Remarks on Marking

This paper calls for a reexamination of the Marking Theory of HPSG, which in its standard form involves a considerable amount of dedicated formal machinery, but which proves to be inapplicable for most types of grammatical marking. As an alternative, it is demonstrated that head-marker phrases can be reanalyzed as head-complement structures, with the marking element treated as the syntactic head. This approach allows the elimination of all markingspecific formal apparatus, with the exception of the attribute MARKING, which percolates as an ordinary HEAD feature, and whose function is significantly expanded. The proposed approach allows marking elements to be related to other lexical heads (prepositions, in particular), and marking constructions are better integrated in the grammar, rather than being grouped into an exceptional class of head-marker phrases. 1 Assumptions and Consequences The Marking Theory in Pollard and Sag (1994) is introduced primarily in order to analyze a single construction—that-marked finite clauses in English—although a few other potentially relevant constructions are mentioned in passing. Focusing on that-clauses for the moment, consider the simple example (1) Kim fears [that Sandy knows the truth]. A satisfactory analysis of the bracketed phrase must meet a number of requirements. First, it must license the binary combination of that (which I will refer to pre-theoretically as the “marking element,” as opposed to “marker”) with the sentence Sandy knows the truth (I ignore other conceivable, but implausible syntactic analyses of this construction): (2) S