Probationary Urdu

Standard Operating Procedure: Probationary Urdu

Document ID: SOP-OPS-URD-409.2
Effective Date: March 30, 2026
Compliance Classification: Enterprise Operational Standard / Linguistic Quality Assurance

---

1. Document Control & Revision History

---

2. Executive Summary & Scope

2.1 Purpose

This Standard Operating Procedure (SOP) establishes the mandatory operational framework for auditing, processing, and validating the linguistic, syntactic, and technical integration of the Urdu language within enterprise workflows, specifically during the probationary evaluation phase of localization assets, NLP training corpora, and customer-facing communications.

2.2 Scope

This procedure applies universally to all internal operations teams, linguistic vendors, quality assurance (QA) engineers, and automated pipeline handlers managing Urdu-script assets. It governs the lifecycle from initial ingestion through rendering verification, linguistic accuracy audits, and final production sign-off. Compliance with this SOP is mandatory to prevent character rendering corruption (Mojibake), RTL (Right-to-Left) layout failures, and semantic degradation in enterprise outputs.

2.3 Regulatory & Compliance Standards

Operations governed under this SOP must adhere to Unicode Standard (v15.0+ for Arabic/Urdu block rendering), ISO 639-1 (Language code: ur), ISO 17100 (Translation Services Management), and internal Enterprise Data Governance Security Policies (EDG-SP-88).

---

3. Prerequisites & Tool Requirements

Execution of this SOP requires mandatory fulfillment of the following technical, credential, and environmental prerequisites prior to workflow initiation.

3.1 Software & System Access

• Localization Management Platform (LMP): Enterprise-tier TMS (e.g., Phrase, Transifex Enterprise) with confirmed RTL layout enablement.
• Linguistic Processing Suite: Open-source or enterprise Urdu NLP toolkit supporting N-gram tokenization and morphological analysis.
• Typography Engine: Certified OpenType font suite supporting complex Urdu ligatures (Must include: Jameel Noori Nastaleeq, Noto Nastaliq Urdu).
• Version Control: Git-based repository access for string-level tracking (git/urdu-localization-main).

3.2 Equipment & Environment

• Workstation: Dual-monitor configuration capable of rendering high-density complex script glyphs without rasterization tearing.
• OS Environment: Windows 11 Enterprise (22H2+) or macOS Ventura+ with explicit right-to-left layout directioning enabled at the kernel level.

3.3 Certifications & Clearances

• Linguistic Lead: Certified Localization Professional (CLP) or ATA-certified English-to-Urdu translator with minimum 5 years enterprise domain experience.
• Auditor: Designated Enterprise Workflow Auditor with clearance level L-2 or higher.

---

4. Roles & Responsibilities (RACI Matrix Table)

• Responsible (R): The role that performs the activity.
• Accountable (A): The sole role with final approval and fiduciary/operational accountability.
• Consulted (C): The role providing subject matter expertise and advisory input.
• Informed (I): The role kept updated on progress and milestone completion.

---

5. Step-by-Step Operational Procedure

Phase 1: Ingestion & Environment Setup

• 1.1 Verify receipt of the raw Urdu asset payload (.xliff, .json, or .tmx) within the secure LMP staging environment.
• 1.2 Validate file encoding parameters to ensure strict adherence to UTF-8 without BOM (Byte Order Mark).
• 1.3 Initialize the local typography rendering test suite and confirm active display of Noto Nastaliq Urdu to prevent fallback-font distortion.
• 1.4 Assign the project ticket to the designated Urdu Lead Linguist (ULL) via the enterprise task management portal.

Phase 2: Linguistic Evaluation & Contextual Audit

• 2.1 Conduct an automated terminology check against the approved enterprise glossary (GLOS-URD-2026-V2).
• 2.2 Perform a manual syntactic audit of the probationary text, ensuring appropriate register (Formal/Shumali vs. Colloquial) is maintained based on brand guidelines.
• 2.3 Verify correct handling of passive voice structures, honorifics (e.g., Aap vs. Tum), and gender-agreement paradigms unique to Urdu grammar.
• 2.4 Log all lexical anomalies, stylistic deviations, and out-of-vocabulary (OOV) terms into the Probationary Audit Ledger (PAL).

Phase 3: Technical Rendering & Layout Verification

• 3.1 Execute the RTL layout stress test across target UI viewports (Web, Mobile iOS/Android, Desktop Client).
• 3.2 Inspect complex ligature formation, specifically verifying the rendering of ZWNJ (Zero-Width Non-Joiner) and ZWJ (Zero-Width Joiner) characters in contextual shaping zones.
• 3.3 Verify truncation, text-wrapping, and line-break behaviors to ensure words do not fracture across inappropriate syllable boundaries.
• 3.4 Confirm bidirectional (BiDi) embedding integrity for mixed-script strings containing alphanumeric digits or English brand terminology within Urdu sentences.

Phase 4: Final Sign-off & Pipeline Deployment

• 4.1 Compile the completed Probationary Quality Report (PQR) containing error density metrics and lexical match rates.
• 4.2 Submit the PQR and audited assets to the Enterprise Workflow Auditor (EWA) for compliance review.
• 4.3 Secure formal sign-off from the EWA upon meeting all threshold criteria.
• 4.4 Push verified assets to the production repository and archive the audit trail in the secure compliance vault.

---

6. Quality Assurance & Verification Criteria

6.1 Audit Checklist

• Zero-Corrupt Ligatures: No visual fragmentation of complex character shaping (e.g., Lam-Alef clusters).
• Glossary Adherence: Minimum 98.5% compliance with mandatory enterprise product terminology.
• BiDi Isolation: Complete prevention of string inversion or punctuation misplacement in mixed alphanumeric contexts.

6.2 Tolerance Thresholds & Acceptance Benchmarks

• Critical Error Limit: Zero (0) tolerance for layout crashes, unrendered glyph blocks (tofu boxes), or severe semantic inversion affecting safety warnings or financial values.
• Minor Linguistic Deviation Allowance: Maximum of 0.5 minor style/punctuation deviations per 1,000 words.
• Acceptance Benchmark: Overall Quality Score must achieve $\ge 96.0%$ on the Enterprise Localization QA Matrix prior to production deployment.

---

7. Practical Troubleshooting Scenarios

Scenario A: Nastaliq Font Fallback Failure (Mojibake / Tofu Blocks)

• Trigger: Asset renders as disjointed, isolated Arabic-script characters or replacement tofu blocks ($\square$) upon ingestion into the UI previewer.
• Immediate Corrective Action Protocol:
  1. Halt downstream processing immediately to prevent corrupted build generation.
  2. Inspect CSS / OpenType styling injections to verify that the primary font stack explicitly declares font-family: 'Jameel Noori Nastaleeq', 'Noto Nastaliq Urdu', sans-serif;.
  3. Run the sanitization script util-utf8-strip-bom.py on the payload file to eliminate hidden byte-order markers.
  4. Re-inject the payload into the staging environment and execute visual verification check 3.2.

Scenario B: BiDi Inversion in Numeric or Mixed-Script Strings

• Trigger: Currency values or telephone numbers (e.g., PKR 5,000) invert their directional sequence when embedded inside an Urdu sentence, rendering as 000,5 PKR.
• Immediate Corrective Action Protocol:
  1. Isolate the affected string in the LMP editor.
  2. Insert explicit Unicode directional formatting characters: Prepend the substring with Left-to-Right Mark (LRM, U+200E) and enclose mixed segments within Unicode directional overrides (RLI/PDI blocks).
  3. Validate correct visual presentation via the rendering engine verification suite.
  4. Log the occurrence in the PAL for developer retrofitting on dynamic string interpolation rules.

Scenario C: Syntactic Over-Formalization / Register Deviation

• Trigger: Automated or translated probationary output utilizes archaic literary Persian/Arabic loan words inappropriate for modern conversational enterprise interfaces.
• Immediate Corrective Action Protocol:
  1. Flag the segment set within the LMP audit view.
  2. Engage the Urdu Lead Linguist (ULL) to execute targeted post-editing, shifting the register to contemporary Standard Urdu (Standardized Islamabad/Lahore media standard).
  3. Update the dynamic project translation memory (TM) to down-rank penalized lexical entries.
  4. Re-run Phase 2 evaluation metrics before proceeding to technical layout validation.