Machine translation is fast and cheap, but a hybrid approach of man and machine is the most effective strategy.
Professional human translators have always been the go-to way of translating documents for legal proceedings. They have the linguistic expertise, a better understanding of cultural context and the ability to make judgment calls on what and how to translate.
However, machine translation (MT) has become a viable alternative for some types of legal content due to faster turnaround and lower cost while retaining a high level of actionability.
For example, cross-border litigation is now widespread as global commerce increases. The COVID-19 pandemic may have dampened global trade in 2020; however, the World Trade Organization forecast a strong growth recovery in 2021 continuing into 2022.
Consequently, as more corporations operate in more countries, the potential of litigation over patent infringements, copyright violations, and product liability rises.
To compound the challenge, corporate attorneys and law firms must now sift through terabytes of electronic data. These are digital documents, emails, databases, spreadsheets, presentations, multimedia files, voicemails, chat transcripts and online content.
Multiply that by the number of languages the data is available in, and it becomes a complex, costly and all-consuming effort.
This is where multilingual e-discovery comes in. Electronic discovery is the process of identifying, collating, organizing and classifying electronically stored information needed for a court case or legal investigation.
The challenge is how best to translate the data and figure out which data needs translation in the first place. What data is relevant to a case?
When it comes to e-discovery, there’s no debate. MT and language identification are crucial in scanning multilingual data, identifying relevant content, and streamlining the document review process in a scalable fashion.
They help legal teams identify which foreign languages are used in documents. Keyword searches allow them to find specific phrases relevant to a case. Combined, they determine the content that needs to be translated by humans.
Machine Translation vs. Human Translation
After the discovery and review process is complete, the decision comes down to how best to translate selected foreign-language content. Three variables need consideration: quality, speed and cost — and there is a trade-off between them.
- Quality. Human translation offers greater accuracy, especially for legal and technical documents. A team of linguists and legal experts can make better decisions on how best to translate content. On the other hand, artificial intelligence is quickly catching up in understanding context and cultural nuances. Plus, MT can produce more consistent translations, particularly of terminology.
- Speed. In legal proceedings, deadlines are critical. If fast turnaround time is a priority, then nothing beats MT. It can rapidly translate volumes of data that humans can’t possibly match.
- Cost. Between human translators and MT, there are significant cost-savings in the latter. MT’s efficiency and speed are unparalleled.
When to Use (Unreviewed) Machine Translation
MT is simply faster and cheaper. Not only can it translate larger volumes of data rapidly, but it can also translate between multiple languages using a single tool.
However, it can introduce mistranslations and will often sound less idiomatic or fluent. If budget and speed are your primary consideration, then MT makes sense. However, it’s not the only reason you should go for MT. These are some situations for choosing MT:
- The content is not crucial in legal proceedings or will be used internally for discovery.
- The translation is only needed for gisting; that is, errors are acceptable if the translation is understandable and actionable.
- The content is straightforward and doesn’t contain cultural nuances or creative expressions.
- The volume of content is huge, and the translation can be basic.
MT can save time and money during the multilingual discovery process and helps identify the content that requires further review, post-editing, and human translation.
When to Use Human Translation
Even with the advances in AI and neural machine translation, linguists are needed to edit and review MT output, depending on the translation purpose and use case.
Here are some situations where opting for a human translation would be best practice:
- The content is a high priority for the legal team, and as such, it demands accuracy.
- The content requires cultural understanding and context.
- The material uses creative language that can’t be literally translated.
- The language to be translated needs specialized human linguists and translators.
- The project is complex and needs collaboration among professional translators.
- Judgment is needed to decide what and how to translate.
Park IP recommends certified human translation for submissions, which achieves the highest accuracy. For additional information or deciding whether the document is suitable for submission, medium-level post-editing of MT suffices.
Combining Machines and Humans
While the situations above warrant human translation, it is not always efficient or cost-effective to rely solely on human translators for large-scale, cross-border litigation. You would need an army of linguists with legal knowledge across multiple languages.
The ideal solution is to use both machine translation and human translation.
The best approach is to leverage the speed and actionability of AI-enabled MT processes, design configurable workflows to streamline automation, and use professional reviewers and translators for blended tasks.
MT can be the first stage in translating huge volumes of data. Incorporating MT in the discovery can determine the documents that require linguists, thus streamlining the process.
For more on HT vs. MT, watch the session “Will Machine Translation Replace Human Translators?” on YouTube.
Subscribe to Attorney at Work
Get really good ideas every day for your law practice: Subscribe to the Daily Dispatch (it’s free). Follow us on Twitter @attnyatwork.