machine translation

Medium
UK/məˈʃiːn trænzˈleɪʃən/US/məˈʃiːn trænzˈleɪʃən/

Technical/Academic/Professional

My Flashcards

Definition

Meaning

The process by which computer software automatically translates text or speech from one natural language to another, without human intervention during the core translation process.

Refers broadly to the field of computational linguistics concerned with automating translation, encompassing the systems (tools like Google Translate, DeepL), the underlying technology (neural MT, statistical MT), and the output itself.

Linguistics

Semantic Notes

Often abbreviated as 'MT'. Can refer to the process, the academic discipline, or the product/output. Contrasts with 'human translation' or 'computer-assisted translation (CAT)', where humans use tools.

Dialectal Variation

British vs American Usage

Differences

No significant lexical differences. Spelling of related terms follows regional conventions (e.g., 'programme' vs. 'program').

Connotations

Identical technical connotations in professional contexts. In casual use, both associate it primarily with free online tools.

Frequency

Equal frequency in technical/academic writing. Slightly more common in American tech journalism.

Vocabulary

Collocations

strong
neural machine translationstatistical machine translationuse machine translationmachine translation systemmachine translation output
medium
advances in machine translationquality of machine translationmachine translation enginepost-edit machine translationmachine translation service
weak
free machine translationonline machine translationreal-time machine translationmachine translation research

Grammar

Valency Patterns

[Verb] + machine translation: employ, leverage, post-edit, rely on, integrate[Adjective] + machine translation: neural, raw, automated, instant, poor-quality

Vocabulary

Synonyms

Strong

MT (abbreviation)

Neutral

automated translationautomatic translation

Weak

computer translationAI translationinstant translation

Vocabulary

Antonyms

human translationmanual translation

Phrases

Idioms & Phrases

  • The raw output of machine translation
  • Garbage in, garbage out (applied to MT quality)
  • To post-edit machine translation

Usage

Context Usage

Business

Used for cost-effective, rapid translation of internal documents, customer support chats, or to get the gist of foreign language content.

Academic

A core subfield of computational linguistics; studied for its models, algorithms, and evaluation metrics like BLEU score.

Everyday

Refers to using free online tools like Google Translate to understand a menu, website, or social media post.

Technical

Specific reference to architectures (encoder-decoder), paradigms (rule-based, neural), and deployment in software localization pipelines.

Examples

By Part of Speech

verb

British English

  • We will machine-translate the manual first, then have it reviewed.
  • The document was machine-translated with moderate accuracy.

American English

  • They machine-translated the entire website overnight.
  • We don't recommend publishing machine-translated content without editing.

adverb

British English

  • The text was translated machine-translatedly, hence the odd syntax. (Rare/awkward)
  • N/A

American English

  • N/A
  • N/A

adjective

British English

  • The machine-translation output was surprisingly fluent.
  • We offer machine-translation post-editing services.

American English

  • They used a machine-translation platform for the initial draft.
  • The report highlighted machine-translation errors.

Examples

By CEFR Level

A2
  • I used machine translation to read a French website.
  • Google Translate is machine translation.
B1
  • Machine translation can help you understand the general idea of a text.
  • The machine translation of this sentence is not perfect.
B2
  • Modern neural machine translation produces more natural-sounding text than older systems.
  • Companies often use machine translation for high-volume, low-stakes content.
C1
  • Despite advances, machine translation still struggles with cultural nuance and complex literary devices.
  • The efficacy of machine translation is contingent upon the linguistic proximity of the source and target languages.

Learning

Memory Aids

Mnemonic

Imagine a vending MACHINE: you input a coin (source text), it works automatically, and out comes a drink (translated text). MACHINE TRANSLATION works automatically like that.

Conceptual Metaphor

LANGUAGE IS A CODE / TRANSLATION IS DECODING. The machine is seen as a decoder breaking and re-encoding meaning.

Watch out

Common Pitfalls

Translation Traps (for Russian speakers)

  • Avoid direct calque 'машинный перевод' as the sole term; it is correct, but know the abbreviation 'МП' (MT) is common in tech contexts.
  • Do not confuse with 'компьютерный перевод' which is less standard.
  • The concept of 'post-editing' ('постредактирование') is a key collocation.

Common Mistakes

  • Using 'machine translation' to refer to translation memory tools (which are CAT tools).
  • Saying 'translation machine' (which refers to a physical device).
  • Treating it as an uncountable noun only (it can be countable: 'various machine translations').

Practice

Quiz

Fill in the gap
For technical documents, we first use to get a rough draft, then a human editor refines it.
Multiple Choice

Which term is a direct antonym of 'machine translation'?

FAQ

Frequently Asked Questions

Google Translate is a specific, widely-known example of a machine translation system. 'Machine translation' is the general term for the technology behind it and similar tools like DeepL, Microsoft Translator, etc.

No, not for content requiring high accuracy, cultural adaptation, creativity, or sensitivity. It excels at speed, scale, and gist translation, but human expertise is crucial for final quality, nuance, and specialised domains.

It refers to systems using artificial neural networks, a type of machine learning model that attempts to mimic aspects of the human brain. NMT models the entire translation process as a single, large neural network, typically producing more fluent outputs than previous statistical methods.

It is the process where a human translator reviews and corrects the raw output from a machine translation system to achieve a quality level fit for its purpose. It blends the speed of MT with the quality control of human expertise.