Research and Thought Leadership Director
Dr Nick Saville has worked for Cambridge Assessment since 1987, becoming Director of the newly formed Research and Thought Leadership Division in 2015. Prior to moving to Cambridge he worked at the University of Cagliari (Facoltà di Magistero) teaching English, and managed a test development project for Cambridge English in Tokyo based at the British Council. He holds a PhD in Language Assessment from the University of Bedfordshire, a degree in Linguistics and an MA TEFL (both from the University of Reading), and an MA Cantab.
The origins of Linguaskill go back to the mid-1990s when we began working on the idea of online, automated, modular tests. The aim was to democratise testing, making it quicker and more efficient. Since then Linguaskill has developed to provide a generic, multilevel test that serves the needs of all audiences, from education to business.
Linguaskill is used to measure what has been learned, as opposed to score-driven alternatives, which can measure how well you have practised for the test. This means it is much more consistent and mission specific across the four skills of listening, reading, writing and speaking.
That ease of use is key. Many of the features within it, such as computer assisted testing, automated writing and automated speaking testing are not necessarily new to us at Cambridge Assessment English, but we’ve learned how to deploy them in ways that are simple to administer, and can scale to hundreds of thousands of people, reliably and seamlessly.
It is clearly early days for the product – we have successfully launched it, and we have created a stable and reliable platform. The holy grail of language testing is being able to personalise tests for each client group, so that they measure what matters to them. Through Linguaskill’s flexibility we are able to move towards this personalisation, so we now have to work with our partners and customers to develop and modify it to meet individual needs.
AI will help close the gap between the actual situations where we use language and the context of taking a test. Test conditions are not real life, so the ideal for language testing is being able to observe people doing the real thing, creating much more specific tests, personalised to them. AI can assist here, particularly when coupled with the mobile devices we all carry around with us. First-generation testing technology just migrated tests from pen and paper – AI lets you go much further and deliver more adaptive, personalised and efficient testing.
I see two key trends. Firstly, merging language testing into learning, so that you can see how people perform in a real-world context will give a greater richness, delivered in a more effective way.
Secondly, use of AI means we’re going to see a virtuous combination of teachers and machines. We will always need teachers to provide human guidance and support, but we can use the robotic strengths of machines to help them deliver more effective learning. For example, if you have a class of 40 people, a teacher clearly can’t mark 40 writing tasks every day. A computer can. The teacher can then look at the analytics and enhance content and support either for the whole class or at an individual level, improving learning outcomes.