Skip to main content
School of Languages, Linguistics and Film

CANCELLED LingLunch | Safir (Rutgers University)

When: Friday, May 13, 2022, 1:00 PM - 2:00 PM
Where: ArtsTwo 3.16

Prof Safir will give a talk entitled Just Pair-merge.


Abstract. In current minimalist theory, there are two structure-building operations, simple merge, which forms sets, and pair-merge, which forms ordered pairs. Chomsky (2004) proposes that syntactic adjuncts are so-defined by the fact that they are always introduced by pair-merge.

However, it is intuitively appealing and consistent with minimalist aspirations to see if linguistic theory can do with just one rather than two merge operations. I will posit that only pair-merge is necessary to generate the full range of syntactic configurations that grammar requires. Labeling will reduce to assigning the syntactic label of the adjoined-to constituent to the constituent formed by pair-merge. No stipulations about the invisibility of copies or adjuncts will be necessary since there will be no labeling algorithm other than what is applied to the output of the ordered pair. (I do not assume that ordered pairs are linearly ordered, unlike Kayne, 1994, whose approach was essentially the first just pair-merge theory).

This means that there is no set/simple merge in the theory at all and no set representation structures. Tree representations are no longer ambiguous between sets and pairs at any given node. In a just-pair-merge theory, it becomes possible to derive the inherent asymmetry of c-command in a derivational definition, i.e., if X adjoins to Y then X c-commands Y (and not vice-versa).

Apparent motivations to cling to the Chomsky (2013,2015) labeling theory will be shown to be based on mistaken generalizations or unmotivated stipulations about adjuncts, agreement, EPP and the supposed “halting problem”.



Back to top