Language barriers in healthcare are usually discussed as something that happens during the visit. An interpreter joins, the conversation works, the patient leaves understood.
Then the clinician sits down to write the note.
That second half rarely gets planned for. The visit happened in Somali or Nepali or Arabic, and the record has to end up in English, often hours later, reconstructed from memory and a few lines jotted between rooms. Family Health Centers, a Federally Qualified Health Center (FQHC, a clinic that receives federal funding to serve a community regardless of ability to pay) running nine sites in Louisville, Kentucky, documents patient visits in five languages.
Two different problems sit inside that sentence. Most practices have arrangements for the first one. Very few have been asked about the second.
The Half of Language Barriers in Healthcare Nobody Plans For
Interpretation is a budgeted line item. Practices contract for it, schedule around it, and account for it in the cost of a visit. Documentation of that same visit is not budgeted at all. It is absorbed by whoever conducted the encounter, after clinic, for free.
Here is what that absorption looks like. A provider finishes a visit conducted through an interpreter. The clinical reasoning happened in real time and in two languages. Now the note has to be written in English, and the provider is translating their own recollection of a conversation they were only half a participant in.
The time cost is the part people notice. The record-quality cost matters more. A note reconstructed six hours later is shorter than one written from the encounter itself, and the detail that drops out first is the specific patient language: how the symptom was described, what the patient had already tried, the hesitation before an answer. That is the material that makes a chart useful to the next clinician who opens it.
Now multiply across nine sites and several language communities. The load is not spread evenly. It concentrates on whichever providers see the most patients with limited English proficiency, usually the same handful of people every week. That uneven distribution will look familiar to anyone who has watched how health centers handle social needs screening.
The result is a quiet inequity in the record itself. Visits in English produce fuller notes than visits in Somali, for reasons that have nothing to do with the patients or the providers.
Family Health Centers, Nine Sites, Five Languages
Family Health Centers runs nine locations across Louisville, Kentucky, offering primary care, dental and behavioral health services.
Its documentation covers multiple languages, including English, Spanish, Somali, Nepali and Arabic, according to the practice’s customer success story. Roughly 40% of the practice’s patients are non-English speaking, as its chief clinical informatics officer described in a 2026 announcement.
The languages after the first two are the interesting part.
Spanish-language support is close to a solved problem in American healthcare technology. Most vendors have it, most practices expect it, and a practice serving a large Spanish-speaking population has options. Somali and Nepali are a different situation entirely. They are the languages of specific resettled communities in specific cities, and Louisville is one of those cities. A practice serving them cannot assume the market has built for it.
That is what makes this a useful example rather than a general one. The question is not whether a health center can document in Spanish. It is what happens to the note when the visit was conducted in Nepali, which is a question most practices in that position have simply had no answer to.
What Changed in the Documentation
Family Health Centers documents these visits using Sunoh.ai, a leading AI medical scribe.
The mechanism is ambient listening. The technology listens to the encounter as it happens and produces a structured draft note from it, in the language configuration the practice has set. The provider is not dictating afterward or working from recollection, because the source material is the conversation itself.
The draft is a draft. The provider reads it, edits what needs editing, and signs it. That review step is not a formality and it is not optional. The technology produces a starting point; the clinician remains responsible for what enters the record.
Adoption is the number worth reading here. Provider usage at Family Health Centers grew from 40% in late 2025 to more than 60%, with 10 providers using Sunoh across the group, per the same customer success story.
Growth like that is a better signal than a satisfaction score. Nobody at a health center is required to use ambient documentation. Providers who try it and find it slower or less accurate than their own habits quietly stop. Usage rising over months means the clinicians who tested it kept it and their colleagues followed, which is the only endorsement of a documentation tool that has ever meant anything.
Where the Technology Stops
This distinction matters enough to state plainly: documenting a visit in the patient’s language is not the same as interpreting the encounter, and it does not change a practice’s language access obligations.
Those obligations are real and they are legal. Section 1557 of the Affordable Care Act governs language access for patients with limited English proficiency, and nothing described here speaks to whether a practice meets them. That is a question for the practice and its counsel, not for a documentation tool.
Qualified interpreters are still required, and the arrangements a practice has already built around them still stand. The interpreter is in the room to make the clinical conversation possible, and that role is untouched by any of this.
What the documentation layer adds sits on top of that. The record reflects the visit that actually happened, in the language it happened in, produced from the encounter rather than assembled afterward. Those are additive things. One does not substitute for the other, and a practice that treats them as interchangeable will end up with a compliance problem and a worse chart.
Drawing that line clearly is in the practice’s interest before it is in anyone else’s. A muddled internal understanding of what the technology does is how a well-meaning scheduler eventually tells a patient an interpreter is not needed.
What Solving Language Barriers in Healthcare Gives Back
The practical effect on the documentation side is that multilingual visits stop carrying a tax that English-language visits do not.
Before, a provider’s after-clinic charting load was a function of how many patients they saw and how many of those visits happened in a second language. Now it is closer to a function of the first number alone. For the providers carrying most of the multilingual schedule, that is not a marginal change.
The equity framing is worth getting right, because it is easy to get wrong. The goal is not to change how patients communicate. It is that a Somali-speaking patient’s record ends up as complete as an English-speaking patient’s. That is a system design outcome, and the system was producing the gap.
Two things are worth measuring. Note completion time, segmented by the language of the visit, is the direct measure, and the gap between languages is what should close. After-clinic charting hours, tracked by provider rather than in aggregate, is the second, because a departmental average hides the four people carrying the load. Both are visible in most systems already.
This transfers to any practice with a concentrated non-English patient population and providers finishing notes after hours. It transfers less well where multilingual visits are occasional, because the documentation tax is small to begin with. Whatever the language, the clinician still reviews and signs before anything enters the record. On what ambient documentation changes for clinicians, the time recovered tends to show up in the evening rather than the workday.
Frequently Asked Questions (FAQ)
What Are the Most Common Language Barriers in Healthcare?
At the operational level there are three: interpretation during the visit, documentation of that visit afterward, and written patient communication such as instructions and portal messages. Practices typically have arrangements for the first, ad hoc handling of the third, and no plan at all for the second.
Does an AI Scribe Replace a Medical Interpreter?
No. An ambient documentation tool produces a draft note from the encounter for a clinician to review and sign. It does not facilitate the conversation between patient and provider, and it has no bearing on a practice’s language access obligations. The interpreter’s role is unchanged.
Which Languages Can Ambient Documentation Handle?
Supported languages depend on the configuration a practice sets. Family Health Centers documents in multiple languages, including English, Spanish, Somali, Nepali and Arabic, which is a useful real-world example rather than a capability list. Any practice evaluating this should confirm its own languages specifically.
Does Documenting in Another Language Slow the Visit Down?
The evidence at Family Health Centers points the other way. Provider usage there grew from 40% in late 2025 to more than 60%, and clinicians do not keep using a tool that costs them time in the room. The time it gives back is time that used to be spent after hours.
What This Looks Like at Your Practice
Language barriers in healthcare do not end when the visit does. They carry into the record, and the practices serving the most linguistically diverse populations absorb the most of that cost without ever putting it on a budget line.
Family Health Centers did not solve interpretation. It already had that. What it changed is what happens to the note afterward, across nine sites and languages most healthcare technology was not built for. Sunoh handles the drafting; the providers still read, edit and sign.
The useful first step is small: pull note completion time by visit language and see how wide the gap is. It is usually wider than expected. From there, how ambient documentation works in practice and how health centers handle social needs screening fill in the operational picture. Schedule a 1:1 discussion to talk through your own language mix.