cs.CLSep 23, 2026

COILD: An Indic-Centric Parallel Corpus and Benchmark for Machine Translation Across Indian Languages

Authors: Kshetrimayum Boynao Singh, Nitin Kumar Mishra, Palash Pratim Dutta, Atai Waris Khan, Aparna Kaushik, Avinash Kumar, Deeksha, Deepak Kumar, +17 more

Organizations: IIT Patna · IIT Delhi · IIT Guwahati · IIIT Delhi · IGDTUW · MIT-MAHE

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

Machine translation (MT) for Indian languages remains constrained by the limited availability of high-quality, Indic-centric parallel corpora and evaluation benchmarks. Existing multilingual resources are largely constructed from English-pivot content and often fail to capture the linguistic diversity, cultural complexity, and domain-specific characteristics of Indian languages. We present COILD, an Indic-centric parallel corpus comprising over 1.16 million human-translated and human-verified sentence pairs, covering 20 Indian language pairs across the Indo-Aryan, Dravidian, Tibeto-Burman, and Austro-Asiatic language families. The corpus is built entirely from original Indian language sources collected from licensed repositories spanning eight domains with direct real-world applicability. Furthermore, we introduce a domain-centric benchmark comprising 2,000 expert-verified sentences to enable consistent multilingual and cross-lingual evaluation across Indian language pairs. To validate the effectiveness of COILD, we fine-tune two representative multilingual neural machine translation models, IndicTrans2-Distilled and NLLB-200. Experimental results demonstrate consistent improvements across language pairs, domains, automatic evaluation metrics, and human evaluation, highlighting the effectiveness of high-quality Indic-centric supervision. COILD provides a valuable training and evaluation resource for advancing multilingual machine translation and future multilingual language models for Indian languages.

Figures & tables

Figure 1: Overview of the COILD framework and evaluation pipeline. The workflow consists of three stages: (1) dataset construction , using a dual-centric and bidirectional strategy with Hindi as the Indo-Aryan pivot and Tamil as the Dravidian pivot, resulting in 20 language pairs and 40 translation directions; (2) model fine-tuning , where IndicTrans2 and NLLB-200 are fine-tuned on the COILD parallel corpus; and (3) evaluation , comprising automatic evaluation using five MT metrics, domain-wise evaluation across eight application domains, and human evaluation based on translation adequacy, fluency, and overall quality using a 5-point Likert scale.
SourceTargetSentence PairsToken CountVocabulary
SourceTargetSourceTarget
HindiAssamese94,87219,86,17815,90,68076,9561,18,170
HindiBengali73,06614,89,73311,76,80458,93779,616
HindiBodo47,5999,88,0367,62,80247,69874,469
HindiDogri58,89611,93,78112,09,04053,38556,814
HindiGujarati96,61220,22,92416,84,92078,3951,15,035
Table 1: Statistics of the COILD corpus. The corpus contains Hindi-centric (Hindi →\rightarrow 17 target languages) and Tamil-centric (Tamil →\rightarrow 3 target languages) parallel corpora. We report the number of aligned sentence pairs, total token counts, and vocabulary sizes for the source and target sides. The COILD corpus contains approximately 1.16 million human-translated sentence pairs, comprising 44.79 million tokens across the source and target sides.
Figure 2: Domain-wise distribution of the COILD benchmark, comprising 2,000 Hindi source sentences across eight domains. The Education domain contains 600 sentences, while each of the remaining seven domains contains 200 sentences.
HyperparameterIndicTrans2NLLB-200
Epochs55
Learning Rate2×10−52\times 10^{-5}2×10−52\times 10^{-5}
LR SchedulerCosineLinear
Warm-up1%3%
OptimizerAdamW (8-bit)AdamW (8-bit)
Effective Batch Size6464
Table 2: Hyperparameter configuration used for fine-tuning IndicTrans2 and NLLB
Figure 3: Overall translation performance after fine-tuning on COILD for IndicTrans2-Distilled (top) and NLLB-200 (bottom). Heatmaps show the improvement over the corresponding pretrained baseline ( Δ\Delta = Fine-tuned −- Baseline) across five evaluation metrics: BLEU, chrF++, TER, BERTScore, and COMET. Columns represent translation directions, rows denote evaluation metrics, and larger positive values indicate better performance achieved through fine-tuning on the COILD corpus.
Appendix figures & tables4 assets

Supplementary material from the paper’s appendix.

Appendix

 IndicTrans2NLLB
DirectionAdequacyFluencyAdequacyFluency
 BaselineFinetuneΔ\DeltaBaselineFinetuneΔ\DeltaBaselineFinetuneΔ\DeltaBaselineFinetuneΔ\Delta
Assamese →\rightarrow Hindi2.162.490.333.013.350.342.522.710.193.423.540.12
Hindi →\rightarrow Assamese2.192.12-0.073.032.82-0.212.022.320.302.662.800.14
Bengali →\rightarrow Hindi3.864.220.364.154.440.293.433.890.473.524.020.49
Hindi →\rightarrow Bengali3.714.090.383.994.270.283.503.850.353.564.020.46
Appendix
Table 3: Human evaluation results comparing the baseline and fine-tuned versions of IndicTrans2 and NLLB. Adequacy and Fluency were rated by human annotators on a 5-point Likert scale (1 = Worst, 2 = Poor, 3 = Average, 4 = Good, 5 = Excellent). Positive Δ\Delta values indicate improvements after fine-tuning and are colour-coded to highlight the magnitude of performance changes after fine-tuning. Green shades indicate improvements, red shades indicate degradations. (Note: Odia Language human evaluation is underway.)
Figure 4: Per-domain change in BLEU after finetuning on COILD, for IndicTrans2 (top) and NLLB-200 (bottom). Columns are translation directions, rows are benchmark domains; cell values are Δ\Delta BLEU (finetune −- baseline).
Figure 5: Plot shows the overall baseline and finetune performance on the COILD Benchmark dataset for IndicTrans2.
Figure 6: Plot shows the overall baseline and finetune performance on the COILD Benchmark dataset for NLLB.

Explore similar work

CardsList

  1. English-to-Prakrit Machine Translation via Multilingual Transfer Learning

    Jun 4, 2026Om Choksi, Smit Kareliya, Shrikant Malviya +1Indian LanguagesText Corpora

    Sardar Vallabhbhai National Institute of Technology, Surat

  2. IndicTalk: A Large-Scale Persona-Based Multilingual Conversational Corpus for Indic Languages

    Jul 25, 2026Sahil Deepak Gawande, Mayank SinghConversational DatasetsIndian Languages

    Lingo Research Group · IIT Gandhinagar

  3. MultiSynt/MT: Trillion-Token Multi-Parallel Pre-Training Data Translated Across 36 Languages

    Jul 1, 2026Maximilian Idahl, Jörg Tiedemann, Sampo Pyysalo +19Multilingual Language ModelsLarge Language Model Pretraining

    ellamind · Leibniz University Hannover · University of Helsinki +14