Tricks of memory

A long time ago, while working what was called a stage (see below) at the European Commission, I got interested in natural language processing, an interest I have been pursuing on and off ever since. An interest prompted in part by the importance of translation in the work of the Commission and their interest in the possibilities of machine support if not machine translation. Maybe they have now cut through all that and just made English the working language, a fact on the ground which will survive Brexit, rather as it did on the Indian subcontinent.

Now, at about that time, I read a small book on that subject, a book which was about half an inch thick and bound in blue hard covers. The words on the covers were printed in white in one of the machine readable typefaces, which being new and exciting, were often used for such purposes at the time. I think the book was called 'Natural Language Processing' and included the idea that a lot of natural language could usefully be described by things called frames. A simple frame might say that it was sensible to talk about large animals sleeping on something for some period of time. So if your computer was processing a chunk of natural language which seemed to fit into this or that frame it was likely that it was on the right track. A notion which combined grammar with semantics.

I only read about half of this book, despite being impressed by what I read, and subsequently kept the book for many years, fully intending to finish reading it. Eventually it succumbed to one of our regular culls.

From time to time the book comes to mind and I try to recover the name of the author, sometimes successfully.

This was the subject to which I fell asleep last night, without making any progress at all. But when I woke up, the name 'Yorick Wilkes' popped into mind. Half waking, I was not at all sure that I was not getting muddled up with Hamlet. Maybe I ought to check with Google. And if this chap is not famous enough to have made it into the world of Google, maybe he had made it into the world of Abebooks.

When I eventually get up, I ask Bing and he turns up a Wikipedia entry for one Yorick Wilks, quite clearly the right person, less the 'e' from his surname, which I imagine I picked up from the John Wilkes about whom I learned at school. Also 'The North Briton'.

But the name of the book that I remember seems to be quite wrong and the best that Wikipedia can do is 'Grammar, Meaning and Machine Analysis of Language', published in 1972, which date is about right and with the whole being confirmed by both Amazon and Abebooks. Sadly, I fail to recover any images of blue books with white machine readable letters on their covers. On the other hand, I did associate to another book from the same era, indeed, from the same year, 'Understanding Natural Language' by Terry Winograd, a book which I think involved a machine language for telling a robot about moving children's building bricks about. Brightly coloured bricks as I recall.

Perhaps the result of all this will be that Mr. Wilks gets lodged in a more accessible part of my memory - even though I suspect that both his and my take on these matters has been overwhelmed by the march of computational statistics and neural networks. Deep thought out, statistical sledge hammers in.

Reference 1: Le Petit Littré: stage: résidence que chaque nouveau chanoine doit faire dans son église pendent six mois, pour jouir des honneurs et des revenues attachés à la prébende. Etc.

Reference 2: https://www.abebooks.co.uk/. An outfit whose range of books overlaps with that of Amazon and which are quite often significantly cheaper.

Reference 3: https://en.wikipedia.org/wiki/Yorick_Wilks.