Veja sites relacionados
Cambridge English Language Assessment
Porquê Cambridge English?
You are here:
Martin Robinson joined Cambridge Assessment in 2002 and is responsible for the development and production of its automated, adaptive assessments, including Linguaskill and BULATS. In addition, Martin manages the Cambridge English Benchmarking service and through his work in these areas he has led national projects for ministries of education in countries including Malaysia, Chile and Malta, as well as managing language test production for the European Commission’s European Survey on Language Competences. Before joining Cambridge English he gained extensive experience in English language teaching and school management in Spain and Japan. He holds an MA in Applied Linguistics, specialising in language assessment, from the University of Reading.
Describe your role and involvement in Linguaskill.
As product owner, I’m responsible for taking customer requirements and developing an end-to-end solution that satisfies those needs. This includes developing the tests, the delivery platform, the automated marking capability, right through to delivering the results to the customer.
What were the needs you were looking to solve with Linguaskill?
Our customers would like assessment that is under their control, is totally on demand and can be taken 24/7/365, and where results are delivered almost instantly.
What makes Linguaskill different from other tests?
The adaptive element is very different from traditional testing. Some organisations provide adaptive testing but the Cambridge English technology is more advanced.
We not only provide one test that covers all levels but the algorithm also ensures coverage of the entire test construct. This means we are testing all the different sub- skills that we need to test, and that different candidates and even the same candidate taking the test more than once receive different tests. We also automark extended writing, such as 200-word pieces, and provide results in terms of the Common European Framework of Reference for Languages (CEFR).
Now that Linguaskill is in the market, what are you most satisfied with in terms of the product and market adoption?
The fact that our customers can buy tests and arrange all their own test sessions, and access reports through one platform. In traditional testing this all takes many months and huge numbers of people across the world. So, when customers say that they can do all this themselves and it’s all really easy, that’s great feedback. We’ve also had lots of comments from students telling us that this is the future of testing. They want to know why all their tests aren’t like this! That’s fantastic to hear.
How do you see Linguaskill developing over the next 2–3 years?
We’ll automate even more so that it’s even easier and quicker for customers, and we’ll provide them with more choice. We’ll also be delivering more feedback to learners and teachers that will help them learn English better.
How do you see computer-based testing changing in the future, particularly with the use of AI?
This is going to be very exciting. We’re going to be doing research into AI-based dialogue systems so that in a few years we might be able to start having learners converse with a computer rather than just answer questions. This is very early days in the concept, and will require a lot of research, but AI does make it potentially possible.
Are there other key trends that you see impacting language learning and testing over the next five years?
Yes. People often talk about ‘localising’ testing, meaning that tests are specially created for a particular country or other group. This is a good first step, but I think we can look to go beyond that. I see personalised learning and testing as the future, so every individual, whether they’re in a classroom or at home, receives a test or a piece of learning that is especially created solely for them as part of a personalised learning path that takes them to where they need to get to. This is all possible with adaptive technologies, machine learning and data (lots and lots of data!).