LakotaandEnglish

From LING073
Revision as of 20:12, 6 May 2022 by Dmahaja1 (talk | contribs) (Final Evaluation)

Jump to: navigation, search

Resources for machine translation between Lakota and English

External resources

lkt -> eng Evaluation

Coverage on Monolingual Transducer : 122/611 (20%)

  • command: ./coverage-lt.sh ../ling073-lkt-eng-corpus/lkt.sentences.txt lkt-eng.automorf.bin

Coverage on Bilingual Transducer : 320/2390 (13%)

  • command: ./coverage-lt.sh ../ling073-lkt-corpus/lkt.corpus.basic.txt lkt-eng.automorf.bin

Our Ten Analyzed Sentences

  • command: echo máni | apertium -d . lkt-eng-biltrans
  • analysing forms: echo mawáni k’éyaš išwíčhatíŋmA | apertium -d . lkt-morph

The original sentence, The intended English translation, The output of your xyz-abc-biltrans (lexical transfer) mode, The output of your xyz-abc (full translation) mode.

1) mawatúkȟa = "I am tired"

  • Output of Lexical Transfer: ^watúkȟa<v><sv><s_sg1>/tired<v><sv><s_sg1>$^.<sent>/.<sent>$
  • Output of Full Translation: #tired

2) uŋwaŋyáŋkepi = "I look at them"

  • Output of Lexical Transfer: ^waŋyáŋkA<v><tv><s_pl1><o_pl3>/look<v><tv><s_pl1><o_pl3>$^.<sent>/.<sent>$
  • Output of Full Translation: #look

3) lé ziŋtkáziwatúkȟa = "This canary is tired"

  • Output of Lexical Transfer: ^*lé/*lé$ ^*ziŋtkáziwatúkȟa/*ziŋtkáziwatúkȟa$^.<sent>/.<sent>$
  • Output of Full Translation: *lé *ziŋtkáziwatúkȟa

4) Šúŋka kiŋ lé waŋyáŋkA. = "I look at the dog"

  • Output of Lexical Transfer: ^*Šúŋka/*Šúŋka$ ^kiŋ<det><def>/the<det><def>$ ^*lé/*lé$ ^waŋyáŋkA<v><tv><s_sg3><o_sg3>/look<v><tv><s_sg3><o_sg3>$^.<sent>/.<sent>$
  • Output of Full Translation: *Šúŋka #the *lé #look

5) Lé úta = "This is an acorn"

  • Output of Lexical Transfer: ^*lé/*lé$ ^úta<n>/acorn<n>$^.<sent>/.<sent>$
  • Output of Full Translation: *lé #acorn

6) mawáni k’éyaš išwíčhatíŋmA = "I walk but they all sleep"

  • Output of Lexical Transfer: ^máni<v><av><s_sg1>/walk<v><av><s_sg1>$ ^k’éyaš<cnjcoo>/but<cnjcoo>$ ^*išwíčhatíŋmA/*išwíčhatíŋmA$^.<sent>/.<sent>$
  • Output of Full Translation: #walk but *išwíčhatíŋmA

7) igmú kiŋ oyúspe skelúta kiŋ masóphiye él = "The cat catches the oriole at the store."

  • Output of Lexical Transfer: ^igmú<n>/cat<n>$ ^kiŋ<det><def>/the<det><def>$ ^oyúspe<v><tv><s_sg3><o_sg3>/catch<v><tv><s_sg3><o_sg3>$ ^skelúta<n>/oriole<n>$ ^kiŋ<det><def>/the<det><def>$ ^*masóphiye/*masóphiye$ ^*él/*él$^.<sent>/.<sent>$
  • Output of Full Translation: #cat #the #catch #oriole #the *masóphiye *él

8) ičú bló waŋ = "He takes a potato."

  • Output of Lexical Transfer: ^ičú<v>/take<v>$ ^bló<n>/potato<n>$ ^waŋ<det><ind>/a<det><ind>$^.<sent>/.<sent>$
  • Output of Full Translation: #take #potato #a

9) šún kiŋ tȟaspáŋ kiŋ itȟókab yaŋké = "The dog sits in front of the apple."

  • Output of Lexical Transfer: ^šún<n>/dog<n>$ ^kiŋ<det><def>/the<det><def>$ ^tȟaspáŋ<n>/apple<n>$ ^kiŋ<det><def>/the<det><def>$ ^*itȟókab/*itȟókab$ ^*yaŋké/*yaŋké$^.<sent>/.<sent>$
  • Output of Full Translation: #dog #the #apple #the *itȟókab *yaŋké

10) wówapi kiŋ wíčazo kiŋ isákhib = "The book is beside the pencil."

  • Output of Lexical Transfer: ^wówapi<n>/book<n>$ ^kiŋ<det><def>/the<det><def>$ ^*wíčazo/*wíčazo$ ^kiŋ<det><def>/the<det><def>$ ^isákhib<n>/pencil<n>$^.<sent>/.<sent>$
  • Output of Full Translation: #book #the *wíčazo #the #pencil

Additions

  • Expand your morphology to cover at least 5 more elements of some paradigm(s). This can be anything from additional verb or noun morphology, to adding all the forms of all the determiners (articles, demonstratives, etc.), to implementing nominal morphology on adjectives (e.g., if your language allows adjectives to be substantivised, which you'll want to add a tag for too).
  • At least 3 new disambiguation rules that make the output of your tagger more accurate.
  • At least 2 new transfer rules that make more of the output of your MT system closer to an acceptable target translation.

Final Evaluation

  • Coverage on Monolingual Transducer : 162/610 (27%)
  • Coverage on Bilingual Transducer : 394/2390 (16%)