Why Would a Small To Mid-Sized Language Service Provider (LSP) Have a Software Development Team?
We can’t stand still while translation technology is moving at lightning speed
Everyone in this business uses technology, but not everyone is creating their own solutions. We have been tech forward for many years and have developed various applications for functions like client project management, collaborative review and edit, automation, API and API connections, etc. These tools are both internal tools and tools that give our clients a window into our processes at every stage of a project. For example, with intellireview, our review software, we sought to eliminate version control issues that can mean having versions out there that are old, have not been reviewed, or have not been signed off on. This is not only critical to quality control, it saves time and money for our clients.
The translation automation holy grail
Though we are a smaller LSP, our capabilities mirror many of the tech offerings that our larger competitors are touting, including a path to automation. That automation can be defined as connecting our systems, via APIs, to the authoring tools our clients use everyday. When those connections are in place, creators can send a project out for a translation quote, approve the quote and get it started with a few clicks in their familiar working environment. After those clicks, a number of processes are initiated on our end:
The project is evaluated for word count, languages required, special requirements, including formatting translations back into the final file formats, and other factors that determine pricing.
A quote is generated and returned to the client. The pricing is competitive with current per word costs but many of the projects we do, like eLearning courseware, require complex reformatting, voiceovers and subtitles for video and audio, desktop publishing for languages that are longer or shorter when translated, or that use different character sets.
We use translation memory technology that scans the content and finds repeated phrases and sections that we have translated previously. This eliminates redundant translation because the memory saves the content that is previously translated and reuses it. This reuse creates a database of translated phrases for each client, for future use, saving them time and money.
A translator selection process is initiated that matches native speaking translators, who are subject matter experts, with the appropriate projects for their expertise. They translate the project and the translation sent to another translator for review and editing.
Subject matter expertise is critical to assure quality control and cultural relevance.
The reviewed translation is returned to us for any final formatting or engineering required to integrate the translated content into the required file format. This could be a survey, course, document, software string or UI element, slide deck, etc.
The translation is delivered and invoiced.
All of these stages involve technology, both our proprietary software and industry standard tools like translation memory and DTP applications. Our developers build plug-ins and connectors to automate as much of these processes as possible, however human translation is critical.
A note about machine translation
We use neural machine translation, like Google Translate, as part of our workflows. However, pure machine translation, while useful as a starting point, still has its limitations. Solely relying on it will only work in situations where accuracy is not critical and where cultural issues are not relevant. It can provide some context but it is till a long way from being a replacement for human translators and reviewers. Right now, and in the foreseeable future, it is just one more tool to save time and money, not a panacea.
Despite claims of full translation automation, there are always humans pulling the levers
And there have to be. Most claims of fully automated translation clouds are actually existing software tied together, much like the processes described above. Except for very rudimentary projects, translation is a complex business. Our technology goal is to make it as seamless as possible from the client point of view, with highly accurate translations and fast turnarounds. Having developers as part of our team is part of our pursuit of these quality goals.
Note: We don’t sell software. The software we develop is for our clients and our internal use.