next up previous
Next: Binding constraints on ARG-S Up: Verb Clusters and Binding Previous: Binding Constraints

Some linguistic data

It is typically assumed ([18]) that verb-cluster constructions do not witness clause-union effects as far as binding phenomena are concerned. This should imply that sentences of the kind dat NP NP NP ziet wassen witness the same binding facts as sentences of the kind dat NP ziet dat NP NP wast, where ziet (sees) is a verb-raiser and wassen/wast (to wash/washes) is an ordinary transitive verb. This is not true. Consider the following examples:4

\eenumsentence{
\item \hspace{8pt}\shortex{6}
{ dat & Jan & Piet$_i$\ & zichzelf...
...n$_i$\ hem$_i$\ Piet ziet wassen}
{\em that Jan$_i$\ sees him$_i$\ wash Piet}
}

In these examples, the impossibility of (33h) illustrates a clause-union effect:

\enumsentence{
\shortex{7}
{dat & Jan$_i$\ & ziet & dat & hij$_i$\ & Piet & wast...
... & he$_i$\ & Piet & washes}
{\em that Jan$_i$\ sees that he$_i$\ washes Piet}
}

Let us attempt to analyse the sentences above solely in terms of SUBCAT. Since we are assuming that non-finite verbs do not contain a subject on their SUBCAT list, such an account is utterly impossible. However, we will illustrate that problems remain, even if we tentatively assume that subjects are present on the subcat list of non-finite verbs.

In (33a-d) there are two relevant SUBCAT lists: the matrix one essentially is a list containing three noun phrases, the most oblique one being the anaphor: $ \langle$Jan, Piet, pro$ \rangle$; and the embedded one which is similar except that the (matrix) subject is not there: $ \langle$Piet, pro$ \rangle$.

(33a) is predicted to be grammatical: principle A is satisfied (in both lists). (33b) is ruled out, because principle B is violated (in both lists). In order to properly rule out (33c) we must assume that principle A holds universally because only in the embedded SUBCAT list principle A is violated. In (33d) we must assume that principle B applies in an existential way: in the embedded SUBCAT list the pronoun is locally free.

In (33e-h) the two SUBCAT lists are $ \langle$Jan, pro, Piet$ \rangle$ and $ \langle$ pro, Piet$ \rangle$, respectively. (33e) and (33f) are ungrammatical (principle C violation). (33g) is allowed in both a universal and existential interpretation of the binding constraints. Finally (33h) can only be ruled out under a universal interpretation of principle B.

Thus we have constructed a paradox concerning principle B. (33d) requires an existential interpretation of principle B, but (33h) requires a universal interpretation. Note that principle A requires a universal interpretation.


next up previous
Next: Binding constraints on ARG-S Up: Verb Clusters and Binding Previous: Binding Constraints
Noord G.J.M. van
1998-09-29