Machine Translation

Machine Translation (MT) aims at emulating the human translation process, while offering considerable time and process efficiencies. It is the process by which computer software is used to translate a text from one natural language (such as English) to another (such as German). Alongside our fully human translation services, Global Lingo is also able to utilise MT technologies in order to cater for specific projects with very high volumes and particularly tight deadlines.

Current industry Machine Translation tools use a number of systems and approaches:

  • Statistical machine translation: This type of MT works by using algorithms to analyse existing human translations, known as bilingual text corpora, to find similar segments or phrasing. Statistical machine translation relies heavily on existing bilingual corpora, with a minimum of two million words for a specific domain. They can be trained for specific disciplines or industries, using data relative to that sector.
  • Neural machine translation: This is a deep learning system, which uses the concept of artificial intelligence, and is designed to imitate the neurons of the human brain to predict the most likely sequence of words. Neural machine translation systems are quickly moving to the forefront of machine translation, recently outcompeting traditional forms of translation systems.
  • Rule-based machine translation: This form of MT was developed by language and semantic experts and is based on including grammar, semantic, morphological, and syntactic rules. These systems can deliver consistent translations with accurate terminology when trained in specialist dictionaries.

.

Global Lingo is able to assess a client’s requirement to suggest the most appropriate MT system and approach. Not all content is suitable for machine translation. We can help you make an informed decision about the right machine translation software for you and the benefits it may offer. The right form of MT will depend on the type of content, volume, speed, and quality.

This form of translation is particularly suited to a variety of content types, including very-high-volume projects with tight deadlines, or where a lot of content needs to be first approximately understood to decide what requires full quality translation or post-editing. Machine Translation services are also particularly suited to content which is highly prescriptive, instructional and repetitive, where consistency and technical language are prominent, for instance in technical content such as instruction manuals.

Machine translation software: Features and benefits

At Global Lingo, a cyclical and constant process of learning and training is applied to ensure the output of MT improves over time.
This advanced technical process consists of:

  • Utilising legacy content and previously translated material to actively tailor and train the MT engine to develop an engine bespoke to the client and subject matter.
  • Reintroducing post-edited translations into the MT engine in real time, ensuring a cyclical learning process and continuous improvement of the output quality as a project progresses.

Machine translation: Processes and solutions

In suggesting the most appropriate MT solution Global Lingo will first assess requirements, the industry sector in which a client operates, and topics that are likely to be covered, before suggesting the most appropriate MT engine and process.

Machine Translation can be implemented in a number of ways, including:

  • MT only: If you need to translate large text, content is passed through an MT engine and automatically produced translation is available immediately. For clients requiring a high volume of content translated to understand the overall information, this is ideal. This is used for information purposes.
  • MT plus post-editing: For fully refined, accurate, and linguistically consistent translation, the automatic translation draft undergoes a process of post-editing by a human linguist, ensuring the output matches the quality level of a fully human-translated document. Post-editing is carried out by expert linguists who are specifically experienced in machine translation post-editing to deliver excellence. Machine translation post editing is used for client-facing, high-quality translations.
  • Direct MT integration via API: In addition to content being shared manually, Global Lingo can also support a fully automated MT process via the use of API feeds, allowing content to be pushed directly from a client database (e.g. CMS, DMS, PIM) into our system. This creates a highly automated process with minimal manual input until the post-editing stage.

To see how machine translation and machine translation post-editing can add value to your current translation processes, saving time and money, visit our case study section.

For further information about our machine translation services, get in touch

Contact us for a free consultation and further information or request a quote via the button below.

TECHNOLOGY