Machine Translation between Hebrew and Arabic: Needs, Challenges and Preliminary Solutions

What it is about

Modern Hebrew and Standard Arabic share orthography, root-and-pattern morphology, clitics and much of their vocabulary, but they differ in case, dual number, broken plurals, the feminine ending in genitive constructions and the agreement of irrational plural nouns. No parallel corpus exists for the pair. The authors show that Google's Hebrew-to-Arabic output goes through English as a pivot, and that it loses information on the English "you", on plural gender and on English-ambiguous nouns. They argue for a direct transfer approach and build one in the Stat-XFER framework, with 42 hand-written rules, a Hebrew analyser that does no disambiguation and an Arabic generator. The generator returns 109 forms per verb and 72 per noun before clitics, so the lattice grows fast. The results are preliminary: sample translations only, with no robust evaluation.

Ideas

I1. Irrational plural agreement needs a rationality flag on each noun, which decides between singular feminine agreement and ordinary plural agreement. Build that table for Arabic nouns first, and test it on the paper's examples, before any generator work.
I2. The Hebrew double genitive (a noun with a possessive suffix, then the marker šel, then the possessor, as in sfr+w šl h+ild, "the boy's book") is a construction the parser must recognise as one genitive. Add it as a pattern and test it on the paper's example.
I3. Count the candidate forms a generator returns per lemma, and the lattice size when the analyser does not disambiguate. Use these counts as a budget before adding any generation target.

For Mova

I1 goes into treebanks/ar as a lexical table. I2 goes into Nouns, adjectives, pronouns and numerals. I3 is a scope note for 12. Mova and Modern Hebrew: where it stands, what it would need, since Mova does not generate Arabic.

Topics: Machine translation

Cite

The paper itself, not this card. BibTeX from the library's record of the work.

@inproceedings{shilon2010machine,
  title = {{Machine Translation between Hebrew and Arabic: Needs, Challenges and Preliminary Solutions}},
  author = {Shilon, Reshef and Habash, Nizar and Lavie, Alon and Wintner, Shuly},
  booktitle = {Proceedings of the 9th Conference of the Association for Machine Translation in the Americas: Student Research Workshop},
  year = {2010},
  url = {https://aclanthology.org/2010.amta-srw.4/}
}

References

16 references, 1 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

P. F. Brown, J. Cocke, S. A. Della Pietra, V. J. Della Pietra, F. Jelinek, J. D. Lafferty, R. L. Mercer, and P. S. Roossin. 1990. A statistical approach to machine translation. Computational Linguistics, 16(2):79–85. find
Ahmed El Kholy and Nizar Habash. 2010. Techniques for Arabic morphological detokenization and orthographic denormalization. In Proc. of LREC-2010. find
David Graff, Ke Chen, Junbo Kong and Kazuaki Maeda. 2006. Arabic Gigaword. Linguistic Data Consortium. find
Nizar Habash and Fatiha Sadat. 2006. Arabic preprocessing schemes for statistical machine translation. In Proc. of HLT-NAACL. find
Nizar Habash, Abdelhadi Soudi, and Timothy Buckwalter. 2007. On Arabic transliteration. In Abdelhadi Soudi, (20)(a) mkwniwt ipwt wšwtrwt ipwt car.pl.f.indef pretty.pl.f.indef and+policewomen.f.indef pretty.pl.def.indef ‘pretty cars and pretty policewomen’ (b) syArAt jmylh̄ wšrTyAt jmylAt car.pl.f.indef pretty.sg.f and+policewomen.f.indef pretty.pl.f.indef ‘pretty cars and pretty policewomen’ (c) syArAt AlšrTh̄ lTyf lTyf car.pl.f.def police.sg.f.def pretty.sg.m.indef pretty.sg.m.indef ‘The police’s cars pretty pretty’ (21)(a) hncigim šlkm nkxw bišibh representative.pl.m.def you.pl.m.poss attend.past.3.pl in+meeting.sg.f.def ‘your representatives attended the meeting’ (b) HDr mmvlwkm Aljlsh̄ attend.past.sg.m representative.pl.m.nom+you.pl.m.poss meeting.def ‘your representatives attended the meeting’ (c) wmmvlwkm AlHADryn fy AlAjtmAς and+representative.pl.m.nom+you.pl.m.poss attend.participle.pl.m.def.acc/gen in meeting.def ‘And your representatives that attended the meeting’ (22)(a) mkwnith šl hmnhlt gdwlh car.sg.f.def+she.poss of principal.sg.f.def big.sg.f.indef ‘The principal’s car is big’ (b) syArh̄ Almdyrh̄ kbyrh̄ car.sg.f.indef principal.sg.f.def big.sg.f.indef ‘The principal’s car is big’ (c) Alf AlrAysy llsyArAt thousand main.sg.m.indef to+the+car.pl.f.def ‘The cars’ thousand main’ Figure 2: Example Hebrew to Arabic translations: (a) Hebrew; (b) our Arabic output; (c) Google’s output Guenter Neumann, and Antal van den Bosch, editors, Arabic Computational Morphology. Springer. find
Nizar Habash. 2004. Large scale lexeme based arabic morphological generation. In Proc. of Traitement Automatique du Langage Naturel (TALN-04). find
Greg Hanneman, Vamshi Ambati, Jonathan H. Clark, Alok Parlikar, and Alon Lavie. 2009. An improved statistical transfer system for French–English machine translation. In Proc. of StatMT ’09: the Workshop on Statistical Machine Translation. find
Alon Itai and Shuly Wintner. 2008. Language resources for Hebrew. Language Resources and Evaluation, 42:75–98. find
Shankar Kumar, Franz J. Och, and Wolfgang Macherey. 2007. Improving word alignment with bridge languages. In Proc. of EMNLP-CoNLL. find
Alon Lavie, Stephan Vogel, Lori Levin, Erik Peterson, Katharina Probst, Ariadna Font Llitjós, Rachel Reynolds, Jaime Carbonell, and Richard Cohen. 2003. Experiments with a Hindi-to-English transfer-based MT system under a miserly data scenario. ACM Transactions on Asian Language Information Processing (TALIP), 2(2):143–163. find
Alon Lavie, Shuly Wintner, Yaniv Eytani, Erik Peterson, and Katharina Probst. 2004. Rapid prototyping of a transfer-based Hebrew-to-English machine translation system. In Proc. of TMI-2004: Theoretical and Methodological Issues in Machine Translation. find our card
Alon Lavie. 2008. Stat-XFER: A general search-based syntax-driven framework for machine translation. In Alexander F. Gelbukh, editor, CICLing, volume 4919 of Lecture Notes in Computer Science, pages 362–375. Springer. find
Dragos Stefan Munteanu and Daniel Marcu. 2005. Improving machine translation performance by exploiting non-parallel corpora. Computational Linguistics, 31(4):477–504. find
Kazunori Muraki. 1987. PIVOT: Two-phase machine translation system. In MT Summit Manuscripts and Program, pages 81–83. find
Dániel Varga, Péter Halácsy, András Kornai, Viktor Nagy, László Németh, and Viktor Trón. 2005. Parallel corpora for medium density languages. In Proc. of RANLP. find
Hua Wu and Haifeng Wang. 2007. Pivot language approach for phrase-based statistical machine translation. In Proc. of ACL. find

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/lingbulk-2010-shilon-machine-translation-between-hebrew-arabic/