Chan and Shen (2026): Conditions on wh-the-hell licensing #

A Singlish single wh-question fronts its wh-phrase, moves it to an intermediate Spec-CP, or leaves it in situ, and Chan and Shen's acceptability experiment, a pair of 2×2 designs in the manner of Sprouse et al., finds that the-hell survives the first two and not the third: the in-situ comparison shows the superadditive interaction of a penalty, the partial-movement one only additive costs. This follows from two independent pieces. The-hell carries an unvalued point-of-view feature checked against the operator in matrix C, which ascribes its negative attitude (surprise, ignorance, doubt of every answer) to the speaker, after Chou; and as a modifier adjoined to the wh-head it moves only with the wh-phrase, after Merchant. The three strategies are three structures (7), (14), (8), each a chain of copies of the wh-phrase: full movement pronounces the top of the chain in matrix Spec-CP; partial movement pronounces a copy in the intermediate Spec-CP below a deleted one there, the covert second step; and an in-situ wh-phrase is bound unselectively, a single copy that never leaves, after Sato and Ngui. The-hell is licensed exactly when the top of its host's chain is matrix Spec-CP. Islands constrain the links of a chain, so extraction from a complex NP fails under full and partial movement and not in situ ((11), (15), and the Malay pattern Cole and Hermon observe); covert movement of the in-situ wh-phrase in Huang's sense, a deleted copy in matrix Spec-CP over the pronounced one, would cross the island and license the-hell in situ, against (11b) and (4d). Mandarin daodi moves on its own and so tolerates an in-situ host.

The intervention account of den Dikken and Giannakidou licenses the-hell in the immediate scope of the question operator, after Linebarger, and so admits Singlish in-situ questions, where nothing intervenes; the attitude-phrase account of Vu and Lohiniva, after Huang and Ochi, base-generates the-hell in the matrix clause and so cannot generate the partial-movement order. The paper's Table 5 is the three accounts against the data.

Implementation notes #

The rows are mapped to structures by their strategy and whether they sit in a complex NP; the structures are schematic Singlish questions (2), (11), and the English, Malay and Mandarin rows share them, since what the accounts read, the chain of the wh-phrase and where it is pronounced, is the same. The-hell adjoined to the wh-head is not built: it is carried by the wh-phrase, so its position is its host's. The intervention account reads the number of interveners the rows record. The schematic clauses have no T, so C selects the verb phrase and each verb selects its subject as well as its complement, and every constituent has a selection or raising head.

Chan and Shen (2026): Conditions on wh-the-hell licensing #

Implementation notes #

References #

The three strategies (7), (8), (14) #

Negative attitude ascription (§3.2–3.3) #

Rival accounts (§3.4) #

The data #