As many have expected for some time, we are beginning to see an increase in the amount of data collected from foreign countries. Even when most of a document collection takes place within the United States, it is not uncommon for a small number of foreign sources to contain information potentially responsive to a matter or investigation. The purpose of this post is not to discuss or debate the numerous security and privacy regulations that may govern foreign collections, but to discuss what is done with the data once all of those requirements have been met and the data is collected. What are the requirements around production and how is the information to be reviewed and potentially used as evidence?

First, there are a number of translation options. Most of these can result in fairly significant expense, and so parties are wise to assess the costs prior to “meet and confer” discussions to determine if the cost outweighs the potential benefit, if the same information has already been translated for other purposes and, if not, what level of translation may be useful. As a first step, many translation companies offer machine translation. While these machine translations offer more accuracy than free software options such as the Google Translate API (whose days seem to be numbered anyway), machine translation is often referred to as “gist translation.” The purpose is not to perform a translation with 100% accuracy, but instead to give readers the “gist” of the document or correspondence. The result of this process is often the ability to separate documents, identify the documents containing responsive information and take those documents to the next step.

That next step, in most cases, is human translation. Human translation will remain more expensive and time-consuming than machine translation, at least until computers can do a better job understanding context, dialect and other language variations which can radically alter the translation. Because human translation is slower and more expensive, it often makes more sense to go through the machine-translated documents first and then send only a subset of those documents through the human translation process. Human translation is much more accurate because human beings are much better than machines at determining context and are fluent in the native dialect required. Whether engaging in machine translation or human translation, it is important to partner with service providers who focus on translation as their core business. With constant focus come formalized processes, including quality control delivered through software and additional human reviewers.

Another option is to engage with attorneys in the country where the documents are collected to conduct review in the original language. This process may result in the highest level of accuracy in determining which documents are responsive (and certainly, the highest cost), but there are questions as to what happens next. Can the documents be produced to opposition or used in trial in the original language? If not, translation will still be required. If translation is viewed as a “first-pass” review, how will the attorneys working on the case use the document contents? The reviewing attorneys may be able to provide some additional commentary on a document-by-document basis, but that will not always be sufficient for case teams preparing motions or trial strategy.

via Foreign Language e-Discovery – An Introduction | E-Discovery Resources & Information – DiscoveryResources.org.