Explaining the Mysterious O-Level Oral Exam Discrepancy



The Blue Cats’ website:
The Blue Cats’ Instagram:

Singapore’s O-Level English oral examination has followed a revised format since 2023, replacing the Reading Aloud component with a Planned Response. Candidates spend 10 minutes at a silent preparation station watching a video clip, reading a question on screen and making notes. They then enter another room where one of two examiners reads the same question aloud, and they have up to two minutes to deliver their response, followed by a Spoken Interaction segment on a broader related topic.

On the morning of 15 July 2026, around 5,000 candidates sat the examination. The video featured street barbers at work, and the Ministry of Education stated the Planned Response question was: “Who do you think would go to a place like this for a haircut? Why do you say so?” Some candidates believed the question on their screens had been phrased personally instead, recalling wording such as “Would you go”, “Would you visit”, “Would you be interested” or “Would you want”. The issue surfaced first on Reddit, and several candidates reported it to their schools or the Singapore Examinations and Assessment Board.

SEAB said on 18 July that it found no discrepancy, noting that only one version of the video and question was released to all centres through its e-Examination system, with checks carried out before each session. On 5 August 2026, Senior Minister of State for Education Janil Puthucheary told Parliament that 173 candidates from 76 schools reported a discrepancy on examination day, rising to 432 candidates from 102 schools, examined by 200 different examiner teams. Reports were not concentrated in any one school, centre or time slot, and candidates examined by the same team recalled different versions. Cybersecurity checks found no hacking or alteration of materials, and a GovTech audit reached the same conclusion. Some recovered planning sheets showed candidates had initially misread the question, with one crossing out the wrong version and writing the correct one. Responding to Workers’ Party MP Jamus Lim, Dr Puthucheary rejected the description of a “collective hallucination”, saying candidates had reported a wide variety of different phrases rather than one identical alternative.

Preliminary results showed the mean oral score for the 15 July session was comparable to the other four sessions, and the 432 candidates who reported discrepancies scored comparably to their peers that day. Examiners use positive marking, assessing vocabulary, sentence structure, grammar, delivery and organisation of ideas rather than adherence to a model answer. Schools are helping distressed candidates apply for special consideration, assessed case by case, and SEAB will review its oral examination processes. Explanations raised include the Mandela Effect, though that typically involves many people recalling the same incorrect detail; mass psychogenic illness, which involves shared physical symptoms not present here; and memory conformity, where recollections shift after discussion with others, currently the strongest theory though unproven.

source

23 Comments

  1. Much ado about nothing. Only in rigid Singapore would this be an issue.
    It's a test of oral efficiency. Not comprehension. The official response that a 'correct' answer is not required but the fluency of the response was counted should have ended all complaints of unfairness.

  2. In 2024, I remembered many students brought up about the EL LC for O levels. I wasn't affected by it, or at least I know that my school wasn't really affected by it, but I remembered there were some schools where the students mentioned that the audio were very soft or like too fast paced that you can't really hear/understand it and even some said the audio were buggy to catch anything, so some schools replayed certain parts including the last part where it's only one time.

  3. my daughter also got the wrong qn but she was not one of the 432 who reported the error as she thought she saw the qn wrongly. The oral exam process was – Student will be shown the qn on a screen. On the screen, there was also the student name and student number. The student will also be given a piece of paper to write down and organize their answers during the prep time. Then the student will go into another room for the actual oral exam and also had to submit the paper which they used to write down their thoughts. My daughter got a different qn but did not raise the alarm. Seems like the school checked through all the collected papers and identified all those who prepped based on "Would you go to…". They invited these students for a consultation whereby they were shown screengrabs of the qn "Who do you think would go…" with the student's names on it, presumably as a "proof" that the qn they were shown was "Would do you think…". With that, case close: there was no error, school/SEAB got proof that the qn was correct. Well, seems like only 2 students in her class were affected. I believe my daughter. She is not the blur type who can read such simple question wrongly. She is a student leader with high ability and integrity. But what can we do about this? If anything, it just fueled her resolve to do well so she can be in the position to make a difference in the future.

  4. No matter 'would you' 'who would', or whatever the question type is, it still surrounds the street barber and the reason behind is pretty similar. A person who is effectively fluent in the language would be able to start a conversation using the content he/she brainstormed prior. Before this change, students do not even have access to the conversation questions and was just given a picture to do a very general brain storming.

  5. Kids need this.. if the general topic is still on the barber, asking thier personal opinion would have been a simpler question. It is not like showing a video of the barber and then ask, who do you think will buy chicken rice from the store..

  6. im amused our politicians would use “mass hallucination” to describe this issue. it’s really laughable that they picked this term from the reddit threads.

    that said, i do think it’s a berenstain effect since the main topic of the question is consistent throughout all recounts, with the inconsistency being the subject in the question. and with increasing number of students coming out to share their experience, it could also lead to the creation of false memories for others. there has been a few psychological studies showing that our memory is not as dependable as we think, unfortunately

  7. The students remembered wrongly. Unless ALL students within the same session, same school or time slot saw the "wrong" question, then it's the student who remembered wrongly. Simple logic. Who would want to waste time screwing with O-level exam question?

  8. It's called mass wayang. No way so many 16-year-olds experienced this "phenomenon". It's a technical glitch that no one wants to admit… hence, the "powers that be" just claim it is everyone's fault but their own.

  9. 我在想是不是曼德拉效应。 我跟我老公bto的时候也发生过。 我们都记得我们的祖屋有20几层, 我们买了15层, 还想着算中高层, 挺不错的。 结果收屋的时候, 最高是18层。 我们整个傻眼。 我们都清楚的记得在HDB选屋子的时候, 一共20几层。

Leave a Reply

Your email address will not be published. Required fields are marked *

You might like

© 2026 Cantinho do Vídeo - WordPress Video Theme by WPEnjoy