By: Yoyo Rita
Workplace investigations are replete with multilingual contexts—from non-English speaking witnesses who provide text messages in their native language, to a multinational company that publishes bilingual corporate briefs, to community pamphlets distributed to multilingual neighborhoods to investigate a school district, say. Thus arises the need for translating text from other languages into English in order to complete a given investigation. And in such cases, machine translation services, such as Google Translate, are readily available for our use. However, we must consider its advantages, and major drawbacks, before using these tools. In doing so, we can better utilize these tools and ensure that our investigations are thorough and accurate.
It is first of utmost importance to consider the reliance on machine translation in high-impact contexts, such as criminal accusations, reporting on world events, or interpreting/translating between world leaders. For example, in 2017, “a Palestinian man was arrested by Israeli police for a post saying “good morning” that Facebook’s automatic-translation service erroneously translated as “attack them” in Hebrew and “hurt them” in English”[1] (Facebook apologizes after wrong translation sees Palestinian man arrested for posting ‘good morning’). We also must consider the possible mass loss of meaning, because the predominant language English can then be translated into almost any other language, thus leading to less authors writing in their native languages. While Facebook’s new program called No Language Left Behind (NLLB)[2] aims to improve low-resource language translations with natural language processing AI technology, I feel that this technology could further lead to cultural erasure and loss of meaning on a mass scale. The only viable benefit I could see from NLLB would be to foment good paying jobs for translators/interpreters with those low-resource language specialties to oversee and maintain the creation of a public, free translation resource based on literature, film, and other artistic/cultural activities.
As an interpreter of spoken language myself, I have witnessed machine translation take over most textual translation jobs in the last several years, leaving many translation professionals in the industry with no choice but to do “post-editing” of machine translation output. I was vindicated to recently uncover a video entitled “Pro Interpreters vs. AI Challenge: Who Translates Faster and Better?”[3], which proved that while AI machine translation is improving linguistically, it still lacks human empathy and emotional intelligence, which is crucial in order to capture the intended meaning and impact of an utterance into your source language.
After all, language is not just about words, but rather about what one is trying to convey through those words. Humans, indeed, will always have the upper hand in this regard when understanding the subtle nuances of word choice, slang, and cultural context.
So, in deciding when and how to use machine translation tools, consider the following:
- What is the impact the text has on your investigation? If the answer is little to none, using a tool like Google Translate is acceptable. If it’s more important to the investigation and your ultimate findings, consider using machine translation, then consulting with a translator after the fact.
- How common is the language you need to translate from? If the answer is common, such as Spanish, Arabic, or French, machine translation tools are often more accurate. However, if you are looking to translate from a less widely spoken language, such as Hungarian, Cantonese dialect, or Nepali, machine translation tools are still not reliably accurate.
- Can a qualified translator be hired to quickly and accurately translate the text? If so, consider supporting trained, professional translators in their craft, to ensure a complete investigation.
[1] https://www.theverge.com/us-world/2017/10/24/16533496/facebook-apology-wrong-translation-palestinian-arrested-post-good-morning
[2] https://ai.meta.com/research/no-language-left-behind/
[3] https://www.youtube.com/watch?v=pwOxlpGYJAY