Do I really have to do this? Isn't translation overly burdensome?

Not all cases will require translation of foreign documents. This is something which should be discussed and negotiated during pre-discovery meet & confer sessions. Before making a decision on translation, attorneys should have a good idea what percentage of the collection would contain foreign language data - also, how many different languages would be represented? Translation has traditionally lent itself well to "burdensome" arguments due to complexity, expense and time constraints. There are several reasons why these arguments are getting tougher to make:

1. Machine translation good enough for pre-cull, filtering.
2. E-discovery best practices (applied to translation) eliminate duplicates and help to target only the responsive data.
3. Employees are increasingly found to be using foreign language email to avoid detection or "red flagging" by compliance monitoring systems.
   

How accurate is Machine Translation? How is accuracy measured?

Depending on the language sets required, MT accuracy can range from 40-97%. Although many translation engines are commercially available and relatively inexpensive, MT alone is a risky solution. At LitTrans, we combine the best MT engines available with multilingual project managers who can score, measure and test translation results. Sampling and testing is expected in the "functional requirements definition" phase of our engagements. Although MT is not a certifiable translation and you would never want to produce unedited MT, it should be good enough to greatly reduce the data set prior to the more costly processes of issue coding, translating, etc.


What is the easiest (fastest and cheapest) process for translation? How can we ensure that costs are contained while not sacrificing quality?

Searching is native language is always a best practice when it is an option. LitTrans can provide foreign language search terms to match your English search terms, which will ensure that entire collection receives equal treatment. If searching in native language is not an option, then a machine translation can be linked to native files and searched in English. Again, multilingual project managers are required to measure deliverables and ensure that nothing gets "lost in translation." Once potentially responsive data has been targeted (and data set reduced), then issue coding and translation should be considered (but not before!).


Should I hire translators and manage translation projects internally?

There are many independent contractors available for translation work and costs can be reduced by hiring contractors directly. However, most law firms, corporate counsel or service providers lack in-house multi-lingual project management teams to apply QA processes. What if the translations are not "good enough?" or if there are issues with the translation consistency? Will it have to be done again? Good translation can cost a premium, but bad translation can cost you your case. Another important consideration is that most legal professionals prefer to use a neutral 3rd Party for translation, because it prevents any argument of a "favorable translation." In these days of spoliation attacks and sensitivity to evidence authenticity, having a neutral party who can defend their translation processes will greatly reduce exposure and stress in the discovery process.


What if the foreign language collection is total junk? Nothing relevant or responsive - do I still have to pay?

Yes, but not much. You should be able to use machine translation to pre-cull data set and identify potentially responsive data at a fraction of the cost of traditional translating, coding, reviewing, etc.