O-Level English oral topics: the themes behind the video prompts

Which themes recur in the video prompts today?

Across 31 recalled current-format rows, 8 theme families account for all of them. Community and the environment are the two largest, each appearing in at least three of the four years since the video-clip format began.

These families are ourgrouping of the recalled record, not SEAB’s, and we reused the same family names the Chinese side of this corpus uses — a Singapore examination board is drawing prompts from the same national conversation in either language, and a shared vocabulary for revision is more useful than two invented ones. SEAB publishes no topic list for either paper. 18 of the 31 current-format rows behind this table are corroborated by two or more independent sources; the rest are single-sourced, as the examples further down show row by row.

What kind of question does the video clip actually ask?

In Planned Response, almost always a reactive question about what the clip shows — how the people in it might be feeling, or what a candidate thinks about the activity — rather than a position to argue for or against. That is a real difference from the Chinese oral’s 会话, which is built almost entirely from opinion prompts.

As recalled for 2023: “Who are the people watching this event, and what do you think they are feeling?” As recalled for 2025: “How do you think people's plans may be affected by the weather?” Both ask a candidate to read the scene and respond to it, not to take a side on a social question the way the Chinese conversation typically does.

Where a Spoken Interaction discussion was recorded separately from the Planned Response, its questions lean the other way — closer to the opinion-prompt shape. As recalled for 2023, alongside a clip on BMT / National Service (follow-up discussion): “What can schools and parents do to develop the character of their children? / Do you think all students should join uniform groups as their co-curricular activities?” As recalled for 2024, alongside a clip on Recreational space design (follow-up discussion): “How would you design an ideal recreational area? / Do you agree that people these days do not have enough time for recreational activities?” A candidate who has only rehearsed describing a scene is not ready for that second question.

Day numbering is the least reliable part of this whole corpus, even where the theme itself is well corroborated. A 2024 clip on Elderly participation in a marathon (a 101-year-old runner) is recorded in its own source only as day “4-or-5 (disputed)”— compilers who agree on the theme still disagree on which day carried it. This page does not print day numbers as fact anywhere, for that reason.

How good is the evidence, year by year?

Uneven, and stronger than it looks in places. 6 of the 8 years covered have at least one theme carried by two or more independent sources — 3 of the 4 current-format years among them, where at least four tuition operators publish detailed recall within days of the sitting. The earliest and the most recent year are the weak points.

The 2026 row is the newest and thinnest: one publisher, recorded about two months after the July sitting, with nothing yet to corroborate it against. The 2019 row is the oldest and thinnest for a different reason — a single compiler’s blog is the only record we could find for that whole year. Everything in between is in noticeably better shape, which is itself worth knowing before leaning on any single row.

What about the pre-2023 paper?

A separate exam, kept here for one reason only: the themes it covered still show what this board asks about, even though the format around them has changed. Its stimulus was a photograph, not a video clip, and it included a Reading Aloud component that 1184 dropped entirely.

These 31 rows describe 1128, the paper Singapore sat from 2019 through 2022 before 1184 replaced it in 2023. Read them for the subject matter only — family, sport, technology, the environment, school life — never for what the exam looks like today, because the stimulus, the timing and one whole component are different.

13 of these 31 pre-2023 rows are corroborated by two or more independent sources; the rest, mostly the earliest years, rest on a single compiler. None of them changes anything about the current paper — they sit here so the record is not quietly deleted, not so a Sec 4 student prepares for a photograph that will not appear.

What has the current paper actually shown?

Below is the corroborated part of the current-format record — clips where two or more independent sources agreed on the theme. Wording is as recalled, so read it for the substance of the question rather than the exact phrasing.

• 2025A boy using a zip-line toy to launch a dice into a LEGO box, with his fathertwo or more sources

  “Do you think the activity shown in the video is meaningful? Why or why not?”

• 2025Young children in a guided gymnastics sessiontwo or more sources

  “What are some considerations when working with young children? Why do you say so?”

• 2025A rainy day, heavy traffic, a crowded bus stoptwo or more sources

  “How do you think people's plans may be affected by the weather?”

• 2025Cleaning robots operating at Changi Airporttwo or more sources

  “Do you think machines like the one shown in the video are useful?”

• 2025Two teenagers doing pottery togethertwo or more sources

  “Who will be attracted to this activity? Why do you say so?”

• 2024Teenagers skateboarding at an outdoor skate parktwo or more sources

  “Would you recommend such a recreational activity for young people? Why or why not?”

• 2024Jewel Changi Airport's Rain Vortex / nature-themed spacestwo or more sources

  “Do you think there should be more nature-themed spots as shown in the video? Why or why not?”

• 2024Children's Corner of a librarytwo or more sources

  “Do you think this particular section of the library appeals to visitors?”

• 2024A rubbish truck collecting wastetwo or more sources

  “Is this a good way to manage and collect waste? Why or why not?”

• 2024A street hawker preparing fruit juice with a manual presstwo or more sources

  “Do you think there should be such street hawkers in Singapore? Why or why not?”

And the discussion that sometimes follows

• 2024Recreational space design (follow-up discussion)single source

  “How would you design an ideal recreational area? / Do you agree that people these days do not have enough time for recreational activities?”

• 2024Youth and nature (follow-up discussion)single source

  “What can be done to get teens to spend more time in nature? / Have teens done enough to protect nature?”

• 2024Reading habits and school libraries (follow-up discussion)single source

  “What suggestions do you have to improve your school library? / In today's world, people tend to read less as they grow older - your thoughts?”

• 2024Waste reduction (follow-up discussion)single source

  “What actions can individuals take to minimise waste? / Is it important for people to prioritise waste reduction?”

• 2024Hawker culture / eating out (follow-up discussion)single source

  “Do you agree that more people in Singapore are eating out these days? / Are there increasingly fewer youths willing to work as hawkers?”

The format page sets out how Planned Response and Spoken Interaction are marked, and why there is no longer a passage to read aloud.

Where did these recollections come from?

10 independent sources across both eras: post-exam recall published by tuition centres and compilers in the days and weeks after each July sitting. A theme is marked corroborated only when two sources that are genuinely independent of each other agree on it.

Two pages from the same publisher count once. One page quoting another counts once. Where a list circulates with no stated source at all, we have not reproduced it and have not counted it.

One fabricated item was caught and excluded while this corpus was being built: a web-search summary described a “2019 Day 1 skateboarding” clip that does not exist in any source page — it had conflated a real 2019 row with the genuine 2024 skateboarding clip. It was caught by checking the summary against the actual source pages rather than trusting the summary, and it is recorded here so it is never reintroduced. A search summary is not a source; the page it summarises is.

This corpus turned out stronger than the Chinese-language one it sits alongside, which is the opposite of what was expected going in. The reason looks commercial rather than pedagogical: several tuition operators run detailed, day-by-day “what came up” posts as content marketing every July, within days of the sitting, which is exactly the kind of fast, checkable publication that corroboration needs.