If all merge is pair-merge
Ken Safir
August 2019

Current versions of minimalist theorizing assume two structure-building operations, simple merge and pair-merge and a means of providing a syntactic label to the output of simple merge. Simple merge takes two syntactic objects (SOs) and includes them in the same set to form a single SO. The labeling algorithm (LA) of Chomsky (2013) assigns the label of a head to the SO formed by simple merge with a non-head. When two non-terminals are simple-merged, there is no determinate label possible, unless the two SOs simple-merged share features that can constitute a label. Pair-merge forms ordered pairs of SOs, with the asymmetry that the first member of the pair SO1 is adjoined to the second SO2 and the label of SO2 becomes the label of the single SO that is formed. I propose that if we assume just a simplified version of pair-merge, both simple merge and LA are unnecessary.
Format: [ docx ]
Reference: lingbuzz/004762
(please use that when you cite this article)
Published in:
keywords: merge, pair-merge, labeling algorithm, adjunction, syntax
Downloaded:133 times


[ edit this article | back to article list ]