2002-10-04 17:20 (UTC+1)

Infelicitous

The Danish Royal Baby has been duly moistened, and may - nay, should - now be referred to as Felix.

He'll always be Knud to us at Desbladet, though. (Although, since he's not technically a prinsess, he may not feature heavily. We can't even be bothered with the King of Sweden, and he's King and everything.)

2002-10-04 13:24 (UTC+1)

Upptagen

I'm think I've figured out how to get LaTeX to do Swedish punctuation, now. Don't be fooled by the home-page: pretty much all pro-league mathematics, and plenty of other stuff, is typeset with this package. It can be a dog to use, but it's our dog, and we love it.

When, during my exile in the world of alleged realness, I had to write reports for $OLDCOMPANY, which insisted upon the use of Word(TM) instead, I retaliated by leaving out all of the equations. Hardly professional behaviour, you might say, but Word(TM) is hardly a professional-level writing tool.

Anyway, with the inclusion of usepackage[swedish]{babel} it all seems to work quite nicely: the auto-dater knows it's oktober, and the auto-contentser knows to announce the Inneh�l and we seem to be in business.

We of the Swedish class also seem to have volunteered ourselves to give presentations on the many fascinating cultural aspects of Sweden we shall be studying, and if I deign to create printed slides, I'll use it for that, too. (Mathematical tradition holds that in fact felt-tipped pens are usually preferably. Powerpoint(TM) would be a lynching offence - even mentioning it provokes ritual exclamations of horror and disgust.)

2002-10-04 09:32

Nei, ekki meiri fisk!

I first read Thor Heyerdal's The Kon-Tiki Expedition in English during a trip on the Hurtigruten boat from Bergen to Lofoton ( Svolv�rgeitir picture here).

I *second read an excerpt in Swedish for last night's class, and very intertextual it all was, too. I could almost taste the gin (we finished our duty-free gin on the boat, drinking it neat for want of mixers), although I suppose that may have been the gin I was drinking (with water, for lack of mixers) and feel the Warm Glow Of Disapproval of the Luxury Class Passengers on staggering out, unwashed and hungover from the room over the engines where we slept on the floor in sleeping bags, but still ready - even keen - to mock their pronunciations of "GLAYsher" (it took me ages to figure out that was "glacier") and "FEEord", I could almost feel the breakfast-herring slithering down my throat. (It is an ancient rule, or custom, in my land that all accounts of Nordic travel must mention, at least once, the ancient rule, or custom, of partaking of herring at breakfast.)

Ah, yes. Indeed, fish and floating vehicular entities featured prominently, also, in Mr Heyerdahl's account of his own trip or expedition, no less so in Swedish than in English: "Ibland svor n�gon av oss till, d� han fick en kall flygfisk mitt i ansiktet". (I can imagine:

"By Odin's hairy arse!" exclaimed Olle, "Again have I caught, or received, a flying fish smack in the kisser!"

You may, of course, imagine differently.

I did mention that my November novel was mostly going to be in the form of dialogue, didn't I? It's not too soon to start shivering in horror and revulsion at the thought!)

The hastily convened svenskkv�llskurskulturkommitt� (Ooh! Love the accent! Is it French?) briskly settled the question of whether or not one can have too much fish in the affirmative, however, and unanimously passed a motion that you'd have to be barking mad to try to cross the Pacific on a balsa-wood raft. (Heyerdahl's journey is not a refution of this position, since he clearly was barking mad, although he remains a hero of mine, more because than in spite of this. It is not a small thing to accomplish - as he surely did - great things, while being - as he surely was - completely and utterly hatstand.)

Incidentally, in this year's class (for the first time!) we have actual writing assignments (skrivuppgifter) and there is a non-negligible risk that some of them will be either recycled or previewed as Desbladet posts. You have been warned.

2002-10-03 12:45

Whoops-a-daisy!

My climbing excursions this year were limited to a single trip to Portishead quarry, which is bad.

On the other hand, I did not plunge to my death in a climbing accident, which is good.

The late G�ran Kropp was rather more active, and the odds caught up with him. It looks like a terrific crag in those photos, though but.

2002-10-03 09:21 (UTC+1)

Bogotic Language Voodoo (BLV).

Gazdar and Mellish's Natural Language Processing in PROLOG is not a good book. But it's not dishonest: the first page of the preface states:

The book is specifically intended to teach NLP [Natural Language Processing - des] and computational linguistics: it does not attempt to teach programming or computer science to linguists, or to provide more than an implicit introduction to linguistics for computer scientists.

Computational linguistics for the authors, then, is not a serious attempt to find out how language works. It does not connect with the work of Chomsky and his followers on syntax - although even people who think Chomsky is wildly misguided must surely acknowledge that he's spent a lot of time thinking about syntax - presumably because that's linguistics and thus ineligible. (It doesn't connect with any descriptive grammatical work on English, either, and yes, it is almost entirely concerned with English.)

Neither does it provide a satisfactory account of the computer science view of languages, automata and grammars, presumably because that's computer science. (Go to the Dragon book instead to get started with this.)

If, however, you want an account of computational linguistics which is ungrounded in linguistics, slights computer science, uses but doesn't teach Prolog and shies away from the state-of-the-art (c. 1989) in favour of "areas that are beginning to be well understood", then this is surely the book for you.

As a bonus, the prose style is strictly drying-paint. I have to read it anyway, but you don't. This is vastly to your advantage, and I urge you to exploit it fully.

(If, on the other hand, you just want to learn some Prolog, have some fun, and have a peek at some toy natural language parsers, go play with Learn Prolog Now!.)

2002-10-03 07:56 (UT+1)

Nobbar Bill Clinton

isn't the most reassuring of sub-headings in a (non-)story about Vickan, but it turns out she can't make it after all: she's doing door-to-door collecting for static cling meeting the Pope.

And it turns out that de-facto Kronprinsess elect Mary isn't invited to the dunking.

We will, of course, bring you news of stuff that actually is happening just as soon as we get some.

2002-10-02 11:29 (UTC+1)

See?

Old Low Moldovan Slovenian [Moldovan is (a dialect of) Romanian and thus not really Slavic enough for this] it is, then:

Since Slavic languages are to a large extent mutually intelligible, many Slavs just do not bother learning Russian because they feel they can communicate adequately with Russians by speaking their own native language. For example, a Czech or Bulgarian may use Czech or Bulgarian, respectively, when speaking to a Russian and be reasonably sure that much of what they say will be understood by the latter. (In addition, not learning Russian or learning it badly was one of the ways that people in the Eastern bloc countries used to show opposition to the dominance of their countries by the Soviet Union.)

[Lyovin An Introduction to the Languages of the World, pp 61-2.]

Now if you'll excuse me, I have a Mentat trance to drop into for today. Alex Golub is doing historical linguistics and Graham Leuschke has stuff about Markov chains so you needn't be bored while I'm busy.

2002-10-02 09:10 (UTC+1)

Oh, Herregud! Jag har en yxa i huvudet!

The InterWebNet also has resources for people interested in a variety of varieties of language: the axe in my head page [via Esther] and the celebrated I can eat glass project are both important contributions.

To something. Probably.

2002-10-01 17:05 (UTC+1)

Wine me, cheese me, cohomologise me.

Um. It was just the department's cheese and wine for incoming postgraduates. I've always held that one glass of wine and I'm anybody's, but I'm prepared to make an exception for maths departments.

OTOH, red wine and thinking are a really terrifically bad combination, don't you find? I'd just go home if it wasn't raining.

2002-10-01 15:36

Big Book Freeze

All my outstanding online bookshop orders are here at last, and it's time for the Big Book Freeze: no new books for a whole year.

The chances of this actually happening are negligible, of course, but I certainly have more than a years' worth of stuff to read and there's the University library to play with as well. (Library books don't count: this is an ownership ban.)

I've finished Harry Potter � l'�cole des sorciers (in less than a week, without using a dictionary, wuhoo!) and I think Chomsky57 is going to have to be next. It's only ickle, and it's surely going to go well with a side-order of definite clause grammars in Prolog, which has been my recreational coding activity this week.

(Incidentally, I don't think this is how language works at all, in case you're wondering, but I also don't think it's worthless, and it's certainly great fun.)

2002-10-01 13:34

Overview Overview

By request, some notes on various very different books which review the world's languages in one way or another.

Comrie's The Worlds Major Languages deals in some depth with the 40 languages with the most speakers. It's currently on loan from the library here so I haven't seen it, but the Amazon blurbs and reviews make it clear that it's a reference book with detailed summaries of the chosen languages by experts.

Campbells' Concise Compendium of the World's Languages has short (5 pages is typical) linguistic sketchlets of almost a hundred languages. Amazon reviews note that the phonologies are often incorrect, but still, it's a lot of good stuff.

Dalby's Dictionary of Languages has lots of languages, no scary technicalities, plenty of interesting information, and is unashamedly browsable. It can suck up whole evenings with a single opening, I can assure you.

By contrast with all of the above Lyovin's An Introduction to the Languages of the World is an actual textbook: you're meant to read it in order and do the exercises and actually learn things about how lots of different languages do things - from my brisk flipping it all looks terrific fun. I'm as chronic a dilettante as the next language enthusiast, of course, and I love Dalby's book to death, but this is a book I would love to have time to read properly. It assumes some prior knowledge of linguistics in general and phonetics in particular.

2002-10-01 09:25 (UTC+1)

Sweet nothings

Ladies and Gentlemen, for the last time as he's on his way back to the library, I have the honour to present Mr Henry Sweet:

Phonetics, of course, should be begun in the nursery. The time will come when ignorance of practical phonetics will be held to incapacitate a nurse as much any other incapacity. If the infant's attempts to speak were guided into the channel of systematic all-round phonetic drill, it would on entering into school-life be already a thorouhgh practical phonetician: all it would have to learn would be the use of a phonetic notation. The pronunciation of foreign languages would then offer no initial difficulties whatever: it would simply be a question of remembering what particular sounds occurred in the foreign language, and associating them with the symbols of the phonetic alphabet for that language.

That was in 1897. In 2002 Lonely Planet's Baltic Phrasebook describes an Estonian "p" as "halfway between English 'p' and 'b'". What they (probably) mean is that it's an unaspirated 'p', but it's as well that they didn't say so: phrase-books have been a fertile source of amusement to people who actually care how languages are pronounced for as long as they have existed, and Civilisation Might Crumble if this became impossible. (Sigh.)

On a happier note:

Such a language as Swedish, on the other hand, with its simplicity, its clearness and harmony of phonetic structure, and its few, but clear, simple, symmetrical inflections, really combines, to a great extent, the advantages of the ancient and modern languages.

This is preposterous, of course, but not less charming for that.

2002-09-30 12:53

Autumn in New York: now available world-wide

Yow! The Computational Syntax and Semantics group at NYU really does seem to believe that linguistics can be reduced to a question of implementing Chomsky's ideas in Prolog, and they want you to believe it, too.

Why not try their free questionnaire Natural Language Computing Workbook?

No? Well, MIT is making all its teaching materials available on the web for free; there's sure to be something you've always wanted to know there. (Although the first item on the required reading for the Linguistic theory and the Japanese Language class is an as-yet-unpublished article only available from the 24.946 cabinet locker in front of the lecturer's office.)

2002-09-30 10:19 (UTC+1)

Proverb.

Food before beer: nothing to fear,
Beer before food: vomiting's rude.

2002-09-30 10:18(UTC+1)

The Lost Generation

No student of Prolog should miss out on the Fifth Generation, an accidental comedy classic of the AI summer. Written in 1983, it describes how the Japanese were then embarking on a fifth generation of computers, based on Prolog, which would become ubiquitous intelligent assistants in the coming knowledge economy, and fulminates frothily against the failure of the American government to organise programmes and taskforces and the like to meet the challenge.

This was in the days when people thought expert systems could be made to work, obviously, and the book has some delightfully impressionistic accounts of the ideas behind them. More seriously, the outrageous over-selling of AI in the eighties inflated a modest bubble of unreasonable expectations, and lead to a credibility recession that we're still in, and I could read the book as a chapter in that tragedy, if I so chose. Happily, I prefer to get drunk and mock it without mercy.

From my perspective, though, the oddest claim in the book is that we would be interacting with these computers in natural language, which is unaccompanied by even implausible explanations of how this is going to be achieved. The authors don't even bother to exhibit anyone who claims to have almost cracked the problem; they don't even seem to acknowledge that there is or could be a problem, which is either immensely arrogant or immensely stupid.