Psycholinguistics II The Dynamics of Language

1. Psycholinguistics IIThe Dynamics of Language LING 641 Colin Phillips, Jeff Lidz

2. Last semester… Abstraction

3. Abstraction • Abstraction is valuable • Provides representational power • Provides representational freedom • Abstraction is costly • Linguistic representations are more distant from experience • This places a burden on the learner - motivation for innate knowledge • This places a burden on comprehension/production systems • (and it makes it harder to know what to look for in the brain)

4. This semester… • Progression in focus • ‘What are the mental representations?’ • ‘How do the representations change?’ • Inference and dynamics

5. Multiple Time Resolutions days - years seconds milliseconds

6. Recurring Themes • How does information/input lead to change? • Highly general inferences using abstract categories … or not • Inferences based on robust generalizations … or not • Risk-taking in inferences … and recovery from error • Integration of information/input across levels of representationDoes certain information have priority?

7. So what does this all have to do with linguistics?

8. Language Learning • Taken seriously as an evaluation measure for theories • Children are good at language learning (and adults are not) • Much data is similar to data of linguistics (binary classification) • Logical Problem of Language Acquisition • Inferences • Abstract, powerful system learned from finite input • Linguists are impressed by intricate details that must be learned … and unlearned • Inferences across constructions [‘parameter setting’] • Inferences across levels of representation [‘bootstrapping’] • Inferences across multiple exemplars [‘statistical/bayesian learning’]

9. Language Learning • Current challenges • Much work on learning focuses on ‘toy’ language systems • Parametric learning models don’t work very well (yet) • Parametric descriptions of language receive little attention • Much known about individual time-slices of language development; far less about learning

11. Speaking and Understanding • Taken less seriously as an evaluation measure for theories

12. “It has sometimes been argued that linguistic theory must meet the empirical condition that it account for the ease and rapidity of parsing. But parsing does not, in fact, have these properties. […] In general, it is not the case that language is readily usable or ‘designed for use.’” (Chomsky & Lasnik, 1993, p. 18)

13. Chomsky (1965) • “Linguistic theory is concerned primarily with an ideal speaker-listener […] who knows its language perfectly, and is unaffected by such grammatically irrelevant conditions as memory limitations, distractions, shifts of attention, and interest, and errors (random or characteristic) in applying his knowledge of language in actual performance. […]We thus make a fundamental distinction between competence (the speaker-hearer’s knowledge of his language) and performance (the actual use of language in concrete situations). Only under the idealization set forth in the preceding paragraph is performance a direct reflection of competence.” (pp. 3-4) • “When we say that a sentence has a certain derivation with respect to a particular generative grammar, we say nothing about how the speaker or hearer might proceed, in some practical or efficient way, to construct such a derivation. These questions belong to the theory of language use - the theory of performance.” (p. 9)

14. Speaking and Understanding • Taken less seriously as an evaluation measure for theories • Are humans good at it? • The boat floated down the river sank. • The man who the senator that the reporter praised resigned.

15. Speaking and Understanding • Taken less seriously as an evaluation measure for theories • Are humans good at it? • The boat floated down the river sank. • The man who the senator that the reporter praised resigned. • How directly is the grammar implicated in speaking and understanding?

16. Standard View 324 697+ ? 217 x 32 = ? arithmetic

17. Standard View specialized algorithm specialized algorithm 324 697+ ? 217 x 32 = ? arithmetic

18. Standard View specialized algorithm specialized algorithm 324 697+ ? 217 x 32 = ? ? arithmetic something deeper

19. Standard View specialized algorithm specialized algorithm understanding speaking grammaticalknowledge,competence language recursive characterization ofwell-formed expressions

20. Standard View specialized algorithm specialized algorithm understanding speaking precisebut ill-adapted toreal-time operation grammaticalknowledge,competence language recursive characterization ofwell-formed expressions

21. Standard View specialized algorithm specialized algorithm understanding speaking well-adapted toreal-time operationbut maybe inaccurate grammaticalknowledge,competence language recursive characterization ofwell-formed expressions

22. Origins of the Standard View

23. J.A. Fodor, M.F. Garrett, T.G. Bever. (1974). The Psychology of Language. New York: McGraw Hill.

25. Townsend & Bever (2001, ch. 2) • “Linguists made a firm point of insisting that, at most, a grammar was a model of competence - that is, what the speaker knows. This was contrasted with effects of performance, actual systems of language behaviors such as speaking and understanding. Part of the motive for this distinction was the observation that sentences can be intuitively ‘grammatical’ while being difficult to understand, and conversely.”

26. Townsend & Bever (2001, ch. 2) • “…Despite this distinction the syntactic model had great appeal as a model of the processes we carry out when we talk and listen. It was tempting to postulate that the theory of what we know is a theory of what we do, thus answering two questions simultaneously.1. What do we know when we know a language?2. What do we do when we use what we know?

27. Townsend & Bever (2001, ch. 2) • “…It was assumed that this knowledge is linked to behavior in such a way that every syntactic operation corresponds to a psychological process. The hypothesis linking language behavior and knowledge was that they are identical.

28. Townsend & Bever (2001, ch. 2) • “…It was assumed that this knowledge is linked to behavior in such a way that every syntactic operation corresponds to a psychological process. The hypothesis linking language behavior and knowledge was that they are identical.” Note:(i) No indication of how psychological processes realized (ii) No commitment on what syntactic operations count (iii) No claim that syntactic processes are the only processesin language comprehension.

29. Miller & Chomsky (1963) • ‘The psychological plausibility of a transformational model of the language user would be strengthened, of course, if it could be shown that our performance on tasks requiring an appreciation of the structure of transformed sentences is some function of the nature, number and complexity of the grammatical transformations involved.’ (Miller & Chomsky 1963: p. 481)

30. Miller (1962) 1. Mary hit Mark. K(ernel)2. Mary did not hit Mark. N3. Mark was hit by Mary. P4. Did Mary hit Mark? Q5. Mark was not hit by Mary. NP6. Didn’t Mary hit Mark? NQ7. Was Mark hit by Mary? PQ8. Wasn’t Mark hit by Mary? PNQ

31. Miller (1962) Transformational Cube

32. Townsend & Bever (2001, ch. 2) • “The initial results were breathtaking. The amount of time it takes to produce a sentence, given another variant of it, is a function of the distance between them on the sentence cube. (Miller & McKean 1964).”“…It is hard to convey how exciting these developments were. It appeared that there was to be a continuing direct connection between linguistic and psychological research. […] The golden age had arrived.”

33. Derivational Theory of Complexity • Miller & McKean (1964): Matching sentences with the same meaning or ‘kernel’ • Joe warned the old woman. KThe old woman was warned by Joe. P 1.65s • Joe warned the old woman. KJoe didn’t warn the old woman. N 1.40s • Joe warned the old woman. KThe old woman wasn’t warned by Joe. PN 3.12s

34. “The moral of this experience is clear. Cognitive science made progress by separating the question of what people understand and say from how they understand and say it. The straightforward attempt to use the grammatical model directly as a processing model failed. The question of what humans know about language is not only distinct from how children learn it, it is distinct from how adults use it.” (Townsend & Bever, 2001)

35. … and aren’t production and comprehension different from one another?

36. Grammar in Parsing • Real-time application of island constraints • Real-time application of binding constraintsetc.

37. Speaking and Understanding • Taken less seriously as an evaluation measure for theories • Are humans good at it? • The boat floated down the river sank. • The man who the senator that the reporter praised resigned. • How directly is the grammar implicated in speaking and understanding? • Is there a Logical Problem of Language Use

38. Powerful Inferences in Parsing • Japanese numeral classifiers • san-satsu hon3-cl book • san-nin gakusei3-cl students • Numeral classifiers and Relative Clauses • John-ga san-satsu-no [RC … ] hon-o yondaJohn-nom 3-cl [RC … ] book-acc read

39. Powerful Inferences in Parsing • Japanese numeral classifiers • san-satsu hon3-cl book • san-nin gakusei3-cl students • Numeral classifiers and Relative Clauses • John-ga san-satsu-no [RC gakusei-ga … ] hon-o yondaJohn-nom 3-cl [RCstudent-nom… ] book-acc read

40. NP CL NP san satsu-no CP NP N IP C NP VP N gakusei-ga

41. Speaking and Understanding • Taken less seriously as an evaluation measure for theories • Are humans good at it? • The boat floated down the river sank. • The man who the senator that the reporter praised resigned. • How directly is the grammar implicated in speaking and understanding? • Is there a Logical Problem of Language Use • Data from speaking/understanding looks quite different from standard data of linguistics

42. Speaking & Understanding • My view… • The classic arguments are serious, but inaccurate • It seems feasible to build models of language that capture the speed and accuracy of parsing and production • We have the tools to (begin to…) track detailed real-time computational operations • If, on the other hand, the standard view is correct, then a detailed, testable account of linguistic competence as a ‘computational system’ may lie beyond our reach

43. Speaking & Understanding • Similar concerns to those found in learning • How do speakers use information/input to update representations? • How abstract/general are the inferential processes? • How risky are the inferences, and how can a speaker recover from error?

44. Memory for Language • Why it has received little theoretical interest • Long-term memory thought of as a list, lacking generality or generativity • Working memory identified with resource limitations that mask the underlying system • Why it deserves more interest • Learning: what is stored affects what generalizations are formed • Encoding: do speakers have one or multiple ways of encoding structure? • Abstraction: studies of memory (‘syntactic priming’) turn out to have important implications for questions of abstraction • Recovery from error: is the evidence destroyed?