Semitic Morphological Analysis and Generation Using Finite State Transducers with Feature Structures

What it is about

Gasser builds a finite-state analyser and generator for Tigrinya verbs, an Ethio-Semitic language with no digitised root dictionary. A stem is a root plus a derivational pattern, and the pattern is described by four binary features (passive, transitive, iterative, reciprocal), giving eight categories. Instead of a multi-tape machine, or a precompiled list of stems, the transducer's arcs carry feature-structure weights that unify along a path. The root and the pattern stay separate, and long-distance agreement, such as the negative circumfix that wraps the verb, is enforced by unification. The full orthographic network has 22,313 states and 118,927 arcs. On 400 random wordforms from a web word list, 308 of the 329 unambiguous verbs were analysed correctly, and the generator produced the correct form in all 272 tests. Without a root dictionary, one word can receive up to 15 analyses.

Ideas

• I1. Agreement as weights: a feature set written at the start of a word and checked at its end, instead of duplicated subnetworks for the two branches. The closest case in our languages is the prefix and suffix agreement of Aramaic verbs (The verb in Biblical Aramaic), and the prefix chain of the tools/hbo segmenter (conjunction, preposition and article before a word) is a test of the same mechanism.
• I2. A root filter: once a list of roots exists, compose a root acceptor onto the analyser so that analyses with non-existent roots are rejected. Hebrew has no root list yet (macula dict has none), so the paper's gap is ours too; a roots table from tools/macula or the Wiktionary-derived tools would close it.
• I3. A measure: a random sample of 400 wordforms, the share of verbs analysed correctly, and the number of analyses per form as an ambiguity figure. Use the same design, with a human check, to report unpointed tools/hbo output.

For Mova

I1 goes into The verb in Biblical Aramaic and the prefix chain of tools/hbo. I2 goes into The lexicon and tools/macula (roots). I3 is an evaluation for tools/hbo, dated when run.

Cite

The paper itself, not this card. BibTeX from the library's record of the work. DOI 10.3115/1609067.1609101

@inproceedings{gasser2009semitic,
  title = {{Semitic Morphological Analysis and Generation Using Finite State Transducers with Feature Structures}},
  author = {Gasser, Michael},
  booktitle = {Proceedings of the 12th Conference of the European Chapter of the ACL (EACL 2009)},
  year = {2009},
  doi = {10.3115/1609067.1609101},
  url = {https://doi.org/10.3115/1609067.1609101}
}

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.

1. Saba Amsalu and Girma A. Demeke. 2006. Non-concatenative finite-state morphotactics of Amharic simple verbs. ELRC Working Papers, 2(3). find
2. Jan Amtrup. 2003. Morphology in machine translation systems: Efficient integration of finite state transducers and feature structure descriptions. Machine Translation, 18:213–235. find
3. Kenneth R. Beesley and Lauri Karttunen. 2003. Finite State Morphology. CSLI Publications, Stanford, CA, USA. find
4. Eugene Buckley. 2000. Alignment and weight in the Tigrinya verb stem. In Vicki Carstens and Frederick Parkinson, editors, Advances in African Linguistics, pages 165–176. Africa World Press, Lawrenceville, NJ, USA. find
5. Bob Carpenter. 1992. The Logic of Typed Feature Structures. Cambridge University Press, Cambridge. find
6. Noam Chomsky and Morris Halle. 1968. The Sound Pattern of English. Harper and Row, New York. find
7. Yael Cohen-Sygal and Shuly Wintner. 2006. Finite-state registered automata for non-concatenative morphology. Computational Linguistics, 32:49–82. find
8. Ann Copestake. 2002. Implementing Typed Feature Structure Grammars. CSLI Publications, Stanford, CA, USA. find
9. Yitna Firdyiwek and Daniel Yaqob. 1997. The system for Ethiopic representation in ascii. URL: citeseer.ist.psu.edu/56365.html. find
10. C. Douglas Johnson. 1972. Formal Aspects of Phonological Description. Mouton, The Hague. find
11. Ronald M. Kaplan and Martin Kay. 1994. Regular models of phonological rule systems. Computational Linguistics, 20:331–378. find
12. Lauri Karttunen, Ronald M. Kaplan, and Annie Zaenen. 1992. Two-level morphology with composition. In Proceedings of the International Conference on Computational Linguistics, volume 14, pages 141–148. find
13. George A. Kiraz. 2000. Multitiered nonlinear morphology using multitape finite automata: a case study on Syriac and Arabic. Computational Linguistics, 26(1):77–105. find our card
14. Kimmo Koskenniemi. 1983. Two-level morphology: a general computational model for word-form recognition and production. Technical Report Publication No. 11, Department of General Linguistics, University of Helsinki. find
15. Wolf Leslau. 1941. Documents Tigrigna: Grammaire et Textes. Libraire C. Klincksieck, Paris. find
16. Mehryar Mohri, Fernando Pereira, and Michael Riley. 2000. Weighted finite-state transducers in speech recognition. In Proceedings of ISCA ITRW on Automatic Speech Recognition: Challenges for the Millenium, pages 97–106, Paris. find

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/aclbulk-2009-gasser-semitic-morphological-analysis-generation-finite/