cs.CLAug 25, 2026

EnSiTa - A Trilingual Multi-Domain Parallel Dataset and Benchmark for Domain-Specific Machine Translation

Authors: Surangika Ranathunga, Nisansa de Silva, Aloka Fernando, Kavindu Warnakulasuriya, Isuru Wijesiri, Menan Velayuthan, Charitha Rathnayaka, Thivaharan Varatharajan, 

Organizations: Massey University · University of Moratuwa · National University of Singapore · WSO2 · 

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

Machine Translation (MT) for low-resource languages remains far behind that of high-resource languages, and the gap is widest in specialised domains, where parallel data is scarce or entirely absent. We present EnSiTa, a trilingual multi-domain parallel dataset and benchmark for English, Sinhala and Tamil. EnSiTa provides human post-edited training data for seven domains, plus manually translated test sets for those and one additional domain, all produced by professional translators under a multi-year, rigorously quality-controlled process. Using this dataset, we conduct an extensive study of domain-specific MT for all six language directions, fine-tuning a from-scratch Transformer, a pre-trained translation model (NLLB-600M), and decoder-only LLMs (Gemma 3 family, 1B-12B, and TranslateGemma) across training-data sizes, model scales, and in-domain, cross-domain, multilingual and multi-domain settings. To the best of our knowledge, this is the most extensive systematically documented multi-domain parallel data creation and benchmarking effort for low-resource MT. Our data and models will be publicly released.

Figures & tables

Table 1: The sources from which the parallel and monolingual data were obtained for corpus preparation
#DomainDir.TrainDevTest
3 en-si1,000406890
 en-ta2,0006301,000
Healthen-si-ta––784 †
5 en-si6,6741,0001,000
 en-ta3,500511864
 si-ta––831 †
Table 2: Statistics of corpora used for experiments. Last three have been borrowed from other sources. Totals exclude trilingual (en-si-ta) subsets marked † , which are derived from or largely overlapping with the bilingual pairs listed for the same domain and split
Figure 1: Distribution of English Sentence Lengths by Domain (Logarithmic Scale)
Figure 2: Pairwise JSD values between domains, computed on English training samples
Figure 3: KL Divergence Across All Language Pair Directions by Dataset
Figure 4: Source-domain (columns) vs cross-domain (rows) results for EnSi direction (chrF++)
Figure 5: Language direction-wise results of models trained and tested with Government data (chrF++)
Figure 6: Cross-domain results of Government-trained models for EnSi direction (chrF++)
Figure 7: Strategy comparison across training data (chrF++) for EnSi
Appendix figures & tables6 assets

Supplementary material from the paper’s appendix.

Appendix

 Gemma 3Translate
 (1B/4B/12B)Gemma (4B/12B)
LoRA rank rr3216
LoRA α\alpha3216
LoRA dropout0.00.05
Adapted modulesattn + MLPattn + MLP
Learning rate2×10−52\times 10^{-5}1×10−51\times 10^{-5}
Appendix
Table 3: Training hyperparameters for LoRA fine-tuning. The Gemma 3 recipe is held fixed across the 1B/4B/12B variants for the model-scaling study; TranslateGemma uses a lighter recipe to limit erosion of its pre-existing translation tuning. The effective batch size (32) is kept constant across model sizes by trading per-device batch against gradient accumulation.
Figure 8: Guideline document given to the translators for the Wikipedia English-Sinhala task. - Page 1
Figure 9: Guideline document given to the translators for the Wikipedia English-Sinhala task. - Page 2
Figure 10: Guideline document given to the translators for the Wikipedia English-Sinhala task. - Page 3
Figure 11: Guideline document given to the translators for the Wikipedia English-Sinhala task. - Page 4
Figure 12: Guideline document given to the translators for the Wikipedia English-Sinhala task. - Page 5
More in this topic

Machine Translation

Machine translation develops computational methods for converting text, speech, and visual content between languages while preserving meaning and purpose.

See all 267 papers

Explore similar work

CardsList

  1. BhashaSetu: A Data-Centric Approach to Low-Resource Machine Translation

    May 26, 2026Param Thakkar, Anushka Yadav, Michael Tiemann +3Low-Resource MTMachine Translation

    Department of Computer Engineering and Information Technology, Veermata Jijabai Technological Institute, Mumbai · Tübingen AI Center, University of Tübingen, Germany

  2. Systematic Analysis of Large Language Models and Transformer-Based Machine Translation for English-Tamil and Tamil-English Across Diverse Datasets

    Jul 27, 2026Sriharshaa S, Sangeetha Sivanesan, Jaya Nirmala SMultilingual Language ModelsLow-Resource MT

    SASTRA Deemed to be University · National Institute of Technology

  3. Omnilingual MT: Machine Translation for 1,600 Languages

    Mar 17, 2026Omnilingual MT Team, Belen Alastruey, Niyati Bafna +29Machine TranslationLow-Resource MT

    FAIR at Meta