[Lingtyp] wordhood

Dryer, Matthew dryer at buffalo.edu
Fri Nov 17 12:22:43 UTC 2017

I think that these five phenomena can be considered a single comparative concept only if they are exhaustive in the sense that there cannot be any other phenomena that are arguably instances of the intended higher order concept they are instances of. In other words, it is not enough that the phenomena share something in common. Otherwise, we really have a disjunctive definition in disguise. After all, Martin’s meggle people all share the property of being people. It is not immediately obvious, however, that these five phenomena share some property to the exclusion of all others.

But I am puzzled by Martin’s asking whether there is evidence for such an abstract property. It’s not clear what it would mean for there to be evidence for such a property. One needs to give an argument, perhaps, but surely not evidence.


From: Lingtyp <lingtyp-bounces at listserv.linguistlist.org<mailto:lingtyp-bounces at listserv.linguistlist.org>> on behalf of Martin Haspelmath <haspelmath at shh.mpg.de<mailto:haspelmath at shh.mpg.de>>
Date: Wednesday, November 15, 2017 at 7:53 PM
To: "lingtyp at listserv.linguistlist.org<mailto:lingtyp at listserv.linguistlist.org>" <lingtyp at listserv.linguistlist.org<mailto:lingtyp at listserv.linguistlist.org>>
Subject: Re: [Lingtyp] wordhood

Christian Lehmann's paper on concepts and categories can be read as
supporting the idea that comparative concepts can also be used for
describing languages, but in a recent extended exchange with him, I
understood that he actually supports the idea that structural
descriptions and tertia comparationis (= comparative concepts) are on a
different level.

In any event, we seem to agree mostly that disjunctive definitions are
"pointless" or even "incoherent" (as I have put it).

David wants to argue that the diverse phenomena in different languages
can be seen as reflecting some more abstract formal property. So for
example, all of (1)-(5) could be seen to reflect phonological
"bondedness" (or "same prosodic domain"):

(1) vowel harmony
(2) tone sandhi
(3) progressive ATR assimilation
(4) vowel reduction
(5) preoralization of final nasals

But my question is: Is there independent evidence for the existence of
this abstract formal property?

To go back to my absurd example of "meggle people": I could set up an
abstract feature [megglehood] and claim that the various observable
properties (Huawei smartphone, Mendelssohn music, taxi-driving) are just
manifestations of this abstract feature.

So is there reason to think that (1)-(5) are different? Well, for
(1)-(3), one can give a substantive reason for their coherence: They are
all assimilatory (though I'm not sure about tone sandhi). Intuitively,
elements that are assimilated to each other have a strong "bond", so I
would find it OK to regard them as manifestations of an abstract feature

But (4)-(5) are not – they are merely restricted to particular domains,
like all other grammatical rules. To the extent that different
phonological rules appeal to the same domain, this gives us reasons to
say that the domain has a privileged existence within the language
(though even domains that are relevant only for one rule must be part of
the grammar). But I don't see how such domains can serve as comparative
concepts, if the reasons for setting them up are different from language
to language.

(Again, as I said earlier: Comparative concepts must eventually be based
on substance – phonetic substance or semantic substance.
Language-particular abstract concepts such as domains are not good for


On 15.11.17 10:25, Peter Arkadiev wrote:
Thank you, David, this methodological (and theoretical) point of is utmost importance. I guess the point Christian Lehmann makes in his recent paper https://www.christianlehmann.eu/publ/lehmann_ling_concepts_categories.pdf is very similar in spirit.

Best regards,


Peter Arkadiev, PhD
Institute of Slavic Studies
Russian Academy of Sciences
Leninsky prospekt 32-A 119991 Moscow
peterarkadiev at yandex.ru<mailto:peterarkadiev at yandex.ru>

15.11.2017, 09:07, "David Gil" <gil at shh.mpg.de<mailto:gil at shh.mpg.de>>:
In response to Bill's ...

On 14/11/2017 23:37, William Croft wrote:
   A definition “variably interpreted in each language” is a disjunctive
   definition. If I use fact A to define ‘word’ in Language X, fact B to
   define ‘word’ in Language Y, and fact C to define ‘word’ in Language
   Z, then ‘word’ is defined as “defined by either A or B or C”. Or else
   ‘word’ means something different in Languages X, Y and Z, i.e. it is a
   language-specific concept, and the fact that it’s called ‘word’ in
   each language is just a coincidence.
Sorry, but I just don't get this. If language X has a significant
pattern involving, say, vowel harmony and some idiosyncratic rule
preoralizing final nasals, language Y has a structurally somewhat
different pattern involving tone sandhi and progressive ATR
assimilation, while language Z makes use of patterns of stress and vowel
reduction to define particular phonological domains, then they're
obviously as different from each other as we all know languages to be.
So yes, if John describes X as having an X-Word, Mary describes Y as
having a Y-Word, and Bill describes Z as having a Z-Word, then these are
indeed three language-specific and (in one sense of the word)
incommensurate notions.

And sure, defining a would-be comparative concept of word disjunctively,
as X-Word OR Y-Word OR Z-Word OR ... would be unrevealing and rather
pointless. (I was going to say "uninteresting", but that sounded too
Chomskyan.) However, and here's the rub, there is no principled reason
why it should not be possible to take John, Mary and Bill's descriptions
of X, Y and Z and abstract away from them a shared formal property which
we then might choose to refer to as a comparative concept of word. Yes,
the comparative concept of word would be "variably interpreted in each
language", but no, the definition of the comparative concept would not
involve disjunctions; it would simply obtain at a higher level of
abstraction than the language-specific phenomena that formed the basis
for the original three language-specific descriptions. (Such
abstractions are the bread and butter of our work as typologists, just
stop and think for a moment how many cycles of abstraction are involved
in a comparative concept such as "passive".) And crucially, it need not
necessarily involve the kind of "clustering" that Martin was taking about.

This is what I am trying to do with my proposed definition of
comparative-concept word. Granted, the proof of the pudding is in the
eating ... and I'm still a bit of a way from getting my definition to
work, by which I mean being both implementable and interesting. But my
point here and now is not to defend my (or any other) definition of
word, but merely to argue that there is nothing incoherent in the
attempt to define a comparative concept of word — even for those of us
(myself included) who share a radically relativist view of linguistic


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<mailto:gil at shh.mpg.de>
Office Phone (Germany): +49-3641686834
Mobile Phone (Indonesia): +62-81281162816

Lingtyp mailing list
Lingtyp at listserv.linguistlist.org<mailto:Lingtyp at listserv.linguistlist.org>
Lingtyp mailing list
Lingtyp at listserv.linguistlist.org<mailto:Lingtyp at listserv.linguistlist.org>

Martin Haspelmath (haspelmath at shh.mpg.de<mailto:haspelmath at shh.mpg.de>)
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

Lingtyp mailing list
Lingtyp at listserv.linguistlist.org<mailto:Lingtyp at listserv.linguistlist.org>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/lingtyp/attachments/20171117/8c75f205/attachment.htm>

More information about the Lingtyp mailing list