NP coordination

[Edith A Moravcsik]

On January 30, I posted a query on LINGTYP asking

   (a) whether there are languages where noun phrase coordination is
grammatical only if it involves certain classes of nouns (e.g., only
animates); or

   (b) whether there are languages where the shape of noun phrase
coordination constructions differs depending on the class
of the nouns involved (e.g. different conjunctions used with animates and
inanimates).

While, as pointed out by Martin Haspelmath, I failed to make this clear in
my message, what I meant by NP coordinations is meaning equivalents of
English "NP and NP".

I have received five responses, for which I would like to thank their
senders:

       Michael Daniel
       Alexandre Francois
       Martin Haspelmath
       Eva Lindstroem
       Paolo Ramat.

Here is a report on the contents of the messages followed by a summary and
the rationale for the query.

MISHA DANIEL pointed out that there is a comitative construction in
RUSSIAN involving two NP-s connected with the preposition _s_ 'with' which
has differring characteristics depending on whether both NP-s are
animate or only the first. In particular, the version involving two
animate noun phrases resembles NP conjoining in various ways. He referred
to three sources for further information:
      - an article by Paul Garde
      - an article by Misha Daniel himself
      - a talk by Kostja Kazenin which shows this both for Russian and for
        CHUVASH.
Kazenin's talk talk was given last week on February 3 2002 at the Third
Winter Typological School in Russia. I do not have the bibliographic
reference for the other two items.

ALEXANDRE FRANCOIS provided data on MWOTLAP (=Motlav), a member of the
North-Central Vanuatu group of Austronesian. (He has recently completed
a grammar of this language as his doctoral dissertation.) Mwotlap has
several ways of conjoining two noun phrases; one of these is restricted
to human noun phrases only and it is in fact the default construction when
the referents are humans. Other constraints on this construction:

    - it can only be used with two NPO-s, not three or more
    - each NP has to have a singular referent
    - the referents must be third person.

Here is schema of the construction::

      NP _koyo_ NP,

where _koyo_ is a third-person dual pronoun. For example:

      John koyo Mick
      John 3DU  Mick

meaning 'John and Mick'.

The word _koyo_ can also be used with a single noun phrase, in which case
it forms an associative dual construction:

      imam   koyo
      father 3DU

meaning 'Dad and Mom' or 'Dad and his friend'.

MARTIN HASPELMATH noted that there are construction types related to NP
coordination that do observe some constraints having to do with the
class of the nouns involved. Thus, in many languages, comitatives and
other inclusory constructions seem to be restricted to people or certain
persons. Martin referred to work on these by Linda Schwartz.

EVA LINDSTROEM reported on KUOT, a language of New Ireland in Papua New
Guinea, which she is working on for her PhD dissertation. In this
language, there are two conjunctions used to link coordinate NP-s. One is
used only when both NP-s have the same degree of animacy - whether high or
low. This conjunction is the preposition _ga_. E.g.:

    (a) James ga Joshua  'James and Joshua'

    (b) ua-ip        ga sagur-up           'taros and vegetables'
        taro-nonSING    vegetable-nonSING

Eva noted that _ga_ can also be translated as 'with'.

The other conjunction, used with NP-s of unequal animacy, is _un_,
a general coordinator which can also be translated as 'with' and it
usually implies that the item following _un_ is carried or held by the
first item. For example, in a story it is described that a bench with a
person sitting on it breaks and the description uses _un_: "the bench
_un_ the man".

PAOLO RAMAT commented on the more general issue of how similar two
conjuncts have to be. In PORTUGUESE, SPANISH, and CATALAN, when two
adverbs ending in _-ment(e)_ are conjoined, it is possible to leave out
one of the two (either first or second) identical endings; such as in
Portuguese:

    larga e fogosamente (for _largamente e fogosamente_)

or in Catalan:

    humilment e devota (for _humilment e devotament_)

The same is not possible in ITALIAN or FRENCH; e.g. Italian

   *umilmente e devota

Paolo pointed out that the elliptability of _mente_ in Portuguese,
Spanish, and Catalan is a sign of the original NP status of these adverbs.

In sum:

   - Re the first query:
     No language has surfaced in which NP coordination would be  possible
only for certain classes of nouns; although similar constructions, such as
comitatives, do observe animacy constraints in certain languages.

   - Re the second query:
     Yes, there are languages (Kuot, Mwotlap) where NP coordination is
formally different depending on the classes of the noun phrases involved.

My hypothesis which prompted the query, was that

   - there would be languages where only human or only animate noun
phrases could be cnojoined; and that
   - there would be languages where NP coordination would take different
forms depending on whether humans/animates are involved or not.

The reason I thought this is as follows. As Smith-Stark originally
proposed in his article in CLS 1974, pluralizability may not hold for all
kinds of nominals in some languages; and when this is the case, there is a
single hierarchy which predicts the cut-off points. On the top of the
hierarchy are first person pronouns, followed by second person pronouns,
followed by kin terms, then rational beings all the way down to
inanimates. However, as has been noted by many linguists, most recently
by DNS Bhat in his new book on pronouns, what is called the "plural" of
first person pronouns is not the same as nominal plurals (e.g. _tables_)
because, rather than involving a homogeneous set of members, it involves
the conjunction of different individuals: 'I and X...'.

But if so, then the top end of the Animacy Hierarchy speaks not about
pluralizability in the usual sense but about semantic conjoinability
saying that first (and then second) person pronouns are the most
conjoinable entities. If this is correct, then one might hupothesize that
conjoinability - just as pluralizability - may also be restricted to
certain classes of nominals in some languages with the splits also defined
by the Animacy Hierarchy so that the conjoinability of, say, inanimates would
imply the conjoinability of animates in a language; and the conjoinability
of, say, non-kin human terms would imply the conjoinability of kin terms.
Similary, one might expect that, if different formal devices are used for
NP coordination in a language, the classes of NP-s with which one or the
other these devices is used would also be animacy-related classes.

The fact that group-forming, fully or partially enumerative
semi-coordinative constructions (such as comitatives and associative
plurals) are (often or always) restricted to the upper end of the animacy
scale would appear to point in this direction.

Edith



   ************************************************************************
			 Edith A. Moravcsik
			 Department of Foreign Languages and Linguistics
			 University of Wisconsin-Milwaukee
		         Milwaukee, WI 53201-0413
                         USA

			 E-mail: edith at uwm.edu
		         Telephone: (414) 229-6794 /office/
				    (414) 332-0141 /home/
		         Fax: (414) 229-2741