Loading config…
Change →
Bilingual Glossary (optional)
Upload spreadsheet
Drop or click ·
.xlsx, .xls, .csv, .tsv
No glossary loaded. Upload a spreadsheet or add rows manually.
Input Texts
Glossary Adherence (local pre-check)
LLM Rubric Scores
LLM Commentary
Sentence-Level Review
Reference Guide — Rubric Criteria
| Criterion | Scale | Description | Frameworks |
|---|---|---|---|
| Accuracy & Faithfulness | 1 – 5 | Faithful meaning transfer. No omissions, additions, or distortions. Correct handling of numbers, names, dates. | NAATI meaning transfer; ATA Misunderstanding / Addition / Omission |
| Target Language Quality | 1 – 5 | Grammar, natural expression, syntax, spelling, punctuation. | NAATI language proficiency; ATA Grammar / Syntax / SP / P |
| Terminology & Word Choice | 1 – 5 | Consistent terminology. Correct domain language. No false cognates. | ATA Terminology; MQM Terminology; NAATI domain knowledge |
| Style, Register & Cohesion | 1 – 5 | Readability, tone, formality, coherence. No excessive literalness. | ATA Cohesion / Literalness; NAATI register; MQM Fluency |
| Cultural & Pragmatic Adequacy | 1 – 5 | Culturally appropriate rendering, idioms, conventions, localisation. | NAATI cultural competence; MQM Locale convention |
| Glossary Adherence (when glossary provided) | 1 – 5 | Strict compliance with bilingual glossary. Expected target terms used wherever source terms appear. | Custom / project-specific terminology compliance |
Document mode returns per-criterion numeric scores (1–5) plus qualitative commentary. Best for holistic quality assessment.
Sentence mode uses MQM-style error annotation: each aligned sentence pair is reviewed for specific issues, tagged with severity (none / minor / major / critical) and category (accuracy, fluency, terminology, style, glossary, omission, addition). Best for localising defects and for translator feedback.
Per-criterion scale: 5 Excellent · 4 Good · 3 Adequate · 2 Poor · 1 Inadequate
Overall percentage bands: 90%+ Professional · 75–89% High quality · 60–74% Acceptable · 40–59% Below standard · <40% Inadequate
Sentence mode uses MQM-style error annotation: each aligned sentence pair is reviewed for specific issues, tagged with severity (none / minor / major / critical) and category (accuracy, fluency, terminology, style, glossary, omission, addition). Best for localising defects and for translator feedback.
Per-criterion scale: 5 Excellent · 4 Good · 3 Adequate · 2 Poor · 1 Inadequate
Overall percentage bands: 90%+ Professional · 75–89% High quality · 60–74% Acceptable · 40–59% Below standard · <40% Inadequate