CuidadoDeSalud.gov: an example of machine translation misused

Roughly 12 percent of the 317 million people in the U.S. speak only Spanish. For many states with Spanish-only residents, the healthcare act represents the first time that many of the underserved will have affordable access to coverage. Unfortunately, sign-ups in states such as New Mexico, California, and Florida have fallen far short of expectations. Many blame the Spanish website’s shoddy translation.

Problems start with the website’s Spanish title, “for the caution of health.” From there visitors have to navigate “Spanglish” plan details, links to English-only pages and awkward translations.

So what’s making the CuidadoDeSalud.gov sick? The culprit appears to be the significant use of machine translation (MT) without the requisite human oversight and editing, according to research students in New Mexico. Glaring errors in verbs and word order point to automated tools.

It’s a perfect example of how and when not to use MT on a project. Given the complexities of the healthcare plans and the need for absolute clarity when it comes to proof of residency, sign-up deadlines, and coverage, MT is probably not the best tool for the job. Mission-critical content with a sales/marketing goal is almost always best left to the gold standard in translation: human linguists. The translation of the word “premium” is a perfect example of how jargon must be decoded by those most familiar with the nuances of meaning. Odds are the healthcare website also contained fewer than a million words, considered by many experts to be a crucial threshold to capitalize on the efficiencies of MT.

The prescription for a bad translation is the TLC that a quality assurance team can give. In CuidadoDeSalud.gov’s case, those trying to sign up were the beta testers. A native-speaker review by those hired to assist users with registration would have revealed many of the problems prior to launch.

In translation, good intentions don’t make a great safety net. When clients realize you haven’t paid attention to the fundamentals of language, it sends a message that you aren’t truly interested in serving them. Don’t make the same mistake with your next major translation project!