NP Movement Passives, Raising: When NPs are not in their theta positions.

Locality restriction on theta roles Leave(obligatory agent) agent i Adrian left Jo left her pencil *it left (where it is an expletive) Must be in same clause *John i thinks [that left] In particular can't mean: John i thinks he i left.

Locality Condition on Theta Roles Theta roles must be assigned within the clause same clause as the predicate that assigns them.

A Problem [John i is likely [ to leave]]. John is the subject of is likely. Is it theta marked by is likely???? –NO! (cf. it is like that John left) –more arguments for this later It is theta marked by leave!!! But it isn’t in the same clause! Yikes!

is likely –[[That John will leave] j is likely ] –It is likely [that John will leave] j Proposition j it is likely [ CP that john will leave ] No theta role on the subject of ‘is likely’  

In the wrong place! John is likely to leave John is theta marked by leave, but appears in the subject position of is likely, in violation of the locality constraint. The NP [John] is displaced from its theta position.

CP C’ C TP T’ T AP A’ A CP Ø likely is C’ C TP T’ Ø to VP John leave John gets its theta role in the specifier of the lower TP, but moves to the specifier of the higher TP. This is called Raising

WHY??? Well one thing we can observe, is the EPP holds. (the requirement that every sentence have a subject). The NP John moves to satisfy this requirement. This doesn’t account for examples like –*John is likely [that left]. –*It is likely John to leave. –Hmm. Movement correlates with finiteness.

Case Theory Case is a licenser. In order for the sentence to be grammatical, an NP must get case –Nominative case is assigned in the specifier of finite TP (note: FINITE) –Accusative case is assigned as the sister to the verb or preposition. These are the only two places you can get case

The Case Filter The case filter (a constraint that filters trees) –All NPs must be in case positions (spec,TP or sister to V/P)

CP C’ C TP T’ T AP A’ A CP Ø likely is C’ C TP T’ Ø to VP John leave Non-Finite: NOT a case assigner can’t get case here Finite: so can assign nominative case to John moves to get case in this position

Raising vs. ?? John is likely to leave John is eager to leave –John gets a theta role from leave –John also gets a theta role from is eager! (agent) *It is eager that John will leave. *It is eager for John to leave. –Violation of Theta Criterion???

PRO An empty NP which gets a Theta-role Distinguishing Raising from PRO can be difficult [More tests coming] –John is eager to leave. –John is likely to leave. Similar structures –John i is eager [ CP PRO i [T to ] leave ] –John i is likely [ CP t i [T to] leave]

Summary of Raising vs PRO Some NPs appear to be displaced from their theta assigners. This is caused by raising. Motivated by Case non-finite Infl can’t assign case NP moves to specifier of finite INFL Not all NP V [ ___ to leave] constructions are raising. Some involve PRO. it depends upon the theta properties of the main verb.

Passives Active [The linguist] kissed [the kitten] Agent theme Passive The kitten was kissed (by the linguist) Theme (agent) Active has agent and patient. Passive requires only a theme which is the subject

Passive Morphology The difference between passives and actives comes from the morphology. The addition of the passive morpheme seems to suppress the agent. kiss  kiss+en (  kissed) agent themetheme

Passive Morphology Hypothesis: The other thing the passive morphology does is suppress the verb’s ability to assign accusative case V’ V NP Acc V’ V+en NP Acc

Burzio's Generalization Burzio 1986 –A predicate that assigns no external theta role can assign no accusative case. –* It was kissed her. [no expletive rescue] EPP satisfied theta criterion satisfied Case Filter violated Possible Polish counterexamples: –Przeczitano ksiaszke. (Read book) –Read-3-sg-neuter Book-accusative

An Active CP C TP T’ T VP Ø NP V’ V NP agent theme assigned nom case by finite T assigned acc case by verb

A passive CP C TP T’ T VP Ø V’ V+en NP theme Moves to this position cannot be assigned accusative case underlyingly empty due to passive morpheme

Passives: A summary The passive morpheme Suppresses agent theta role Suppresses V’s ability to assign accusative case The theme NP can’t get case from the passive verb, so it moves (to the specifier of TP, where it can get nominative case.)

Passive: a Hybrid Lexical Component: An affix that changes Theta-roles A movement component Why BOTH? Why not just have the affix directly make the argument external? Passive out of embedded clauses –Wilma considers [John to be foolish] –John is considered [ t to be foolish]

Non-arguments can be Passive Subjects Passive out of embedded clauses –Wilma considers [John to be foolish] –John is considered [ t to be foolish] Consider does not assign a theta-role to John Foolish does So if the passive affix made John an external argument, how it could assign the right role in the lexicon?

NP Movement With both raising and passives, you are moving NPs, and in both situations you do this to get case on a caseless NP. This transformation is called “NP movement” The filter that forces NP movement is the case filter.