Accessible Intake Research
Testing how call teams capture language preference
A study design for comparing open language-preference prompts with assumptions based on accent, name, or prior records.
Finding and scope
Federal language-access and accessibility resources support meaningful communication and accessible processes. They do not show that one phone prompt works for every caller. This article proposes a service-level study and makes no claim about legal compliance or measured outcomes.
Evidence base
The HHS guidance portal on limited English proficiency collects federal language-access resources. Executive Order 13166 addresses access to federally conducted and assisted programs for people with limited English proficiency. W3C guidance on consistent help explains why help mechanisms should appear predictably. Applicability depends on the organization and service; none of these sources validates a commercial call script.
Proposed method
Include eligible callers who reach a queue where approved language help is available. Randomize between the current process and a plain, open prompt that asks which language the caller prefers for the conversation. Do not use name, accent, country, or area code to assign language.
Compare the recorded preference with the language the caller confirms at the end of the interaction or in a short independent follow-up when consent and policy allow. Measure corrected language selections, interpreter connection completion, caller requests to change language, abandoned handoffs, handling time, and records with no confirmed preference. Review whether the prompt itself caused confusion.
Inference limits and decision boundary
Language preference can change by topic, channel, or day. Agreement with a saved record does not prove that communication was effective. The assistant may offer available choices and record the caller's answer. It should not infer proficiency, disability, nationality, literacy, or immigration status.
Limitations
Callers who disconnect may be missing for reasons related to the prompt, which can bias results. A follow-up can favor people who are easier to reach. Results depend on available interpreters, queue design, languages offered, and staff training. The study cannot establish civil-rights compliance or accessibility by itself.
Sources
1. HHS, Limited English Proficiency resources 2. U.S. Department of Justice, Executive Order 13166 3. W3C, Understanding Consistent Help