Editor for this issue: <>
ilf.uio.no>Paul Kershaw and David Powers had some interesting observations on NP coordination a few days ago. The first one concerns the impossibility of a plural determiner to occur in front of a coordinated NP consisting of singular conjuncts: (1) *these man and woman I do not know of any language in which this kind of construction is acceptable, and believe the reason must be that the plurality of a nominal phrase is a functional feature which, if not already present in the individual conjuncts, can only be resolved at a functional projection level. In other words, a DP can be interpreted as plural even if it contains a conjunction phrase consisting of singular conjuncts, but an NP can only be interpreted in terms of the actual features of its daughters. It is not surprising, then, that a plural determiner like *these* cannot take as its complement a singular NP, even if this NP is of the category CoP[NP, sg, sg]. The whole nominal construction is only plural at the functional projection level. Similarly, a determiner with a feature X cannot have a complement of conflicting features since a non-functional category is not the place for resolution of feature conflicts. Thus, the Norwegian (2a) is not possible, because there is a gender conflict between the determiner and the second conjunct. (2) a. *en (M) stol (M) og bord (N) a chair and table b. en (M) stol (M) og lampe (M) a chair and lamp I have discussed elsewhere (e.g. in my dissertation) the phenomenon of Unbalanced Coordination. Languages vary a lot with respect to the extent that they allow it. However, the examples of comparable constructions there, i.e. constructions in which there is a resolved feature conflict, all happen at a level above the functional, nominal internal projection, as in the following example from the Cushitic language Qafar (taken from Hayward and Corbett 1988), which can choose between unbalanced coordination (agreement with one conjunct only) or balanced coordination (agreement with both conjuncts via number resolution). It therefore seems reasonable to suggest that resolution of feature conflicts can only occur at a functional level, making sub-functional conflicts impossible. (3) a. lubak-kee yanguli yumbulle /yumbullen lion.ABS.-and hyena.NOM.M.SG. were.seen.M.SG/PL The second observation is that two conjuncts cannot always share the same determiner, even when there is no feature conflict, as in (4). (4) a. *a man and woman b. the man and woman c. this man and woman I do not know why there is a difference here. It certainly is language specific (since the Norwegian equivalent of (4a) is fine), but it is also lexically determined, since (4b,c) are better or even fine. The third observation concerns the fact that there is an unresolved conflict between the conjuncts with respect to verb agreement in some cases: (5) a. ?John or I am happy b. ?John or I is happy c. *John or I are happy As I have mentioned to Paul Kershaw earlier, I don't think this case is quite comparable with the above ones, since here, there is disjunction instead of conjunction, so that number resolution typically picks out one conjunct only. However, one should expect either conjunct to be allowed to agree with the verb in a fully acceptable way. Instead, all alternatives are odd or out. I agree with him that the reason (5a) is better than (5b) is "phonological", in that it is a string that we are used to hearing. However, I don't think this is the reason that (1) is bad, since (6) is also unacceptable: (6) *these men and woman More generally, I do not think adjacency is a syntactic factor that plays a role in coordination, since (as I have shown in my dissertation) what is important in unbalanced coordination is what is specifier and what is complement in CoP, and not what happens to be adjacent to what. Of course there is often a great deal of overlap, but not always. Janne B.J.Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue