liveBook · Manning

Machine translation

As explained in LLM Evaluation and Alignment, The Foundational Ideas

Machine translation is the automated translation of text or speech from one human language into another.

Machine translation automatically produces translated text. Its output can be compared with human reference translations to support faster iteration and system comparisons. section 2.0 section 2.1

BLEU measures lexical overlap, not meaning, synonyms, or paraphrases. Treat its score as a useful but limited signal of translation quality. section 2.0

Benchmarks and low-resource languages

BLEU remains a primary metric for comparing systems on WMT benchmarks. For languages such as Arabic and Telugu, token matching can provide a more reliable signal than an LLM judge. section 2.3.1 section 2.3.1

How LLM Evaluation and Alignment, The Foundational Ideas covers it

How deep it goes
Explains BLEU's lexical-overlap approach and practical role, but does not detail translation-system internals.
Examples
Uses BLEU benchmark comparisons, WMT, and low-resource Arabic and Telugu evaluation as examples.
What you can do after
Evaluate and compare translation systems with BLEU while treating its score as a limited signal rather than semantic judgment.
How it connects
Connects machine translation to BLEU, human reference translations, and manual judgments of adequacy, fidelity, and fluency.

Also explained in

Compare how 12 books explain machine translation →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of LLM Evaluation and Alignment, The Foundational Ideas. The section links point to the parts of the book it draws on.