TQA stands for Translation Quality Assessment. It is the structured process of evaluating a translated document, categorising errors by type and severity, and producing a measurable quality score. Here is everything Indian language translators need to know — including how t09n.com's built-in TQA tool works.
TQA stands for Translation Quality Assessment. It is the process of systematically evaluating a translation against its source text, identifying errors, categorising them by type and severity, and computing a numerical quality score.
TQA is different from proofreading. Proofreading looks for errors in the target text only — grammar, spelling, style. TQA is a bilingual comparison. It measures how accurately, completely, and naturally the source meaning has been carried into the target language. Every segment of the translated document is checked against the corresponding source segment.
TQA is used in professional translation workflows for three reasons:
Most professional TQA tools, including t09n.com's, use the MQM framework — Multidimensional Quality Metrics. MQM is the ISO-aligned international standard for translation quality evaluation developed by the translation industry and used by major enterprises and agencies globally.
MQM works by defining a taxonomy of error types and assigning severity weights to each. Errors are counted across the source word volume to produce a normalised score:
TQA Score = (Critical errors × weight + Major errors × weight + Minor errors × weight) ÷ word count × 1,000
A score below a defined threshold (often 5–25 errors per 1,000 words depending on the client) is considered a pass. Above the threshold, the translation is returned for revision.
The target text does not accurately reflect the source. Includes mistranslation, omission (content from the source missing entirely), addition (content not in the source added to the target), and untranslated text.
The target text is not natural in the target language. Includes grammar errors, spelling and orthography errors, punctuation issues, and register inconsistency (formal vs informal).
The wrong term was used. Either a term from the client's approved glossary was not used, or an industry-standard term was substituted with a less precise equivalent.
The translation is technically accurate but does not match the client's style guide. Includes tone mismatches, sentence structure preferences, and formatting conventions.
Numbers, dates, units, currencies, or address formats are not adapted to the target locale. For Indian languages this includes correct use of Indian number formatting (lakh/crore), date formats, and currency symbols.
Translating a drug dosage incorrectly. Omitting a safety warning. Changing a legal obligation. Adding content not in the source. Critical errors result in automatic TQA failure regardless of score and require immediate revision before delivery.
Mistranslating a product feature. Using the wrong honorific throughout a document. Wrong register (formal/informal) used for the audience. Major errors are weighted at 5–10× minor errors in the quality score calculation.
Punctuation inconsistency. A slightly awkward but technically correct phrasing. Minor style preference not followed. Minor errors are tracked and counted but weighted lightly in the final score — they rarely cause a TQA failure on their own.
In Indian language translation, the most common critical and major errors are honorific misuse (using tum instead of aap in a formal context in Hindi), omission of negation (which completely reverses meaning), and untranslated segments left in English in the middle of an otherwise-translated document.
Indian language translation has quality challenges that are less visible in European language pairs — and more damaging when undetected.
Honorifics. Hindi has aap, tum, and tu — three levels of formality. Using tum in a formal business document intended for senior management is a major error. It reads as disrespectful and unprofessional to a native speaker, even if the meaning is otherwise correct. TQA catches this explicitly as a register error.
Gender agreement. Hindi, Gujarati, and Marathi require noun-adjective-verb gender agreement throughout sentences. A document that mixes masculine and feminine forms across a paragraph — a common MT artifact — will have numerous minor and major fluency errors that human reviewers catch but that can slip through informal proofreading.
SOV word order artifacts. English is SVO (subject-verb-object). Indian languages are SOV (subject-object-verb). MT systems frequently produce English word order in Indian language output. This creates sentences that are technically composed of correct words but read unnaturally to native speakers — a fluency major error in MQM terms.
Number formatting. Indian conventions use lakh and crore, not million and billion. A translated document that renders ₹10,000,000 rather than ₹1,00,00,000 is a locale convention error — and in financial documents, that level of imprecision is unacceptable.
TQA makes these issues visible, counted, and reportable — which is why more Indian language translation clients are requiring TQA reports as standard delivery documentation.
t09n.com includes TQA built in for all users on the Professional plan — no add-ons, no third-party tools, no manual spreadsheet scoring.
Upload your source document and the completed translation. Supported formats include DOCX, XLSX, PPTX, SRT, XLIFF, SDLXLIFF, MQXLIFF, TXML, and TTX. t09n.com aligns the segments automatically.
Select the target language, error threshold (errors per 1,000 words), and the error type weights relevant to your project. Client-specific style guides and glossaries can be linked for terminology checking.
t09n.com generates a full MQM-style report with each error flagged by segment, type, and severity. Critical errors are highlighted at the top. The overall quality score is computed against your threshold.
Click any flagged segment to jump directly to the CAT editor and correct the error. Re-run TQA after corrections to get an updated score. Iterate until the document passes.
Export the completed TQA report as PDF or CSV to include with your client delivery. The report includes the quality score, error breakdown by type and severity, and pass/fail status against the agreed threshold.
TQA is not only for large agencies. Freelance translators working in Indian languages benefit from running TQA on every significant project for these reasons:
t09n.com's TQA tool is available on the Professional plan (₹2,499/month) — no separate quality management software required.
TQA stands for Translation Quality Assessment. It is the structured process of evaluating a translation against its source, categorising errors by type and severity, and producing a numerical quality score.
MQM stands for Multidimensional Quality Metrics — the international standard framework for TQA. MQM defines error types (accuracy, fluency, terminology, style, locale convention) and severity levels (critical, major, minor). Errors are weighted and counted against word volume to produce a quality score.
A critical error is one that causes harm, legal risk, safety issues, or completely changes the source meaning — such as omitting a safety warning, translating a dosage incorrectly, or inserting content not in the source. Critical errors cause automatic TQA failure.
No. Proofreading checks the target text only for grammar, spelling, and style. TQA is a bilingual comparison — it checks every segment against the source, categorises errors by type and severity, and produces a numerical score. TQA is more thorough and produces defensible quality documentation.
Yes. t09n.com TQA is built in for all users on the Professional plan (₹2,499/month). Upload source and target, get a full MQM-style report with critical/major/minor error breakdown, then export the report for client delivery. No add-on or third-party tool required.
Full MQM-style TQA for Hindi, Gujarati, Tamil, and 10+ Indian languages. No add-ons, no spreadsheets.
Start Free on t09n.com →