Keywords: constraint-based approach, unification, construction, Move, Merge, HPSG, Simpler Syntax
To appear in Language 89(4).
This paper compares several current linguistic theories: at the extreme ends of the current views are Minimalist theories, which restrict themselves to the binary branching operations Move and Merge, and Simpler Syntax, which assumes flat structures and a surface oriented mapping between syntactic structures and grammatical functions.
I show that purely surface-oriented theories have problems in accounting for the relatedness of syntactic and morphological structures and for the iteration of valence changing processes and argue for a lexical analysis, as it can be found in current Minimalist theories, HPSG, and Categorial Grammar. I furthermore show that the Chomskian view on label computation is problematic for several reasons and should be given up in favour of explicit accounts like the one used in HPSG. I discuss problems for the analysis of complements and specifiers in Minimalist theories with special focus on Stabler's Minimalist Grammars. I argue that once all problems are fixed the resulting combinatorial rules are rather similar to what HPSG does.
As various proponents of more surface-oriented theories like Construction Grammar, Simpler Syntax, and HPSG pointed out, two types of binary branching, headed rules are not sufficient to account for the entirety of language, which leads to the conclusion that both research directions are right to a certain extend: there is need for (constraint-based versions of) Move and Merge and there is need for special phrasal constructions.
Draft of May 09, 2013: