Whamit!

The Weekly Newsletter of MIT Linguistics

Syntax Square 10/22 - Yiannis Katochoritis (MIT)

Speaker: Yiannis Katochoritis (MIT)

Title: Long-distance pivot movement measures Phase Unlocking: Malagasy vs. Dinka

Time: Tuesday, October 22, 1pm – 2pm

Location: 32-D461


Abstract: Austronesian Malagasy and Nilotic Dinka Bor share non-trivial parallels: (i) an Austronesian voice/pivot system, where one (any) DP argument of the clause is promoted to a syntactically and discourse-wise prominent pivot position; (ii) clausebound promotion to pivot exhibits core properties of A-movement; (iii) promotion to pivot may be long-distance, so that a matrix pivot is thematically linked to an embedded gap; (iv) long-distance pivot movement proceeds via Phase Unlocking.

Based on fieldwork in Malagasy, I present two empirical findings: (i) The Malagasy-Dinka parallelism breaks down in long-distance pivot movement: in Dinka it retains its A-properties, whereas in Malagasy it suddenly only exhibits core A’-properties relative to arguments of the matrix clause; (ii) In Malagasy, embedded CP complements that either become the matrix pivot themselves or allow their own pivot to be extracted to the matrix clause, must first undergo (covert) A-movement to the matrix pivot position.

To explain their difference, I make the two following proposals: (i) Dinka and Malagasy differ in the structural height of Phase Unlocking: it happens by matrix v in Dinka, but by matrix C in Malagasy; (ii) their difference in the locus of Phase Unlocking is a conspiracy of three factors: (1) head V-movement in Dinka vs. phrasal roll-up VP-movement resulting in smuggling of the theme pivot in Malagasy; (2) different ’alignment’, contingent on the order of Agree and Merge at the v/Voice cycle: ACC-like in Dinka, split-ERG-like in Malagasy; (iii) Dinka’s composite pivot probe is not sensitive to partial A-intervention, whereas Malagasy’s is.

The analysis has several implications: (i) the Dinka pivot is a clausal topic with A-properties, dissociated from subjecthood, whereas the Malagasy pivot is both a topic and surface subject; (ii) Phase Unlocking may exceptionally undo the scope-island status of CPs for (covert) QR, allowing cross-clausal inverse scope; (iii) Phase Unlocking is not an ‘altruistic’ primitive of the grammar, but only occurs if the unlocking Agree operation is independently available locally: unlike Dinka, Malagasy v/Voice does not unlock the CP complement phase because it never gets the chance to directly Agree-interact with the internal argument during the derivation.