The Karamel System and Semitic Languages: Structured Multi-Tiered Morphology

What it is about

Karamel is a finite-state system for multi-tape transducers. A typed Cartesian product ("unit") groups the tapes into a tree, so that a root, a template, affixes and clitics can each sit on their own tape and in their own node. Feature structures are compiled statically into the automaton, and generalized restriction rules (contextual constraints applied at once) define the surface forms. Karamel is compared with Xerox's finite-state tools, with MAGEAD (Arabic dialects, one level of structure) and with MMORPH. The Akkadian verb is the worked example: a description with five tapes (feature structures for big and small grains, a canonical form, and two dialect surface forms), in 27 regular expressions, 22 let definitions and 6 rules. The grammar covers strong verbs and verbs with one weak radical, and it is not yet type-checked. The paper gives no accuracy figure.

Ideas

• I1. Make the parse of a verb form a tree of units rather than a flat string of splits: a big grain for the core (root and vowels), one for the personal affix, one for the clitics, and small grains inside the core. Today the segmenter in tools/hbo splits into a flat list; the same tree could carry the root and pattern as separate nodes for the verb analysis.
• I2. Compile the person, gender and number suffix as a typed feature structure with a finite domain, and write the vowel of a stem as a finite table over aspect and stem. This is a rule-based table, easy to check against a paradigm, and it can be the first verb model in The verb.
• I3. Write the weak-verb surface changes as contextual rules that look at the lexical and the surface tape at once (the paper's deletion of support vowels before a vowel is the model case), and test them on the paradigm forms of MACULA trees.

For Mova

I1 is a design note for the output of tools/hbo (segmenter before gparse) and for 13. Biblical Hebrew in Mova: the tools. I2 goes into The verb. I3 goes into The weak verbs as a set of rules, with a test set from tools/macula.

Cite

The paper itself, not this card. BibTeX from the library's record of the work. DOI 10.3115/1621774.1621778

@inproceedings{barthelemy2009karamel,
  title = {{The Karamel System and Semitic Languages: Structured Multi-Tiered Morphology}},
  author = {Barthélemy, François},
  booktitle = {Proceedings of the EACL 2009 Workshop on Computational Approaches to Semitic Languages},
  year = {2009},
  doi = {10.3115/1621774.1621778},
  url = {https://doi.org/10.3115/1621774.1621778}
}

References

16 references, 3 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Cyril Allauzen, Michael Riley, Johan Schalkwyk, Wojciech Skut, and Mehryar Mohri. 2007. Openfst: A general and efficient weighted finite-state transducer library. In Implementation and Application of Automata, 12th International Conference, CIAA, volume 4783 of LNC, pages 11–23, Prague, Czech Republic. find
2. François Barthélemy. 2007a. Finite-state compilation of feature structures for two-level morphology. In International Workshop on Finite State Methods in Natural Language Processing (FSMNLP), Potsdam, Germany. find
3. François Barthélemy. 2007b. Multi-grain relations. In Implementation and Application of Automata, 12th International Conference (CIAA), pages 243–252, Prague, Czech Republic. find
4. Kenneth R. Beesley and Lauri Karttunen. 2003. Finite State Morphology. CSLI Publications. find
5. Kenneth R. Beesley. 1998. Arabic morphology using only finite-state operations. find our card
6. Edmund Grimley-Evans, George Kiraz, and Stephen Pulman. 1996. Compiling a partition-based two-level formalism. In COLING, pages 454–459, Copenhagen, Denmark. find
7. Nizar Habash and Owen Rambow. 2006. Magead: a morphological analyzer and generator for the Arabic dialects. In ACL: Proceedings of the 21st International Conference on Computational Linguistics and the 44th annual meeting of the Association for Computational Linguistics, pages 681–688. find our card
8. Lauri Karttunen and Kenneth R. Beesley. 2000. Finite-state non-concatenative morphotactics. In Fifth Workshop of the ACL Special Interest Group in Computational Phonology, pages 1–12, Luxembourg (Luxembourg). find
9. Lauri Karttunen. 1995. The replace operator. In ACL-95, pages 16–23, Boston, Massachusetts. Association for Computational Linguistics. find
10. George Anton Kiraz. 2000. Multitiered nonlinear morphology using multitape finite automata: a case study on Syriac and Arabic. Comput. Linguist., 26(1):77–105. find our card
11. Kimmo Koskenniemi. 1983. Two-level model for morphological analysis. In IJCAI-83, pages 683– 685, Karlsruhe, Germany. find
12. Florence Malbran-Labat. 2001. Manuel de langue akkadienne. Publications de l’institut Orientaliste de Louvain (50), Peeters. find
13. John J. McCarthy. 1981. A prosodic theory of nonconcatenative morphology. Linguistic Inquiry, 12:373– 418. find
14. D. Petitpierre and G. Russel. 1995. Mmorph: the mul-tex morphology program. Technical report, ISSCO technical report, Geneva, Switzerland. find
15. S. Pulman and M. Hepple. 1993. A feature-based formalism for two-level phonology. Computer Speech and Language, 7. find
16. Anssi Mikael Yli-Jyrä and Kimmo Koskenniemi. 2004. Compiling contextual restrictions on strings into finite-state automata. In Proceedings of the Eindhoven FASTAR Days 2004 (September 3–4), Eindhoven, The Netherlands, December. find

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/aclbulk-2009-barthelemy-karamel-system-semitic-languages-structured/