Difference between revisions of "Munsee and English/Structural Transfer"

From LING073
Jump to: navigation, search
(→‎Structural Transfer Rules)
(→‎Post-Evaluation)
Line 40: Line 40:
  
 
== Post-Evaluation ==
 
== Post-Evaluation ==
 +
 +
Statistics about input files
 +
-------------------------------------------------------
 +
Number of words in reference: 35
 +
Number of words in test: 21
 +
Number of unknown words (marked with a star) in test: 5
 +
Percentage of unknown words: 23.81 %
 +
 +
Results when removing unknown-word marks (stars)
 +
-------------------------------------------------------
 +
Edit distance: 24
 +
Word error rate (WER): 68.57 %
 +
Number of position-independent correct words: 11
 +
Position-independent word error rate (PER): 68.57 %
  
 
[[Category:Sp25 StructuralTransfer]]
 
[[Category:Sp25 StructuralTransfer]]
 
[[Category:Munsee]]
 
[[Category:Munsee]]
 
[[Category:English]]
 
[[Category:English]]

Revision as of 21:44, 24 April 2025

Pre-Evaluation

Statistics about input files


Number of words in reference: 35

Number of words in test: 14

Number of unknown words (marked with a star) in test: 10

Percentage of unknown words: 71.43 %

Results when removing unknown-word marks (stars)


Edit distance: 33

Word error rate (WER): 94.29 %

Number of position-independent correct words: 2

Position-independent word error rate (PER): 94.29 %

Structural Transfer Rules

Rule Description

Absence Suffix

  • Noun phrases in English (noun w/ adjective "late") get information from noun phrases in Munsee (noun w/ absence suffix "aya").
  • Example Translation: (umu) nŭmoxoomsáya → (eng) my late grandfather
  • Example Input: ^moxóomus<n><aa><px1sg><abs>
  • Example Output: ^prpers<det><pos><p1><mf><sg> late<adj><sint> grandfather<n><sg>
  • Implemented Rule:

Possessive Personal Pronouns

Post-Evaluation

Statistics about input files


Number of words in reference: 35 Number of words in test: 21 Number of unknown words (marked with a star) in test: 5 Percentage of unknown words: 23.81 %

Results when removing unknown-word marks (stars)


Edit distance: 24 Word error rate (WER): 68.57 % Number of position-independent correct words: 11 Position-independent word error rate (PER): 68.57 %