ݺߣ

ݺߣShare a Scribd company logo

Class17 & 18 - Language Comprehention

•
•
Nathacia Lucena
Nathacia Lucena

The document discusses various parsing strategies and principles used in language comprehension, including: 1) PARSER refers to the structural processor that analyzes syntax using operations like Merge, Labeling, and Movement. 2) Ambiguities like global ambiguity, local ambiguity, and garden path sentences make parsing difficult. 3) Strategies like minimal attachment, late closure, low attachment, and good enough processing help the parser analyze sentences despite ambiguities and processing limitations. 4) Active filler licensing allows relating gaps and filler elements in long-distance dependencies.

1 of 57
Language Comprehension LCD 105 - Class 17 - Professor Nathacia
We’ve seen that…
Class17 & 18 - Language Comprehention
Jane did not go to school
Jane did not go to school PARSER
[TP[NP Jane] did [NegP not [vP[VP go to school]]]]. PARSER
PARSER = Structural Processor
PARSER = Structural Processor STRUCTURE Merge Labeling Movement
PARSER = Structural Processor STRUCTURE Merge Labeling Movement PARSER Constituents Labels Traces and indexes (gaps)
PARSER = Structural Processor STRUCTURE Merge Labeling Movement PARSER Constituents Labels Traces and indexes HOW?
Do we really compute (abstract) syntax structure?
The elected official hope to succeed.
*The elected official hope to succeed.
*The elected official hope to succeed. The elected official hopes to succeed.
*The elected official hope to succeed. The elected official hopes to succeed. And a huge number of experiments have being testifying that hypothesis.
By the way… Experiment over other ways of doing linguistics gives us the possibility of testing the Faculty of Language beyond of what it can generate only, by testing impossible outputs as well. Testing objects that cannot be generated by the system helps understanding what the restrictions of the system are, instead of testing just the way it operates.
Language Comprehension LCD 105 - Class 18 - Professor Nathacia
[TP[NP Jane] did [NegP not [vP[VP go to school]]]]. PARSER
PARSER = Structural Processor
PARSER = Structural Processor STRUCTURE Merge Labeling Movement PARSER Constituents Labels Traces and indexes (gaps)
PARSER
PARSER
The man saw the boy with the binoculars.
The man saw the boy with the binoculars.
The man saw the boy with the binoculars.
The man saw the boy with the binoculars. ?? ?
The man saw the boy with the binoculars. ?? ? Global Ambiguity
The student told the professor that…
The student told the professor that… ?? ?
The student told the professor that… he wanted a better grade. ?? ?
The student told the professor that… …he wanted a better grade. … taught the course that he wanted a better grade. ?? ?
The student told the professor that… …he wanted a better grade. … taught the course that he wanted a better grade. … really unbelievable story. ?? ?
The student told the professor that… …he wanted a better grade. … taught the course that he wanted a better grade. … really unbelievable story. ?? ? Local Ambiguity
The horse raced past the barn fell.
The horse raced past the barn fell. The old man the boat. ?? ?
The horse raced past the barn fell. The old man the boat. The complex houses married and single soldiers and their families. ?? ?
The horse raced past the barn fell. The old man the boat. The complex houses married and single soldiers and their families. ?? ? Garden Path
The horse raced past the barn fell. The old man the boat. The complex houses married and single soldiers and their families. ?? ? Garden Path The less the work the better. “(…) the parser makes immediate decisions about locally ambiguous constituents, always opting for the analyses that requires fewer computational resources.” (Fernández & Cairns, 2011:214)
The horse raced past the barn fell. The old man the boat. The complex houses married and single soldiers and their families. ?? ? Garden Path The less the work the better. “(…) the parser makes immediate decisions about locally ambiguous constituents, always opting for the analyses that requires fewer computational resources.” (Fernández & Cairns, 2011:214) ] Minimal Attachment
The horse raced past the barn fell. The old man the boat. The complex houses married and single soldiers and their families. ?? ? Garden Path Minimal Attachment
The horse raced past the barn fell. The old man the boat. The complex houses married and single soldiers and their families. ?? ? Garden Path Minimal AttachmentX
The horse raced past the barn fell. The old man the boat. The complex houses married and single soldiers and their families. ?? ? Garden Path Minimal AttachmentX Reanalysis
The horse raced past the barn fell. The old man the boat. The complex houses married and single soldiers and their families. ?? ? Garden Path Minimal AttachmentX Reanalysis Processing costs
Physicists are thrilled to explain what they are doing to people.
Physicists are thrilled to explain what they are doing to people. ?? ?
Physicists are thrilled to explain what they are doing to people. ?? ? Late Closure
Physicists are thrilled to explain what they are doing to people. ?? ? Late Closure Put everything you can in the opened box. “This strategy prompts the parser to integrate new words received from the lexical processor into the syntactic constituent that is currently being processed.” (Fernández & Cairns, 2011:218)
Physicists are thrilled to explain what they are doing to people. ?? ? Late Closure Put everything you can in the opened box. “This strategy prompts the parser to integrate new words received from the lexical processor into the syntactic constituent that is currently being processed.” (Fernández & Cairns, 2011:218) ] Low Attachment
Physicists are thrilled to explain what they are doing to people. ?? ? Late Closure Put everything you can in the opened box. “This strategy prompts the parser to integrate new words received from the lexical processor into the syntactic constituent that is currently being processed.” (Fernández & Cairns, 2011:218) ] Low Attachment NOT universal!
While Mary bathed the baby played in the crib.
Did Mary bathe the baby?
Did Mary bathe the baby? ?? ? Good Enough
Did Mary bathe the baby? ?? ? Good Enough Reanalysis
?? ? Good Enough Reanalysis While Mary bathed the baby played in the crib.
We didn’t know which book the teacher read to the children from.
We didn’t know [which book]i the teacher read ti to the children from ti. ?? ?
We didn’t know [which book]i the teacher read ti to the children from ti. ?? ? Active Filler

More Related Content

Class17 & 18 - Language Comprehention

  • 1. Language Comprehension LCD 105 - Class 17 - Professor Nathacia
  • 4. Jane did not go to school
  • 5. Jane did not go to school PARSER
  • 6. [TP[NP Jane] did [NegP not [vP[VP go to school]]]]. PARSER
  • 8. PARSER = Structural Processor STRUCTURE Merge Labeling Movement
  • 9. PARSER = Structural Processor STRUCTURE Merge Labeling Movement PARSER Constituents Labels Traces and indexes (gaps)
  • 10. PARSER = Structural Processor STRUCTURE Merge Labeling Movement PARSER Constituents Labels Traces and indexes HOW?
  • 11. Do we really compute (abstract) syntax structure?
  • 12. The elected official hope to succeed.
  • 13. *The elected official hope to succeed.
  • 14. *The elected official hope to succeed. The elected official hopes to succeed.
  • 15. *The elected official hope to succeed. The elected official hopes to succeed. And a huge number of experiments have being testifying that hypothesis.
  • 16. By the way… Experiment over other ways of doing linguistics gives us the possibility of testing the Faculty of Language beyond of what it can generate only, by testing impossible outputs as well. Testing objects that cannot be generated by the system helps understanding what the restrictions of the system are, instead of testing just the way it operates.
  • 17. Language Comprehension LCD 105 - Class 18 - Professor Nathacia
  • 18. [TP[NP Jane] did [NegP not [vP[VP go to school]]]]. PARSER
  • 19. PARSER = Structural Processor
  • 20. PARSER = Structural Processor STRUCTURE Merge Labeling Movement PARSER Constituents Labels Traces and indexes (gaps)
  • 23. The man saw the boy with the binoculars.
  • 24. The man saw the boy with the binoculars.
  • 25. The man saw the boy with the binoculars.
  • 26. The man saw the boy with the binoculars. ?? ?
  • 27. The man saw the boy with the binoculars. ?? ? Global Ambiguity
  • 28. The student told the professor that…
  • 29. The student told the professor that… ?? ?
  • 30. The student told the professor that… he wanted a better grade. ?? ?
  • 31. The student told the professor that… …he wanted a better grade. … taught the course that he wanted a better grade. ?? ?
  • 32. The student told the professor that… …he wanted a better grade. … taught the course that he wanted a better grade. … really unbelievable story. ?? ?
  • 33. The student told the professor that… …he wanted a better grade. … taught the course that he wanted a better grade. … really unbelievable story. ?? ? Local Ambiguity
  • 34. The horse raced past the barn fell.
  • 35. The horse raced past the barn fell. The old man the boat. ?? ?
  • 36. The horse raced past the barn fell. The old man the boat. The complex houses married and single soldiers and their families. ?? ?
  • 37. The horse raced past the barn fell. The old man the boat. The complex houses married and single soldiers and their families. ?? ? Garden Path
  • 38. The horse raced past the barn fell. The old man the boat. The complex houses married and single soldiers and their families. ?? ? Garden Path The less the work the better. “(…) the parser makes immediate decisions about locally ambiguous constituents, always opting for the analyses that requires fewer computational resources.” (Fernández & Cairns, 2011:214)
  • 39. The horse raced past the barn fell. The old man the boat. The complex houses married and single soldiers and their families. ?? ? Garden Path The less the work the better. “(…) the parser makes immediate decisions about locally ambiguous constituents, always opting for the analyses that requires fewer computational resources.” (Fernández & Cairns, 2011:214) ] Minimal Attachment
  • 40. The horse raced past the barn fell. The old man the boat. The complex houses married and single soldiers and their families. ?? ? Garden Path Minimal Attachment
  • 41. The horse raced past the barn fell. The old man the boat. The complex houses married and single soldiers and their families. ?? ? Garden Path Minimal AttachmentX
  • 42. The horse raced past the barn fell. The old man the boat. The complex houses married and single soldiers and their families. ?? ? Garden Path Minimal AttachmentX Reanalysis
  • 43. The horse raced past the barn fell. The old man the boat. The complex houses married and single soldiers and their families. ?? ? Garden Path Minimal AttachmentX Reanalysis Processing costs
  • 44. Physicists are thrilled to explain what they are doing to people.
  • 45. Physicists are thrilled to explain what they are doing to people. ?? ?
  • 46. Physicists are thrilled to explain what they are doing to people. ?? ? Late Closure
  • 47. Physicists are thrilled to explain what they are doing to people. ?? ? Late Closure Put everything you can in the opened box. “This strategy prompts the parser to integrate new words received from the lexical processor into the syntactic constituent that is currently being processed.” (Fernández & Cairns, 2011:218)
  • 48. Physicists are thrilled to explain what they are doing to people. ?? ? Late Closure Put everything you can in the opened box. “This strategy prompts the parser to integrate new words received from the lexical processor into the syntactic constituent that is currently being processed.” (Fernández & Cairns, 2011:218) ] Low Attachment
  • 49. Physicists are thrilled to explain what they are doing to people. ?? ? Late Closure Put everything you can in the opened box. “This strategy prompts the parser to integrate new words received from the lexical processor into the syntactic constituent that is currently being processed.” (Fernández & Cairns, 2011:218) ] Low Attachment NOT universal!
  • 50. While Mary bathed the baby played in the crib.
  • 51. Did Mary bathe the baby?
  • 52. Did Mary bathe the baby? ?? ? Good Enough
  • 53. Did Mary bathe the baby? ?? ? Good Enough Reanalysis
  • 54. ?? ? Good Enough Reanalysis While Mary bathed the baby played in the crib.
  • 55. We didn’t know which book the teacher read to the children from.
  • 56. We didn’t know [which book]i the teacher read ti to the children from ti. ?? ?
  • 57. We didn’t know [which book]i the teacher read ti to the children from ti. ?? ? Active Filler