Head of Automated Assessment and Learning
Dr Ardeshir Geranpayeh is Head of Automated Assessment and Learning at Cambridge English, which he joined in 2001. Since 2009, he has been leading on the development and implementation of the automated assessment of writing and speaking. He has 29 years’ experience of test validation and has contributed to the design, development, validation, revision and evaluation of almost all Cambridge English language proficiency tests. Additionally, he is a regular presenter at international conferences where he has led workshops on statistics, computer-based testing, listening tests and test security. He holds a PhD in Applied Linguistics from the University of Edinburgh on the comparability of language proficiency testing.
I was involved at the initial conceptual design stage, dating back to 2002. More recently Linguaskill was re-invented, putting automatic assessment at the heart of the product.
I was involved at the design stage for the technical model, creating the Computer Adaptive Test (CAT) for Reading and Listening. I was also responsible for preparing the automated assessment engine for marking Linguaskill’s Writing and Speaking components.
A valid, reliable, quick and cost-effective test of proficiency that could be delivered online to candidates with immediate reporting of results, all at an affordable price.
There are two main features. Firstly, its adaptiveness, which means that each candidate receives a different set of test items, personalised to their true ability level. Secondly, using AI technology to deliver immediate marking of the Writing and Speaking tests.
It is important to keep the customer at the heart of planning any new test. We learned how to use sophisticated technology to improve candidates’ experience of taking the test, while ensuring that results are reported to stakeholders in a fast and efficient way. Everyone in the team has been working towards that goal of meeting customer needs and making the product simple to use.
The ease with which candidates can sit the test, and the simplicity to run the test locally are the most satisfying features we managed to achieve in Linguaskill. We’ve had lots of positive feedback around these areas, which is extremely rewarding for us.
The Linguaskill family tests will grow rapidly in terms of its use by multiple stakeholders and will become, I hope, the default benchmarking proficiency test of choice for various test purposes.
I’m continuing to work on the improvement of the automarker for writing and speaking, introducing new diagnostic features to provide efficient feedback to candidates to improve their English.
I believe AI will become part and parcel of any computer-based test in the future. We are happy that we are leading in this area.
It is very clear that AI will have a big impact on language learning and testing moving forward. The Cambridge Learning Oriented Assessment (LOA) approach, which has AI at its heart, will significantly change language learning and testing by bringing these two areas closer, helping drive more effective, individual learning.