The potential of automated translation in multilingual litigation
During a recent webcast presented by Inventus, eDiscovery experts discussed ways firms and organizations can overcome the obstacles of multilingual litigation – a growing trend in the legal field as litigation increasingly goes global.
“Accelerating International eDiscovery: Overcoming the challenges of multilingual litigation” featured the insights of Jerome Torres Lozano, director of professional services at Inventus; John Tinsley, chief executive officer at Iconic Machine Translations; and Dominic Piernot, an eDiscovery consultant.
Multilingual Litigation is on the Rise
Multilingual litigation is a mainstay of litigation and continues to trend upwards. By 2020, about 80% of litigation will be multilingual, according to Gartner. Not only is litigation going to be largely multilingual, there is going to be a lot more electronic content associated with it. In tandem with the growth in multilingual litigation is the explosion of online data, which is growing at a 40% annual compound rate.
Another challenge on the horizon is an increase in cross-border and multijurisdictional matters. Nearly 40% of all companies surveyed in 2019 by Statista reported involvement in cross-border litigation.
Translations can be Costly and Complex
The magnitude of multilingual litigation often makes certified translators a cost- and time-prohibitive choice during the discovery process. Law firms have been using machine translation, or automated translation, for several years, but until recently, the poor quality of the translations presented its own set of problems. In the early days of automated translation, systems were unable to differentiate between a surname and a common word. In one Japanese matter, for instance, the translation software translated the name of a plaintiff, Mr. Suzuki, into something like “Mr. Sea Bass.” In each instance, the name had to be manually overwritten.
Machine Translation During Cross-Border Discovery
Recent technological advances, however, have enhanced the reliability and accuracy of machine translation.
Automated translation software often refers to online applications, such as Google Translate.
Law firms also call on the services known as “professional machine translation” from providers such as Iconic Machine Translations, a Dublin, Ireland-based translation software provider that serves Inventus.
How are professional machine translation providers different?
Professional machine translation uses neural translation, which has helped to prevent misinterpretation, and the software can be “taught” to correct past mistranslations.
While certified translators are still considered the gold standard in accuracy, they take time, are costly, and impractical for reviewing large volumes of discovery documents.
In the discovery stage, most litigation practitioners need a translation that’s effective enough to determine whether a document is relevant, and that often means some type of automated translation.
Online machine translation tools have the benefit of speed and low cost but have a number of limitations. The software tends to be one-size-fits-all, the type of application that someone might use to translate a menu in Chinese.
The main drawback of online machine translation is its lack of information security - what’s happening to the data and where is it being processed, and it’s largely the reason that online translation is ruled out for legal discovery.
Professional Machine Translation: Balancing Cost, Security, and Accuracy
During discovery, professional machine translation gives the best balance between the two options: It’s quick, affordable, secure, and flexible in how it can be integrated.
As far as cost, certified translators typically charge by the word, whereas the price is usually by the page in automated translation.
Iconic runs its cloud platform on Amazon Web Services and can use a Relativity parlance on a workspace-by-workspace basis to configure where information is processed. A client could specify on a matter-by-matter basis to process the information in different countries, depending on the requirements of a particular case.
Iconic also has a non-retention policy to protect information security. When something appears on Iconic’s cloud servers, it happens over a secure connection. After the information is processed, it’s sent to the client’s Relativity workspace, and none of the information is retained by Iconic.
Litigators can have the translation application configured for their Relativity workspace and as part of the configuration, direct the application to a particular data center. Then, the litigator selects the documents they want to translate and hits the translate button. The user will be prompted to either specify the language, or have it auto-detected. That sends a request to Iconic’s Amazon web servers in a specific data center, such as London, and starts up a dedicated server instance running the machine translation software for the respective language. The information is sent over a secure connection over hypertext transfer protocol (HTTP) and to that dedicated instance.
Once the translations are completed file-by-file, the documents are sent back into Relativity and deleted from the server. As an extra precaution to make sure all the data is expunged once the translation has been completed, that server is shut off.
In the first half of this year, Iconic translated over half a billion words of content to English from more than 30 foreign languages.
Ultimately, different types of translation serve as tools for different purposes. While professional machine translation might be most practical for the discovery process, a litigator might agree to certified translators for specific documents to be presented in court.
To listen to the full webinar, visit https://register.gotowebinar.com/recording/4960951784749693195