One can be forgiven for thinking legal holds only exist in the U.S. It is, after all, a legal responsibility under the Federal Rules of Civil Procedure and characteristic of the United States' e-discovery culture and tradition.

But U.S.-based multinational companies often have foreign subsidiaries, meaning legal holds can be issued for data stored far from U.S. shores. Managing legal holds in different countries, however, can be a difficult process, not in the least because of language barriers.

Just ask e-discovery and legal hold technology provider Cicayda, whose clients have international offices and legal hold needs beyond U.S. borders. Cicayda recently released an updated 3.0 version of its Fermata legal hold solution, which offers foreign language support for a variety of languages, including Japanese, Chinese, Cyrillic and Spanish.

Before Cicayda could add foreign language functionality to its software, however, it first had to understand how its customers went about managing legal holds at their foreign offices. “It's kind of a complex process,” said Clark Rickman, chief of product development at Cicayda.

The process involves both technology and human manual review. Rickman noted that while clients will often use machine learning software to initially translate text, they will have a native speaker review the translation for accuracy and certain nuances the technology might not pick up, such as idiomatic expressions.

An employee at an overseas office, for instance, will receive a legal hold notification “in the exact same manner—with the questionnaire and attachments—as the custodians would receive it,” Rickman said. “Then they can quality control it, provide changes, and provide approval before it is sent out to, in a lot of cases, hundreds of custodians.”

Where Cicayda's Fermata enters is in uploading and sending out the translated notices to various international offices and employees. The solution, therefore, has to be able to register and display each different language's characters and punctuation.

From a developer's perspective, this isn't as difficult as it may sound. To handle multiple languages, the Fermata database “had to be structured in a manner to handle those characters,” which meant simply updating the database with Unicode scripts, Rickman said.

Unicode, a computing standard for the encoding and representation of text, “gives us support for hundreds of languages,” he added. “And we know because of international standards with the character systems and the work that has already been done in technology that our Unicode support is going to support all those languages.”

Rickman noted that without such Unicode scripts, when trying to process a foreign language, “you basically get gibberish characters.”

Still, onboarding Unicode scripts isn't always a seamless process. Since Western languages rely on similar alphabetic characters, each of which take up one byte of space in a Unicode script, having multiple Western languages within a database may cause some similar files to overlap or replace one another.

“When we made the change, we had millions of records in the software previously, so part of the testing was making sure that we did not adversely affect those single byte records in a manner that was unexpected,” Rickman said.

Developers also had to make sure the software was able to display the distinct syntax and punctuation of certain languages as well. Rickman noted that languages like Thai, Lao, and Arabic don't have breaks at the end of each word, and Arabic in particular is read and written from right to left.

Overall, however, updating the legal hold solution to handle multiple languages did not take Cicayda a long period of time. “We had a pretty rapid development,” Rickman said, explaining that it the whole process took “around six months of development, counting testing.”