Towards a structure-based typology
Hubert Haider
January 2021

The predictive power of Greenbergian word order typology can be strengthened. The type assignment of a language ought to amount to an empirically valid prediction. The type-assigned language is predicted to display the core properties of the type. The accuracy of such predictions can be enhanced if word order patterns are joined with properties of phrase structures that underlie the respective patterns. Presently, Greenbergian types are weak predictors, for several reasons. First, they are defined in terms of the linearization patterns of minimal clauses. Such patterns are structurally ambiguous and therefore, cross-linguistically, associated with incompatible typological correlates. Second, the crucial notion "subject" needs to be defined structurally, not semantically, in order to correctly assort the corresponding patterns in different alignment types. Third, the clause type does not fully determine the phrase-structure type. A more reliable predictor is the phrase-structure type, that is, head-final, head-initial, and crucially, alterable positioning of the head within its phrase in combination with clause-structure types. The paper lists and analyses eight syntactic properties. These properties correlate directly with the grammatically determined, canonical positioning of the head within its phrase. Hence they can serve as diagnostics for more accurate type assignments, with SOV, SVO, VSO, and {S,V,O} (viz. freely head-positioning) as major syntactic clause types. The predictive accuracy of phrase-structure-based taxonomy is demonstrably higher than linearization-based ones.
Format: [ pdf ]
Reference: lingbuzz/005617
(please use that when you cite this article)
Published in: (submitted)
keywords: typology, sov, osv, svo, greenberg, slavic, russian, free word order, no dominant order, adverbials, adjuncts, ergative, alignment, superiority, verbal cluster, definition of subject, syntactic universals, head positions, epp, expletive
previous versions: v3 [January 2021]
v2 [December 2020]
v1 [December 2020]
Downloaded:188 times


[ edit this article | back to article list ]