Back to Articles
Advanced English and Exams

AEE Adaptive Placement Test

Vladimir Pavlovich 225 views

Most students dread tests: by their very nature, they are often depressing and artificial, leaning towards a black-and-white approach. Even worse, the vast majority of standardised examinations do not aim to reward what a candidate can do, but penalise each and every mistake. The Russian community knows this very well; the notorious Unified State Exam is a case in point.

Cambridge is different. Designed around the CEFR, its examinations focus on can-do statements rather than cannot-do statements, thereby providing meaningful and universally recognised results. IELTS is not entirely different in this respect, although the approaches of the two systems are fundamentally distinct.

Cambridge examinations assess communicative competence within a specific level. This is why they are built around the skills expected at B2, C1 or C2. If a candidate does not demonstrate sufficient competence at the target level, the level below may still be awarded; if the examination proves too easy, the candidate may be placed at the level above. One target level and two possible alternatives.

IELTS, by contrast, is a predictive test. Its purpose is different, since it aims to determine whether a candidate is likely to cope in an academic or professional environment. The “level” itself is secondary.

For years, here at AEE, we followed the principles of communicative competence testing. This resulted in separate tests for different levels. They were shortened adaptations of Cambridge examinations: instead of eight questions in an open cloze, for example, we might give only four.

And this worked; I cannot say that it did not. However, the testing process was lengthy, and more often than not we had to send a candidate another test when the original results did not appear particularly valid, especially if the test had proved far too easy or far too difficult.

This led us to begin work on a new and rather massive project: an adaptive entry test.

The idea was simple in theory: depending on the answers given, the level of the questions should change automatically. However, once I began working on it, I realised that it would require months of work, both in methodology and coding.

Now, after weeks of trials and validation, the test comprises between 25 and 30 questions. The first five are the same for every candidate, but then the magic begins.

There is a database of approximately 500 questions covering grammar (structure, modality and aspect), vocabulary (collocation, idioms, phraseology and precision), organisation (linking and discourse patterns), and practical English usage (register, appropriacy and tone). All of these reflect CEFR expectations at different levels.

If a candidate answers the questions successfully, they become progressively more challenging. If mistakes are made, the test adapts to the candidate’s needs. This creates a more holistic testing environment: candidates feel relatively comfortable while taking the test because it remains challenging without becoming insurmountable.

The teacher then sees the result — the predictive result. It is absolutely essential to emphasise this.

The test is methodologically designed to establish the “floor”: the lowest level at which the candidate can function consistently. This is deliberate. The vast majority of our students are English teachers, and we need to ensure that we do not place them in a group whose demands they are not yet ready to meet.

For a similar reason, the test also establishes a “ceiling”. If questions from the level above are consistently answered incorrectly, this indicates that, despite the candidate’s ambitions, it would be far more prudent to complete the lower level first.

The new entry test will therefore become a central part of our placement process. It will not replace the teacher’s judgement but provide a more reliable starting point for it. By combining adaptive testing with a follow-up interview, we can form a fuller picture of each candidate’s current ability, identify the most appropriate course and reduce the risk of placing someone at a level that is either insufficiently challenging or unnecessarily demanding.

Ultimately, the aim is not simply to produce a score. It is to make placement more accurate, more consistent and more closely aligned with the way our courses are structured.

References

  • Bachman, L. F., & Palmer, A. S. (2010). Language assessment in practice: Developing language assessments and justifying their use in the real world. Oxford University Press.
  • Cambridge English. (n.d.-a). Results for Cambridge English Qualifications. Cambridge University Press & Assessment.
  • Cambridge English. (n.d.-b). The methodology behind the Cambridge English Scale. Cambridge University Press & Assessment.
  • Council of Europe. (2020). Common European Framework of Reference for Languages: Learning, teaching, assessment—Companion volume. Council of Europe Publishing.
  • Jones, N., & Saville, N. (2009). European language policy: Assessment, learning, and the CEFR. Annual Review of Applied Linguistics, 29, 51–63. https://doi.org/10.1017/S0267190509090059
  • Kerstjens, M., & Nery, C. (2000). Predictive validity in the IELTS test: A study of the relationship between IELTS scores and students’ subsequent academic performance. IELTS Research Reports, 3, 85–108.
  • Wudthayagorn, J., & Nipaspong, P. (2025). Enhancing predictive validity in the English placement test: Evidence and insights from a Thai public university. Language Testing in Asia, 15, Article 69. https://doi.org/10.1186/s40468-025-00408-2