Summary

  • Decision 2026-151 gives certain providers until 15 January 2027 to select a language-routing solution for each call type sent to an intermediary 9-1-1 centre, and requires bilingual coverage plus contingency handling.
  • Three permitted selectors infer a caller’s language from an account or region. The complaint baseline can describe reported cases, but its listed fields do not establish a mismatch rate across all intermediary-routed calls.

The route is a safety boundary

Most Canadian 9-1-1 calls go directly to a public safety answering point, or PSAP, run under provincial, territorial or municipal responsibility. An intermediary call centre enters the path in more limited circumstances: a place may lack a PSAP, or a service such as nomadic VoIP may not supply a reliable caller location. The intermediary’s job is to locate the caller and pass the call to the appropriate PSAP. Language is therefore decided at a point between the caller and the public responder, not necessarily at the destination itself.

The CRTC’s 30 June decision applies to that handoff. It orders providers required to use intermediaries to select at least one language-routing method for each relevant call type by 15 January 2027. Providers that use intermediaries voluntarily are encouraged to adopt the same routing methods. Separately, all providers whose calls pass through an intermediary, whether required or voluntary, must report official-language complaints. This is an English-and-French framework; it does not promise emergency service in every language spoken in Canada.

Four clues, four different callers

The decision allows four inputs. A provider can use a customer’s reported language preference for billing or 9-1-1 service; the subscriber’s billing address; the calling number’s area code; or the serving cell site. The CRTC describes reported preference as the most accurate option, but says it is complex and costly, not supported for every call type, problematic for business and multi-user accounts, and assumes that the subscriber is the caller. The underlying working-group report said no provider had implemented it at that time.

The other three options are faster to deploy, but each describes a proxy. An address can be missing or inaccurate for a default-routed or nomadic VoIP call. An area code may reflect where a number was first issued rather than where its holder now lives or where a call originates; number portability and shared business lines weaken the link further. A cell site describes a service area, not the language of the person using the phone. These signals can help choose a queue. They cannot, on their own, identify who is speaking.

That distinction is not a reason to dismiss interim routing. It is a reason to judge each method against its call type and to keep a human recovery path. Decision 2026-151 requires contracts with intermediary centres to provide English- and French-language support around the clock, through bilingual staff, language queues combined with bilingual staff, or both. If the first queue is wrong, the process must allow a bilingual greeting and language switch, transfer to a qualified telecommunicator or a priority bilingual queue, and interpretation as a last resort.

The order’s own architecture recognizes that a routing choice is not the same thing as a completed conversation.

A complaint baseline is not a failure rate

By 2 March 2027, providers must file a Routing Solutions Report describing their method by call type, their contingency process and their customer notices. They must also file an initial complaint baseline covering complaints received from 1 January 2020 through 1 February 2027. The required details include the call type, answering intermediary, language used, caller’s preferred official language, options offered and option used. A follow-up report is due 1 March 2028 for complaints through 1 February 2028. The Emergency Services Working Group is due to prepare public-facing material by 15 January 2027.

Those fields can make a reported case more useful: an operator or regulator can see where a call entered, which language was used and what remedy followed. But the list does not request the total number of calls sent to intermediaries by call type. Without that denominator, complaint counts alone cannot show what share of calls reached the wrong language queue. They are a baseline of reported complaints, not a measured mismatch rate. That is an inference from the reporting fields, not a finding that the CRTC has already made.

Commissioner Bram Abramson’s dissent accepts the immediate steps while warning against treating them as the destination. He argues that the decision does not define the longer-term architecture for language-aware routing and does not explain why area-code routing was accepted without testing whether a more granular number prefix might be a better, still imperfect, proxy. His disagreement points to a practical review question: which information should travel with a call, who is responsible for maintaining it, and how can a future route reduce dependence on geography without claiming to know more about the caller than the evidence supports?

Section 26 of the Official Languages Act frames the CRTC’s responsibility as a regulatory duty, where reasonable, to help the public communicate with and obtain available services in both official languages from organizations carrying out activities related to health, safety or security. The decision is a concrete response within that mandate. Whether it works will depend on observable handling after the rule is adopted: the selected input, the queue reached, the language in which a person answered, the fallback used and the handoff to the PSAP.

Heng Lu’s Note 65, “Running-Code Primacy,” is an editorial lens here, not evidence about Canadian law or this order: distinguish an institutional declaration from what a running system actually does. For 9-1-1, the decisive record is not only which method a provider announces, but whether a caller can complete the language handoff when an imperfect proxy chooses the first queue.

Sources