John Sinclair, who has died of cancer aged 73, was a giant in English language studies and a world leader in three fields, two of which he helped create, as well as a contributor to many others. Birmingham University's professor of modern English language (1965-2000), he was a pioneer in the field of discourse analysis, a superb lexicographer, and an outstanding corpus linguist, whose books are compulsory reading for any serious language student.
John was Scottish and proud of it - I remember him piping in the haggis on Burns Night in full kilted regalia. He went to George Heriot's school in Edinburgh, and took a first in English language and literature at Edinburgh University.
After RAF national service he returned to the city as a research student in 1958 and was quickly appointed to a lecturership in English language and general linguistics.
Then, at just 31 and with barely a publication to his name, he was appointed to the foundation chair at Birmingham. This turned out to be the making of the university's reputation as a world centre for English language studies. A paper in a volume in memory of JR Firth in 1966 showed the direction his work would take. Strikingly up to date, it defines the collocation - the tendency for words to associate with each other in partly arbitrary ways - now deemed central in descriptive linguistics but counter to most of the then current theories of language.
He ran one of the first research projects in computational linguistics, and its continued relevance is indicated by the fact that the 1970 report on that project was republished three years ago. At the same time he directed a research team producing innovative task-based language development materials for the children of West Indian immigrants.
In 1972, he published a competent but unoriginal account of English grammar. It was a commissioned work, but he never again allowed his research agenda to be set for him.
In 1975 he published, with Malcolm Coulthard, one of his greatest works, Towards the Analysis of Discourse, the first truly systematic account of a genre of spoken discourse. He and Coulthard show that the hierarchy of structures that operate in grammar can be applied effectively to exchanges in the classroom.
In the late 1970s John became chief adviser to Collins on its English Dictionary, and on the strength of this, and his work on collocation, he proposed a new kind of dictionary for advanced learners of English, written from scratch on computers and involving the creation of the largest corpus of English language texts in the world. He collected a team of gifted linguists and lexicographers, and the result was the Collins Cobuild English Language Dictionary (1987).
This dictionary was the first to show how the collocations of a word affect its meaning, the first to use authentic and representative examples drawn from a corpus, and the first to use a natural defining language. It was also the first to take seriously the possibility that a dictionary might have an important part to play in improving students' writing and speaking as well as helping with their reading and listening. Today all major advanced learners' dictionaries have engaged with these features.
In 1991, John published his second masterwork, Corpus, Concordance, Collocation, the bible for a generation of corpus linguists. In it he explains in deceptively simple English how collocations can be identified and what their implications are for a proper description of English. Here too he proposed his revolutionary "idiom principle". According to this, when we speak or write, we make use of semi-preconstructed phrases, and these account for fluency in speech, with grammar being used as a fall-back resource when we run out of suitable semi-preconstructed phrases.
In 1995, John took partial retirement from the university, but there was a continued flow of publications. In Reading Concordances (2003), he gave practical advice in corpus linguistics, but it was Trust the Text (2004) his third masterwork, that took this area of research forward. In this book he proposed a new way of describing lexical items, identifying their grammatical and semantic patterns and their "semantic prosodies" (crudely, their communicative effect on the reader or listener).
Linear Unit Grammar (2006), co-authored with Anna Mauranen, revisits the idiom principle and absorbs it into an integrated theory of grammar. It is likely that this will posthumously be recognised as a further major contribution to our understanding of English.
Helped by his second wife, Elena Tognini Bonelli, he founded the Tuscan Word Centre. Located in Certosa di Pontignano, it became famous for corpus-based language research and ran courses for budding corpus linguists, thereby training a generation of researchers.
John was an honorary life member of the Linguistics Association of Great Britain, a member of the Academia Europaea, the holder of an honorary doctorate from Gothenburg University, and honorary professorships at Glasgow, Shanghai and Jiao Tong Shanghai, universities. John could be robust with equals but was unfailingly generous with his time and kind to anyone who sought his help. There are linguists all around the world who readily acknowledge their deep debt to him in this respect.
He leaves Elena, their two children, and three from his first marriage.
· John McHardy Sinclair, academic and lexicographer, born June 14 1933; died March 13 2007