query re dyad constructions

[Nick Evans]

Dear colleagues, I have been overwhelmed with 
very interesting comments and discussion on 
yesterday's query, some posted to lingtyp, some 
just to me - so much so that I need to rework my 
article to accommodate many of the comments. And 
I would be grateful for further comments and 
examples.

Meanwhile, though, I am posting this provisional 
reply, concentrating particularly on Paolo Ramat 
and Gideon Goldenberg's discussion posted on the 
list, so as to clarify the relationship of dyad 
constructions to others such as associative dual 
and plurals, dvandva constructions, and various 
types of elliptical constructions, such as the 
'dualis a potiori' [al-muthannà 'alà l-taghlib] 
from Classical Arabic.

To deal first with Gideon's reply, that dyads are 
simply a subset of the various dualis a potiori. 
It is certainly the case that the referential 
range of dyads may overlap with these, as in an 
elliptical dualis based on 'father(DU)', for 
example, which may mean 'father and child' 
(dyadic) but also 'father and mother', the latter 
NOT dyadic because the designated relationship 
(be father to) does not hold within the pair 
(i.e. it doesn't fit my definition 'two such that 
one is [in social relationship] K to the other'). 
In particular languages, some such 'elliptic 
dual' examples may then be translation 
equivalents of dyads in languages that have 
these, but others will not be. Simply classifying 
them as dualis misses the point that many 
languages have constructions specialized for 
exactly the meaning I characterise as dyads, and 
which could not be used for other pairs (sun + 
moon, father + mother) expressible by elliptic 
duals, associative duals (e.g. father-AssocDu for 
'father and characteristically accompanying 
other, incl. father and mother'). (I discuss the 
relationship of these constructions to dyads in 
my Cologne Arbeitspapier (p. 37-38), which can be 
downloaded from
http://www.uni-koeln.de/phil-fak/ifl/asw/forschung/ap/dyad.pdf )

Relating them to duals also biases our view of 
dyad constructions by focussing on ONE 
construction type which they overlap with, at the 
expense of others (e.g. reciprocals in some 
languages, possessives in others, proprietives in 
others, and so forth).

Rather similar points can be made with respect to 
the Vedic dvandva constructions, which Paolo 
Ramat has mentioned in his reply, e.g. mitr 
...váru...(u) 'Mitra and Varuna'. Again, in 
principle, this construction could well include 
combinations (if you got it with e.g. 
father...son) that would be translated with dyads 
in languages that have them. However, again, this 
would be a matter of overlapping semantic ranges, 
rather than of either one being a subtype of the 
other: the dvandva construction is 
characterisable as something like 'X and Y, 
characteristically paired', while the dyad is 
'two, such that one is K to the other'. And by 
defining dyads in a way that is independent of 
form, we can remain non-committal about how the 
dyad construction is built up: it might be 
derived from a  lexical root referring to one of 
the entities (e.g. Kayardild ngamathu-ngarrba), 
be non-segmentable  (in a lexical dyad like the 
Mianmin lum example), or both (in Cantonese 
constructions like the following, courtesy of 
Stephen Matthews:
  loeng5 hing1-dai6   (from hing1 'elder brother' + dai6-dai6 'younger brother')
two brother-brother
'two brothers'

I hope this clarifies why I see it as worthwhile 
to characterise dyad constructions as a distinct 
type (though obviously related to a range of 
other phenomena): the exact semantic 
configuration recurs in a large enough number of 
languages to make it worthwhile to have a crisp 
semantic definition that covers exactly this 
semantic range.

I will post a revised version of the article once 
I have had time to assimilate more of the 
information that is pouring in; meanwhile, thanks 
to all contributors and please send me more 
information if you have it.

Best, Nick Evans

>Dear Nick,
>enclosed you may find the text of your article 
>with a suggestion I have added to it.
>
>Best wishes,
>Paolo
>
>----- Original Message -----
>From: <mailto:nrde at UNIMELB.EDU.AU>Nick Evans
>To: 
><mailto:LINGTYP at LISTSERV.LINGUISTLIST.ORG>LINGTYP at LISTSERV.LINGUISTLIST.ORG
>Sent: Sunday, May 23, 2004 4:56 AM
>Subject: query re dyad constructions
>
>Dear Colleagues, I am currently preparing an 
>article on 'Dyad constructions' for the 
>Encyclopaedia of Language and Linguistics. I 
>enclose pdf and Word versions of the draft of 
>this article, as well as a pdf file listing the 
>languages known to me so far in which Dyad 
>constructions are attested. The paragraph pasted 
>in below summarizes what dyad constructions are.
>
>So far I have the impression that the worldwide 
>distribution of dyad constructions is heavily 
>skewed, with a strong concentration in the 
>Western Pacific (especially Australia, 
>Austronesian, Papuan) with sporadic attestation 
>in western north America, the Amazon, the 
>Caucasus, Siberia and Khoisan. However, this may 
>be an artefact of my own areas of knowledge, of 
>the book holdings in our library here, or of 
>particular descriptive traditions. I would 
>therefore be very grateful if any subscribers to 
>this list were able to extend the list of 
>languages (and of course further data on 
>construction types etc.) beyond what I have 
>included in these files.
>
>Yours gratefully, Nick Evans
>
>Dyad constructions denote relationally-linked 
>groups of the type 'pair/group of brothers' or 
>'mother and child(ren)'. They may be formed by 
>morphological derivation, as with Kayardild 
>(Australian) ngamathu-ngarrb 'mother and child' 
>< ngamathu 'mother', or they may be 
>unanalyseable lexical roots , such as Mianmin 
>(Papuan) lum  'father and child'. Though they 
>most commonly refer to pairs, as in the above 
>examples, they may also refer to larger groups, 
>e.g. Mianmin lum-wal  'father and children'. 
>Where a dual-plural contrast exists, the dual 
>dyad is usually formally unmarked (§3). Though 
>the above languages have dedicated dyad forms, 
>it is more common for dyadic constructions to 
>overlap formally with other categories, most 
>commonly reciprocals, proprietive or possessive 
>constructions, or pair markers (§4). Dyad 
>constructions display a notably skewed 
>geographical distribution, being concentrated in 
>the language families of the Western Pacific, 
>with only scattered occurrences elsewhere (§5).
>
>
>Attachment converted: Macintosh 
>HD:dyads.encycl.art (1).doc (WDBN/MSWD) 
>(0006DCF4)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/lingtyp/attachments/20040524/10976819/attachment.htm>