Wang and Sun (2026): Detaching Mandarin Classifiers from Nouns #

This file formalizes [WS26]'s account of three restrictions on Mandarin classifier phrases in [Adg25]'s mereological syntax. A classifier is not a projection of the noun but an object of its own, (7a): it subjoins to Q as Q's 1-part, the numeral is Q's 2-part, and the noun, if present, is the classifier's 1-part, (7b), (26). Dimensionality then leaves no room at Q, so a degree-modified adjective cannot sit between the numeral and the classifier, (29a) (subjoin_modifier_q), although it subjoins to the classifier as its 2-part, (28a) (subjoin_modifier_cl), or to D, (28b) (subjoin_modifier_d). The classifier spells out with Q along Q's complementation line, (26b) (nPart_one_q), so the classifier and the noun cannot be dislocated without the numeral. The particle de realises Mod, (7c): with de after the classifier the quantity phrase is the 1-part of a Mod that is D's 2-part, and the classifier is reached from D only across dimensions, (34)–(35) (not_clVisible_numClDeN). Visibility from D is the paper's structural correlate of the sortal reading, glasses as entities, against the mensural reading, glassfuls, (33) (reading). Nothing more, such as a second de-modifier, can subjoin to the filled D, (37a) (subjoin_numClDeN). In the measure use the classifier has no noun and Q is the 2-part of Deg, off the complementation line of the predicate, (45) (nPart_one_pred).

Extraction runs by Angular Locality, (40), on the clausal structure of (41): the object D is the 2-part of O, and O, v and Asp are the 1-part chain of C. A wh-numeral first subjoins to D and from there to C, (41) (wh_numeral); with a de-modifier in D's 2-part the numeral can neither subjoin to D nor, in any derivation, reach C, (39b), (42) (wh_numeral_modifier). Topicalisation of the whole D reaches C at once, (43c) (topic_d); the noun must subjoin to D first, stranding the numeral and classifier, (43d) (topic_n), which a modifier as the classifier's 2-part permits, (44a) (topic_n_modifier_cl), and a modifier as D's 2-part blocks for good, (44b) (topic_n_modifier_d).

Implementation notes #

The nominal structures and the clause of (41) are Parthood values on one type of objects, so visibility from D, Dimensionality at Q and D, and extraction are read off the same parthood relations, and each claim is checked by decide. In the clause, Mod stands for the whole modifier, whose internal structure, (27), is given only in the nominal structures.
Linearisation, (7e), and the collective spell-out of Q with Cl are not represented; the adjacency of the numeral and the classifier is stated as Dimensionality at Q, which is what the paper derives it from, and the ban on modifiers between D and Num, (31), is not derived.
The topicalisation of the whole D in (43c) is proved for the structure without a modifier.

Wang and Sun (2026): Detaching Mandarin Classifiers from Nouns #

Implementation notes #

References #

Structures, (26)–(28), (34), (45) #

Modification, (28)–(29), (37) #

Spell-out lines, (26b), (45b) #

Visibility and the reading of a classifier, (33)–(36) #

Extraction by Angular Locality, (39)–(44) #