In Polish a single category-less root may be found in two or more verbs
which belong to different verbal classes. This observation suggests that
the established approach to deriving the exponents of verbal classes in
Polish, i.e. Czaykowska-Higgins (1998), was on the wrong track in
assuming that the identity of thematic vowels is dependent on the
specification of the root. This paper discusses several types of root ambiguity
and puts forward an alternative account in which it is the
specification of the category-defining V-heads as well as the aspectual
characteristics and argument structure that drive the exponence of Polish
verbs. It is also postulated that the traditional account which assumes the
existence of 8 verbal classes may be replaced with an approach which
postulates the existence of 5 classes. I show that not only does such an
approach capture certain affinities between Polish verbal classes, but also
predicts which verb (sub)classes will never share one and the same root.