Docs / Studies / Studies.AlexeyenkoZeijlstra2025

Linearization of Complex Modifiers: Ways of (Dis)obeying the Head-Final Filter

Alexeyenko, S. & Zeijlstra, H. (2025). Linguistic Inquiry, 1–64

Source Bibliography Paper

The Head-Final Filter — prenominal modifiers contain no post-head material — both overgenerates (Greek and Russian allow A–XP–N) and undergenerates (Basque, Chácobo and Eastern Oromo bar the mirror order N–XP–A while Farsi, Atong and Kalaallisut allow it). The paper's replacement, the Modifier-Noun Adjacency Generalization, lets an XP separate an attributive adjective from its noun only if the adjective's agreement marker is shared with the predicative form and covers every nominal feature, or its attributivizer is morphophonologically independent of the adjective.

Both conditions are derived here from the paper's analysis: direct modification is available only to adjectives that are φ/κ-complete, some form of theirs being used both attributively and predicatively with an overt marker specified for every feature of the DP, case included (Language.PhiKappaComplete); every other language needs an attributivizer, which forces adjacency exactly when it is an adjectival affix, overt or null, under the Input Correspondence Principle (AttrStatus.Adjacent). Russian's forms are the long and short adjectives of the Fragment. The resulting Language.Possible encodes the decision trees and agrees with the sample of Table 3 on every language the paper discusses. The example rows carry only their lexical anchors (adjective, noun, dependent); the linear orders are computed from token positions (orderOf) and checked against the languages and the judgments.

References

Agreement marking

Attributivizers

L84–88

The languages of the paper

L103–107
L117–122
L127–142
L144–150
L152–158
L160–164
L177–181

Table 3

L196–202

The enough exception

L250–257

Rows

L272–279
L286–288
L305–316
L326–335