📄 Une description morphologique structurée en arbre du verbe akkadien qui utilise des structures de traits et des transducteurs multirubans [A Tree-Structured morphological description of the Akkadian verb which uses Feature Structures and Multi-tape Transducers]

What it is about

This is the French, long version of the Karamel grammar of the Akkadian verb. Each verb form is a tree: a core (root, stem gemination, infixes, aspect vowels) and, around it, personal affixes and clitics. Inside the core, small parts are separated by the rule that a part must be significant and determined by a set of features. The author distinguishes three kinds of vowel: aspectual, lexical, and support vowels, which vanish in weak contexts. Morphotactics are simultaneous constraints on the structure; surface changes are a cascade of rewrite rules. The grammar has 71 regular expressions, 65 computations and 6 rules. The transducer without lexicon has about 496,000 states, and with a 799-root lexicon about 3 million. Coverage is partial: strong triliteral verbs and verbs with one weak radical; the test data are conjugation tables from textbooks and a few dozen forms from the Code of Hammurabi, which the author says is not a representative coverage measure. The paper argues that the two-vowel scheme of older handbooks is not well supported, and that support vowels are better inserted in a fixed priority order than deleted from a full form.

Ideas

• I1. Insert support vowels by an ordered priority list instead of writing all vowels and deleting the useless ones. The author reports that the deletion approach gave catastrophic results in his trial, so the test is a comparison: both directions run on the same forms, with the error counted for each.

For Mova

• I2. Test the claim that only one vowel of the Akkadian verb depends on all four features (aspect, voice, sub-voice, lexical class), and that it is usually the last one. A small Hebrew check on the binyan vowels in languages/hbo/06-verb.md can say whether the same holds for Biblical forms in MACULA.
• I3. Run the transducer twice, with and without a root lexicon. Without the lexicon, the grammar proposes candidate roots for an unknown form; with it, the output is limited to known roots. For tools/hbo, a no-lexicon mode that lists candidate roots for unknown words would be a useful and explainable fallback.

I1 and I2 go into languages/hbo/03-phonology.md and languages/hbo/06-verb.md (vocalisation and the vowel positions of the verb). I3 is a mode for the lexicon step, languages/hbo/10-lexicon.md, and a test in tools/hbo. The weak-verb section of the paper is a reference for languages/hbo/07-weak-verbs.md.

🏛️ Institutions

🔬Analyse Linguistique Profonde à Grande EchelleParis · France11 works · 1,087 citations

Topics𒀭 Akkadian and cuneiform NLP🧩 Morphology🐪 Semitic languages

Cite

Cite as François Barthélémy. 2009. Une description morphologique structurée en arbre du verbe akkadien qui utilise des structures de traits et des transducteurs multirubans [A Tree-Structured morphological description of the Akkadian verb which uses Feature Structures and Multi-tape Transducers]. Traitement Automatique des Langues, Volume 50, Numéro 2 : Langues anciennes [Ancient Languages].

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

@inproceedings{barthelemy2009une,
  title = {{Une description morphologique structurée en arbre du verbe akkadien qui utilise des structures de traits et des transducteurs multirubans [A Tree-Structured morphological description of the Akkadian verb which uses Feature Structures and Multi-tape Transducers]}},
  author = {François Barthélémy},
  booktitle = {Traitement Automatique des Langues, Volume 50, Numéro 2 : Langues anciennes [Ancient Languages]},
  year = {2009},
  url = {https://aclanthology.org/2009.tal-2.8/}
}

References

26 references, 2 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Barthélemy F., « Finite-State Compilation of Feature Structures for Two-Level Morphology », International Workshop on Finite State Methods in Natural Language Processing (FSMNLP), Potsdam, Germany, 2007a. find
2. Barthélemy F., « Multi-grain Relations », Implementation and Application of Automata, 12th International Conference (CIAA), Prague, Czech Republic, p. 243-252, 2007b. find
3. Barthélemy F., « Using Mazurkiewicz Trace Languages for Partition-Based Morphology », ACL, Prague (Czech Republic), 2007c. find
4. Beesley K., « Finite-state morphological analysis and generation of Arabic at Xerox Research : Status and plans in 2001 », ACL Workshop on Arabic Language Processing, 2001. find
5. Beesley K. R., « Arabic morphology using only finite-state operations », in M. Rosner (ed.), Proceedings of the Workshop on Computational Approaches to Semitic languages, 1998a. our card
6. Beesley K. R., « Arabic morphology using only finite-state operations », Proceedings of the ACL Workshop on Computational Approaches to Semitic languages, p. 50-57, 1998b. our card
7. Beesley K. R., Karttunen L., Finite State Morphology, CSLI Publications, 2003. find
8. Breckwoldt T., Cunningham G., Black J., George A., Postgate N., A Concise Dictionary of Akkadian, Harrassowitz Verlag, 2000. find
9. Buccellati G., A Structural Grammar of Babylonian, Harrassowitz Verlag, 1996. find
10. Grimley-Evans E., Kiraz G., Pulman S., « Compiling a Partition-Based Two-Level Formalism », COLING, Copenhagen, Denmark, p. 454-459, 1996. find
11. Heise J., « The Akkadian language », , http ://www.sron.nl/˜jheise/akkadian/, 1995. find
12. Huehnergard J., A Grammar of Akkadian, Eisenbrauns, Harvard Semitic Studies, 2005. find
13. Kaplan R. M., Kay M., « Regular Models of Phonological Rule Systems », Computational Linguistics, vol. 20 :3, p. 331-378, 1994. find
14. Karttunen L., « Constructing Lexical Transducers », COLING-94, Kyoto, Japan, p. 406-411, 1994. find
15. Karttunen L., Beesley K. R., « Finite-State Non-Concatenative Morphotactics », Fifth Workshop of the ACL Special Interest Group in Computational Phonology, Luxembourg (Luxembourg), p. 1-12, 2000. find
16. Kataja L., Koskenniemi K., « Finite-state description of semitic morphology : a case study of Ancient Akkadian », Proceedings of the 12th conference on Computational linguistics, Association for Computational Linguistics, Morristown, NJ, USA, p. 313-315, 1988. find
17. Kay M., « Nonconcatenative finite-state morphology », ACL Proceedings, Third European Conference, p. 2-10, 1987. find
18. Kiraz G. A., « Compiling Regular Formalisms with Rule Features into Finite-State Automata », ACL, Madrid, Spain, 1997. find
19. Kiraz G. A., Computational Nonlinear Morphology, Cambridge University Press, 2001. find
20. Koskenniemi K., « Two-Level Model for Morphological Analysis », IJCAI-83, Karlsruhe, Germany, p. 683-685, 1983. find
21. Malbran-Labat F., Manuel de langue akkadienne, Publications de l’institut Orientaliste de Louvain (50), Peeters, 2001. find
22. McCarthy J. J., « A Prosodic Theory of Nonconcatenative Morphology », Linguistic Inquiry, vol. 12, p. 373-418, 1981. find
23. Sanchez R., « Cours d’akkadien », , disponible sur http ://gedomia.ens-lsh.fr/, 2005. find
24. Wintner S., « Strengths and weaknesses of finite-state technology : A case study in morphological grammar development », Nat. Lang. Eng., vol. 14, n° 4, p. 457-469, 2008. find
25. Yli-Jyrä A. M., Koskenniemi K., « Compiling contextual restrictions on strings into finite-state automata », Proceedings of the Eindhoven FASTAR Days 2004 (September 3–4), Eindhoven, The Netherlands, December, 2004. find
26. Zajac R., « Feature Structures, Unification and Finite-State Transducers », In FSMNLP’98 : International Workshop, on Finite State Methods in Natural Language Processing., 1998. find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/barthelemy-2009-une-description-morphologique-structuree-en/