Learner corpora and writing skills
Helping learners to build their vocabulary is one of the two main goals of our new edition. The other is to give students the tools they need to become better, more confident writers. All our research into dictionary-users’ needs (including our user survey) convinced us that dictionaries should do more to help students acquire writing skills, and this is an area we have really focussed on in the new Edition.
It’s clear that there is a big demand; anyone using English in the workplace or in their studies must be able to write a report, an essay, or a dissertation in good, natural-sounding language. Becoming a good writer is a complex business, but it’s helpful to think of this in terms of two key objectives: accuracy and fluency.
Producing text which is accurate means avoiding mistakes in areas such as grammar, spelling, and the use of articles. But there is more to good writing than just being ‘correct’. It’s quite possible to write text which is free of errors, yet at the same time dull, repetitive, and lacking in naturalness. Achieving fluency is a more difficult task; a fluent writer is one who has a good command of all those aspects of language which contribute to interesting and natural-sounding text, including:
Above all, fluency implies an ability to select the most appropriate ways of expressing ideas and performing common writing functions such as introducing new topics, expressing opinions, and drawing conclusions.
The big challenge for us at Macmillan was to figure out how a dictionary could help its users achieve the twin goals of accuracy and fluency. We decided that the best way to start was by comparing texts written by learners (in essays, dissertations and reports) with similar texts produced by expert native-speaker writers. The idea was that this would help us to discover those areas of the language that caused most difficulties for learners and created obstacles to their success as writers. To do this, we began a collaboration with the Centre for English Corpus Linguistics (CECL), which is based in the Université catholique de Louvain, in Belgium. CECL is the world-leading centre in the study of learner corpora. Unlike a ‘standard’ corpus of the kind dictionary-makers have used for many years, a learner corpus consists of text produced by people learning a language. The researchers at CECL have a large corpus of learners’ writing, and an enormous amount of expertise in analysing this kind of data, so we were very fortunate to work with them.
In a two-year collaboration with CECL, we compared native and learner writing, and in this way developed a better understanding of learners’ problems. As a result, we have added three major features to the new edition of MED, namely:

