"Recursion in the base"Katz-Postal HypothesisKP Hypothesis in the natural order of thingsThe Katz-Postal hypothesis says yes to both.GB theory said no to both.Chomsky's 1995 Minimalism yes to 1, and no to 2. StipulatesLexical itemsCausative predicates show uniformities of selection for theiThis means certain morphemes are marked as obligatorily coalWhy not discard the minor rules, and identify the predicatesPredicate Raising "applies to trees which terminate in semanConsequence: semantics might be interpretive (off Deep StruLakoff: "On generative semantics" (pub. 1970). An argument Claim: Not synonymous.Claim: Only reading is one in which NEG > many. This folloClaim: Scope of NEG and many corresponds to linear order. Claim: These data show that (17) holds of a block of rules tExample (24) shows NEG > many, but (25) cannot.Makes sense, if the negative component of dissuade starts ouBut the negation-raising rule now follows Shallow Structure.Lakoff's Conclusion: Lexical insertion is late, and Deep StLakoff "Global Rules"Global rules: Classical Greek agreementGlobal rules: One-pronominalizationWhere's the evidence that the rule in (35) exists and is latIn fact, that/those can pronominalize the plus a mass noun, ContractionSluicingPostal: "A Global Constraint on PronominalizationWhat ifsWhat if GS had developed the notion of a trigger for QR rathWhat if GS had developed trace theory and the copy theory ofWhat if GS had interspersed move and merge and tried to holdWhat if GS had investigated more carefully the nature of infSyntactic Models 4/19/06 Back to the Future: 1960-1970 1. "Recursion in the base" Suppose generalized transformations that embedded clauses in larger structures are strictly ordered w.r.t. other transformations. We find an ordering paradox: (1) Passive > there a. A teacher was reading a book . b. There was a teacher reading a book. c. A book was being read by a teacher. d. There was a book being read by a teacher. (2) Passive > there > Raising to object a. John believed there to be a teacher reading a book . b. John believed there to be a book being read by a teacher. (3) Passive > there > Raising to object > Passive a. There was believed (by John) to be a teacher reading a book . b. There was believed (by John) to be a book being read by a teacher (4) Passive > Raising to object > Passive > there [John believed someone to be reading a book -> Someone was believed to be reading a book -->] There was someone believed to be reading a book. Solution 1: cyclic application of transformations, with clausal embedding the last rule of the cycle, or... Solution 2: Clausal cycle + recursion in the base 2. Katz-Postal Hypothesis Selectional restrictions imposed before transformations such as passive ==> a pre-transformational, lexically governed "pure" representation of semantic properties projected from lexical items. Possibility: Some semantics is projected from Deep Structure (PS tree + inserted lexical items) Katz-Postal Hypothesis: All semantics is projected from Deep Structure. Therefore (corollary): Transformations do not "change meaning" (cf. Chomsky's 1990s criterion of "inclusiveness") Therefore (corollary): If a transformation seems to change meaning, either: (i) it doesn't exist (e.g. TNEG); or (ii) a trigger for the transformation, present in the base is what introduced the extra semantic content. Example: wh-movement (something->what + movement) triggered by a Q-morpheme. "Given a sentence for which a syntactic derivation is needed; look for simple paraphrases of the sentence which are not paraphrases by virtue of synonymous expressions; on finding them, construct grammatical rules that relate the original sentence and its paraphrases in such a way that each of these sentences has the same sequence of underlying P-markers. Of course, having constructed such rules, it is still necessary to find independent syntactic justification for them." (K&P 1964, 157; Newmeyer p. 71) 3. KP Hypothesis in the natural order of things Since it is a fact that the presence/absence of movement often correlates with semantic properties, there is a natural link between 1 and 2: 1. movement is triggered <--> 2. interpretation is pre-movement. • • • The Katz-Postal hypothesis says yes to both. GB theory said no to both. Chomsky's 1995 Minimalism yes to 1, and no to 2. Stipulates inclusiveness. Other possibilities? Cite as: David Pesetsky, course materials for 24.960 Syntactic Models, Spring 2006. MIT OpenCourseWare (http://ocw.mit.edu/), Massachusetts Institute of Technology. Downloaded on [DD Month YYYY].-2- 4. Lexical items Lakoff, Irregularity in Syntax (published1970, but written in 1965) • • Causative predicates show uniformities of selection for their subjects s.t. an Aspects model is simplified if these predicates are bimorphemic. This means certain morphemes are marked as obligatorily coalesced with others. The rules that do this are a subset of the "Minor Rules" (rules that "apply only to exceptions") - a category that also includes allomorphy rules (foot/feet) etc. Example: persuade means 'intend', but is constrained so it must meet the SD of a CAUSATIVE rule and must undergo it. (5) John [+V +CAUSATIVE +PRO] [NP [S Harry [persuade 'intend' m SD CAUSATIVE, m R CAUSATIVE. m ID-NP DEL ]] Example: dissuade means 'not intend'. (6) John [+V +CAUSATIVE +PRO] [NP [S Harry [persuade 'intend', NEGATIVE, m SD CAUSATIVE, m R CAUSATIVE. m ID-NP DEL ]] Example: kill is inchoative 'dead' marked to undergo obligatory causativization. Pesetsky (1994) Zero Syntax revives some of this. (7) Against biclausal analysis #John killed Bill on Tuesday by poisoning him on Sunday. (Fodor) (8) For biclausal analysis Sue closed the window for three minutes. (Klein) McCawley (1968) [among others] • • • Why not discard the minor rules, and identify the predicates of Deep Structure with elements more abstract than the words of the language: Predicate Raising "applies to trees which terminate in semantic matter rather than in lexical matter". Consequence: semantics might be interpretive (off Deep Structure) in name only. Where once we had Deep Structrure, now we have a semantic representation. "Lexical decomposition" (9) Sam likes lox more than herring or whitefish. (or = and) (10) Sam prefers lox to herring or whitefish (or = and) Therefore: