1 / 14

Dependency Locality Theory: A Distance-based Theory of Linguistic Complexity

Dependency Locality Theory: A Distance-based Theory of Linguistic Complexity. Boris Krupa 5/4/2001. Motivation. How does the brain construct an interpretation for a sentence?. However it does it, it must:.

colin
Download Presentation

Dependency Locality Theory: A Distance-based Theory of Linguistic Complexity

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. Dependency Locality Theory: A Distance-based Theory of Linguistic Complexity Boris Krupa 5/4/2001

  2. Motivation • How does the brain construct an interpretation for a sentence? However it does it, it must: • Perform structural integrations: connecting a word into the structure for the input thus far • Keep the structure in memory, which includes keeping track of incomplete dependencies

  3. Motivation • How does the brain construct an interpretation for a sentence? However it does it, it must: • Perform structural integrations: connecting a word into the structure for the input thus far • Keep the structure in memory, which includes keeping track of incomplete dependencies PROBLEM !! - constrained by limited computational resources (working memory span, attention, etc.)

  4. Dependency Locality Theory • A theory of human computational resources in sentence parsing • Key feature: LOCALITY – the cost of integrating two elements depends on the distance between them

  5. A Case Study A Case Study to investigate computational resource constraints on sentence processing - nested (or center-embedded) structures Example: 1a. The reporter disliked the editor. 1b. The reporter [who the senator attacked] disliked the editor. 1c. The reporter [who the senator [who John met] attacked] disliked the editor. The difficulty to understand 1c is not because of a local ambiguity (there is none), or lexical infrequency/implausibility, as the following sentence proves: 2. John met the senator [who attacked the reporter [who disliked the editor]].

  6. A Case Study • In Japanese the same holds: 3a. Aniga imooto ijimeta. 3b. Bebiisitaa-ga [ani-ga imooto-o ijimeta to] itta. 3c. Obasan-ga [bebiisitaa-ga [ani-ga imooto-o ijimeta to] itta to] omotteiru.

  7. How Can This Be Explained? • The difficulty in comprehension of nested structures is determined by the maximal number of incomplete syntactic dependencies that the processor has to keep track of during the course of processing the sentence 1a. The reporter disliked the editor. 1b. The reporter [whothe senatorattacked] disliked the editor. 1c. The reporter [who the senator [who John met] attacked] disliked the editor.

  8. Great! Problem solved! WRONG!! Compare the following sentences: 1c. The reporterwho the senatorwho John metattackeddisliked the editor. 3a. A bookthat some Italianthat I have never heard ofwrotewill be published by MIT press. 3b. The reporterwho everyonethat I mettrusts said the president won’t resign yet.

  9. Why? 1c. The reporterwho the senatorwho John metattackeddisliked the editor. 3a. A bookthat some Italianthat I have never heard ofwrotewill be published by MIT press. 3b. The reporterwho everyonethat I mettrusts said the president won’t resign yet. 3a and 3b contain pronouns in the most embedded relative clauses!!

  10. Test Compare the following sentences: 1. The reporter who the senator who the professor met attacked disliked the editor. 2. The reporter who the senator who John met attacked disliked the editor. 3. The reporter who the senator who I met attacked disliked the editor.

  11. Theory • This can be explained by the dependency locality theory • Parsing of sentences consists of two major tasks: • Integration of the current word into the structure built thus far • Storage of the structure build thus far Both are associated with resource costs Integration costs – observations: - difficulty of processing an NP depends on the accessibility of the referent of the NP: focused entities, usually referred to with pronouns, are highly accessible vs. new discourse objects consume substantial resources - integration depends on the distance between the heads of the two phrases

  12. Theory • Integration costs : a) 1 energy unit (EU) is consumed for every new discourse element b) When connecting a new input head h2 to a projection of an existing head h1, count 1 EU for every new referent in the intervening region • Can account for the complexities and reading times of the above sentences, and compare difficulties of two grammatical formulations Example: The reporter who the photographer sent to the editor hoped for a good story a) 0 1 0 0 1 1 0 0 1 1 0 0 0 1 b) 0 0 0 0 0 2 0 0 0 3 0 0 0 0 The reporter who sent the photographer to the editor hoped for a good story a) 0 1 0 1 0 1 0 0 1 1 0 0 0 1 b) 0 0 0 0 0 0 0 0 0 3 0 0 0 0

  13. Theory • Storage costs: 1 memory unit (MU) is associated with each syntactic head required to complete the current input as a grammatical sentence Example: The reporter who the senator attacked disliked the editor. 2 1 3 4 3 1 1 1 0 Integration costs combined with storage costs can resolve ambiguities in sentences, etc.

  14. Conclusions • A unified theory for a large number of disparate phenomena: • Matches online reading times of relative clauses • Explains the complexity of doubly nested relative clause constructions • Explains lower complexity of multiply embedded structures with pronouns

More Related