What is Machine Translation?

Machine translation (MT) is the application of computers to the task of translating texts from one natural language to another. One of the very earliest pursuits in computer science, MT has proved to be an elusive goal, but today a reasonable number of systems are available which produce output which, if not perfect, is of sufficient quality to be useful in a number of specific domains. (European Association of Machine Translation)

Why use machine translation?

The two main uses of MT are:
- Productivity
- Communication
- In the translation and localization industry, MT is part of the workflow and it is used in conjunction with - translation memory.
- The three main uses of MT in our industry are:
- Productivity
- Turnaround times
- Cost

What types of materials can be processed with machine translation?

Basically, most content can be processed by our engine. The benefits of this technology is improved when: Sentences are short There is no redundancy There is consistent use of terminology The text avoids jargon There is correct use of grammar The active voice prevails over passive voice

Best Results

*Documentation*
Manuals
Product description
E-commerce platform content
Online reviews
Online training
Agreements, patents
Knowledge base
Internal communications
User-generated content

Worst Results

Creative texts (advertising).

Is quality affected by the use of machine translation?

If done professionally, using a good engine and prepared post-editors, definitely not. We have seen amazing results for even the most complicated accounts.

Which MT engine do we use at Latamways?

We use a paid engine, Google Translate API, which has the following benefits:
1. Neural
2. Integration with professional CAT tools;
3. Confidentiality: Google does not have the right to keep or use your data so you do not need to worry about it. We got you covered!

How does it work?

The source text is checked to confirm the content is suitable for machine translation (subject matter, grammar constructions, type of content).
Pretranslation with translation memory (from high fuzzies up to context match words).
Pretranslation with the MT engine.
Post-editing work performed by a professional (human) editor.
Editing work performed by a professional (human) editor.

What happens after content is machine translated?

Post-editing, the process by which a linguist modifies the translation generated by the machine to achieve a final product that matches the required quality standards. Post-editors need to have specific skills to properly post-edit a machine-translated text. Our vendors are carefully selected and trained to perform according to our standards.

How do I know what kind of post-editing service I need?

The better the structure of the source text, the better the output.
The type of output and the intended use of this content will define the level of editing needed.
Post-editing services can be "light" or "full" and the difference lies in the depth of the editing process required according to client’s needs.

Light post-editing
(most suitable for content that has a short lifespan, internal communications)
The target text is comprehensible and easily readable.
No information has been added or omitted.
Semantic factors take precedence over syntactic and stylistic factors.
Correct spelling, punctuation and grammar.
No untranslated segments are present.

Full post-editing
(most suitable for customer facing materials and content that has a medium to long lifespan)
Terminology is appropriate for the subject matter and target audience.
The target text is comprehensible and easily readable.
No untranslated segments are present.
Semantic, syntactic, stylistic and terminological factors are all of primary importance.
General quality-control procedures are performed.
Appropriate formatting is performed.
Final product is of publishable quality.
Final product has no difference with a human translation.