[Lingtyp] Right node raising cross-linguistically

Martin Haspelmath martin_haspelmath at eva.mpg.de
Wed Sep 29 19:18:53 UTC 2021

Dear all,

Ellipsis in coordination was a hot topic in typology in the 1970s 
(following Ross's famous 1970 paper where the term "gapping" was 
introduced), and there was a chapter on this in Mallinson & Blake (1981).

The most recent treatment that I am aware of is in my 2007 paper on 
coordination (https://zenodo.org/record/1133876), where I discuss 
ellipsis in §6 on 9 pages, trying to avoid decisions on constituent 
structure. I report on Sanders's (1977) paper which made some 
interesting claims about types of ellipsis in different languages. 
Unfortunately, neither Sanders's old paper nor this part of my more 
recent paper have been followed up on – maybe because "ellipsis" is not 
easy to identify to begin with in many languages?

As Adam Tallman has noted (here and in various other venues), 
constituent structure is not always determinate because different 
phenomena may argue for different structures – this is of course an 
experience familiar to typologists. So we'd need comparative concepts 
that do not make reference to abstract and language-dependent notions 
such as constituent structure.

Now Adam Singerman might object: "it's not clear that RNR is a "thing" 
to be explained in of itself. So the issue might be moot."

I completely agree that RNR (what I call /right periphery ellipsis/ 
[RPE] in 2007) may not be a "unified phenomenon" in the sense that the 
*mechanisms* that different languages (or different speakers) use to 
state the generalizations may be rather different. And this is a 
familiar type of claim – people have argued for the "disunity" of VSO 
order, for the disunity of ergativity, and so on.

But this is the very basis for the observation that cross-linguistic 
comparison cannot be based on language-particular mechanisms, but must 
be based on comparative concepts that abstract away from the mechanisms. 
Generative linguists have typically had the hunch that we need to figure 
out these mechanisms if we want to do useful typology, because whatever 
explanations we come up with must reside in these mechanisms ("universal 
grammar" etc.).

But there is an alternative view/option, which has recently also been 
advanced by some (nominally) generative linguists such as Julie Anne 
Legate and Amy Rose Deal (as discussed in this blogpost: 
https://dlc.hypotheses.org/2481). For example, Deal (2016) says, in her 
discussion of the disunity of split ergativity, that "hierarchy effects 
*ultimately must arise external to the grammar itself*, from the 
organization of human cognition and communication – a conclusion in line 
with various approaches that locate the origin of these effects 

So if we find cross-linguistic generalizations about coordination 
ellipsis that are not reflections of a unique mechanism, they may still 
have a unified explanation ("external to the grammar", i.e. independent 
of language-particular mechanisms).


Am 29.09.21 um 18:23 schrieb Adam Singerman:
> Dear Adam,
> I don't believe that the RNR [right node raising] issue is as damning for syntactic theory
> as you do (this shouldn't surprise anyone who's talked with the both
> of us, as we generally have different takes on the successes/failures
> of syntactic theory). On the contrary, I think the RNR issue nicely
> illustrates the fact that there will be empirical generalizations in
> individual languages that require some kind of formalization to be
> stated in the first place, whatever the ultimate verdict is on the
> formal apparatus that the generalizations are first couched in. Even
> if we take Chaves's analysis as correct (namely, that RNR is really
> the conflation of three separate phenomena: VP/N' deletion,
> extraposition, backward periphery deletion), we will still need a
> formal apparatus to define these phenomena. A formal syntactician
> would probably say that the crosslinguistic version of RNR is not
> capturable as a comparative concept in the sense of Haspelmath because
> what defines RNR — whether it's taken to be a phenomenon in of itself
> or a conflation of three separate phenomena, as Chaves argues — is
> still ultimately formal. So this is probably a good example of how
> there are parts of speakers' linguistic competence which need to be
> described, analyzed and theorized, but which do not lend themselves in
> an obvious way to cross-linguistic comparison outside of a formal
> framework. Unless of course I have misunderstood Martin's point about
> comparative concepts...?
> Abraços,
> Adam
> _______________________________________________
> Lingtyp mailing list
> Lingtyp at listserv.linguistlist.org
> http://listserv.linguistlist.org/mailman/listinfo/lingtyp

Martin Haspelmath
Max Planck Institute for Evolutionary Anthropology
Deutscher Platz 6
D-04103 Leipzig

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/lingtyp/attachments/20210929/384b4fdc/attachment.htm>

More information about the Lingtyp mailing list