Sorry I’ve been offline again for a long time. Promise to keep on track now I’m back from la France profonde.

Let’s start back with a spot of fun – at least for English-speakers.

Don’t expect a superior lexeme counter for your translations if you go to WordCount . It’s actually an ‘artistic experiment’ carried out by Jonathan Harris, a designer with Number27. He presents the “86,800 most frequently used English words (from proper nouns to prepositions), ranked in order of commonality” along a line on diminishing font size from left to right. What does he mean by commonality here? Dictionaries suggest it means “The possession, along with another or others, of a certain attribute or set of attributes: a political movement’s commonality of purpose, or A shared feature or attribute”. For WordCount this presumably means that the words share the feature of being “scaled to reflect its frequency relative to the words that precede and follow it”.

What you can do with this artistic toy is either find out how “frequent” a word is (e.g. the last word in the whole list is oddly, conquistador, ahead of items like conflas and fwag), or enter any digit up to 86,800 and see which word has that position in the rankings. Harris takes the data from the 100 million word British National Corpus that covers a very wide range of spoken and written sources. He plans to use WordCount to track “word usage within any desired text, website, and eventually the entire Internet.”

But what people really like about WordCount, apparently, is the fact that the arbitrary word line of relative frequencies sometimes generates surreal phrases: e.g. around translate you get “pussy patting translates geomorphology impasse”. Quite. And around globalization, you find “BASF Hokkaido repugnance globalization Sunderby”. Odd stuff. In other words, we keep on seeing meaning where the machine simply lines up arbitrary forms.

Most people these days know how to smile at the ‘five year’s away’ promises from the IT sector, especially in the language technology industry. In Europe (and no doubt elsewhere), the language and speech technology community has started building various roadmaps to add a bit more rigor to specifying developmental milestones in such areas as speech recognition and MT. It will be interesting to see whether, like certain other roadmaps, progress follows the vision. Recently, one of the bodies that advises the EU on future R&D programs (the European Information Society Technologies Advisory Group (ISTAG)) published a draft report on the Grand Challenges in the Evolution of the Information Society which tries to crystallize possible futures (circa 2010) into a number of easy-to-picture challenges and solutions.

One of them is devoted to the “serious linguistic challenges” of a very large Union, and proposes “The Multilingual Companion” – “a small portable device making it easy for each person (i.e. in an encounter) to understand the others.” Here’s the vision:

A device of this sort…would rapidly and accurately render translations in text and speech in any other European language desired. Hence, it could be used to take voice dictation, immediately generating the text in any number of languages, thereby streaming the connection between humans and computers. The companion would be extremely useful for disseminating the minutes of meetings to many countries simultaneously, to each in its own language.

The idea behind this challenge is, of course, to focus research in all areas of language technology (speech, learning, cognitive clues for meanings, summarizing, translation, cross-lingual search, etc) towards a concrete practical goal rather than towards a scatter of purely short-term R&D targets, as has often been the case in the past. The model here is clearly the successful VERBMOBIL project run by the German government to produce a mobile real-time translation device for a limited language set.

As the report points out, European IT giants like Philips and France Telecom have been reducing their R&D effort in these areas in recent years, while IBM and Microsoft have boosted research in ‘natural language’ technology. ISTAG’s hope is that the new generation of language technology companies, that in several cases have been spun off from larger organizations, or which emerged from earlier EU language tech projects, will now carry the work forward in Europe to produce the sort of a polyglot wristwatch that will solve our face to face communication problems.

Eurolang reports that the Irish Government is to “initiate a process of discussions with the other EU Member States and the EU Commission with a view to seeking official and working language status for the Irish language in the EU under EEC Regulation 1/1958.”

Pádraig Ó Laighin, who led the Stádas campaign for full working status told Eurolang:

“More people will speak and use Irish in the future as a result of this decision, and the whole raison d’être for learning the language in school, as a first or second language, has been transformed. No longer will Irish speakers be second-class citizens in their own country. There are no guarantees for the future of any language such as Irish, which is under threat, but this decision restores hope and strengthens self-esteem”.

I’m blogging from Bonn, John Le Carre’s famous small town in Germany for a certain generation, where I’m attending a very dynamic second edition of Localization World. The sun is shining, barges slowly plow the Rhine outside the windows of the Beethovenhalle, and the post-event beers are satisfyingly cool. The event is well-attended, well-organized and off to a good start. New term of the day: reverse localization, used by keynote speaker Peter Williamson, Professor of International Studies and Asian Business at Insead France and Singapore. His idea is that not only will a knowledge-driven economy need what we normally think of as localization (adapting products to locale/language) but it will increasingly require the reverse process of ‘localizing’ back to the manufacturer the kind of local(e) knowledge that their developers need about locale sensitivity to design better products. Williamson ended up by challenging the industry to “raise their game”. Meaning presumably that bidirectional localization of the sort he was suggesting will position translation more than ever as the knowledge broker.