1 / 32

Linguistic Theory

Linguistic Theory. Lecture 6 Generalisation. Recap: generalisation from 60s to 70s. The first generative model had rules and phenomena virtually in a one to one relation with each other: Phrase structure rules produced basic sentences (kernel sentences)

reed-dale
Download Presentation

Linguistic Theory

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Linguistic Theory Lecture 6 Generalisation

  2. Recap: generalisation from 60s to 70s • The first generative model had rules and phenomena virtually in a one to one relation with each other: • Phrase structure rules produced basic sentences (kernel sentences) • Transformations produced more complex sentneces

  3. Grammar Language PS rules kernel sentences T1 T2 T3 T4 … sentence type 1 sentence type 2 sentence type 3 sentence type 4 …

  4. Developments in the 1960s allowed rules to become more general: • Phrase structure grammar did not produce a specific set of sentences, but was applicable to all sentences • Transformations, by being constrained, were not construction specific but applied to a larger number of structures

  5. Grammar Language PS Rules T1 T2 T3 constraints

  6. Advantages • The grammar is simpler that the phenomena that it accounts for: it is not just a reflection of the facts and so helps us to understand the phenomena • Generality  explanation • Helps to account for language acquisition • Less to be learned • More data to learn from • Restricted hypothesis space

  7. Transition to the 80s – the final steps • There were further generalisations made from Extended Standard Theory at the end of the 1970s • In the phrase structure component • In the transformational component

  8. Complete generalisation of phrase structure • The structuralists had noted that some phrases have heads (endocentric): • A word which can replace a whole phrase: • Poor John left • John left • On the basis of this evidence the noun was taken to be the head and hence the phrase was a noun phrase • But not all phrases are endocentric • In particular sentences are exocentric • Poor John left = sentence • Poor John  sentence • Left  sentence

  9. Chomsky’s phrase structure rules were meant to formalise the structuralists notion of constituent structure • But they did not capture the notion of a head: • NP  Det N • NP  P V • As both of these are possible rules, the fact that only the first is an actual rule has no explanation • Thus the fact that phrases contain a word of the same category is just accidental

  10. The X-bar notation, introduced in 1970, solved this problem: • XP  ... X ... • But it was assumed that the structuralists were right in that not all phrases are endocentric • Thus X-bar theory was only applied to NP/N, VP/V and AP/A • Soon after it was extended to PP/P • S and S were seen as different (exocentric) • Also, Det, INFL and COMP were seen as non-X-bar elements

  11. S COMP S NP INFL VP Det N’ may V’ the N V man leave

  12. Thus in 1970 the situation was assumed to be thus: • In one way this is fine as the X-bar elements have traditionally been seen as a separate class from the non-X-bar elements • major categories : minor categories • But this distinction has tended to be made on semantic grounds • Major categories carry more semantic information than minor categories • Given that both the structuralists and generativists claim that syntax and semantics are independent, this should have been worrying

  13. Moreover, the status of X-bar theory was taken to be a restriction on Phrase Structure rules, rather than the basis of the phrase structure component. • This was because there were unexplained peculiarities of certain phrases • these were not lexical – too general • therefore they were grammatical, but restricted to certain categories • E.g. V and P take NP complements, N and A do not: • see [John] • after [John] • * a picture [John] • * fond [John]

  14. If phrase structure rules were as general as X-bar rules, then all these would be predicted to be grammatical • So, we assume that there are category specific rules: • V’  V NP • P’  P NP • These rules are acceptable because they fit the general X-bar template: • X’  X YP • Other rules are out because they don’t: • V’  Det N • V’  V’’ NP • Yet other rules are out because they are not part of the grammar (though they could be): • N’  N NP • A’  A NP

  15. But some rules are OK even though they don’t conform to X-bar, because they are exceptions: • S  COMP S • S  NP INFL VP • So things are not entirely general

  16. Stowell 1981 argued that we can get rid of the phrase structure part of the grammar because category specific phenomena can be explained by some other (more general) part of the grammar: • * picture John • * fond John • These can both be explained on the assumption that the complement of a noun or an adjective are not Case positions • Hence these constructions violate the Case filter • * NP, if NP is not in a Case position

  17. Lexicon PS Rules X-bar Deep Structure Transformations Constraints Surface Structure • Thus the grammar went from:

  18. Lexicon X-bar Deep Structure Transformations Constraints Surface Structure • To: • But the fact that X-bar rules are not completely general is even more suspicious now Case Filter

  19. The head of the sentence • During the 1970s, the idea that sentences were not exocentric had been considered • Jackendoff 1977 proposed that the head of the sentence was V and so sentences were really VPs: • V’’’ NP V’’John Aux V’ may V NP meet Mary

  20. Chomsky proposed that INFL was the head of the clause • if the inflection is finite the clause is finite • If the inflection is non-finite the clause is non-finite • Thus sentences are IPs (first propose in Stowell 1981) • IPNP I’ I VP • This has a certain appeal

  21. But, look what it does to our theory: • This is worse than before as the set of X-bar elements do not form a natural class (e.g. Major categories) • Therefore the pressure is on to make X-bar completely general

  22. COMP and S • X-bar can be generalised to the S and the complementiser through the assumption that the former is the head of the latter: • CPwhP C’ C IP • Again, this has some appeal • Recall that wh-phrases did not go in the same position as the complementiser • Also complementisers play a role in determining the declarative or interrogative nature of the clause

  23. Determiners and DP • In 1986 it was first proposed that determiners also fall within X-bar theory and are heads • NP DPDet N’  D’ a N PP D NP picture of John a N’ N PP picture of John

  24. This addresses an original problem with the structuralist assumptions: • Recall that the head was defined as a word which can replace the phrase (has the same distribution) • Poor John left • John left • Thus the noun is the head of the phrase (= NP) • But: • That dog left • * dog left • That left • Thus the determiner is the head of the phrase (= DP)

  25. Completely generalising transformations • By the 1970s there were two transformations: • NP movement • Wh-movement • These could not be reduced as they conformed to different principles: • NP movement: • Tensed S condition • Specified Subject condition • Wh-movement • Crossover constraint

  26. But, with the introduction of trace theory, a new approach to these phenomena became possible

  27. Getting rid of TSC and SSC • Note the following similarities: • John1 seems [ t1 to like Mary] • * John1 seems [ t1 likes Mary] • * John1 seems [Mary to like t1] • John1 expects [himself1 to like Mary] • * John1 expects [himself1 will like Mary] • * John1 expects [Mary to like himself1]

  28. We can capture this similarity if we assume that the trace of an NP has the same properties as reflexive pronouns • Whatever governs the behaviour of pronouns (Binding theory) will govern the behaviour of NP traces and we do not need transformation specific constraints

  29. Note the following similarities: • Who t1 thinks [he1 likes Mary] • * who does he1 think [ t1 likes Mary] • John1 thinks [he1 likes Mary] • He1 thinks [John1 likes Mary]

  30. We account for this similarity by assuming that wh-traces have similar properties to referential phrases. • Whatever governs the behaviour of referential phrases (Binding theory) will govern the behaviour of wh-traces and we do not need transformation specific constraints

  31. Thus if there are no transformation specific constraints, we no longer need specific transformations: • NP movement: • Move an NP to an NP position • Wh-movement: • Move a wh-phrase to a wh-position • Move  • Move any category to any position

  32. X-bar Theory Lexicon D-structure Move  Constraints S-structure Case Theory Binding Theory

More Related