Cross-Border edisCovery
Machine translation use cases
Given the increasing rate of globalization and risk of
regulatory action, the likelihood that organizations
will have to contend with cross-border eDiscovery has
never been greater.
1As a result, the likelihood that
counsel may encounter foreign language content
during the eDiscovery process continues to rise.
These foreign language documents can add a layer of
complexity and cost to the already challenging process
of cross-border eDiscovery.
Recently, there has been larger adoption of
Machine Translations in the eDiscovery and
litigation process. Machine Translation has
become not just more popular, but a necessity
depending on the use-case or business need.
For example, according to some estimates, reviewing a document is approximately 10 times as expensive as the technical work performed on it, and translating a document is at least 10 times as expensive as reviewing it. As a result, if the producing party is required to translate any foreign documents this can dwarf all other costs associated with the disclosure.2
However with good planning and effective use of
machine translation, most of the potential headaches and exponential costs regarding foreign language documents can be avoided.
Although there are a number of unique ways to use MT in global business, the examples that VIA encounters most frequently are as follows:
1) Machine Translations as a production tool for opposing counsel.
As mentioned above, the cost for the producing party to translate foreign language documents using traditional human translation can be impractical. Not to mention, time consuming.
While, MT is not necessarily a substitute for human
translation, it can be significantly more cost effective and efficient. As a result, producing Machine Translations for opposing counsel has become more common in large cross border matters.
For example, VIA recently translated 1.7 terabytes of documents from multiple law firms with multiple litigations for a prominent audit, tax and advisory firm. Furthermore, there is now legal precedence for MT as a production tool.
VIA has worked on a number of cases where machine translations were allowed and even encouraged by the courts as a production for opposing counsel. In the case of Chinese-Manufactured Drywall Products Liability
1 https://bol.bna.com/balancing-cross-border-discovery-and-data-protection/
Litigation, the Court stated, in pertinent part:
The best you can do is machine translations, and then if the plaintiff needs some additional translations, the plaintiff can employ someone to translate those specific documents. . . I want the defendants to do it. Do it by machine. If the plaintiffs want to specifically have somebody look at that particular document because it seems like a really hot document, and you
need a better translation, get somebody to do it.3
2) Quality review of in-region counsel.
Frequently, inside counsel will rely on in-region counsel to collect and review cross-border data on their behalf. In these situations, a common best practice is to establish a check and balance for the in-region counsel to understand what they are looking at and evaluate their decisions. Machine Translation can be an effective way to facilitate this process, particularly when you are looking for certain sets of documents.
For example, if the matter is an FCPA investigation you know that the government is going to want to look at
certain documents such as invoices or payment to vendors. MT can be beneficial in this scenario because you don’t care about the details, you just need to know which documents are invoices.
This not only allows you to quickly identify potentially responsive documents, but enables you to spot check in-region counsel to make sure they are on the right track and are not getting side tracked looking at vacation time, when you are scoped to look at payments.
For example, the head of litigation for a large globally distributed corporation was looking at and analyzing the discovery efforts of his regional counsel across the world. Because a lot of global discovery involves supplier contracts, he began to notice that he would continually have the same documents coming up, from the same vendor, across the world. He was then able to determine if this was something he needed to look into further by asking questions like:
1. Is one supplier causing a major percentage of our organization’s cases?
2. Is it a vendor, for example, in a certain geography that we have to pay cash?
3. If that is the case, do I need to validate?
3) Normalizing the data.
Think of it from a data analytics perspective. If you have multiple languages, you can use MT to get all of the data into the same “range,” and have all documents into English.
Normalizing the data into one language, makes the
review process more efficient and enables one resource to evaluate the content across the data sets. Instead of relying on in-region counsel to read a document and tell you what it says, you can use MT to get a basic understanding of the content.
While MT can be a fantastic tool for managing
foreign language data efficiently and cost
effectively, there are a few situations where it may
NOT be the best solution.
1) Culling down a foreign language data set.
Using MT to rule out obviously non-responsive content in foreign language documents can be time consumptive and expensive. Several of our clients, when faced with this challenge, have been tempted to use machine translations to help cull down the data set.
However, it’s best to identify potentially responsive docs in the source language using TAR and then use MT to normalize the content into one language pair. This makes the review process more efficient and eliminates the need to have multilingual parties at every part in the process. That said, if you have a significant number of foreign language documents in your data set, it may be worth having in-region counsel identify key phrases or terms in
the source language and run a search for those terms in your foreign language data set. Then identify the keywords or phrases for the matter or case in English and search the translated content for the English keywords. Compare the two results sets to identify any additional documents that may need to be translated and placed in the English review set for the team. Failure to do so could result in responsive documents being excluded from review and disclosure before anyone has even looked at them.
Each country has its own innuendo or ways of referring to different things. For example, if you search for the term “non-disclosure agreement” in a case regarding your Australian client’s joint project with a German company, you may well miss documents in German discussing the “Geheimhaltungsvereinbarung”, which technically translates as “confidentially agreement”.
2) Using Machine Translation as a substitute for human translation.
Remember, MT only provides a gist and should not be evaluated on the same criteria as human translation. So, once you have normalized your data you may need to do a more formal human translation to get a better understanding of the context of the content.
For example, after you’ve had a chance to review a contract that’s been translated from Chinese to English using machine translation, you may want to ask a Chinese
attorney about the tone or idiomatic issues in a particular section.
This allows you to approach your colleagues in China with a deeper knowledge or a more specific question. As opposed to simply asking them to “translate this and tell me what it says,” you’re now able to say, “Please look at the second paragraph, are there any nuances or colloquialisms
that I should be aware of when evaluating this?”
Conclusion
With the increase in unstructured data and the growing number of cross-border investigations and litigation matters, machine translation will continue to be a viable option for supporting your discovery process and as a production tool for the courts.
Have questions? Let’s talk it over.
In addition to world-class machine technology capabilities, VIA has a global network of over 5,000 legal translators. VIA uses its in-depth legal industry knowledge, legal specific process, experience and scalability to seamlessly partner with law firms and in-house legal departments of large multinational corporations, including Grant Thornton, Cisco, Oracle, HP, Walmart, Expedia, Nike, St. Jude Medical and more. To learn more about how VIA can help support your cross-border eDiscovery needs, call us at 1-800-737-8481 or visit www.viadelivers.com.