Larry > Recent Status Updates

You can’t derive *He thinks John VP (C violations), because moving the pronoun pied-pipes [John,he] to SpecTP, and then John must be extracted, and moved back, impossible because of the rightward movement to a non-c-commanding position ban. But that account of C predicts Which pictures that John took yesterday did he destroy today is bad (because it starts He destroy which pictures John took), but it’s fine. So?

You can’t derive John thinks he’s intelligent and John considers him intelligent in the same way. Plausibly, John needs to stop at an intermediate position on its way out of [John,he] to SpecTP: SpecCP provides such a stop in the former, but nothing equivalent in the latter. In (1) John praises him, no intermediate stop before SpecTP for John, and above it, it can no longer be the subject, so John ≠ him in (1)

Even better: He is smart is UNGRAMMATICAL in isolation if “he” is left unstressed (stress being taken as equivalent to pronouncing a demonstrative, “this guy”). In John is famous. He’s smart too, the antecedent is John (as if the two sentences were conjoined by “and”). What do we make of the fact that it doesn’t c-command the pronoun?

Reading pronouns and their antecedents. What a paper. All pronoun-antecedent relations are established derivationally by leftward movement of the Spec of a doubling constituent [John,he]. This accounts for Control. It also reduces C (since *He(i) thinks John(i) is smart is ruled out simply because of the ban on movement of a non-Spec into a c-commanding position, violating the PIC).

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

The data on particles vs As indicate that “the complex form acts like a single unit with respect to aktionsart and argument structure, but is not a single lexical word”.

Some relevant difference between particles and result adverbs or adjectives: you can throw out the dog (to throw the dog out) but you can’t *paint red the barn (to paint the barn red). When a lexical item can’t “identify” Res°, that means that it can’t be far from Res°. Hence it seems that particles, unlike As, can identify Res°. Something like that

The solution? You get a direct resultative wwhen the subevents are identified by the same lexical item. For instance, “freeze” identifies solidity. You get an indirect one when the result is lexically foreign to the verb. For instance, “hoarse” is alien to “sing”.

It’s not that unaccusatives alone enter into resultative constructions: it’s that unaccusatives give rise to direct resultatives, where the process and the result are contemporaneous (“the lake froze solid”), while unergatives give rise to indirect resultatives, where not (“John sang himself hoarse”). What about “John bled Bill dry”? 4 this +other reasons, this dichotomy need revision (Ramchand)

Successive cyclic movement allows us to save the subjacency condition!

The difference btw Raising to object and Control is whether the surface DP complement is selected by the verb (Control: eg “convince”), or surfaces post-verbally as the embedded subject of the TP-complement (Raising: eg “believe”)

Object control vs ECM verbs: OCs select an (inf-)TP-, ECMs, an (inf-)CP-complement

Selection test: is “time” selected by “seems” or “elapse” in “Time seems to elapse AdvP”? Change the embedded VP: “*Time seems to swim AdvP”: bad. Is it because of “seems”? Compare: “Sharks seem to swim AdvP”: fine. Conclusion: “Time” seems to be selected by “elapse”, not “seems”.

On the 1hand, VPE strands Neg°, so NegP can’t adjoin to VP (ie [VP [Neg] [VP]]), it must take VP as its complement (ie [NegP [Neg’ [Neg] [VP]]]), but on the other, Neg doesn’t change the category of what it applies to (“alienable” is Adj, so is “inalienable”), arguing for adjunction. Tough

I didn’t know there was controversy on whether adjuncts are sisters to X’ or XP: interesting!!