Larry’s Reviews > Elementary Syntactic Structures: Prospects of a Feature-Free Syntax > Status Update

If you take a and b and a labels b (ie a is the edge and b the complement), you can now introduce g, and no need to transfer (because a already has a distinct identity from g by virtue of labeling b), and d, but now you need to transfer so that g ≠ d. this makes {g, a} the complement domain of d (since b is now impenetrable, per PIC). It is a transitive phase. {b} is/was an intransitive phase

Richards 2010: no 2 elements of the same syntactic category can be transferred at the same time. So for a and b = 2 same category elements, the external systems will tolerate a syntax that combines them as long as they’re separated by a phase boundary. So phases (or cycles) must be small, otherwise the external systems will be overflown with more elements than they can distinguish per spell out point

Cyclic transfer or spell out allows the syntax to keep minimum structure in WM. What is spelled out is the complement of the phase or complement domain; what is not is the edge domain: the phase head, and the specifiers and adjuncts above. Therefore, what is available for further syntactic computation is a function of derivational timing, as previously spelled out phase complements become “impenetrable” (PIC)

Phases are cyclic transfers of unconstrainedly merged structure to externalization/interpretation -> if all asymmetries in UG are phasal, they will all be interpretive

The transformation of lexical into functional categories is documented by grammaticalization. Conclusion: all lexical categories are configurational flavors of transitive or intransitive phase derivations

CB shares with DM the idea that all selection facts must be reduced to configuration facts determined by simple computational principles. Syntax does not see anything about “sell” taking n arguments. Form and meaning are radically distinct. Link to Pietroski, according to whom lexicalization turns all concepts (of various addicities) into monadic concepts, combinable in boring pair-like ways

CB proposes a bare syntax: Merge operates on roots. The asymmetry is introduced by phases, with their edges sent to LTM. Categories are functional, and parametrically lexically realized. Eg verbs are nouns (so they have no real internal argument), merged as little ‘v’s upstairs.

But it’s also fascinating: CB thinks that the standard model of syntax with Merge driven by features doesn’t work, because there is no theory of features, and also because Merge likely builds up words, which are (or must start) featureless.

This is a very dense and insanely badly edited and written book