8 articles in this category
Forget the academic jargon. Here's how translation theory can help you pick tools, avoid disasters, and make better decisions—based on real-world expe...
BLEU is still everywhere, but it's a poor stand-in for human quality. Newer metrics like COMET and BLEURT are better—and here's why you should trust t...
Neural MT slashes costs. But for legal, medical, or marketing text, raw machine output is a liability. A head-to-head comparison of human, MT, and hyb...
We compare BLEU, TER, and COMET for judging MT output in a working translation workflow, then argue that COMET and HTER together offer the most practi...
BLEU is dead. If you're still using it to judge translation quality, you're flying blind. Here's what to use instead—and why post-editing is your real...
That BLEU number you're chasing? It doesn't measure what you think. If you're choosing a translation tool or shipping output, you need a metric that r...
You think a high BLEU score means good translation? Think again. Here's why translation theory is failing you and what to do about it.
BLEU is dead. If your post-editors are burning out, the fix isn't a new metric—it's translation theory. Here's how to use it today.