About: the UNGA 81 speech tracker
An independent way to explore the speeches, speakers and themes of the United Nations General Debate.
Why this project exists
UNGA is the United Nations General Assembly; the 81st session brings together a remarkable range of national perspectives. This explorer is intended to make the full General Debate easier to understand and navigate: follow the speakers, find what was said, and see the subjects and names that appear across the speeches.
IGO Jobs (launching soon) is a careers platform for jobs at the United Nations, international organisations, NGOs and development agencies. It brings openings from across the international system into one place, with clear search and filters and tools that match each role to your profile. This independent UNGA explorer is part of that work: making the international system easier to follow.
Coverage and counting
The schedule contains 197 addresses, including the UN opening and closing addresses. Analysis and search also include 26 rights-of-reply statements: 223 analysed statements in total. Replies are separate interventions, so one delegation can appear more than once. These totals count statements, not unique people or countries.
The saved transcript includes 2 longer chair interventions that were previously included in analysis. These procedural interventions are excluded from topic, mention and keyword statistics. The daily recording timelines retain all UN transcript segments, including chair introductions and meeting instructions; a segment count is therefore different from an address or analysed-statement count.
The General Debate ended on 28 September 2026. The official source snapshot was captured on 2026-09-29; this is a frozen archive, not a live feed.
Methodology: How the data is classified
Source material comes from official United Nations websites and public UN data services and APIs where available. Speech text is presented from official UN sources; IGO Jobs does not write or creatively rewrite transcript content. See the Sources below.
Issues are broad themes grouped by this project, informed by the UNGA 81 agenda and subjects raised in the debate. They are useful ways to explore the speeches, not official UN categories or labels.
Transcript mentions are identified through keyword matching. Issue classification combines keyword matching with AI-assisted context analysis to help identify the subject being discussed. These tools classify and organize source material; they do not add text to the speeches.
A topic total counts each analysed statement once if that topic was identified anywhere in it. Regional percentages use the analysed statements in the selected region and day as their denominator. They do not measure support for a position or the topic’s importance.
AI classification assigns topics to passages of up to four sentences within a UN paragraph. Where those labels are unavailable or no longer match the transcript, keyword matching operates on individual sentences. Statement pages label the units as passages or matching sentences. Topic results are ordered by the number of labelled sentences, rather than mixing passage and sentence totals; the underlying classification methods still differ.
Topic previews retain complete classified passages and prefer passages with explicit topic wording. A passage can cover more than one subject, and a sentence included for context may not itself name the topic. Select the timestamp to check the surrounding transcript.
Mentions rankings count statements containing a matched name, once per statement. A country’s references to itself are excluded from country-mention counts. The keyword list instead counts matching sentences, not word occurrences; comparison word lists count word uses in the selected scheduled address. Repeated replies are not substituted for that address. CSV matrix headers identify each statement and its count unit; quote downloads contain individual timed transcript sentences.
Accuracy and corrections
The English timed transcripts are supplied by the UN and may contain automated transcription errors. Names and aliases can be ambiguous, and automated topic classification can miss context or assign an imperfect label. A mention is not an endorsement, and an absent label is not proof that a subject was never discussed.
Check quotations against the linked official transcript and recording before citing them. For a correction, include the statement URL, timestamp and official source when contacting [email protected].
Sources
Official United Nations sources for the General Debate schedule, speeches, transcripts, video and meeting coverage.
- General DebateSchedule, speakers and official statements.gadebate.un.org/en
- UN TranscriptsTranscript text and timing data.transcripts.un.org/en
- UN Web TVOfficial General Debate video.webtv.un.org/en
- UN Meetings CoverageOfficial reporting on United Nations meetings.press.un.org/en