Navajo/Transducer

From LING073
Jump to: navigation, search

Link to Github Transducer

https://github.swarthmore.edu/Ling073-sp22/ling073-nav

Analyser Evaluation

nav.corpus.basic.txt

Date Evaluated Coverage Remaining unknown forms
Thu 10 Mar 2022 10:01:08 AM EDT - 1 470 / 1639 (~0.28676021964612568639) 1169
Mon 14 Mar 2022 11:21:19 EDT - 2 481 / 1634 (~0.29436964504283965728) 1153
Sun Mar 20 2022 11:28:39 EDT - 3 650 / 1676 (~0.38782816229116945107) 1026
Sun Mar 20 2022 19:21:52 EDT - 4 804 / 1676 (~0.47971360381861575179) 872
Sun Mar 25 2022 19:25:41 EDT - 5 693 / 1364 (~0.50806451612903225806) 671
Tue Apr 26 2022 10:58:12 AM EDT - 6 698 / 1364 (~0.51173020527859237537) 666
  • Initial Evaluation: 1
  • After adding Perfective-Person-Ø/Ł prefixes: 2
  • After removing initial glottal stop from all non-affixes: 3
    • Significantly changed most common unknown words, showing that they must be written without them often
  • After adding first group of analyses from uncommon words: 4
    • Several high frequency words we have definition, but not sure underlying form/how to generate
  • Replace apostrophe with glottal stop character and add more unknown words: 5
  • Add a large amount of twol rules for much of possessive noun phonology, accept various types of input characters, finish implementing verb-nominalization and adjectival-noun pattern: 6

Lexical Information

Lexicons: 67

Lexicon entries: 240

Patterns: 1

Pattern entries: 31

Counts for individual lexicon

Verb-Stems: 4

Imperfective-Person: 5

Perfective-Person-0/Ł: 5

Future-Person: 5

NounRoot: 42

RegPossInfl: 4

Verb-Derived-Noun: 3

Nominalized-Verbs: 2

Nominalizing-Suffix: 2

Adjective-NounRoot: 3

Adjectival-Suffix: 3

Determiner: 1

First-Person-Pronoun: 1

Second-Person-Pronoun: 1

Third-Person-Pronoun: 1

Third-Person-Indirect-Pronoun: 1

Demonstrative-Pronoun: 5

Postposition: 16

Conjunction: 13

Adverb: 26

Auxiliary: 7

Adjective: 4

Negation: 1

Numeral: 10

Punctuation: 22

Iterative-Person: 5

Perfective-Person-D/L: 5

Optative-Person: 5

All anonymous lexicons: 38

Tests

As of April 26, 2022

  • nav.yaml: Total passes: 90, Total fails: 28, Total: 118
  • commonWords.yaml: Total passes 6, Total fails: 13, Total: 19 -- Actual pass rate higher, possible errors in characters/generation as they are in transducer with analyses

Current Unknown Words

TOP UNKNOWN WORDS:

     6 ^áyiilaa/*áyiilaa$
     6 ^naakai/*naakai$
     6 ^hazlį́į́ʼ/*hazlį́į́ʼ$
     6 ^adinídínígíí/*adinídínígíí$
     5 ^ádaatʼéii/*ádaatʼéii$
     5 ^yáʼíítʼééh/*yáʼíítʼééh$
     5 ^yiyíízhiʼ/*yiyíízhiʼ$
     5 ^yinééłʼį́įʼgo/*yinééłʼį́įʼgo$
     5 ^hazlį́įʼjįʼ/*hazlį́įʼjįʼ$
     5 ^dahólǫ́/*dahólǫ́$
     5 ^bineestʼąʼ/*bineestʼąʼ$
     5 ^azlį́į́ʼ/*azlį́į́ʼ$
     4 ^ééʼdeetįįhii/*ééʼdeetįįhii$
     4 ^átʼé/*átʼé$
     4 ^ánáádooʼniid/*ánáádooʼniid$
     4 ^ádaadin/*ádaadin$
     4 ^deiłníigo/*deiłníigo$
     4 ^bílástsiiʼ/*bílástsiiʼ$
     4 ^bideeʼ/*bideeʼ$
     4 ^ashdlaʼii/*ashdlaʼii$

Notes

  • Currently, our transducer is missing rules for the iterative mode, the perfective D/L classifier mode, and the optative mode. This is because we have had extreme difficulty in finding the underlying forms of the pronominal prefixes in these modes. Their stems are listed in our transducer, but we are unable to make a rule for their generation because of this missing information.
  • While the forms are being analyzed, several adverbs/conjunctions appear to behave like postpositions, having 3rd person inflected morphology, though we are unsure if they may undergo all the inflection that a postposition or possessed noun does, as they only appear inflected for 3rd person. We are also unsure of their underlying forms.
  • A large percentage of our unknown words are verbs which we have yet to analyze.

Generator Evaluation

Analysis

morph-test -csi tests/nav3.yaml

Generation

morph-test -cl tests/nav3.yaml

Date Evaluated Evaluation Total passes Total fails Total
Apr 27 2022* Analyser 90 26 116
Generator 90 28 118

Notes

  • Currently, we have implemented a large amount of phonology involving possession of nouns and postpositions.