Linguistics in the Netherlands - Volume 14, Issue 1, 1997

AbstractIn early, simultaneous analyses of signs, [a_contact] is a multivalent feature pertaining to the movement parameter (cf. Friedman 1976). In models that make use of sequential units (Liddell and Johnson 1989, Sandler 1989, Perlmutter 1989, van der Hulst 1993) the valence of [contact] can be reduced to one. In comparing two types of sequential models I will show that one of them - the No-Movement model -is more adequate in accounting for the contact types proposed in Friedman (1976). By examining the consequences of the representation of the contact types in the No-Movement model of van der Hulst (1993) and further developments thereof (Crasborn 1995,1996; van der Hulst 1995, 1996; van der Kooij 1994, 1996; van der Kooij and Crasborn 1996). I will show that contact is a redundant property, predictable from the place specification of the sign. Being phonologically redundant, variation and non-distinctiveness of contact is correctly predicted.