Machine Translation

What is machine translation?

Machine translation (MT) is the process of translating content from one language to another without any human involvement. The original text (source) is split into words and phrases, and the MT technology will find and replace these with words and phrases in another language (target). Machine translation works by using different algorithms and large databases of existing translations to predict new translations. This process is often used hand in hand with human translation since MT isn’t guaranteed to be 100% accurate.

Machine translation software uses MT technology to offer translations. This software is often used by translation service providers or integrated with translation management software. 

Logiciel Machine Translation

Logiciel mentionnant machine translation comme fonctionnalité ou terme.

Phrase Localization Platform

Crowdin

Transifex

Smartcat

TextUnited

Lokalise

Types of machine translation

While most modern tools utilize neural machine translation technology, older forms of translation may be used in some cases. 

• Rule-based machine translation (RBMT): The oldest form of machine translation, RBMT, is based on programmed information that dictates how a word or phrase in the source language should be read in the target language. This programmed information is based on linguistic rules developed by human language experts and programmers who have extensively studied the rules between two languages. Because language is dynamic and evolves over time, the effectiveness of these RBMTs is limited.
• Statistical machine translation (SMT): SMT learns how to translate by analyzing existing human translations (known as bilingual text corpora). While RBMT takes a word-based approach, SMT uses a phrase-based approach to assemble translations using overlapping phrases. It does this by using statistical models whose parameters are derived through the analysis of bilingual text corpora (both source and target languages) and monolingual corpora (target language).
• Neural machine translation (NMT): NMT is based on neural networks designed to mimic how the human brain learns. These systems aim to learn from each translation task and improve with each subsequent translation. This helps them to recognize patterns in the source language and understand the context of what is being translated to better predict the correct word or phrase choices.

Benefits of using machine translation

The benefits of using machine translation include:

• Faster translations: Machine translation offers a quicker option to translate text from one language to another. Users input text in the source language, and with the click of a button, they receive the translated text in the target language.
• Supplements human translations: Machine translation tools are often used in parallel with human translation. MT is used first, then a human will evaluate the translation for errors and edit for accuracy.
• Consistency in terminology: Because machine translation works based on set algorithms and databases, using MT tools ensures that consistent terminology is used in all translated content using these tools.
• Ability to translate into numerous languages: While human translation might be doable when translating content to a few languages, it gets complicated and time consuming when content needs to be translated into several languages. MT streamlines this process, making it more efficient and easier to manage.

Machine translation best practices

In order to best utilize machine translation, follow these best practices:

• Error-free source text: It’s important to input an error-free source text to get as accurate output as possible. The text should be grammatically correct and have the correct spelling, so make sure to review and edit the source text before entering it into the machine translation system.
• Post-translation check: It’s important to remember that while machine translation technologies have significantly improved in recent years, 100% translation accuracy isn’t guaranteed. If possible, all translations should be checked for accuracy.
• Translation management system: Larger translation projects should be handled with the aid of a translation management system (TMS). TMSs integrate with machine translation tools and offer additional features to reduce redundant tasks and manual steps. TMSs can also help with project management by keeping all team members on track with translation tasks.

Machine translation vs. computer-assisted translation

Machine translation utilizes a system that completes translation without human involvement. It’s entirely done by technology. On the other hand, computer-assisted translation (CAT) utilizes tools designed for translators to use throughout the translation process. Humans do the actual translating, reviewing, and proofreading, but they use CAT tools to improve their productivity and the quality of their translations.