What Section 1557 Requires for Afghan Languages

The rule, precisely

Section 1557 prohibits discrimination on the basis of race, color, national origin, sex, age, or disability in health programs and activities that receive federal financial assistance. The implementing regulation — the 2024 final rule, published May 6, 2024 at 89 FR 37692 and codified at 45 CFR Part 92 — took effect July 5, 2024, with compliance dates staggered through July 5, 2025 (HHS, 89 FR 37692; Davis Wright Tremaine).

The language-access core sits at 45 CFR 92.201. A covered entity “must take reasonable steps to provide meaningful access to each individual with limited English proficiency (including companions with limited English proficiency)” eligible to be served or likely to be directly affected by its health programs (§ 92.201(a)). The services provided under that duty “must be provided free of charge, be accurate and timely, and protect the privacy and the independent decision-making ability of the individual” (§ 92.201(b)). The regulation’s definitions make the connection to national origin explicit: national origin includes “a person’s manifestation of the physical, cultural, or linguistic characteristics of a national origin group” (45 CFR 92.4).

Two features of this design matter for Afghan languages. First, the rule names no languages. The obligation attaches to the individual in the room, not to a tier of “supported” languages — a patient who speaks Pashayi or Prasun is owed the same reasonable steps as a patient who speaks Spanish. Second, the obligation is individual-level: “each individual,” not “each language group.” An interpreter in a related language is not automatically meaningful access in the patient’s language.

How many languages does that implicate for Afghan populations? The governed reference is the Ariana 24-Language Atlas: 24 languages across five families, from Dari and Pashto through Uzbeki, Turkmeni, Balochi, and Hazaragi to the low-resource Nuristani group (Kati and Prasun among them). Speaker figures belong to the Atlas’s per-language profiles, which publish estimates only after per-language verification.

Where the rule stands in 2026

Compliance officers reading about Section 1557 litigation should separate what changed from what did not.

On October 22, 2025, the U.S. District Court for the Southern District of Mississippi issued final judgment in Tennessee v. Kennedy, vacating provisions of the 2024 rule to the extent they extended sex discrimination to gender identity. On June 2, 2026, HHS published a Notice of Vacatur confirming that the vacated gender-identity provisions are void — and that “the other provisions of the Section 1557 Rule remain in force,” with OCR continuing to enforce protections against discrimination based on race, color, national origin, age, and disability (Federal Register, 91 FR 32887; HHS press release). The language-access provisions were not among those vacated.

Separately, Executive Order 14224 (March 1, 2025) designated English the official language of the United States and revoked Executive Order 13166; the Department of Justice rescinded its 2002 LEP guidance in April 2025 and suspended LEP.gov materials in July 2025 (Harvard EELP tracker). An executive order does not amend Title VI of the Civil Rights Act or Section 1557, and neither statute has been amended: the regulatory duties in Part 92 remain binding on covered entities (NILC; KFF). Courts have also recognized private enforcement under Section 1557, so the exposure does not depend on OCR’s docket alone (NHeLP). Federal enforcement priorities can shift, and further litigation over the 2024 rule is possible; the prudent reading is that the obligations stand and should be verified against the current eCFR at each decision point.

“Qualified interpreter” when no certification exam exists

The rule’s standard is competence-based, not credential-based. A qualified interpreter under § 92.4 is one who: has demonstrated proficiency in speaking and understanding both spoken English and at least one other spoken language; can interpret “effectively, accurately, and impartially,” using necessary specialized vocabulary “without changes, omissions, or additions and while preserving the tone, sentiment, and emotional level of the original oral statement”; and adheres to generally accepted interpreter ethics principles, including client confidentiality (45 CFR 92.4).

This matters for Afghan languages because, for most of them, there is no certification to hold. The national medical-interpreter certifying bodies — CCHI and NBCMI — offer full language-specific oral certification only in a short list of high-volume languages, and no Afghan language is on either list. On the court side, federal certification exists only for Spanish; interpreters of other languages appear in federal and state registries as “qualified” or “language-skilled,” a court-conferred standing, not a certificate (U.S. Courts).

The rule neither requires a certification nor excuses its absence. When no exam exists, the covered entity carries the qualification burden itself, and the file has to show it:

• Dual-language proficiency, assessed and documented — in the language the interpreter will actually work in, named precisely (Dari is not Iranian Farsi; Hazaragi is not standard Dari).
• Interpreting skill, distinct from bilingualism: consecutive interpreting, sight translation, register control, medical terminology.
• Ethics adherence, referencing a recognized code such as the NCIHC National Code of Ethics.
• Preparation toward external credentials where they exist — CCHI, NBCMI, court registries — as the benchmark the assessment points to. Qualification is demonstrated against external standards; it is not something a provider or agency awards itself.

Section 92.201(e) then closes the fallback routes that scarce languages invite. A covered entity must not require an LEP individual to provide or pay for their own interpreter; must not rely on an accompanying adult except in an emergency or on the individual’s specific, privately made and documented request; must not rely on a minor except in an emergency; and must not rely on staff who are not qualified interpreters, qualified translators, or qualified bilingual/multilingual staff (45 CFR 92.201(e)). Where remote modalities are used, § 92.201(f)–(g) sets quality floors: real-time full-motion video, clear audio over dedicated high-speed connections, and adequate training for staff who operate the technology.

Machine translation under § 92.201(c)(3)

The 2024 rule addressed machine translation directly, for the first time: “If a covered entity uses machine translation when the underlying text is critical to the rights, benefits, or meaningful access of an individual with limited English proficiency, when accuracy is essential, or when the source documents or materials contain complex, non-literal or technical language, the translation must be reviewed by a qualified human translator” (45 CFR 92.201(c)(3)). Machine translation is defined as automated, text-based translation “without the assistance of or review by a qualified human translator” (§ 92.4).

Read the three triggers against what a hospital actually translates — consent forms, discharge instructions, eligibility notices, medication guidance — and human review is effectively the default for patient-facing text. The reviewer must meet the § 92.4 qualified-translator bar in the language pair: demonstrated written proficiency, accuracy without changes, omissions, or additions, and translator ethics.

Three Afghan-language realities sharpen this:

• Coverage cannot be assumed. Machine translation is an option only where an engine actually supports the language in question — a per-language determination to make before the tool is reached for; where no engine exists, § 92.201(c)(2) simply requires a qualified translator from the start.
• Engines substitute dialects silently. A system offering “Persian” typically produces the Iranian standard, which the Atlas documents as distinct from Dari. Machine output can be fluent and still be the wrong language for the reader.
• The reviewer is the constraint. The compliance question is rarely “may we use the engine” — it is “who reviews the output, and what documents their qualification in this language.”

A defensible practice logs each reviewed item: the document, the language and variety, the reviewer, the basis of the reviewer’s qualification, and the date.

Notice duties: nondiscrimination and availability

Two notices carry the public-facing duty.

Notice of nondiscrimination (§ 92.10). Covered entities must tell participants, beneficiaries, enrollees, applicants, and the public that they do not discriminate — including on the basis of “national origin (including limited English proficiency and primary language)” — that language assistance services are provided free of charge, how to obtain them, how to reach the Section 1557 coordinator where one is required, and how to file a grievance and an OCR complaint (45 CFR 92.10). Compliance was due November 2, 2024 (Morgan Lewis).

Notice of availability of language assistance (§ 92.11). Due by July 5, 2025, this notice must state that language assistance and auxiliary aids are available free of charge, and must be provided “in English and at least the 15 languages most commonly spoken by individuals with limited English proficiency of the relevant State or States” (45 CFR 92.11). It must appear annually, upon request, conspicuously on the entity’s website, in prominent physical locations in no smaller than 20-point sans serif font, and attached to a specified list of communications: intake and application forms, denial and termination notices including EOBs, consent forms and instructions for procedures, discharge papers, public-health-emergency communications, and more.

The Afghan-language complication is upstream of the notice itself: the top-15 determination runs on state LEP data, and the most widely used federal tabulations group Dari with Farsi under a single “Persian (incl. Farsi, Dari)” category (U.S. Census Bureau, ACS language tables). A state list can therefore carry “Persian” while the population being served reads Dari — or speaks Hazaragi, which the tables do not surface at all. Because § 92.11(b) sets a floor (“at least the 15 languages”), an entity with a significant Afghan service population can add taglines beyond the state list, and its own documented service-area analysis is the defensible basis for doing so. HHS pointed entities to LEP.gov and Census data for the top-15 determination (Morgan Lewis); with LEP.gov materials suspended since July 2025, record the analysis and its sources explicitly.

These are the enforceable duties. A companion brief examines the National CLAS Standards, and how they differ from Section 1557.

What compliance officers should document

OCR’s evaluation standard rewards paper. In assessing whether meaningful access was provided, the Director gives “substantial weight” to the nature and importance of the communication at issue — and takes into account “the effectiveness of the covered entity’s written language access procedures” adopted under § 92.8(d) (45 CFR 92.201(d)). The written program is not overhead; it is the evidence the rule itself says will be weighed.

The file, by regulation:

• Section 1557 coordinator — required for entities with 15 or more employees; designation was due November 2, 2024 (§ 92.7; DWT).
• Written language access procedures (§ 92.8(d), due July 5, 2025) — at minimum: coordinator contact information; how an employee identifies that an individual has limited English proficiency; how an employee obtains qualified interpreters and translators; the names of any qualified bilingual staff; and a list of translated materials held, the languages, dates of issuance, and how to access electronic versions (45 CFR 92.8).
• Training records on those policies and procedures (§ 92.9, due July 5, 2025).
• Interpreter and translator qualification records — the per-person, per-language assessments described above.
• Machine-translation review logs under § 92.201(c)(3).
• Documented exceptions — when an accompanying adult interprets at the individual’s request, § 92.201(e)(2)(ii) requires the request and the adult’s agreement to be documented.
• Grievance records — retained no less than three calendar years (§ 92.8(c)(2)).
• The service-area language analysis behind the notice list and the staffing plan.

For an entity serving Afghan populations, one column matters most in every record above: which language, named precisely. A qualification file that says “Farsi/Dari” has already conceded the question an investigator will ask.

Where Afghan languages create exposure

Two failure patterns account for most Afghan-language compliance risk, and both are structural rather than negligent.

Aggregation masking. Intake systems, EHR language fields, and federal statistics tend to inherit a category — “Persian,” “Farsi,” sometimes “Afghan” — that dissolves the distinctions the rule turns on. Dari speakers are recorded as Farsi; Hazaragi speakers as Dari; Uzbeki and Turkmeni speakers vanish into “Other.” The consequences compound: demand is invisible in the data, so the language never reaches the notice list, the interpreter panel, or the translated-materials inventory — and the entity’s own records now understate the population it is obligated to serve. Disaggregation starts with a registry that names the languages correctly; that is the function of the Ariana 24-Language Atlas.

Dialect substitution. The reciprocal failure: the language is visible, and a neighboring one is supplied — an Iranian Farsi interpreter for a Dari-speaking patient, a Dari interpreter for a Hazaragi or Pashayi speaker. Section 92.201(a) runs to “each individual,” and the § 92.4 definition requires interpretation to and from “such language(s)” — the language the individual speaks, not a relative of it. Mutual intelligibility is a spectrum; informed consent is not. An encounter record showing a qualified interpreter in the wrong variety documents the gap rather than closing it.

The remaining exposures follow from scarcity: the drift toward family members, unassessed bilingual staff, or raw machine output when no vendor covers the language — each specifically restricted by § 92.201(e) and § 92.201(c)(3) — and the absence of qualification records for the interpreters actually used. None of these is cured by effort at the moment of the encounter. They are cured earlier, in the registry, the roster, and the file.

For the data instrument built on these requirements — an audit-ready index of language-access conformance for Afghan-speaking patients — see the Section 1557 Compliance Reality Index. Institutions that need the requirements operationalized — qualified interpretation, governed translation, and the documentation layer, scheduled and by arrangement — can review the firm’s Section 1557 language-access practice.