1018_tpdn_bottup

1018_tpdn_bottup - Language comprehension Lecture 5:...

Info iconThis preview shows pages 1–9. Sign up to view the full content.

View Full Document Right Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: Language comprehension Lecture 5: Experience / frequency and ambiguity resolution; the serial / parallel issue 9.591; 24.945 October 18, 2004 Ted Gibson 9.591 Course so far Lecture 1: Experimental methods; Informational constraints affecting sentence comprehension: Lexical frequency, plausibility, context and syntax; Modularity in sentence comprehension. Lecture 2: Resources and sentence complexity. The complexity of unambiguous sentences. Lecture 3: Working memory and sentence comprehension. Lecture 4: Ambiguity resolution: Resources; structural frequencies. Lecture 5: Experience / frequency and ambiguity resolution; the serial / parallel issue. Sentence processing: Recap thus far Multiple factors are involved in processing unambiguous sentences and in ambiguity resolution. How exactly the factors are represented and processed are open questions: Minimal Attachment & Late Closure vs. Storage and Integration (DLT)? Evidence supporting the DLT in ambiguity resolution Frequency: What kind of frequency is the human sentence processor sensitive to? We dont know yet. Generalization: Corpus frequencies correlate with RTs, when the appropriate level of corpus analysis is used. (e.g., Desmet & Gibson, 2003) What is the time course of information integration? Modular (syntax-first)? Or non-modular? Is the parser serial or constrained parallel? To do today: The serial / parallel question Lexical / structural frequencies and sentence comprehension: Is the human parser sensitive to structural frequencies? Is the human parser sensitive to syntactic / lexical frequencies contingent on a syntactic / discourse / world knowledge context? The serial / parallel question Does the human parser retain exactly one structure for the input at each parse state (serial), or does it sometimes retain more than one (parallel) ? See e.g., Gibson & Pearlmutter (2000); Lewis (2000) for discussion. Serial versus parallel parsing Frazier (1978), Frazier & Rayner (1982), etc.: Comprehenders show clear preference asymmetries. Bill argued the position... ...forcefully. ...was incorrect. An unlimited parallel processor predicts no asymmetry. Alternative to unlimited parallel parsing: Multiple alternatives are ranked according to various criteria ( Ranked-Parallel Processing ) Evidence for (Ranked-) Parallelism (?) Effects on reading time at disambiguation, of lexical frequencies and plausibility of inappropriate alternatives. E.g. Garnsey et al. (1997): NP/S ambiguity: Bill argued the position... ...forcefully. ...was incorrect. The more plausible the NP interpretation, the greater the ambiguity effect for the S continuation. Similarly, the more frequent the NP subcategorization, the greater the ambiguity effect for the S continuation. Evidence for Serialism / Parallelism Ranked parallel explanation: Both interpretations are calculated in parallel, and size of the ambiguity effect reflects the effort to re-rank the structures.reflects the effort to re-rank the structures....
View Full Document

This note was uploaded on 11/11/2011 for the course BIO 9.07 taught by Professor Ruthrosenholtz during the Spring '04 term at MIT.

Page1 / 87

1018_tpdn_bottup - Language comprehension Lecture 5:...

This preview shows document pages 1 - 9. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online