A Morphological Analyzer for Akkadian Verbal Forms with a Model of Phonetic Transformations

What it is about

A two-level analyser for Akkadian verb forms. The first level is a Prolog definite clause grammar (162 rules) that describes one paradigm for strong and weak verbs across the stems, slicing each form into nine parts (personal prefix, stem prefix, infix, first radical, second radical reduplication, and so on). The second level is a non-deterministic transducer of phonetic rules (assimilation, mutation, dissimilation) that is run backwards from an attested form to candidate roots. It was tested on 122 forms from a textbook (87 strong, 35 weak) and 54 forms from the Hammurabi code: all were recognised with the right reading, but the answers also include extra roots, and the transducer is exponential in length, taking hours for forms of about ten characters. The main weakness is preterite against imperfect for verbs whose second radical is weak, where only the vowel tells them apart.

Ideas

• I1. Use the vowel as the discriminator for weak second radicals: where the reduplication is lost, the stem vowel decides preterite or imperfect (the paper's example of two readings of one form). In the Hebrew verb module, keep the stem vowel as a feature of the candidate and not as a free choice.
• I2. Replace backtracking enumeration of weak-letter insertions with one shared structure for all candidates. The paper suggests a regular-expression intersection. For tools/hbo the candidate generation for unpointed tokens could be one automaton intersected with the verb paradigm, not a list of strings.
• I3. Build a gold list of forms with their recorded ambiguity (the paper's 60 strong forms, and the form with 15 candidate roots) and measure whether the true root is ranked first in tools/hbo.

For Mova

I1 goes into The weak verbs (weak second radical, aspect by vowel). I2 is a design idea for the candidate generation in tools/hbo. I3 is a test set: the Akkadian gold list would go to treebanks/akk if Akkadian is added, and the same measure runs on tools/hbo.

Topics: Akkadian and cuneiform NLP

Cite

The paper itself, not this card. BibTeX from the library's record of the work. DOI 10.3115/1621753.1621766

@inproceedings{barthelemy1998morphological,
  title = {{A Morphological Analyzer for Akkadian Verbal Forms with a Model of Phonetic Transformations}},
  author = {Barthelemy, Francois},
  booktitle = {Computational Approaches to Semitic Languages},
  year = {1998},
  doi = {10.3115/1621753.1621766},
  url = {https://doi.org/10.3115/1621753.1621766}
}

References

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

1. Kenneth R. Beesley. 1990. Finite-state description of Arabic morphology. In Proceedings of the Second Cambridge Conference on Bilingual Computing in Arabic and English, September 5-7. No pagination. find
2. Kenneth R. Beesley. 1996. Arabic finite-state morphological analysis and generation. In COLING'96, volume 1, pages 89-94, Copenhagen, August 5-9. Center for Sprogteknologi. The 16th International Conference on Computational Linguistics. find
3. Laura Kataja and Kimmo Koskenniemi. 1988. Finite-state description of Semitic morphology: A case study of Ancient Akkadian. In COLING'88, pages 313-315. find
4. Martin Kay. 1987. Nonconcatenative finite-state morphology. In Proceedings of the Third Conference of the European Chapter of the Association for Computational Linguistics, pages 2-10. find
5. George Kiraz. 1994. Multi-tape two-level morphology: a case study in Semitic non- linear morphology. In COLING'9,~, volume 1, pages 180-186. find our card
6. George Anton Kiraz. 1996. Computing prosodic morphology. In COLING'96. find

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/aclbulk-1998-barthelemy-morphological-analyzer-akkadian-verbal-forms/