Difference between revisions of "Eastern Burushaski and English"
From LING073
(→2 lexical selection rules) |
|||
| (15 intermediate revisions by 2 users not shown) | |||
| Line 1: | Line 1: | ||
| − | + | =Initial evaluation= | |
==Resources for machine translation between [[Eastern_Burushaski]] and English== | ==Resources for machine translation between [[Eastern_Burushaski]] and English== | ||
| Line 31: | Line 31: | ||
|han buċ bila. || There is an oriental plane. || ^hin<num><clx>/one<num><clx>$ ^buc̣<n>/oriental.plane.(tree.species)<n>$ ^b<cop><pres><ind><p3><sg><cly>/be<cop><pres><ind><p3><sg><cly>$. || #one #oriental.plane.(tree.species) #be. | |han buċ bila. || There is an oriental plane. || ^hin<num><clx>/one<num><clx>$ ^buc̣<n>/oriental.plane.(tree.species)<n>$ ^b<cop><pres><ind><p3><sg><cly>/be<cop><pres><ind><p3><sg><cly>$. || #one #oriental.plane.(tree.species) #be. | ||
|- | |- | ||
| − | |d̪aa šaam meey bila. || It gets | + | |d̪aa šaam meey bila. || It gets dark again. || ^d̪aa<conn>/and<conn>$ ^šaam<n>/evening<n>$ ^meey<v>/become.IPFV<v>$ ^b<cop><pres><ind><p3><sg><cly>/be<cop><pres><ind><p3><sg><cly>$. || #and #evening #become.IPFV #be. |
| + | |- | ||
| + | |ine gušpurate es baas basimi.|| That falcon sat on the prince. || ^in<prn><dem><dst><sg><mf>/that<dem><dst><sg><mf>$ ^gušpur<n><loc>/prince<n><loc>$ ^es<dem>/that.[−h,.I]<dem>$ ^baas<n>/times<n>/falcon<n>$ ^bas<v><iv><p3><sg><m>/sit<v><iv><p3><sg><m>$ || #that #prince #that.[−h,.I] #times #sit. | ||
|} | |} | ||
| + | =Final evaluation= | ||
| + | |||
| + | ==Additions== | ||
| + | ===5 More Morphological Elements=== | ||
| + | *-i and -u/o final nouns add -y- and -w- respectively before prefixes that begin with a vowel | ||
| + | *-(a)n indefinite noun suffix | ||
| + | *added most common plural endings | ||
| + | ===Fix 100 Stems=== | ||
| + | *lexc and dix files (bsk and bsk-eng respectively) | ||
| + | **910 --> 842 "UNK" tags (added tags to unknown words), so a reduction of 68 | ||
| + | **replaced instances of "." in the middle of entries with spaces (eg. "because.of" --> "because of") | ||
[[Category:Eastern_Burushaski]][[Category:Sp19_TranslationPairs]] | [[Category:Eastern_Burushaski]][[Category:Sp19_TranslationPairs]] | ||
| + | |||
| + | ===2 lexical selection rules=== | ||
| + | **ulu "in" (adposition) and "inside" (adverb) | ||
| + | *** it̪e bulaar ulu gimo "She entered in the polo match" | ||
| + | *** ulu hiran bai "A man was inside" | ||
| + | ** | ||
| + | ***but int̪izaam eer et̪umo. | ||
| + | ***but uyam baran | ||
Latest revision as of 01:20, 14 May 2019
Contents
Initial evaluation
Resources for machine translation between Eastern_Burushaski and English
- Apertium basic rule-based machine translation [1]
- Corpus for MT [2]
- Developed bilingual corpus https://github.swarthmore.edu/Ling073-sp19/ling073-bsk-eng-corpus
- Corpus data courtesy of Dr. Sadaf Munshi at the University of North Texas
bsk --> eng Evaluation
Monolingual transducer finds 27 forms, misses 0. Bilingual transducer finds 135 forms, misses 0. This difference is likely due to the monolingual transducer classifying multiple forms of the same word as only one, as in "to be."
| Burushaski Text | English free translation | Lexical transfer output | Full translation output |
|---|---|---|---|
| tis eču bo. | She is making a mistake. | ^tis<n>/pit<n>/mistake<n>$ ^eč<v><p3><sg><f>/do.IPFV#it <v><p3><sg><f>$ ^b<cop><pres><ind><p3><sg><f>/be<cop><pres><ind><p3><sg><f>$. | #pit #do.IPFV#it #be. |
| ine paad̪šaa t̪alo yuu bam. | That king had seven sons. | ^in<prn><dem><dst><sg><mf>/that<dem><dst><sg><mf>$ ^paad̪šaa<n>/king<n>$ ^t̪ʰalo<num><mfclx>/seven<num><mfclx>$ ^yuu<n><pl>/sons<n><pl>$ ^b<cop><past><ind><p1><pl>/be<cop><past><ind><p1><pl>$. | #that #king #seven #sons #be. |
| kuult̪o eer ečam. | Today I will do it for him. | ^kuult̪o<n>/today<n>$ ^e<prn><pers><p1><sg><dat>/I<prn><pers><p1><sg><dat>$ ^eč<v><p1><sg>/do.IPFV#it <v><p1><sg>$. | #today #I #do.IPFV#it . |
| aa ine gane t̪ayaar baa. | I am ready for him. | ^e<prn><pers><p1><sg><abs>/I<prn><pers><p1><sg><abs>$ ^in<prn><dem><dst><sg><mf>/that<dem><dst><sg><mf>$ ^gane<post>/for.the.sake.of<post>$ ^t̪ayaar<n>/ready<n>$ ^b<cop><pres><ind><p1><sg>/be<cop><pres><ind><p1><sg>$. | #I #that #for.the.sake.of #ready #be. |
| d̪aa it̪eer šaan e naa. | And please be careful of that. | ^d̪aa<conn>/and<conn>$ ^in<prn><dem><dst><sg><cly><dat>/that<dem><dst><sg><cly><dat>$ ^šaan<adj>/careful<adj>$ ^e<prn><pers><p1><sg><erg>/I<prn><pers><p1><sg><erg>$ ^naa/$. |
#and #that careful #I . |
| it̪e paad̪šaa buc̣ bila. | That is the oriental plane of the king. | ^in<prn><dem><dst><sg><cly><erg>/that<dem><dst><sg><cly><erg>$ ^paad̪šaa<n>/king<n>$ ^buc̣<n>/oriental.plane.(tree.species)<n>$ ^b<cop><pres><ind><p3><sg><cly>/be<cop><pres><ind><p3><sg><cly>$. | #that #king #oriental.plane.(tree.species) #be. |
| hin paad̪šaan bam. | There was a king. | ^hin<num><mf>/one<num><mf>$ ^paad̪šaa<n><indef>/king<n><indef>$ ^b<cop><past><ind><p1><pl>/be<cop><past><ind><p1><pl>$. | #one #king #be. |
| e tis amana baa. | I made a mistake. | ^e<prn><pers><p1><sg><erg>/I<prn><pers><p1><sg><erg>$ ^tis<n>/pit<n>/mistake<n>$ ^amana<v>/I# become<v>$ ^b<cop><pres><ind><p1><sg>/be<cop><pres><ind><p1><sg>$. | #I #pit #I# become #be. |
| han buċ bila. | There is an oriental plane. | ^hin<num><clx>/one<num><clx>$ ^buc̣<n>/oriental.plane.(tree.species)<n>$ ^b<cop><pres><ind><p3><sg><cly>/be<cop><pres><ind><p3><sg><cly>$. | #one #oriental.plane.(tree.species) #be. |
| d̪aa šaam meey bila. | It gets dark again. | ^d̪aa<conn>/and<conn>$ ^šaam<n>/evening<n>$ ^meey<v>/become.IPFV<v>$ ^b<cop><pres><ind><p3><sg><cly>/be<cop><pres><ind><p3><sg><cly>$. | #and #evening #become.IPFV #be. |
| ine gušpurate es baas basimi. | That falcon sat on the prince. | ^in<prn><dem><dst><sg><mf>/that<dem><dst><sg><mf>$ ^gušpur<n><loc>/prince<n><loc>$ ^es<dem>/that.[−h,.I]<dem>$ ^baas<n>/times<n>/falcon<n>$ ^bas<v><iv><p3><sg><m>/sit<v><iv><p3><sg><m>$ | #that #prince #that.[−h,.I] #times #sit. |
Final evaluation
Additions
5 More Morphological Elements
- -i and -u/o final nouns add -y- and -w- respectively before prefixes that begin with a vowel
- -(a)n indefinite noun suffix
- added most common plural endings
Fix 100 Stems
- lexc and dix files (bsk and bsk-eng respectively)
- 910 --> 842 "UNK" tags (added tags to unknown words), so a reduction of 68
- replaced instances of "." in the middle of entries with spaces (eg. "because.of" --> "because of")
2 lexical selection rules
- ulu "in" (adposition) and "inside" (adverb)
- it̪e bulaar ulu gimo "She entered in the polo match"
- ulu hiran bai "A man was inside"
-
- but int̪izaam eer et̪umo.
- but uyam baran
- ulu "in" (adposition) and "inside" (adverb)
