Nearly 5,000 candidates sat for an O-Level English oral examination on 15 July 2026.
In the days that followed, 432 of them reported that the question displayed during their preparation time appeared different from the question later read aloud by their examiner.
However, the Singapore Examinations and Assessment Board, or SEAB, found that candidates and examiners had accessed the same examination material.
Senior Minister of State for Education Janil Puthucheary told Parliament on 5 August that there was no evidence of a system malfunction, cybersecurity incident or deliberate act.
GovTech also audited the technical platform and processes and reached the same conclusion.
How the English Oral Examination Now Works
The English oral examination may be different from what some older Singaporeans remember.
Previously, candidates had to read a passage aloud. That component was replaced by a Planned Response component in 2023.
Candidates are first seated at a silent preparation station, where they watch a video and read a question displayed on a computer screen.
They are given 10 minutes to plan their answer.
They are then brought before two examiners. One examiner reads the same question aloud, after which the candidate has up to two minutes to deliver the prepared response.
This is followed by a Spoken Interaction component, during which the examiners discuss a topic broadly related to the video with the candidate.
The Planned Response and Spoken Interaction components are each worth 15 marks.
Candidates Said They Prepared for a Different Question
The video shown during the 15 July session featured street barbers at work.
The official question was: “Who do you think would go to a place like this for a haircut? Why do you say so?”
However, some candidates later said that they remembered seeing different versions during their preparation time.
The variations reportedly included questions beginning with phrases such as “Would you go”, “Would you visit”, “Would you be interested” and “Would you cut your hair”.
These were not merely different ways of remembering a few words.
A question asking whether the candidate would personally visit the barber could lead to a different response from one asking what type of people would visit the barber.
Complaints began appearing on social media, including Reddit, after the examination.
According to The Straits Times, 173 candidates reported a discrepancy on the day of the examination. The number subsequently increased to 432.
SEAB Checked What Appeared on Candidates’ Screens
Dr Janil told Parliament that SEAB began investigating on the day of the examination.
It conducted system-level checks and logged into individual candidates’ accounts to confirm the video and question accessed by each account.
SEAB also examined digital records showing how the examination materials were created and delivered.
This included logs recording what appeared on candidates’ screens during the 10-minute preparation period.
Cybersecurity checks found no indication that the system had been compromised.
According to Dr Janil, there was only one approved and encrypted source of examination material, which was accessed by both candidates and examiners.
SEAB therefore concluded that all candidates and examiners had been provided with the same video and question.
Reports Came From 200 Different Examiner Teams
The authorities also examined whether the complaints shared a common pattern.
The 432 candidates came from 200 different oral examiner teams.
There was no consistent candidate order involved. For example, complaints might have come from the second and eighth candidates under one team, but from the third and tenth candidates under another team.
Candidates examined at the same centre by the same team also remembered different versions of the supposed alternative question.
SEAB considered the wide range of phrases reported by candidates to be an important part of its findings.
It concluded that the reported incidents were not technical in nature and were not caused by a deliberate or malicious act.
Some Planning Sheets Indicated That Candidates Had Misread the Question
Workers’ Party MP Kenneth Tiong asked whether SEAB had examined the planning sheets used by candidates during the preparation period.
Dr Janil said many sheets could not be linked to specific candidates because the candidates had not written their names on them.
However, some sheets appeared to show that candidates had initially misread the question.
In one example, a candidate had written down what they initially thought the question was, crossed it out, and then wrote the correct question before continuing with the plan.
This does not mean that every candidate who reported a discrepancy was found to have misread the question. It was one piece of evidence considered alongside the digital records and the different versions remembered by candidates.
Jamus Lim Asked Whether It Was a “Collective Hallucination”
During the Parliament exchange, Workers’ Party MP Jamus Lim asked whether SEAB was suggesting that the incident was a case of “collective hallucination”.
He also asked whether the investigation had placed more weight on computer evidence than the accounts of hundreds of candidates.
Dr Janil rejected that characterisation.
He said both the candidates’ accounts and the technical findings had been considered.
However, the candidates had not collectively reported seeing one identical alternative question. Instead, they remembered a significant variety of phrases.
Dr Janil asked Assoc Prof Lim not to put words in his mouth about “collective hallucinations” or imply that one type of evidence had been prioritised over another.
Candidates Who Reported the Issue Had Comparable Scores
Preliminary results showed that the average score for the 15 July oral examination was comparable to the average scores for the other four oral sessions held in 2026.
The average score of the 432 candidates who reported discrepancies was also comparable to that of other candidates who sat for the examination on 15 July.
Dr Janil said this meant that, as a group, those who reported discrepancies performed as well as candidates who did not.
He explained that oral examiners use “positive marking”.
Candidates are assessed on their oral communication skills, including their vocabulary, grammar, sentence structures and ability to organise and express their ideas.
They are not marked solely on whether their response follows one exact template.
Some Candidates Can Apply for Special Consideration
MOE acknowledged that some candidates may have experienced significant distress, which could have affected their ability to respond properly.
Schools are working with these candidates to submit applications for special consideration.
Special consideration is an evidence-based process used to review the final grade of a candidate whose examination performance may have been affected by circumstances beyond their control.
It does not mean that every candidate will automatically receive additional marks. Each application will be assessed on its own merits.
Some applications had already been submitted as of the Parliament sitting on 5 August.
SEAB Will Review the Oral Examination Process
SEAB will review its oral examination processes following the incident.
This includes looking at how the risk of candidates misreading a question can be reduced and how candidates can be better prepared to read the prompt carefully during the 10-minute preparation period.
For now, the authorities’ position is that every candidate and examiner accessed the same material, and that no technical or cybersecurity fault was found.
However, candidates who were significantly affected can still seek special consideration through their schools.
