Augmented Translation Powers up Language Services

By Donald A. DePalma and Arle R. Lommel

Language services today stand on the cusp of a disruptive transformation that will redefine how professional linguists work. This shift will come from the availability of ubiquitous artificial intelligence (AI) that extends their reach and capability and makes them far more efficient than they could otherwise be. 

CSA Research calls this new professional the “augmented translator.” Just as “augmented reality” uses AI to enrich individuals’ access to relevant information about their surroundings, this transformation provides linguists with more context and guidance for their projects. They work in a technology-rich environment that automatically processes many of the low-value tasks that consume an inordinate amount of their time and energy. It brings relevant information to their attention when needed. This computing power will help language professionals be more consistent, more responsive, and more productive, all the while allowing them to focus on the interesting parts of their jobs rather than on “translating like machines.” 

Until now language technology developers have focused their work on speeding up the process and lowering costs. Those drivers have left many translators feeling alienated from the very aspects of their work that attracted them to the job in the first place – the creativity of language, the challenge of solving difficult problems, and the ability to work on stimulating texts and topics. Translators often find that they spend as much time managing the technology as they do translating, and that their rates are always under pressure. The augmented translation model changes all this by assisting linguists when they need it and getting out of the way when they do not. 

Today we see bits and pieces of this new paradigm, but the outlines are coming into focus. What will this new model look like? We predict it will use the following technologies: 

• Adaptive machine translation. This technology – currently found in Lilt and SDL BeGlobal – learns from translators on the fly. It adapts to the content they work on, automatically learning terminology and style. It remembers what linguists have previously translated at the sub-segment level, and goes beyond translation memory to help translate text it has never seen before in a way that is consistent with how the individual professional works. Rather than post-editing MT output of dubious quality, linguists see the results as suggestions they can choose to use or not. The more they use the system, the better these suggestions will become.
• Neural machine translation. Today NMT requires vast amounts of processing power, but – as the technology matures – it will improve MT’s fluency and ability to “blend in” with human translation. Even if NMT is in the upswing of a hype cycle, it is a major step forward. Many of the major tech players – such as Baidu, Facebook, Google, and Microsoft – and dedicated translation technology providers – such as SYSTRAN and Iconic – are actively developing this technology.
• Lights-out project management. Project management can be time-consuming for both managers and linguists. Manual processes such as invoicing and paperwork that eat up valuable time can be automated. When lights-out systems handle these tasks without the need for human intervention, they free up translators, interpreters, and reviewers to focus on their tasks.
• Automated content enrichment (ACE). This technology is just catching on, driven by projects such as FREME and commercial offerings such as OpenCalais. ACE will benefit linguists by automatically linking terms to authoritative resources and by helping disambiguate them, which will improve MT. It will open new windows in transcreation by helping them find locale-specific content and resources that can make translations more relevant for the target audience.

If this vision is to become a reality, the individual technology components will need to talk to each other and work and learn from the linguist. For example, an ACE system will query a terminology database to identify terms and suggest them. The MT system will interface with both....

Continue reading the rest of this free blog post on our website here: Augmented Translation Powers up Language Services

Looking for data? Here is a recent list of Automation reports, available to members: CSA Automation Research