Pith. sign in

REVIEW 5 major objections 5 minor 1 cited by

Advancing Dialectal Arabic to Modern Standard Arabic Machine Translation

T0 review · 5 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read A 9-billion-parameter open model, fine-tuned with LoRA and 4-bit quantization on a 62,775-example Arabic corpus, outperforms zero-shot GPT-4o on translating three Arabic dialects into Modern Standard Arabic, 49.88 vs. 44.58 chrF++.

desk verdict Useful honest empirical paper on DA-MSA MT, but the headline fine-tuning vs GPT-4o gap is not fully clean because the dev set used for hyperparameter selection overlaps the test set and there are no error bars. read the letter →

arxiv 2507.20301 v2 pith:ZGJSDLNX submitted 2025-07-27 cs.CL

Abdullah Alabdullah , Lifeng Han , Chenghua Lin This is my paper · ORCID
classification cs.CL
keywords MachineTranslationDialectalArabicModernStandardEvaluationLargeLanguageModelFine-TuningLow-RankAdaptationQuantization
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper tries to establish that high-quality machine translation from three Arabic dialects (Levantine, Egyptian, and Gulf) into Modern Standard Arabic is achievable with modest compute: a 9B open model fine-tuned with Low-Rank Adaptation and 4-bit quantization on a 62,775-example curated corpus reaches a chrF++ of 49.88, beating zero-shot GPT-4o (44.58) by 5.3 points. It also demonstrates that, among training-free prompting strategies, few-shot prompting with a few dialect-specific examples consistently beats zero-shot, chain-of-thought, and the paper's own three-stage Ara-TEaR self-refinement method. If these results hold, they matter because dialectal Arabic is the dominant everyday written register across the Arab world, yet most NLP systems handle only MSA; the paper offers a practical, low-resource route to dialectal inclusion, including a 60% memory cut from quantization with negligible quality loss.

What carries the argument

The load-bearing mechanism is a resource-efficient fine-tuning pipeline built on the Gemma-2-9B decoder-only transformer, adapted with Low-Rank Adaptation (LoRA, rank and alpha 16) and 4-bit quantization, and trained for one epoch at learning rate 5e-5 on an Alpaca-style instruction template that omits explicit dialect labels. The data side is a curated three-dialect corpus: MADAR's professionally translated travel-domain pairs, Dial2MSA-Verified's native-speaker-validated social-media pairs, and 5,000 MSA–MSA pairs from the UFAL North Levantine corpus used as same-source-target regularization. Evaluation is carried by chrF++, a character- and word-level F-score chosen because Arabic morphology makes word-overlap metrics like BLEU too rigid; the test set is 1,200 sentences drawn 70% from MADAR and 30% from Dial2MSA-Verified.

What would settle it

Hold out a set of dialectal Arabic sentences from domains and dialects not in the training corpus (for example, Maghrebi tweets or Levantine forum comments), have native speakers rate the fine-tuned Gemma2-9B and GPT-4o outputs for meaning preservation and MSA formality, and check whether the 5.3-point chrF++ advantage survives; if the human ranking reverses or the gap disappears, the core claim fails to generalize.

Watch

Extended reading notes

Core claim

The central discovery, on the authors' terms, is that task-specific fine-tuning of a mid-sized open model outweighs raw scale for DA–MSA translation. A Gemma-2-9B model, fine-tuned for one epoch with LoRA rank 16 and 4-bit quantization on the Extended-Gold corpus (MADAR plus Dial2MSA-Verified, 62,775 pairs), scores 49.88 chrF++ on the Large-Test set, 5.3 points above zero-shot GPT-4o (44.58) and above its own larger sibling Gemma-27B (48.49). The ablation series attributes the gain to joint multi-dialect training (10.6% better than an Egyptian-only model), the addition of the social-media Dial2MSA-Verified data, and a one-epoch schedule to avoid overfitting; 4-bit quantization costs under 1% chrF++ while cutting memory 60% and speeding inference 22%. In contrast, synthetically paraphrased training data and the longest-sentence active-learning subset both degraded quality, and the proposed Ara-TEaR refinement prompt never beat plain few-shot prompting.

Load-bearing premise

The load-bearing premise is that the automatic metric scores on a test set drawn mostly from the same genre as the training data reflect how well the system translates dialectal Arabic that people actually write; if that metric or that test composition favors the fine-tuned model, its 5.3-point lead over GPT-4o could vanish on other dialectal text.

Editorial extensions

If this is right

  • If the central claim holds, an open 9B model fine-tuned on roughly 63,000 gold examples can outperform a much larger closed API model for dialect-to-MSA translation, which lowers the barrier to building dialectal Arabic technology in resource-limited settings.
  • Pooling scarce dialect data across dialects beats separate per-dialect models, so future low-resource efforts should prefer joint training over dialect-specific systems.
  • 4-bit quantization with LoRA makes dialectal MT feasible on consumer hardware: roughly 60% less memory and 22% faster inference for under 1% chrF++ loss.
  • Few-shot prompting with a few dialect-labeled examples is the cheapest reliable training-free strategy, and explicitly naming the source dialect in the prompt helps, especially for smaller models.
  • Synthetic paraphrase augmentation and naive longest-sentence active learning can hurt translation quality, so careful data curation matters more than raw data volume.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The 70% MADAR share of the test set likely favors the fine-tuned model, since its training data comes from the same travel-domain corpus; a dialect-balanced, out-of-domain test could narrow or reverse the 5.3-point gap over GPT-4o.
  • The paper rejects embedding-based metrics like COMET based on unreported preliminary experiments, so a human or semantic-adequacy evaluation might rank the systems differently than chrF++ does.
  • Because the fine-tuned model was trained without explicit dialect labels, explicitly naming the dialect at inference time could improve its outputs further, mirroring the gain observed for prompted models.
  • The public release of the paraphrase-augmented dataset lets others test filtered augmentation ratios (for example, one part synthetic to three parts gold), which the authors could not run; such experiments could still rescue the augmentation idea.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 5 minor

Summary. The paper presents two approaches for dialectal Arabic (DA) to Modern Standard Arabic (MSA) machine translation under low-resource constraints: (i) a systematic evaluation of training-free prompting strategies (zero-shot, zero-shot chain-of-thought, few-shot, and a proposed three-stage self-refinement method Ara-TEaR) across six large language models, and (ii) a resource-efficient fine-tuning pipeline based on quantized LoRA fine-tuning of smaller models, with ablations covering data scale, augmentation, architecture, quantization, and active learning. The headline claim is that a 4-bit quantized Gemma2-9B model fine-tuned on the Extended-Gold dataset (62,775 examples) reaches a chrF++ score of 49.88 on a 1,200-sentence test set, outperforming zero-shot GPT-4o (44.58). The prompting study finds that few-shot prompting is generally best and that Ara-TEaR provides no significant gain, while the fine-tuning ablations report that joint multi-dialect training helps, synthetic augmentation hurts, and 4-bit quantization retains quality at large memory savings.

Significance. If the headline comparison holds, the paper provides a practically valuable result for low-resource dialectal MT, showing that an open 9B model can outperform a much larger closed API model on this task. The paper's strengths include releasing data and code, running a structured set of eight ablations, and honestly reporting negative results (Ara-TEaR, augmentation, active learning). However, the central comparison is currently on unequal footing because of test-set overlap with the development set used for hyperparameter selection, and the statistical analysis of the prompting study contains a reversed sphericity criterion. These issues must be repaired before the main claims can be considered reliable.

major comments (5)
  1. [§3.4.1, §4.2.2, Table 2] The central comparison in Table 7 (Gemma2-9B 49.88 vs. GPT-4o 44.58) is compromised by an overlap between the development and test sets. Section 3.4.1 states that Small-Test (600 MADAR sentences) was used as a development set during hyperparameter optimization, and Section 4.2.2 reports that learning rate and epochs were selected on it. Table 2 shows that Large-Test (1,200 sentences) samples 840 sentences from the same MADAR test split, and the text explicitly acknowledges that the two test sets 'might contain potential overlapping examples' but does not deduplicate them. Any MADAR sentence in Large-Test that was also in Small-Test was used to select hyperparameters before the final score was computed, while GPT-4o received no such tuning. This is a selection-bias asymmetry, not a pure translation-quality difference. Please deduplicate the test sets (or hold out a fresh disjoint test set) and re-evaluate both the fine-tuned and prompted models.
  2. [§4.1.1] The sphericity criterion for Mauchly's test is stated backwards: 'the assumption being met if p≤0.05.' In standard repeated-measures ANOVA, sphericity is assumed when p > 0.05 and violated when p ≤ 0.05. As written, the verification procedure would accept a violation as satisfaction, invalidating the statistical-significance claims in Table 6 and the statements that few-shot 'consistently' outperforms other strategies or that Ara-TEaR fails to reach significance. Re-run the ANOVA with the correct criterion, and report effect sizes or corrected degrees of freedom (e.g., Greenhouse-Geisser) where appropriate.
  3. [§4.1.2 vs. Table 5] The dialect-identification experiment reports GPT-4o zero-shot chrF++ of 40.52 with explicit dialect names, but Table 5 reports GPT-4o zero-shot as 42.18 on the same Small-Test set. These numbers cannot both be correct for the same prompt and test set; either the §4.1.2 experiment used a different subset or prompt, or one of the numbers is wrong. This inconsistency undermines the conclusion that explicit dialect identification improves translation quality. Please clarify the experimental setup and reconcile the scores.
  4. [§4.2.7 and Abstract] The abstract's claim that '4-bit quantization reduced memory usage by 60% with less than 1% performance loss' is based on LLaMA-3.2-3B, not on the headline Gemma2-9B model. Section 4.2.7 explicitly fine-tunes LLaMA-3.2-3B in full precision versus 4-bit, and the memory/inference numbers apply to that 3B model. Generalizing this to the Gemma2-9B pipeline is not supported by the experiments. Report quantization results for the model actually used in the main comparison, or qualify the abstract and conclusions accordingly.
  5. [§4.2.2–§4.2.8] No multi-seed runs, confidence intervals, or significance tests are reported for any fine-tuning result. All differences are point estimates on a single evaluation; for example, the 49.88 vs. 48.49 gap between Gemma2-9B and Gemma2-27B in Table 7 could easily be within noise. At minimum, provide bootstrap confidence intervals for chrF++ on the test sets, or run 2–3 seeds for the central configurations, so the reader can judge the stability of the ranking.
minor comments (5)
  1. [§4.2.1, Tables 8/10/12] Typographical issues: 'acorss' should be 'across'; 'BELU' in the Levantine, Gulf, and Egyptian BLEU tables should be 'BLEU'; 'HypperParam' in the experiment names should be 'HyperParam'.
  2. [§3.3.3] The implementation details mention evaluating Claude 3.5 Sonnet via the Anthropic API, but Claude 3.5 Sonnet is not listed in Table 3 nor in any results table; either add the model and its results or remove the mention.
  3. [§3.2] The dismissal of COMET and BERTScore is based on 'preliminary experiments' with no details; please add a brief description of those experiments or a citation to prior work showing similar limitations for dialectal Arabic.
  4. [Appendix] The appendix numbering is confusing: '6 Challenges and Limitations' appears after the conclusions, followed by '7 Further Work', '8 Detailed Prompting Results', and '9 Prompting Templates'. Renumber or restructure the sections for clarity.
  5. [§3.1] The comparison of test-set sizes with shared tasks (NADI-2024, OSACT-2024) would be more informative if the corresponding chrF++ scores from those shared tasks were also reported, since the reader currently cannot calibrate the absolute numbers.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: all central claims are empirical evaluations against external corpora, with no derivation chain that reduces to its own inputs.

full rationale

This paper is an empirical study, not a derivation. The central claims—(i) few-shot prompting outperforms other prompting strategies, (ii) a fine-tuned 4-bit quantized Gemma2-9B reaches chrF++ 49.88 on Large-Test versus GPT-4o’s 44.58, (iii) joint multi-dialect training and quantization trade-offs—are all measured outcomes on externally sourced corpora (MADAR, Dial2MSA-Verified, UFAL). No step fits a parameter and then renames the fit as a prediction; hyperparameters (learning rate, epochs) are tuned on Small-Test and the final model is evaluated on Large-Test, with the paper explicitly acknowledging that the two sets ‘might contain potential overlapping examples’ (Section 3.1). That is a potential evaluation-validity / data-leakage concern, not circularity: the 49.88 score is not equivalent by construction to the dev-set score, and the overlap is disclosed as a limitation rather than hidden. The proposed Ara-TEaR method is empirically tested and found not to help, so it is not a self-justifying input to any conclusion. Self-citations (e.g., Han et al. 2021, 2024) appear only as background support for metric choice and fine-tuning practice and are not load-bearing for the paper’s headline comparisons. The paper’s own limitation statement that SacreBLEU and chrF++ ‘are likely incapable of fully capturing meaning preservation’ is an honest caveat about metric validity, not a circular step. No equation, definition, or fitted parameter reduces the results to their inputs; the claims therefore stand on independent empirical evaluation rather than on self-reference.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

All listed parameters are empirical design choices rather than theoretical constants; the central claim depends on them only through the specific experimental configuration. The main non-empirical assumptions are corpus reliability and metric validity.

free parameters (6)
  • learning_rate = 5e-5
    Chosen on the Small-Test development set; 1e-4 scored 43.72 vs 43.03 but was deemed less stable (Section 4.2.2).
  • training_epochs = 1
    One epoch (45.64) beat three epochs (40.59) on the validation set; all subsequent experiments used one epoch (Section 4.2.2).
  • LoRA_rank = 16
    LoRA rank and alpha set to 16; treated as a fixed standard value, not tuned (Section 3.4.2).
  • test_set_composition = 70% MADAR / 30% Dial2MSA-Verified
    Large-Test composition chosen because MADAR is considered more reliable; this weighting affects the reported 49.88 vs 44.58 comparison (Section 3.4.1).
  • few_shot_examples = 3 per dialect
    Three representative examples (conversational, descriptive, idiomatic) were hand-picked per dialect; no systematic variation of example count or content was performed (Section 3.3.1).
  • quantization_bits = 4
    4-bit quantization is the key efficiency claim; the ablation compared only 16-bit vs 4-bit on LLaMA-3.2-3B (Section 4.2.7).
assumptions (5)
  • domain assumption MADAR and Dial2MSA-Verified corpora contain correct DA-MSA translations
    Both training and test data come from these corpora; errors in labels or translations would bias all reported scores (Section 3.1).
  • domain assumption chrF++ is a valid proxy for DA-MSA translation quality
    chrF++ is the primary metric; embedding-based metrics were rejected based on unreported preliminary experiments. The authors themselves acknowledge this limitation (Sections 3.2, 6).
  • domain assumption The sampled test sets are representative of the three dialects
    Small-Test (600) and Large-Test (1,200) are random samples from MADAR/Dial2MSA test splits; no power analysis or stratification by text type is given (Section 3.1).
  • domain assumption Single-sampling LLM outputs at temperature 0.3 are stable
    All prompting results are based on one generation per item; no repeated sampling or variance estimate is reported (Section 3.3.3).
  • domain assumption Fixed seed 3407 makes training deterministic and representative
    All fine-tuning uses a single seed; no multiple-seed variance is reported, so differences between configurations may reflect noise (Section 3.4.2).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Advancing Dialectal Arabic to Modern Standard Arabic Machine Translation." pith.science (2026). https://pith.science/paper/ZGJSDLNX

@misc{pith2026250720301,
  author       = {Pith},
  title        = {Pith review of: Advancing Dialectal Arabic to Modern Standard Arabic Machine Translation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZGJSDLNX}},
  note         = {Machine review of arXiv:2507.20301}
}
read the original abstract

Dialectal Arabic (DA) poses a persistent challenge for natural language processing (NLP), as most everyday communication in the Arab world occurs in dialects that diverge significantly from Modern Standard Arabic (MSA). This linguistic divide impedes progress in Arabic machine translation. This paper presents two core contributions to advancing DA-MSA translation for the Levantine, Egyptian, and Gulf dialects, particularly in low-resource and computationally constrained settings: (i) a comprehensive evaluation of training-free prompting techniques, and (ii) the development of a resource-efficient fine-tuning pipeline. Our evaluation of prompting strategies across six large language models (LLMs) found that few-shot prompting consistently outperformed zero-shot, chain-of-thought, and our proposed Ara-TEaR method. Ara-TEaR is designed as a three-stage self-refinement prompting process, targeting frequent meaning-transfer and adaptation errors in DA-MSA translation. In this evaluation, GPT-4o achieved the highest performance across all prompting settings. For fine-tuning LLMs, a quantized Gemma2-9B model achieved a chrF++ score of 49.88, outperforming zero-shot GPT-4o (44.58). Joint multi-dialect trained models outperformed single-dialect counterparts by over 10% chrF++, and 4-bit quantization reduced memory usage by 60% with less than 1% performance loss. The results and insights of our experiments offer a practical blueprint for improving dialectal inclusion in Arabic NLP, showing that high-quality DA-MSA machine translation is achievable even with limited resources and paving the way for more inclusive language technologies.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Biomedical Machine Translation for Low-Resource Arabic-Script Languages via Cross-Lingual Transfer and LoRA Adapter Merging

    cs.CL 2026-07 conditional novelty 6.0 of 10

    Zero-data merging of Arabic and Persian biomedical LoRA adapters comes within 1.4–3.5 CHrF++ points of supervised adaptation for Dari; Pashto and Sorani Kurdish stay below usable quality.

Reference graph

Works this paper leans on

64 extracted references · 42 canonical work pages · cited by 1 Pith paper

  1. [1]

    write newline

    " write newline "" before.all 'output.state := FUNCTION fin.entry add.period write newline FUNCTION new.block output.state before.all = 'skip after.block 'output.state := if FUNCTION new.sentence output.state after.block = 'skip output.state before.all = 'skip after.sentence 'output.state := if if FUNCTION not #0 #1 if FUNCTION and 'skip pop #0 if FUNCTIO...

    work page

  2. [2]

    , " * write output.state after.block = add.period write newline

    ENTRY address archive author booktitle chapter doi edition editor eid eprint howpublished institution journal key month note number organization pages publisher school series title type url volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.all ...

  3. [3]

    write newline

    " write newline "" before.all 'output.state := FUNCTION add.period duplicate empty 'skip "." * add.blank if FUNCTION if.digit duplicate "0" = swap duplicate "1" = swap duplicate "2" = swap duplicate "3" = swap duplicate "4" = swap duplicate "5" = swap duplicate "6" = swap duplicate "7" = swap duplicate "8" = swap "9" = or or or or or or or or or FUNCTION ...

  4. [4]

    , " * write output.state after.block = add.period write newline

    ENTRY address author booktitle chapter doi edition editor eid howpublished institution journal key month note number organization pages publisher school series title type url volume year label INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.all := #1 'mid.sentence := #2 'after.sentence := #3 '...

  5. [5]

    write newline

    " write newline "" before.all 'output.state := FUNCTION if.digit duplicate "0" = swap duplicate "1" = swap duplicate "2" = swap duplicate "3" = swap duplicate "4" = swap duplicate "5" = swap duplicate "6" = swap duplicate "7" = swap duplicate "8" = swap "9" = or or or or or or or or or FUNCTION n.separate 't := "" #0 'numnames := t empty not t #-1 #1 subs...

  6. [6]

    , " * write output.state after.block = add.period write newline

    ENTRY address author booktitle chapter doi edition editor eid howpublished institution journal key month note number organization pages publisher school series title type url volume year label INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.all := #1 'mid.sentence := #2 'after.sentence := #3 '...

  7. [7]

    write newline

    " write newline "" before.all 'output.state := FUNCTION if.digit duplicate "0" = swap duplicate "1" = swap duplicate "2" = swap duplicate "3" = swap duplicate "4" = swap duplicate "5" = swap duplicate "6" = swap duplicate "7" = swap duplicate "8" = swap "9" = or or or or or or or or or FUNCTION n.separate 't := "" #0 'numnames := t empty not t #-1 #1 subs...

  8. [8]

    Abdul-Mageed M, Keleg A, Elmadany A, Zhang C, Hamed I, Magdy W, Bouamor H, Habash N (2024) NADI 2024: The fifth nuanced A rabic dialect identification shared task. In: Habash N, Bouamor H, Eskander R, Tomeh N, Abu Farha I, Abdelali A, Touileb S, Hamed I, Onaizan Y, Alhafni B, Antoun W, Khalifa S, Haddad H, Zitouni I, AlKhamissi B, Almatham R, Mrini K (eds...

    2024

Show all 64 references
  1. [9]

    Al-Khalifa H, Darwish K, Mubarak H, Ali M, Elsayed T (eds) (2024) Proceedings of the 6th Workshop on Open-Source Arabic Corpora and Processing Tools (OSACT) with Shared Tasks on Arabic LLMs Hallucination and Dialect to MSA Machine Translation @ LREC-COLING 2024, ELRA and ICCL,...

    2024

  2. [10]

    Alahmari SS (2024) S irius \_ T ranslators at OSACT 6 2024 shared task: Fin-tuning ara-t5 models for translating A rabic dialectal text to M odern S tandard A rabic. In: Al-Khalifa H, Darwish K, Mubarak H, Ali M, Elsayed T (eds) Proceedings of the 6th Workshop on Open-Source A...

    2024

  3. [11]

    Alhafni B, Al-Towaity S, Fawzy Z, Nassar F, Eryani F, Bouamor H, Habash N (2024) Exploiting dialect identification in automatic dialectal text normalization. In: Habash N, Bouamor H, Eskander R, Tomeh N, Abu Farha I, Abdelali A, Touileb S, Hamed I, Onaizan Y, Alhafni B, Antoun...

    2024

  4. [12]

    In: OSACT 2024 Workshop, pp 98--103

    Atwany H, Rabih N, Mohammed I, Waheed A, Raj B (2024) Osact 2024 task 2: Arabic dialect to msa translation. In: OSACT 2024 Workshop, pp 98--103

    2024

  5. [13]

    Sensors 21(19):6509, doi:10.3390/s21196509, ://doi.org/10.3390/s21196509

    Baniata LH, Ampomah IKE, Park SY (2021) A transformer-based neural machine translation model for arabic dialects that utilizes subword units. Sensors 21(19):6509, doi:10.3390/s21196509, ://doi.org/10.3390/s21196509

    2021 doi

  6. [14]

    Bavaresco A, Bernardi R, Bertolazzi L, Elliott D, Fern \'a ndez R, Gatt A, Ghaleb E, Giulianelli M, Hanna M, Koller A, Martins A, Mondorf P, Neplenbroek V, Pezzelle S, Plank B, Schlangen D, Suglia A, Surikuchi AK, Takmaz E, Testoni A (2025) LLM s instead of human judges? a lar...

    2025

  7. [15]

    Bouamor H, Habash N, Salameh M, Zaghouani W, Rambow O, Abdulrahim D, Obeid O, Khalifa S, Eryani F, Erdmann A, Oflazer K (2018) The MADAR A rabic dialect corpus and lexicon. In: Calzolari N, Choukri K, Cieri C, Declerck T, Goggi S, Hasida K, Isahara H, Maegaard B, Mariani J, Ma...

    2018

  8. [16]

    Brown T, Mann B, Ryder N, Subbiah M, Kaplan JD, Dhariwal P, Neelakantan A, Shyam P, Sastry G, Askell A, Agarwal S, Herbert-Voss A, Krueger G, Henighan T, Child R, Ramesh A, Ziegler D, Wu J, Winter C, Hesse C, Chen M, Sigler E, Litwin M, Gray S, Chess B, Clark J, Berner C, McCa...

    2020

  9. [17]

    Linguistic Data Consortium, LDC2004L02, doi:10.35111/050q-5r95, ://catalog.ldc.upenn.edu/LDC2004L02

    Buckwalter T (2004) Buckwalter arabic morphological analyzer version 2.0. Linguistic Data Consortium, LDC2004L02, doi:10.35111/050q-5r95, ://catalog.ldc.upenn.edu/LDC2004L02

    2004 doi

  10. [18]

    In: Proceedings of the Ninth Conference on Machine Translation (WMT)

    Caillaut G, Qader R, Nakhlé M, Liu J, Barthélemy JG (2024) Scaling laws of decoder-only models on the multilingual machine translation task. In: Proceedings of the Ninth Conference on Machine Translation (WMT)

    2024

  11. [19]

    In: 2023 ACM/IEEE International Conference on Model Driven Engineering Languages and Systems Companion (MODELS-C), pp 588--596, doi:10.1109/MODELS-C59198.2023.00097

    Chen B, Yi F, Varró D (2023) Prompting or fine-tuning? a comparative study of large language models for taxonomy construction. In: 2023 ACM/IEEE International Conference on Model Driven Engineering Languages and Systems Companion (MODELS-C), pp 588--596, doi:10.1109/MODELS-C59...

    2023

  12. [20]

    J Mach Learn Res 25(1)

    Chung HW, Hou L, Longpre S, Zoph B, Tai Y, Fedus W, Li Y, Wang X, Dehghani M, Brahma S, Webson A, Gu SS, Dai Z, Suzgun M, Chen X, Chowdhery A, Castro-Ros A, Pellat M, Robinson K, Valter D, Narang S, Mishra G, Yu A, Zhao V, Huang Y, Dai A, Yu H, Petrov S, Chi EH, Dean J, Devlin...

    2024

  13. [21]

    Cui M, Gao P, Liu W, Luan J, Wang B (2025) Multilingual machine translation with open large language models at practical scale: An empirical study. In: Chiruzzo L, Ritter A, Wang L (eds) Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Associati...

    2025 doi

  14. [22]

    Demidova A, Atwany H, Rabih N, Sha ' ban S (2024) A rabic train at NADI 2024 shared task: LLM s' ability to translate A rabic dialects into M odern S tandard A rabic. In: Habash N, Bouamor H, Eskander R, Tomeh N, Abu Farha I, Abdelali A, Touileb S, Hamed I, Onaizan Y, Alhafni ...

    2024

  15. [23]

    SIL International, Dallas, Texas, ://www.ethnologue.com/insights/ethnologue200

    Eberhard DM, Simons GF, Fennig CD (eds) (2023) Ethnologue: Languages of the World, 26th edn. SIL International, Dallas, Texas, ://www.ethnologue.com/insights/ethnologue200

    2023

  16. [24]

    https://explodingtopics.com/blog/gpt-parameters

    Exploding Topics (2025) Number of parameters in gpt-4. https://explodingtopics.com/blog/gpt-parameters

    2025

  17. [25]

    Feng Z, Zhang Y, Li H, Wu B, Liao J, Liu W, Lang J, Feng Y, Wu J, Liu Z (2025) TE a R : Improving LLM -based machine translation with systematic self-refinement. In: Chiruzzo L, Ritter A, Wang L (eds) Findings of the Association for Computational Linguistics: NAACL 2025, Assoc...

    2025 doi

  18. [26]

    Habash N, Rambow O (2005) A rabic tokenization, part-of-speech tagging and morphological disambiguation in one fell swoop. In: Knight K, Ng HT, Oflazer K (eds) Proceedings of the 43rd Annual Meeting of the Association for Computational Linguistics ( ACL ' 05), Association for ...

    2005

  19. [27]

    Habash N, Diab M, Rambow O (2012) Conventional orthography for dialectal A rabic. In: Calzolari N, Choukri K, Declerck T, Do g an MU, Maegaard B, Mariani J, Moreno A, Odijk J, Piperidis S (eds) Proceedings of the Eighth International Conference on Language Resources and Evalua...

    2012

  20. [28]

    Hadj Mohamed N, Rassem M, Han L, Nenadic G (2023) A lpha MWE - A rabic: A rabic edition of multilingual parallel corpora with multiword expression annotations. In: Mitkov R, Angelova G (eds) Proceedings of the 14th International Conference on Recent Advances in Natural Languag...

    2023

  21. [29]

    Hamed MM, Hreden M, Hennara K, Aldallal Z, Chrouf S, AlModhayan S (2025) Lahjawi: A rabic cross-dialect translator. In: Ezzini S, Alami H, Berrada I, Benlahbib A, El Mahdaouy A, Lamsiyah S, Derrouz H, Haddad Haddad A, Jarrar M, El-Haj M, Mitkov R, Rayson P (eds) Proceedings of...

    2025

  22. [30]

    In: Proceedings for the First Workshop on Modelling Translation: Translatology in the Digital Age, pp 15--33

    Han L, Smeaton A, Jones G (2021) Translation quality assessment: A brief survey on manual and automatic methods. In: Proceedings for the First Workshop on Modelling Translation: Translatology in the Digital Age, pp 15--33

    2021

  23. [31]

    Frontiers in Digital Health 6:1211,564

    Han L, Gladkoff S, Erofeev G, Sorokina I, Galiano B, Nenadic G (2024) Neural machine translation of clinical text: an empirical investigation into multilingual pre-trained language models and transfer-learning. Frontiers in Digital Health 6:1211,564

    2024

  24. [32]

    Howard J, Ruder S (2018) Universal language model fine-tuning for text classification. In: Gurevych I, Miyao Y (eds) Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), Association for Computational Linguistics, Melb...

    2018 doi

  25. [33]

    Ibrahim M (2024) CUFE at NADI 2024 shared task: Fine-tuning llama-3 to translate from A rabic dialects to M odern S tandard A rabic. In: Habash N, Bouamor H, Eskander R, Tomeh N, Abu Farha I, Abdelali A, Touileb S, Hamed I, Onaizan Y, Alhafni B, Antoun W, Khalifa S, Haddad H, ...

    2024 doi

  26. [34]

    Transactions of the Association for Computational Linguistics 5:339--351, doi:10.1162/tacl_a_00065, ://aclanthology.org/Q17-1024/

    Johnson M, Schuster M, Le QV, Krikun M, Wu Y, Chen Z, Thorat N, Vi \'e gas F, Wattenberg M, Corrado G, Hughes M, Dean J (2017) G oogle ' s multilingual neural machine translation system: Enabling zero-shot translation. Transactions of the Association for Computational Linguist...

    2017 doi

  27. [35]

    Kadaoui K, Magdy SM, Waheed A, Khondaker MTI, El-Shangiti AO, Nagoudi EMB, Abdul-Mageed M (2023) TARJAMAT : Evaluation of bard and C hat GPT on machine translation of ten A rabic varieties. In: Sawaf H, El-Beltagy S, Zaghouani W, Magdy W, Abdelali A, Tomeh N, Abu Farha I, Haba...

    2023 doi

  28. [36]

    Khalifa M, Abdul-Mageed M, Shaalan K (2021) Self-training pre-trained language models for zero- and few-shot multi-dialectal A rabic sequence labeling. In: Merlo P, Tiedemann J, Tsarfaty R (eds) Proceedings of the 16th Conference of the European Chapter of the Association for ...

    2021 doi

  29. [37]

    Khered A, Benkhedda Y, Batista-Navarro R (2025) D ial2 MSA -verified: A multi-dialect A rabic social media dataset for neural machine translation to M odern S tandard A rabic. In: Ezzini S, Alami H, Berrada I, Benlahbib A, El Mahdaouy A, Lamsiyah S, Derrouz H, Haddad Haddad A,...

    2025

  30. [38]

    Klementiev A, Irvine A, Callison-Burch C, Yarowsky D (2012) Toward statistical machine translation without parallel corpora. In: Daelemans W (ed) Proceedings of the 13th Conference of the E uropean Chapter of the Association for Computational Linguistics, Association for Compu...

    2012

  31. [39]

    Koehn P, Hoang H, Birch A, Callison-Burch C, Federico M, Bertoldi N, Cowan B, Shen W, Moran C, Zens R, Dyer C, Bojar O, Constantin A, Herbst E (2007) M oses: Open source toolkit for statistical machine translation. In: Ananiadou S (ed) Proceedings of the 45th Annual Meeting of...

    2007

  32. [40]

    Krubi \'n ski M, Sellat H, Saleh S, Posp \'i s il A, Zem \'a nek P, Pecina P (2023) Multi-parallel corpus of N orth L evantine A rabic. In: Sawaf H, El-Beltagy S, Zaghouani W, Magdy W, Abdelali A, Tomeh N, Abu Farha I, Habash N, Khalifa S, Keleg A, Haddad H, Zitouni I, Mrini K...

    2023 doi

  33. [41]

    Kwaik KA, Saad M, Chatzikyriakidis S, Dobnik S (2018) A lexical distance study of arabic dialects. Procedia Computer Science 142:2--13, doi:https://doi.org/10.1016/j.procs.2018.10.456, ://www.sciencedirect.com/science/article/pii/S1877050918321562, arabic Computational Linguistics

    2018 doi

  34. [42]

    Nature 521(7553):436--444, doi:10.1038/nature14539

    LeCun Y, Bengio Y, Hinton G (2015) Deep learning. Nature 521(7553):436--444, doi:10.1038/nature14539

    2015 doi

  35. [43]

    ACM Comput Surv 55(9), doi:10.1145/3560815, ://doi.org/10.1145/3560815

    Liu P, Yuan W, Fu J, Jiang Z, Hayashi H, Neubig G (2023) Pre-train, prompt, and predict: A systematic survey of prompting methods in natural language processing. ACM Comput Surv 55(9), doi:10.1145/3560815, ://doi.org/10.1145/3560815

    2023 doi

  36. [44]

    In: Proceedings of Translating and the Computer 34 (TC34)

    Lommel A, Popovic M, Burchardt A (2012) Multidimensional quality metrics (mqm): A framework for declaring and describing translation quality metrics. In: Proceedings of Translating and the Computer 34 (TC34)

    2012

  37. [45]

    Lommel A, Gladkoff S, Melby A, Wright SE, Strandvik I, Gasova K, Vaasa A, Benzo A, Marazzato Sparano R, Foresi M, Innis J, Han L, Nenadic G (2024) The multi-range theory of translation quality measurement: MQM scoring models and statistical quality control. In: Martindale M, C...

    2024

  38. [46]

    Luccioni S, Jernite Y, Strubell E (2024) Power hungry processing: Watts driving the cost of ai deployment? In: Proceedings of the 2024 ACM Conference on Fairness, Accountability, and Transparency, Association for Computing Machinery, New York, NY, USA, FAccT '24, p 85–99, doi:...

    2024

  39. [47]

    Mager M, Bhatnagar R, Neubig G, Vu NT, Kann K (2023) Neural machine translation for the indigenous languages of the A mericas: An introduction. In: Mager M, Ebrahimi A, Oncevay A, Rice E, Rijhwani S, Palmer A, Kann K (eds) Proceedings of the Workshop on Natural Language Proces...

    2023 doi

  40. [48]

    Mousi B, Durrani N, Ahmad F, Hasan MA, Hasanain M, Kabbani T, Dalvi F, Chowdhury SA, Alam F (2025) A ra D i CE : Benchmarks for dialectal and cultural capabilities in LLM s. In: Rambow O, Wanner L, Apidianaki M, Al-Khalifa H, Eugenio BD, Schockaert S (eds) Proceedings of the 3...

    2025

  41. [49]

    Naik D, Naik I, Naik N (2024) Decoder-only transformers: The brains behind generative ai, large language models and large multimodal models. In: Naik N, Jenkins P, Prajapat S, Grace P (eds) Contributions Presented at The International Conference on Computing, Communication, Cy...

    2024

  42. [50]

    Papineni K, Roukos S, Ward T, Zhu WJ (2002) B leu: a method for automatic evaluation of machine translation. In: Isabelle P, Charniak E, Lin D (eds) Proceedings of the 40th Annual Meeting of the Association for Computational Linguistics, Association for Computational Linguisti...

    2002

  43. [51]

    Pasha A, Al-Badrashiny M, Diab M, El Kholy A, Eskander R, Habash N, Pooleery M, Rambow O, Roth R (2014) MADAMIRA : A fast, comprehensive tool for morphological analysis and disambiguation of A rabic. In: Calzolari N, Choukri K, Declerck T, Loftsson H, Maegaard B, Mariani J, Mo...

    2014

  44. [52]

    Popovi \'c M (2017) chr F ++: words helping character n-grams. In: Bojar O, Buck C, Chatterjee R, Federmann C, Graham Y, Haddow B, Huck M, Yepes AJ, Koehn P, Kreutzer J (eds) Proceedings of the Second Conference on Machine Translation, Association for Computational Linguistics...

    2017 doi

  45. [53]

    Post M (2018) A call for clarity in reporting BLEU scores. In: Bojar O, Chatterjee R, Federmann C, Fishel M, Graham Y, Haddow B, Huck M, Yepes AJ, Koehn P, Monz C, Negri M, N \'e v \'e ol A, Neves M, Post M, Specia L, Turchi M, Verspoor K (eds) Proceedings of the Third Confere...

    2018 doi

  46. [54]

    J Mach Learn Res 21(1)

    Raffel C, Shazeer N, Roberts A, Lee K, Narang S, Matena M, Zhou Y, Li W, Liu PJ (2020) Exploring the limits of transfer learning with a unified text-to-text transformer. J Mach Learn Res 21(1)

    2020

  47. [55]

    Rei R, Stewart C, Farinha AC, Lavie A (2020) COMET : A neural framework for MT evaluation. In: Webber B, Cohn T, He Y, Liu Y (eds) Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), Association for Computational Linguistics, Online,...

    2020 doi

  48. [56]

    In: Findings of the Association for Computational Linguistics: EMNLP 2023, Association for Computational Linguistics

    Sajjad H, Abdelali A, Mubarak H, Al-Kaabi M, Bouamor H (2023) Dialects in the age of large language models: A case study on arabic. In: Findings of the Association for Computational Linguistics: EMNLP 2023, Association for Computational Linguistics

    2023

  49. [57]

    https://github.com/tatsu-lab/stanford_alpaca

    Taori R, Gulrajani I, Zhang T, Dubois Y, Li X, Guestrin C, Liang P, Hashimoto TB (2023) Stanford alpaca: An instruction-following llama model. https://github.com/tatsu-lab/stanford_alpaca

    2023

  50. [58]

    Vaswani A, Shazeer N, Parmar N, Uszkoreit J, Jones L, Gomez AN, Kaiser Lu, Polosukhin I (2017) Attention is all you need. In: Guyon I, Luxburg UV, Bengio S, Wallach H, Fergus R, Vishwanathan S, Garnett R (eds) Advances in Neural Information Processing Systems, Curran Associate...

    2017

  51. [59]

    In: Proceedings of the 36th International Conference on Neural Information Processing Systems, Curran Associates Inc., Red Hook, NY, USA, NIPS '22

    Wei J, Wang X, Schuurmans D, Bosma M, Ichter B, Xia F, Chi EH, Le QV, Zhou D (2022) Chain-of-thought prompting elicits reasoning in large language models. In: Proceedings of the 36th International Conference on Neural Information Processing Systems, Curran Associates Inc., Red...

    2022

  52. [60]

    ACM Trans Asian Low-Resour Lang Inf Process doi:10.1145/3698236, ://doi.org/10.1145/3698236, just Accepted

    Yin Hong K, Han L, Batista-Navarro R, Nenadic G (2024) Cantonmt: Investigating back-translation and model-switch mechanisms for cantonese-english neural machine translation. ACM Trans Asian Low-Resour Lang Inf Process doi:10.1145/3698236, ://doi.org/10.1145/3698236, just Accepted

    2024 doi

  53. [61]

    arXiv preprint arXiv:190409675 ://arxiv.org/abs/1904.09675

    Zhang T, Kishore V, Wu F, Weinberger KQ, Artzi Y (2019) BERTScore : Evaluating text generation with BERT . arXiv preprint arXiv:190409675 ://arxiv.org/abs/1904.09675

    2019 arXiv

  54. [62]

    arXiv preprint arXiv:231002245

    Zhao W, Wang X, Liu J, Zhang T, Zhang B, Fung P (2023) Exploring the limits of zero-shot transfer for arabic dialect processing with llms. arXiv preprint arXiv:231002245

    2023

  55. [63]

    Author, Article title, Journal, Volume, page numbers (year)

  56. [64]

    Publisher, place (year)

    Author, Book title, page numbers. Publisher, place (year)

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.