New version page

# MIT 24 960 - Lecture Notes

Documents in this Course

8 pages

9 pages

8 pages

Unformatted text preview:

What are conditions on transformations?A/A (p.85) as a condition on application of transformations.An example of a condition on the functioning of a specific rOne can also imagine conditions on the form of rules, i.e. cwith wh if possible,otherwise with that if Aux is Tense,otherwise with for.SSC/TSCThe TSC derives effects of the "Insertion Prohibition" (= phAssumption: Each other derived by each lowering The candi...each-movement also fails to apply downward into an NP wit...and wh-movement fails to apply (upward) from an NP with a..It-replacement (Tough Movement) applies out of a clause wi....Disjoint Reference Rule (Principle B): a rule of interp..not-many rule: (17c), if good, means "J's pix of many of Likewise (Lasnik, p.c. to NC) for not-enough association, whThe Strict Cycle / The COMP escape hatchTensed S means Tensed S'!( ranges over the phases cyclic nodes NP and "S"Thus movement to COMP within a single "S" is ok, even if it Example (12b) is ruled out because NP is a node with a "subjExample (22) is excluded because there is only one COMP escaSuperiority Condition/Subjacency ConditionESubjacency extensions: controversial judgmentsExtraction out of subjectsCreates a redundancy with the Subjacency condition where ( is an S contained in an NP, which includes sentential subjects (for Chomsky) as well as CNPC cases.[p. 108]Eliminate the redundancy: restrict the Subject Condition to ...does have an effect on each-movement. Condition (34) unmoLikewise for the RI of Disjoint Reference:"Proper containment" matters. It is ok to move each to a suNo raising to object"There is no necessity for a rule raising the subject of an Perhaps we should forbid string-vacuous movement. [pp. 113-"Specified subject" replaced by reference to a subject's convs. promiseOther cases where controller is relevantArbitrary null subject (() blocks each-movement and RI:But note the still puzzling "honey-dripping" problem on p. 1First use of the word "trace"The moved NP actually replaces PRO as the first step of moveReasonable, since the NP replaces it as its final step.To explain the badness of (45a-b), we might order each-movemSo the men becomes a subject in (45a) (replacing it) after tIn (45b), each cannot lower after Raising/it-replacement forSo far so good -- -but this story messes up on (45c), howeve"We might account for [(45b)] by assuming that when the NP JWe now can drop the ordering assumption about it-replacementSomething else blocks each-movement in (45a).An ordering solution? Each-movement precedes wh-movement on the highest cycle (so each-movement never sees an embedded clause without who in subject position.Alternative: wh-movement also leaves a "trace" [p. 135], anWhy SSC?Left-right asymmetriesThe complementizer systemTowards just two transformationsProposal 0: Each factor in the SD of a transformation is an Proposal 1: A factor changed by a rule must be either a fixProposal 2: Converse of proposal 1 -- If a term i of SD isReciprocals/DRBoth the "reciprocal rule" and DR are now rules of interpretSSC: "a condition on surface structures applying quite generTraces count/"Sentence Grammar""The trace [is] a bound variable, with all of the (relevant)"Movement of a phrase by a transformation leaves behind a trWhy is this ok? The each...other relation is not a propertyLikewise, the rule that relates a pronoun to its antecedent PRO counts/Restructuring (PRO-deletion)PRO countsPRO is deleted (constructions later identified [by Rizzi] asWh-movementSurface Structure! LF!"Free deletion in COMP" also yields the paradigm of English Zeitgeist 1: questioning transformationswe are convinced that movement structures are identifiable awe are convinced that such structures reflect processes thatwe are naturally led to reconsider the formal properties of Redundancies between "antecedent-trace" relations/chain formDistinctions between A-movement and A-bar movement and the c...plus lexical rules?Reassociation of internal argument with subject and reassociShows the same reassociation:involves a morpheme also attachable to nouns (a salaried empchanges category (V->A) -- adjectives disallow NP complementfeeds processes of derivational morphology like un- prefixatCategory-changing rules are morphological processes in the lLexical integrity: Transformations do not look inside lexicTowards LFG...Syntactic Models May 8, 2006 Conditions... "...On Transformations" 1. What are conditions on transformations?  A/A (p.85) as a condition on application of transformations. More precisely, a condition on the choice of factorization in case of ambiguity. Passive SD: X NP V NP Y 1 2 3 4 5 SC: 1 4 be - en 3 by+2 5 (1) A factorization blocked by A/A John's winning — the race — surprised — me — ø ==> 1 2 3 4 5 *John's winning I was surprised by the race.  An example of a condition on the functioning of a specific rule.  One can also imagine conditions on the form of rules, i.e. conditions on the grammars themselves. (2) Comp-substitution universal. No rule may fill COMP by rightward movement. (3) Background on Comp: C empty in the base. Filled by transformation: o with wh if possible, o otherwise with that if Aux is Tense, o otherwise with for. Normal proposal: The rules are language particular, the conditions on functioning universal. Tantalizing Possibility: The rules are universal (or some of them, anyway), and the conditions on functioning are language-specific? This is a key step towards Principles and Parameters, but then he retreats, but most of the paper in fact takes up arguably universal conditions on the functioning of rules.] 2. SSC/TSC Passive should apply blindly to: (4) a. I believe the dog to be hungry. b. I believe the dog is hungry. But it doesn't: (5) a. The dog is believed to be hungry. b. *The dog is believed is hungry. (6) Tensed S Condition: version 1 No rule can involve X, Y in the structure . . .X . . .[α . . . Y . . . ] where α is a tensed sentence. [20]  The TSC derives effects of the "Insertion Prohibition" (= phase impenetrability for insertion of "morphological material") for tensed sentences.  Assumption: Each other derived by each lowering The candidates each hated the others. ==> The candidates hated each other. (7) a. The candidates expected each other to win. b. *The candidates expected that

View Full Document