The binding constraints we assume are essentially those found in chapter 6.4 of [24]:

In [24] the notion o-command is defined in terms of the
SUBCAT list: a referential sign *X*
locally o-commands *Y* iff *X* precedes *Y* on SUBCAT. The general
notion o-command then holds between *X* and *Y* iff *X* locally
o-commands a *Z*, and *Z* dominates *Y*.

The fact that arguments can be shared in SUBCAT lists raises the question how we should interpret the principles of binding. For example, principle B can be understood at least in two ways:

- A personal pronoun must be locally o-free
*in all SUBCAT lists it is a member of* - A personal pronoun must be locally o-free
*in at least one SUBCAT list it is a member of*

We show that if the binding constraints are defined with respect to SUBCAT lists then both interpretations give rise to problems. As an alternative we will then use the feature ARG-S introduced in [11] to account for the binding data from Dutch.

1998-09-29