[Lingtyp] odd clitic behaviours

Mon Dec 6 20:05:47 UTC 2021

Thanks, Arnold and Peter, for the interesting critical comments!

I completely agree with Peter Arkadiev that "if linguistics is to deal 
with complexity and diversity of linguistic structures, its 
terminological apparatus cannot be simplistic" – yes, we need a lot of 
terms for all this complexity, in fact far more than most people make 
use of (which is why I keep proposing new terms).

But I do not fully agree with Arnold Zwicky that "Our job is to discover 
what the relevant concepts are in the domain in question and then to 
provide names for them" – we have tried this, but it turns out that it 
doesn't work well for general linguistics. Different languages have 
fairly different "relevant concepts" (= language-particular categories), 
so the comparison of languages requires a distinct set of comparative 
concepts. For example, we cannot readily describe Arabic or Chinese with 
concepts derived from Ancient Greek (such as "(en)clitic").

De facto, however, linguists do use quite a few Latin-derived (and 
Greek-derived) terms for (comparison of) languages from around the 
world, i.e. as comparative concepts. What should we take these terms 
(e.g. /accusative, plural, preterite, imperative, affix, passive/) to 
mean? They have a fairly clear meaning in Latin, but what do they mean 
*in general*?

It does not seem to make sense to pose this as a research question – we 
cannot study languages in order to find out what "accusative" or 
"passive" means. We attach these labels to languages around the world 
because we think that they are generally understood, but often we are 
not particularly clear about what that meaning is. We know what a 
stereotypical "accusative" or a stereotypical "passive" (or a 
stereotypical "clitic") is, but if there are no boundaries, we cannot 
decide what to do in non-stereotypical cases (e.g. in the case of "odd 
clitic behaviours" in Quechua, as studied by Alexander Rice).

Though Peter Arkadiev says that "we do not need apparently "precise" 
definitions which end up delimiting arbitrary classes of things having 
nothing in common apart from the randomly chosen property "defining" 
them", I do not see what the alternative is – simply *avoiding* the use 
of tradional terms? In practice, this will not happen, as people will 
continue to talk about /accusative, plural, preterite, imperative, 
affix, passive/, etc. So I think it's better to try to provide simple 
and clear definitions that can be used in textbooks. (Often, of course, 
language-particular classes will not map perfectly onto these 
definitions, as is illustrated by Riccardo Giomi's example of the 
Italian promiscuous diminutive /-icchi-/.)

Arnold Zwicky says "What I'd like to avoid is disputes over whether some 
element E in some language variety L is *really, truly" a clitic", and 
indeed, we have too many such fruitless disputes – I have a long list in 
my 2007 paper on pre-established categories 
(https://zenodo.org/record/1133882, §3.3). But why are such disputes 
about "clitics" fruitless? I'd say it's because there is no clear 
definition of "clitic", while at the same time, many people *think* that 
there is some general concept (a building block of UG?) behind this 
term. But this need not be the case: "Clitic" may not be more than a 
term that has been handed down to us by tradition (Ancient Greek 
grammar, and then Nida 1946, which shortened "enclitic" to "clitic"). 
Since this term is an accident of the history of linguistics, giving it 
an arbitrary definition seems just the right step to me – in this way, 
its arbitrariness becomes apparent to everyone. (If "clitic" is used as 
an "umbrella term", by contrast, there is no way in which it can be used 
for precise communication, and we might as well not use it at all.)

It would be great if it turned out that "properties not listed in the 
definition are predictable from the ones that are" (as Arnold notes), 
and in my 2015 paper on clitics (which was greatly inspired by Arnold's 
work), I do discuss this issue (§7, see 
https://zenodo.org/record/4550427). But this is not necessary – "clitic" 
is a commonly used technical term, and as such, it should have a clear 
definition (or should not be used). Quite generally, I do not think that 
vague and stereotype-based "umbrella terms" are needed in science, 
though they are of course ubiquitous in everyday language.

Best wishes,
Martin

P.S. I have more discussion of the general issues of terminological 
precision in my 2021 paper: https://ling.auf.net/lingbuzz/005489

Am 06.12.21 um 20:13 schrieb Arnold M. Zwicky:
>
>> On 06/12/2021 16:25, Martin Haspelmath wrote:
>>> Yes, Zwicky's 1994 idea that "clitic" is an "umbrella term" was adopted by Spencer & Luís (2012) – but this is not a CLAIM.
>>>
>>> If the question is how to use a term, we make *terminological choices* – and my proposal was to make the choice that a clitic is defined as "a non-affix non-root bound form". This would give the term "clitic" a precise meaning (as a general-typological concept).
> This seems incomprehensively bizarre to me. Our job is to discover what the relevant concepts are in the domain in question and then to provide names for them (which could be more or less arbitrary, taken from familiar terms, created via metaphor, or whatever). But I'm baffled by your apparent position that history provides us with a term that has been more or less useful in the past, so our job is to arbitrarily assign it to one of the relevant concepts, with the consequence that this term is then *inapplicable* to -- inappropriately used for -- every one of the other relevant concepts.
>
> If this is an (arbitrary) prescription about how the term should be used within the community of linguists (the relevant set of language users in ths case), it's just terrible -- guaranteed to sow confusion and misunderstanding. It's Humpty-Dumpty's "[a word] means justi what I choose [that is, what *I* choose] it to mean -- more more and no less". I, Arnold Zwicky, am free to declare that what "clitic" means is ''hoofed mammal', so that if you want to talk to me you have to use it that way too. (Actually, I use "ungulate" for that purpose, and some people use "hoofed mammal" and even more people, faced with the task of explaining the concept they're talking about, give an ostentive definition ending with the ominous "etc.".  But nobody's going to buy my insistence that these creatures taken together are called, technically, "clitics" and that the Tagalog second-position elements are *not* clitics.)
>
> I coined the technical term "umbrella term" to provide some sort of continuity with the history of our field for terms like "clitic", covering an assortment of loosely similar concepts -- each of which deserves its own label.
>
> Perhaps you mean to claim that all the things under the "clitic" umbrella are in a family-resemblance relationship with one another (like things under the "game" umbrella)  and that there are central members of the family -- clitics *par excellence*, as iit were. But that's an analysis designed for ordinary language, not technical language, so I'm not sure how the *cognitive* significance of centrality would carry over.
>
> What I'd like to avoid is disputes over whether some element E in some language variety L is *really, truly" a clitic -- with reference to the Martin Haspelmath definition of what a clitic really, truly is.
>
> Perhaps you want to claim that your choice of a definition is not arbitrary, not "merely terminological", but signals that the particular definition you have chosen is one for a concept that is empirically rich, in the sense that ("interesting") properties not listed in the definition are predictable from the ones that are. But you haven't actually claimed that.
>
> I'm afraid that I'm going to have to stop here, with the comments above. At this point in my life I don't have the time for extended dialogue on *anything*, even if it might be fruitful.
>
> Arnold
>
> ******
> No, Martin, we do not need apparently "precise" definitions which end 
> up delimiting arbitrary classes of things having nothing in common 
> apart from the randomly chosen property "defining" them. I find this 
> approach neither productive nor scientific. If linguistics is to deal 
> with complexity and diversity of linguistic structures, its 
> terminological apparatus cannot be as simplistic as that. I apologise 
> for putting it so bluntly.
> Best wishes,
> Peter

-- 
Martin Haspelmath
Max Planck Institute for Evolutionary Anthropology
Deutscher Platz 6
D-04103 Leipzig
https://www.eva.mpg.de/linguistic-and-cultural-evolution/staff/martin-haspelmath/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/lingtyp/attachments/20211206/791bee16/attachment.htm>