IRCS Technical Reports Series

Document Type

Technical Report

Date of this Version

March 1994


University of Pennsylvania Institute for Research in Cognitive Science Technical Report No. IRCS-94-05.


In what proved to be probably the most influential Principles-and-Parameters manuscript of the last year, Kayne (1993) has proposed 1) a Linear Correspondence Axiom which together with a particular definition of (asymmetric) c-command is supposed to allow only SVO and OVS as underlying word orders and 2) an abstract beginning node asymmetrically c-commanding all other nodes which is supposed to further exclude OVS so that one arrives at the conclusion that SVO constitutes the universal underlying word order. Below, I argue against this conclusion on both theoretical and empirical grounds. While the Linear Correspondence Axiom has desirable effects on clause structure (cf. section 3), neither it nor the assumption of an abstract beginning node has any effects on word order.1 In particular, Kayne's system actually allows not only SVO and OVS, but also SOV and VOS (cf. section 4). Moreover, it will not do to simply stipulate SVO as the universal underlying word order since word order in German, a language traditionally analyzed as being underlyingly SOV, cannot be adequately treated in the universal SVO approach, especially when it is compared with word order in Yiddish, a closely related SVO language (cf. section 5). The next section introduces the theoretical machinery of Kayne (1993). It should be read even by those who are already familiar with Kayne's paper, since the exposition of the linear ordering concept given in section 2 will help the reader to understand the central theoretical arguments in section 4.



Date Posted: 18 September 2006