---
title: Translation Agent in Modular LLM Systems
url: https://www.emergentmind.com/topics/translation-agent
type: topic
---

# Translation Agent in Modular LLM Systems

Searching arXiv for the key papers to ground the article in current literature.
A translation agent is a modular, role-specialized component in an LLM-based translation system that performs a bounded subtask within a larger workflow, rather than attempting end-to-end translation in a single pass. In current research, the term covers both natural-language and code-translation settings, including literary translation, document-level machine translation, simultaneous speech translation, multimodal subtitling, legal translation, LaTeX translation, and repository-scale code migration. Across these settings, translation agents are typically embedded in multi-agent systems that decompose translation into drafting, retrieval, grounding, validation, critique, revision, or orchestration stages, with the stated goal of improving semantic fidelity, structural correctness, consistency, and domain conformity under real deployment constraints [2506.08403] [2603.14054] [2410.08143] [2409.19894].

## 1. Concept and scope

In the contemporary literature, a translation agent is not defined by a single model architecture but by functional specialization. In TACTIC, an agent handles one cognitive subtask such as drafting, refinement, evaluation, scoring, context reasoning, or knowledge gathering [2506.08403]. In LegacyTranslate, the “Initial Translation Agent” generates a first Java draft from PL/SQL using retrieved exemplars, while other agents ground the draft in internal APIs and repair it through compiler feedback [2603.14054]. In DelTA and Loong, the translation agent is document-level and memory-centric: it translates incrementally while consulting summaries, entity records, exemplars, and prior sentence pairs to maintain coherence across long documents [2410.08143] [2605.30274]. In CLASI and Agent-SiMT, the translation agent operates under streaming constraints, producing output from partial input while separate components control read/write timing or chunk boundaries [2407.21646] [2406.06910].

This breadth of usage has made “translation agent” a cross-domain term. In code migration, the agent often translates source units under compilation, test, and framework constraints rather than lexical equivalence alone [2508.17720] [2604.07341]. In natural-language translation, the same term often denotes agents that reason over context, style, terminology, or evaluation criteria [2410.21067] [2505.14848]. A plausible implication is that the field treats translation less as a single decoding act and more as a controlled process in which multiple specialized agents expose intermediate structure that monolithic prompting tends to hide.

## 2. Core architectures and coordination patterns

Most translation-agent systems use explicit role decomposition. Some systems are sequential pipelines, some are iterative repair loops, and some introduce routing or adversarial interaction. LegacyTranslate uses three agents: Initial Translation, API Grounding, and Refinement [2603.14054]. RepoTransAgent uses a RAG Agent, a Context Agent, and a Refine Agent [2508.17720]. TRANSAGENT uses four agents—Initial Code Translator, Syntax Error Fixer, Code Aligner, and Semantic Error Fixer—under a controller that alternates between validation and localized repair [2409.19894]. TACTIC uses six agents aligned with Cognitive Translation Studies: DraftAgent, RefinementAgent, EvaluationAgent, ScoreAgent, ContextAgent, and ResearchAgent [2506.08403]. TransLaw uses Translator, Annotator, and Proofreader [2507.00875]. MAATS replaces a single reviewer with multiple MQM-specialized evaluators followed by a synthesis editor [2505.14848].

| System | Domain | Specialized agents |
|---|---|---|
| LegacyTranslate | PL/SQL→Java modernization | Initial Translation, API Grounding, Refinement |
| RepoTransAgent | Repository-aware code translation | RAG, Context, Refine |
| TRANSAGENT | Code translation | Initial Code Translator, Syntax Error Fixer, Code Aligner, Semantic Error Fixer |
| TACTIC | Machine translation | Draft, Refinement, Evaluation, Score, Context, Research |
| TransLaw | Legal translation | Translator, Annotator, Proofreader |
| MAATS | Machine translation | Translator, MQM evaluators, Editor |

These decompositions differ in what they treat as the hard part of translation. Some systems assume generation is easy but timing or context selection is difficult, as in simultaneous translation [2406.06910]. Others assume initial generation is easy but environment integration is difficult, as in enterprise code modernization [2603.14054]. Others treat evaluation itself as a first-class agentic task, as in MAATS and DITING’s AgentEval, where multiple evaluators simulate deliberation rather than relying on a single scalar metric [2505.14848] [2510.09116].

An important architectural distinction concerns control. Some frameworks use fixed workflows, such as Translator→Annotator→Proofreader in TransLaw [2507.00875]. Others use dynamic routing. SAMAS computes an 81-dimensional Stylistic Feature Spectrum and deterministically chooses a “Faulkner-esque” or “Hemingway-esque” workflow based on thresholds $H > 0.85$ and $E_{\text{low}} > 0.6$ [2602.19840]. Loong performs stepwise observe-and-act context selection before translating each segment [2605.30274]. ACToR uses adversarial collaboration: a translator agent repairs Rust code while a discriminator agent searches for new failing tests [2510.03879].

## 3. Context, memory, and grounding mechanisms

A defining property of translation agents is explicit context management. Direct prompting is repeatedly described as inadequate when translation depends on repository structure, document discourse, internal APIs, or multimodal evidence. LegacyTranslate retrieves semantically similar PL/SQL→Java pairs with SFR-Mistral embeddings and uses $k = 3$ exemplars as the best trade-off between relevance and prompt length in experiments [2603.14054]. RepoTransAgent separates retrieval from context acquisition: its RAG Agent queries a pair store and a name store, while its Context Agent iteratively invokes tools such as `get_source_class_info`, `find_target_imports`, and `find_target_method_body` to gather repository-specific facts in unified JSON [2508.17720].

Several systems replace raw history with structured memory. DelTA stores Proper Noun Records, Bilingual Summary, Long-Term Memory, and Short-Term Memory, and updates them sentence by sentence to improve consistency and avoid omissions [2410.08143]. Loong generalizes this design into a 3E memory module—Essence, Exemplar, and Entity—and optimizes context selection from that memory with preference-based reinforcement learning [2605.30274]. ViDove extends memory across modalities: its short-term memory stores translation history and multimodal cues for the current video, and its long-term memory stores domain knowledge and web retrieval results via Tavily, all accessed through LlamaIndex [2507.07306]. CRAT narrows retrieval to flagged “unknown” or context-sensitive terms, then consolidates internal and external evidence into a Translation Knowledge Graph before allowing the Translator to consume it [2410.21067].

Grounding can also be stylistic or institutional rather than factual. SAMAS derives control signals from wavelet-packet statistics over word-length sequences, using Relative Wavelet Energy, Wavelet Entropy, skewness, and kurtosis to route texts through specialized literary agents [2602.19840]. TransLaw grounds legal translation in Combined Department of Justice glossaries and in Proofreading Memory triplets of the form $\langle \text{src}, \text{ref}, \text{err} \rangle$ [2507.00875]. LegacyTranslate grounds code in an API knowledge base of 80 entries extracted from shared Java libraries [2603.14054]. A recurring pattern is that translation agents increasingly externalize context into structured stores rather than relying on latent model memory alone.

## 4. Validation, feedback, and evaluation

Translation-agent systems are unusual in the centrality they give to validation. In code translation, compiler feedback, test execution, and structural checks are often direct inputs to subsequent agents. LegacyTranslate defines Structural Validity, Compilation Rate, and Test-Pass Rate, with $CR = C/N$ and $TPR = T/N$ [2603.14054]. TRANSAGENT uses Computational Accuracy as the fraction of translated programs passing all tests, and narrows semantic repair by aligning source and target execution traces at block granularity [2409.19894]. RepoTransAgent treats compilation errors, runtime errors, functional errors, and non-terminating executions as distinct categories for reflection-driven repair [2508.17720]. ReCodeAgent raises this to repository scale: across 118 projects, it reports 99.4% compilation success and improves validated developer-test pass rate by 60.8% over baselines [2604.07341].

Natural-language translation work likewise shifts evaluation from single scalar metrics to multi-perspective analysis. TACTIC emphasizes XCOMET-XXL, MetricX-24-XXL, and COMETKIWI-23-XXL, and uses an internal ScoreAgent to aggregate faithfulness, expressiveness, and elegance into an acceptance score [2506.08403]. DITING argues that web-novel translation requires genre-specific evaluation across idiom translation, lexical ambiguity, terminology localization, tense consistency, zero-pronoun resolution, and cultural safety; its AgentEval debate framework achieves the highest correlation with human judgments among seven tested automatic metrics, with Spearman $\rho = 0.669$ for the multi-agent debate variant [2510.09116]. TransLaw uses a weighted legal-evaluation score
$$
I = \alpha A + \beta C + \gamma S
$$
with default weights $\alpha = 0.6$, $\beta = 0.3$, and $\gamma = 0.1$, where $A$ is legal semantic accuracy, $C$ is structural coherence and cohesion, and $S$ is stylistic appropriateness [2507.00875].

Several papers explicitly question the adequacy of overlap-based metrics. CLASI introduces Valid Information Proportion, defined as $VIP = K/N$, to measure how much of a simultaneous interpretation is actually usable to the listener; on RealSI it reports VIP of 81.3% for Chinese-to-English and 78.0% for English-to-Chinese, while other systems achieve 35.4% and 41.6% [2407.21646]. TransAgents reports a d-BLEU of 25.0 yet wins 55.9% of bilingual LLM pairwise comparisons against GPT-4-1106-PREVIEW and 66.2% against a human reference, suggesting that literary preference and n-gram overlap can diverge sharply [2405.11804]. DelTA formalizes proper-noun consistency with LTCR-1 and reports consistency gains of up to 4.58 percentage points [2410.08143]. This suggests that translation agents are also forcing a redefinition of what counts as translation quality.

## 5. Domain-specific instantiations

The term “translation agent” now covers a wide spectrum of applied settings.

In enterprise code modernization, LegacyTranslate was developed in a financial-institution migration of approximately 2.5 million lines of PL/SQL to Java. The Initial Translation Agent alone achieved 45.6% compilable outputs and a 30.9% test-pass rate; the full pipeline raised these to 52.9% and 33.8% [2603.14054]. RepoTransAgent, evaluated on hundreds of Java–C# pairs from six open-source repositories, achieved up to 55.34% compile rate and 45.84% pass rate, outperforming UniTrans, PLTranslation, and a no-agent baseline [2508.17720]. TRANSAGENT targets general code translation across Python, Java, and C++, reaching Computational Accuracy as high as 94.5% depending on direction and outperforming UniTrans and TransCoder [2409.19894].

In document and literary translation, DelTA improves consistency and COMET through multi-level memory while remaining sentence-wise and omission-resistant [2410.08143]. Loong targets long-document translation with observe-and-act adaptive context selection and reports average gains of up to 13.0 points across sCOMET, dCOMET, and LLM-as-a-Judge [2605.30274]. SAMAS focuses on style fidelity rather than only semantic correctness and reports FLORES-200 en→xx XCOMET of 96.93 when paired with GPT-5, with statistically significant gains over TACTIC [2602.19840]. TransAgents simulates a literary publication workflow with CEO, Senior Editor, Junior Editor, Translator, Localization Specialist, and Proofreader roles for ultra-long literary texts [2405.11804].

In multimodal and streaming translation, ViDove integrates auditory and visual agents with multimodal memory for video subtitling, achieving BLEU 23.51 and SubER 73.38 on DoveBench, improving BLEU by 28.8% and reducing SubER by 15.5% over Whisper + DelTA [2507.07306]. CLASI treats simultaneous speech translation as an agentic read–retrieve–plan–write loop and reports 70% VIP on an extremely hard dataset where other systems remain under 13% [2407.21646]. Agent-SiMT separates policy selection from translation generation by pairing an HMT-based policy agent with an LLM translation agent, yielding state-of-the-art latency–quality trade-offs in simultaneous machine translation [2406.06910].

In tightly constrained domains, specialized translation agents handle structure or regulation directly. LaTeXTrans uses Parser, Translator, Validator, Summarizer, Terminology Extractor, and Generator agents to preserve compilability and structure in LaTeX documents; on En–Zh with a GPT-4o backbone it reports COMETkiwi 73.59, LLM-score 8.92, and FC-score 71.52 [2508.18791]. TransLaw adapts the pattern to Hong Kong legal judgments and reports that its TransLaw-ChatGPT configuration improves legal semantic accuracy, structural coherence, and stylistic fidelity over GPT-4o, with ACS gains across all tested weightings [2507.00875].

## 6. Limitations, controversies, and research directions

Despite strong results, current work presents a consistent set of limitations. First, multi-agent decomposition is not free. TACTIC notes that multi-stage workflows increase inference time and cost relative to single-pass translation [2506.08403]. Loong requires trajectory sampling for preference construction and uses 4×A800 GPUs for roughly 20 hours of training per model [2605.30274]. LegacyTranslate reports that large prompts and iterative refinement incur computational cost, with experiments using an A100 GPU for about five hours total [2603.14054]. ReCodeAgent averages about \$15.3 and about 57 minutes per project [2604.07341].

Second, retrieval and memory can help or harm. CRAT explicitly argues that straightforward RAG can overload the model and induce hallucinations, motivating targeted retrieval and causal validation [2410.21067]. DelTA and Loong both treat redundant context as actively harmful, not merely unnecessary [2410.08143] [2605.30274]. RepoTransAgent shows that smaller models may see smaller gains or even declines because complex prompts and tool invocation become difficult to follow [2508.17720]. LegacyTranslate states that retrieval quality and API-knowledge-base completeness are pivotal and that hallucinations remain a risk for framework glue code without clear examples [2603.14054].

Third, validation itself remains imperfect. ACToR demonstrates that adversarial testing materially improves C→Rust correctness, but it still depends on discriminating test suites and assumes deterministic, single-threaded programs [2510.03879]. CLASI argues that BLEU, BLEURT, and COMET correlate poorly with human VIP at higher quality [2407.21646]. MAATS improves true-positive error detection substantially but also increases false positives, though the reported excess is mostly minor [2505.14848]. TransLaw observes that multi-turn revision can induce hallucinations and that contextualizing complex legal terminology and achieving stylistic naturalness still trail human experts [2507.00875].

Current research directions therefore converge on a few themes. One is richer grounding: curated termbases, API knowledge bases, multimodal evidence, and legal or domain-specific retrieval [2603.14054] [2507.07306] [2507.00875]. Another is better control over long context, through compact summaries, adaptive retrieval, and learned routing policies [2410.08143] [2605.30274]. A third is stronger validation, including execution alignment, adversarial testing, coverage-guided validation, and repository-scale trajectory analysis [2409.19894] [2510.03879] [2604.07341]. A plausible implication is that the future translation agent will be less a standalone translator than a validated, memory-bearing controller over translation-specific tools, evaluators, and external knowledge.

Source: https://www.emergentmind.com/topics/translation-agent