Back to the Real World


I spent last week in Konstanz, Germany, on the shores of Lake Constance, at a small conference devoted to a theory of syntax called lexical-functional grammar (LFG). Among the attendees were two language engineers who attend almost every year: Ron Kaplan, co-developer of LFG (with the Stanford syntactician Joan Bresnan) and an industrial computational linguist since the 1980s, and Tracy Holloway King, a 1993 Stanford linguistics Ph.D. who also has a long career in Silicon Valley industrial research. Both currently work for A9, the research arm of Amazon. Their mission is to make internet search more intelligent, and ultimately to apply linguistics to the machine understanding of queries on the web.

The LFG conference was pleasingly free of the hostility and acrimony that so often infect theoretical syntax. Even my invited talk, basically a tentative critique by an outsider, was met with warm thanks and interesting discussion rather than raised hackles and animosity.

And there was no trace of the antipathy sometimes found between theoretical linguists and language engineers. Kaplan and King, in particular, closely followed the theoretical content of the presented papers in every session and contributed thoughtful, theoretically-based comments and useful questions. Their day job may be to help speed up online shopping, but that has not even slightly dimmed their interest in linguistic theory. They contribute as linguists, and then go back to their industrial environment, perhaps bearing new ideas for applying linguistics to search problems.

It was a pleasure to spend a peaceful week of constructive thinking with people interested in both linguistic theory and the goal of getting machinery to simulate intelligent linguistic behavior. But eventually one has to rejoin the real world. As if to remind us of the violent nature of that world, while I was traveling home on the Saturday evening, at a Konstanz night club the son-in-law of the owner was thrown out after an argument; he returned with an M16, killed the doorman, wounded a policeman, and was shot dead by the police. Back in Edinburgh I soon learned from the BBC World Service that the White House chief of staff had been replaced by a retired Marine Corps general and a foul-mouthed financier with zero judgment had been fired from the world’s most important public-relations job. Business as usual.

And there were linguistic stories in the press as well. The day after I got back I was asked at short notice to speak on the radio about one of them. It brought me right down to earth as regards what the media think is interesting about language.

On a BBC Radio London talk show I was told that a teenager had just captured the record for the longest word ever uttered in the British parliament. The previous record-holder was Jacob Rees Mogg MP, who found an excuse in 2012 to include the word flocci­nauci­nihili­pili­fic­ation in a speech. (Supposedly meaning “the act of estimating to be worthless,” this jokey coinage was invented by boys at Eton College who had noted a line in their Latin grammar book about nouns such as flocci, nauci, nihilī, and pilī taking the genitive case after verbs like aestimo.)

The schoolboy who outdid Mogg was giving invited testimony about teenagers with psychological problems relating to body image. He found an excuse to contrast physical conditions with mental ones, and as an arbitrary example of the former he chose pneumono­ultra­micro­scopic­silico­volcano­coni­osis. (This is an entirely artificial word supposedly naming a lung disease caused by microscopic silicon particles ejected by volcanic eruptions. It was coined by Everett M. Smith at the 1935 National Puzzlers’ League convention and is not used by the medical profession.)

With these rather unfascinating events as the rationale for getting a linguistics professor on the show, I was asked about how the language we use affects the impression we make on others.

My first point was that a language should not be confused with a list of words, and that nobody with any sense would take utterance of artificially created polysyllabic words as an index of intelligence. Then I tried, with only modest success, to steer the discussion in more sensible directions. I made a few points, but suddenly it was time for the weather and with a hasty word of thanks I was off the air. That’s what radio work is like.

Notice what the news media think is interesting. Nothing to do with the potential application of theoretical syntax to internet search. Some schoolboy used a superlong word, that’s what’s hot. Talk about fake news.

Surprisingly, the next morning the newspapers actually did have a story about language technology. But sadly, it was bizarrely stupid. I don’t have space here to explain to you just how stupid. I will have to leave that until next week. I’ll tell you about it then. Promise.

Return to Top