How can more legal help providers get more of their information & guidance into more languages?
There is a huge language access problem in legal services. So many people who need help have issues with Limited English Proficiency. It would be better to get these LEP users to articles, guides, FAQs, and services in their own native languages. But there is not enough funding, staffing, and capacity to provide robust information & services in all languages needed.
Especially since each jurisdiction or organization is having to do language access on their own — it becomes a huge budget & capacity issue.
A 2013 report for the Legal Services Corporation, “Can Translation Software Help Legal Services Agencies Deliver Legal
Information More Effectively in Foreign Languages and Plain
English?“ by Jeff Hogue & Anna Hineline (pdf at link), outlines different strategies that legal aid groups can use to increase the capacity & accuracy of language access efforts.
They outline various tech strategies that could increase this capacity to serve in multiple languages:
- Machine Translation (like a variation of Google Translate), in which a computer program is receiving the text, and proposing the translation
- Human Translation, in which a person is proposing the translation based on their knowledge of language & the situation
- Translation Memory, in which people record their translations into a database, and then when there is a new text to be translated — they draw on this existing database for the translation
This third category — of a shared database of translations and glossaries — could be a powerful solution to get to scaled, accurate language access. What if legal aid groups & legal help websites shared their multi-lingual (and plain language) translations of paragraphs, sentences, phrases, and words?
If there was a collective, open-source effort to create a Translation Memory database, this could spread the costs out among many groups. Instead of each group translating their content, they could share their past translations and allow other groups to draw from this.
This can also avoid the potential harms of a machine translation solution. In that setup, the providers are hoping that the machine (and its algorithms) can provide accurate & understandable translations. They might have a human to help review this. But the Translation Memory approach prioritizes the expert human translation from the start, and then uses technology to make that approved, hand-crafted translation more accessible and replicable.
The authors of the report highlight that this shared Translation Memory approach could be valuable but costly. Here are some of their recommendations:
- “The amount of time and effort that needs to be put into developing and maintaining a high-quality glossary and translation memory is non-trivial. We recommend that the Legal Services Corporation convene a group of leaders from legal services providers, plain language experts, and court leaders to adopt or discard this approach.” (page 23)
They also recommend gathering a similar group of stakeholders to explore what is ethically & technically possible with combining machine translation with human review or specialized legal glossaries. Could there be an effective way to build on top of Google Translate or Microsoft Translate? It would be important to have a group of stakeholders and expert reviewers decide if this is possible and ethical.
For either a Translation Memory or Machine Translate + Human Review approach, having a shared database of glossaries is a key step. Our team at Legal Design Lab has started gathering glossaries that already exist, to start building an open-source database of legal help-oriented translations.
Please feel free to write or share if you want to work on this project with us! We hope to push language access forward with this infrastructure work, that can lay the groundwork for more accessible and scalable legal help efforts.