Rule-based Surface Realization of Romanian Weak Pronouns

What it is about

Romanian weak pronouns, the clitics, take many surface forms that depend on their neighbours. They also appear in fixed clusters with a fixed order, and the cluster position depends on the sentence type. The paper implements a rule-based model that generates the right form of each pronoun from its context. It covers obligatory sandhi forms and, in addition, optional ones. The author also compiled a database of test inputs that cover the phenomena, and tested the model against it. The main claim is transparency: each generated form can be traced to the rule that produced it. The author notes that the resources were built by hand and may contain errors, and that the approach can be adapted to other constraint-based frameworks.

Ideas

I1. A phenomenon-based test suite as a gate. Compile test inputs by phenomenon, here the cluster order and the sandhi forms, and run the exact-match check before and after each change. This is the same idea as the paper's database, applied to the translator's function words.
I2. Make every surface choice carry its rule. Add the id of the rule that fired to the output of en/src/morph.rs, so that each form can be explained. This follows the transparency goal of the paper and fits Mova's explain requirement.
I3. Context-conditioned forms. Store the function-word forms of the target language by neighbour, in the same way the encoder of profile.rs falls back from concept plus tag plus neighbours to concept alone. A form that depends on the neighbour is then a table entry, not a special case.

For Mova

Romanian is not a Mova language, so only the method carries over. The rules are applied in a fixed order, which fits the stack of tools/translate. The fit is en/src/morph.rs (rule ids), the test-suite gate, and the context-conditioned table of the encoder.

Topics: Machine translation

Cite

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

@inproceedings{gerstenberger2025rule,
  title = {{Rule-based Surface Realization of Romanian Weak Pronouns}},
  author = {Gerstenberger, Ciprian},
  booktitle = {Proceedings of the 9th Workshop on Constraint Grammar and Finite State NLP},
  year = {2025},
  url = {https://aclanthology.org/2025.cgmta-1.2/}
}

References

22 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

Roxana-Maria Barbu and Ida Toivonen. 2018. Romanian Object Clitics: Grammaticalization, agreement and lexical splits. In Proceedings of the LFG18 Conference, page 67–87. Stanford: CSLI Publications. How weak Nordlyd, Bruce Hayes, Bruce Tesar, and Kie Zuraw. 2013. OTSoft 2.5 [software package]. http://www.linguistics.ucla.edu/people/hayes/otsoft/. ↗
Ronald M. Kaplan, John T. Maxwell III, Tracy Holloway King, and Richard Crouch. 2004. Integrating finite-state technology with deep LFG grammars. In Proceedings of the ESSLLI’04 Workshop on Combining Shallow and Deep Processing for NLP. find
Kenneth R. Beesley and Lauri Karttunen. 2003. Finite-State Morphology. CSLI Publications, Stanford, CA. find
Emily M. Bender and D. Terence Langendoen. 2010. Computational linguistics in support of linguistic theory. Linguistic Issues in Language Technology, 3. find
Udo-Michael Klein. 2007. Encoding of argument structure in Romanian and SiSwati. Ph.D. thesis, University of London. Available at http://hdl.handle.net/11858/00-001M-0000-0012-8EEC-D. ↗
Eulàlia Bonet. 1994. The person-case constraint: A morphological approach. MIT Working Papers in Linguistics, 0(22):33–52. find
Kimmo Koskenniemi. 1983. Two-level model for morphological analysis. In IJCAI, volume 83, pages 683–685. find
Anca Chereches , . 2014. A Prosodic Analysis of Romanian Pronominal Clitics. In University of Pennsylvania Working Papers in Linguistics, volume 20. Available at https://repository.upenn.edu/pwpl/vol20/iss1/7/. ↗
Géraldine Legendre. 2001. Positioning Romanian verbal clitics at PF. In (Gerlach and Grijzenhout, 2001). find
Noam Chomsky and Morris Halle. 1968. The Sound Pattern of English. Harper & Row. find
Paola Monachesi. 2001. Clitic placement in the Romanian verbal complex. In (Gerlach and Grijzenhout, 2001). find
Ann A. Copestake. 2001. Implementing typed feature structure grammars. In CSLI lecture notes series. find
Paola Monachesi. 2005. The Verbal Complex in Romance: A Case Study in Grammatical Interfaces. Oxford University Press. find
Carmen Dobrovie-Sorin. 1999. Clitics across categories: The case of Romanian. In Henk C. van Riemsdijk, editor, Clitics in the Languages of Europe. Mouton de Gruyter. find
John J Ohala. 2008. The emergent syllable. In The syllable in speech production. Taylor & Francis. find
Carmen Dobrovie-Sorin and Ion Giurgea, editors. 2013. A Reference Grammar of Romanian, volume 1: The noun phrase. John Benjamins. find
OpenAI. 2024. ChatGPT [Large Language Model]. https://chat.openai.com/chat. ↗
Alexandra Popescu. 2000. The morphophonology of the Romanian clitic sequence. In Lingua, volume 110, pages 773–799. Elsevier. find
Alexandra Popescu. 2003. Morphophonologis-che Phänomene des Rumänischen. Ph.D. thesis, University of Düsseldorf. Available at https://docserv.uni-duesseldorf.de/servlets/-DocumentServlet?id=3187. ↗
Kan Sasaki and Daniela Căluianu. 2000. An Optimality Theoretic Account for the Distribution of Pronominal Clitics in Romanian. Technical report, University of Tsukuba. Available at https://www.academia.edu/33668351/An_optimality_theoretic_account_for_the_distribution_of_pronominal_clitics_in_Romanian. ↗
Stanca Somesfalean. 2007. On the Form and Interpretation of Clitics. Ph.D. thesis, Université du Québec à Montréal. Available at https://ar-chipel.uqam.ca/9628/. ↗
Oana Săvescu Ciucivara. 2009. A Syntactic Analysis of Pronominal Clitic Clusters in Romance - The view from Romanian. Ph.D. thesis, New York University. Available at https://www.proquest.com/docview/304954033. ↗

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/gerstenberger-2025-rule-based-surface-realization-romanian/