Summary

  • APNIC reports 363 survey completions in languages other than English in 2026, about 26% of 1,385 valid responses. The published language table describes the mix, not who those respondents represent.
  • The released tables do not connect response language to respondent relationship, subregion or stated priorities. The EC’s promised strategy response is a chance to explain the translation and decision path without exposing confidential comments.

A language count is an intake measure

The number is useful, but narrower than it first appears. APNIC says 1,385 valid responses were received for its 2026 survey; 363 were completed in a language other than English. That is roughly one in four responses. The report’s language table shows ten non-English options: Bengali, Indonesian, Traditional and Simplified Chinese, Mongolian, Japanese, Thai, Burmese, Vietnamese and Korean. Bengali accounts for 29% of the non-English responses and Indonesian for 23%; the remaining shares range from 1% for Korean to 12% for Traditional Chinese.

Those percentages show which language versions respondents used. They do not say whether respondents were APNIC Members, NIR members or other stakeholders; whether they came from a particular subregion; or how their priorities differed. Nor do they measure the number of people who could not or did not respond. The questionnaire asks respondents where they live, their organization’s relationship with APNIC and their role, but the public release does not connect those fields to the language table.

That boundary matters because APNIC’s survey is a channel into institutional planning. A language choice records how a response entered the process. It is not a count of all people who need that language, and it does not by itself show how much influence the response had.

Three editions, with a changed instrument

The report lists 441 non-English completions in 2022, 347 in 2024 and 363 in 2026. The total survey samples were 1,621, 1,173 and 1,385. The non-English share therefore moved from about 27% to 30% and then 26%. Between 2024 and 2026 the count rose by 16 while the overall sample rose by 212. That is a description of published totals, not a clean trend in participation.

APNIC calls the 2026 edition the fourteenth iteration and says its approach evolved. It now places more weight on strategic direction and future priorities, while operational feedback is gathered continuously through existing engagement channels. The question mix and purpose matter when comparing editions. A changing share cannot be attributed to outreach, language access or respondent preferences from these totals alone.

The composition also shifted. Bengali rose from 26% to 29% of non-English completions between 2024 and 2026; Indonesian rose from 14% to 23%, while Thai fell from 11% to 4%. The table does not supply exact counts for each language, and its percentages are rounded. It gives APNIC and readers a reason to ask what languages were selected, not a basis for explaining why.

Translation is part of the evidence chain

In its 2024 report, APNIC described a specific translation step: non-English free-text feedback was translated into English with ChatGPT and checked by APNIC language specialists. The 2026 report does not say whether that method continued. Its silence is not evidence that translation or review did not happen. It does mean the public record reviewed here cannot confirm the current process.

That process is more than a production detail. Translation can preserve an operational term, flatten a local distinction or make two different comments look alike when they are grouped into a theme. No such distortion is demonstrated in APNIC’s report. The point is that readers cannot evaluate a path they cannot see. The survey itself asks for a respondent’s economy and relationship with APNIC, and the released appendix reports results by relationship, subregion and economy classification. The separate language table does not cross those dimensions with language or strategic priorities.

There is a real privacy constraint. The questionnaire says individual responses are confidential, but free-text comments are passed to APNIC unedited; a respondent who identifies themselves in a comment may be visible. A useful public account should therefore preserve method and disposition, not publish raw comments or small cells that could expose a person.

The next public record

The Executive Council says it will publish a formal response later in 2026 explaining how the survey will inform APNIC’s 2028–2032 Strategic Plan, whose development begins next year. That response can make the evidence chain legible. It could state how the 2026 language versions and free-text material were handled, which safeguards applied to translation and coding, and how the resulting themes were accepted, deferred or left outside the plan.

The released data tables already show that aggregate answers can be organized by respondent relationship, subregion and economy classification. A language cross-tab need not expose identities or every combination. APNIC could publish safe, sufficiently large groupings, suppress small cells, and provide the denominator for each comparison. Where a cross-tab is not safe or reliable, it can say so and explain the limit.

The evidence supports a measured conclusion. APNIC made ten non-English survey languages available in 2026 and counted 363 completions in them. Its report records requests for local-language resources, in-country activities and stronger local partnerships. It does not establish that these requests came only from non-English respondents, that the responses represent every language community, or that the EC has yet acted on them. The count shows an input channel; the EC response can show how that channel was carried into a decision.

Sources