
</blockquote></blockquote>
<br>
<br>
<br>
______________________________<wbr>_________________<br>
Lingtyp mailing list<br>
<a href="mailto:Lingtyp@listserv.linguistlist.org" target="_blank">Lingtyp@listserv.linguistlist.<wbr>org</a><br>
<a href="http://listserv.linguistlist.org/mailman/listinfo/lingtyp" rel="noreferrer" target="_blank">http://listserv.linguistlist.o<wbr>rg/mailman/listinfo/lingtyp</a><br>
</div></div></blockquote><span class="m_451588021156008535HOEnZb"><font color="#888888">
<br>
-- <br>
Martin Haspelmath (<a href="mailto:haspelmath@shh.mpg.de" target="_blank">haspelmath@shh.mpg.de</a>)</font></span><span class="m_451588021156008535im m_451588021156008535HOEnZb"><br>
Max Planck Institute for the Science of Human History<br></span><span class="m_451588021156008535im m_451588021156008535HOEnZb">
Kahlaische Strasse 10 <br>
D-07745 Jena<br>
&<br>
Leipzig University<br>
IPF 141199<br>
Nikolaistrasse 6-10<br>
D-04109 Leipzig<br>
<br>
<br>
<br>
<br>
<br></span><div class="m_451588021156008535HOEnZb"><div class="m_451588021156008535h5">
______________________________<wbr>_________________<br>
Lingtyp mailing list<br>
<a href="mailto:Lingtyp@listserv.linguistlist.org" target="_blank">Lingtyp@listserv.linguistlist.<wbr>org</a><br>
<a href="http://listserv.linguistlist.org/mailman/listinfo/lingtyp" rel="noreferrer" target="_blank">http://listserv.linguistlist.o<wbr>rg/mailman/listinfo/lingtyp</a><br>
</div></div></blockquote></div><br></div></div>