Workflow — Foreign Language Review

Responsive without translating everything.

Mandarin, Japanese, Korean, German, Portuguese, Arabic, and beyond — responsiveness review without full translation, with hot-doc translation only on the shortlist. English summaries and tagged production into Relativity, Reveal, Everlaw, or Lighthouse. Replaces native-speaker contract attorneys at a fraction of the per-word and per-hour cost.

Take the AI Assessment Get a Per-Document ROI Model

$0.18–$0.32

Per word, contract translation at the ALSP

$50–$120

Per hour, native-speaker contract attorney (billed)

60–85%

Volume off the native-speaker line after cutover

What This Replaces

The Native-Speaker Review Center on Every Cross-Border Matter

The work the native-speaker reviewer does on every foreign-language document — and the cost of leaving it there.

The labor

Foreign-language review today moves through native-speaker contract attorneys at Morae, Lighthouse, Consilio, Epiq, KLDiscovery, UnitedLex, and Cobra Legal Solutions. Pricing combines per-word translation rates ($0.18–$0.32 per word for contract-grade translation) with hourly review billing ($50–$120 per hour for native-speaker reviewers). A 100K-document Mandarin or Japanese set with full translation can run into seven figures before the case team has a hot-doc shortlist.

The cycle time

Cross-border discovery typically runs 8–20 weeks at the native-speaker review center, with native-speaker recruiting, language calibration, and per-language quality control eating into the schedule. Every week the foreign-language portion of the production isn't tagged is a week the case team can't see the cross-border hot-doc clusters and can't refine its theory of the case for international depositions or treaty requests.

The Workflow

Input · Analysis · Output

What goes into the foreign-language review, what we do to it, and what shows up in the review platform.

Input

Foreign-language document set

Documents in Mandarin, Japanese, Korean, German, Portuguese, Arabic
Mixed-language documents (foreign body + English exhibits)
OCR-extracted text from scans and image PDFs
Custodian metadata and email-thread structure
Case-specific issue taxonomy (English-defined)
Native-speaker glossaries for case-specific terms
Less common languages flagged for native-speaker queue

Analysis

Triage without full translation

Responsiveness coding without full translation
Privilege and hot-doc identification in source language
Issue tagging per English-defined taxonomy
Key-term identification across languages
Hot-doc translation queue — only the shortlist
English summaries for hot docs and witness exhibits
Confidence score per coding decision; exceptions to native-speaker queue

Output

Tagged set + translations into the platform

Relativity (REST API and RDOs)
Reveal (REST API)
Everlaw (REST API)
Lighthouse (Smart Workflows)
Translated hot docs as English-language production
English summary memo for the case team
Per-language coding-decision audit trail

Side by Side

Foreign Language Review Today vs. With Last Rev

The numbers that matter: cycle time, per-document cost, accuracy, and audit posture.

Dimension	Native-Speaker Review Center	Last Rev Foreign Language Review
Cycle time, set received to first-pass tagged	8–20 weeks (recruiting + calibration)	7–14 days
Translation cost on the full set	$0.18–$0.32 per word, applied broadly	Translation only on hot-doc shortlist — typically 1–5% of the set
Per-hour native-speaker review cost	$50–$120/hour billed	Per-document, benchmarked at 20–40% of native-speaker unit cost
Coding consistency across languages	Variable — per-language reviewer pools, calibration drift	Same case taxonomy applied across all languages identically
Audit log per coding decision	Reviewer notes in source language, no machine-readable lineage	Source paragraph + coding decision + model version + confidence per language
Mixed-language document handling	Routed by primary language, slower QC	Multi-language extraction in a single pass
Renegotiation leverage at next review-vendor renewal	None — you're locked in	60–85% of foreign-language first-pass volume off the contract

How It Works

From Foreign-Language Set to Tagged Production

Five steps. Every one logged. Every one reversible if your confidence threshold isn't met.

Submission Lands

Foreign-language document set in Relativity, Reveal, Everlaw, or Lighthouse — with custodian metadata, email-thread structure, English-defined case taxonomy, and any case-specific glossaries the team has built.

Extraction & Classification

Responsiveness, privilege, and hot-doc coding applied in the source language without full translation. Issue tags applied per the English-defined taxonomy. Key-term identification across languages. Mixed-language documents handled in a single pass.

Validation Against Case-Team Bar

Coding decisions validated against your case-team's per-language calibration sample. Anything below your confidence threshold per coding decision is routed to a native-speaker exception queue — your call which queue, ours or yours.

Push to Review Platform

Tagged production set, translated hot docs, and English summary memo delivered into Relativity, Reveal, Everlaw, or Lighthouse via the documented integration. Translation cost applied only to the hot-doc shortlist — typically 1–5% of the original set.

Audit Log Persisted

Every coding decision, hot-doc tag, and translation request logged with the source paragraph, language, model version, prompt, and confidence score. Defensible chain of custody for cross-border productions and treaty-based document requests.

Compliance & Defensibility

Built to Meet the Quality Bar Cross-Border Litigation Already Runs On

Cross-border discovery posture

FRCP Rule 26(g) certification supported by a per-language audit trail of coding decisions. Hague Convention, EU Evidence Regulation, and treaty-based document-request workflows accommodated through the same source-paragraph-cited audit log.

Per-language calibration discipline

Each language has its own calibration sample and seed coding maintained by the case team. The audit log records which calibration version applied to which document, so per-language QC remains defensible across the lifecycle of the matter.

Translation chain of custody

Translation requests trace from the source-language hot-doc to the English production version with the model version, prompt, and confidence score recorded. When opposing counsel challenges a translation, the chain of custody produces the basis on demand.

Data residency and PRC / GDPR posture

Cross-border matters often have data-residency constraints (PRC PIPL, GDPR transfer rules, sectoral data laws). Deployable in your VPC or our SOC 2 environment, including in-region deployments where the matter requires it. Encryption in transit and at rest; retention policies tied to your matter and treaty obligations.

Common Questions

What Cross-Border Litigation Teams Ask About Foreign Language Review

Each of those review platforms has improving multilingual features and we don't compete head-to-head. The competitor on this page is the native-speaker contract-attorney line on your matter budget — Morae, Lighthouse, Consilio, Epiq, KLDiscovery, UnitedLex, or a captive offshore center charging $0.18–$0.32 per word for translation plus $50–$120 per hour for native-speaker review. We undercut that labor cost, integrate directly into your existing review platform, and deliver tagged production sets and English summaries into the system of record.

eDiscovery first-pass is the high-volume English-language review across the post-cull set. This page is the focused workflow on the foreign-language portion of the same set — responsiveness coding without full translation, hot-doc translation only on the shortlist, and English summaries for the case team. They share infrastructure but the unit economics, the bar, and the deliverables are different. We built each as a separate page so the workflow stays specific to what cross-border case teams buy.

Mandarin, Japanese, Korean, German, Portuguese, Arabic, French, Spanish, Italian, Russian, Hindi, and most major commercial languages are well-supported with high-confidence coding. Less common languages (Indonesian, Vietnamese, Turkish, Hebrew, Polish, etc.) route to a native-speaker exception queue — your call which queue, ours or yours. We are honest about which languages perform best and which need native-speaker QC; the audit log makes the per-language confidence visible to the case team.

Our pilot success threshold is responsiveness, privilege, and hot-doc coding accuracy at parity with or above your incumbent native-speaker reviewer, measured on the same shadow-data sample of foreign-language documents and validated against the case-team's per-language calibration set. Anything below your defined confidence threshold per coding decision is routed to a native-speaker exception queue — your call which queue, ours or yours.

Case-specific glossaries are built during onboarding with native-speaker counsel from the matter team. Legally-loaded terms (e.g., specific PRC contract terms, German GmbH governance language, Japanese keiretsu relationships) are flagged with the original source paragraph and a glossary-aware English summary so the case team makes the call on a richer file than the standard machine-translation output. We don't make the legal call on culturally-loaded terms — we surface the original-language evidence and the contextual gloss.

Data-residency constraints are accommodated through in-region deployments. PRC PIPL matters can run in PRC infrastructure; EU GDPR transfer constraints can be satisfied with EU-region deployments. Your IT team and outside counsel determine the compliance posture; we deploy to match. Audit log persists per the matter's retention policy and treaty obligations.

Foreign-language review pilots typically run 4–6 weeks: 1 week of integration, language-specific calibration, and case-glossary build with case counsel and matter-team native speakers; 2–3 weeks of shadow-mode running on a constrained subset of the foreign-language set; 1–2 weeks of supervised cutover. Production rollout is staged after the pilot meets your accuracy and SLA bar per language.

We benchmark against your blended per-word translation cost ($0.18–$0.32) plus per-hour native-speaker review cost ($50–$120) on the historical baseline. Our target is 20–40% of that combined unit cost, with the savings driven primarily by translating only the hot-doc shortlist and applying per-language coding consistently. Pricing structures around volume tiers and outcome SLAs, not hourly billable rates.

Two Ways to Start

Take the AI assessment for a structured read on foreign-language-review feasibility on your typical cross-border matters. Or talk to us if you already know which language is bleeding the most native-speaker labor cost.

Self-serve · 8 minutes

Take the AI Assessment

A short structured assessment that maps your typical foreign-language volume, the languages involved, and your native-speaker arrangement to AI feasibility and ROI.

Talk to a specialist

Get a Per-Document ROI Model

Send us your typical foreign-language volume, the languages involved, your review platform, and your current native-speaker arrangement. We'll come back with a per-document unit-cost comparison and a pilot plan in 5 business days.

Other Workflows

More Legal Workflows We Replace

The same approach, applied to the other document-heavy labor lines on your legal-ops or ALSP budget.

eDiscovery First-Pass

Responsiveness, privilege, and hot-doc tagging across 200K–5M documents in Relativity, Reveal, or Everlaw.

Privilege Review

Attorney-client and work-product determinations with metadata-ready privilege log generation — defensible, claw-back-aware.

M&A Due Diligence Review

Data-room contracts triaged for change-of-control, MAC/MAE, and indemnification — issues list into Intralinks or Datasite in days.

Patent Prior Art Search

Element-by-element claim charts and invalidity contentions — anticipatory and obvious references located faster than a $1.5K–$5K offshore chart.