The Voice in the Wilderness

Stating that the implementation of management and translation technology are the legacies of language fever does not mean that the role of linguists and translators is dead. It just means that in the high-volume documentation environment that exists today, having good translators is simply not enough.

But there are linguists who are fighting back. The most notable example is Douglas Hofstadter, a comparative literature professor and the author of an enormous new book (632 pages) Le Ton Beau de Marot: In Praise of the Music of Language (Basic Books, New York 1997).

He wages an attack on how today's computational models of languages and thought stack up next to the human mind and the skills of a literary translator. The author takes on Systran's automated software in particular.

He first comments that a "major error of Systran" was that it "renders the French phrase nous avions (we had) as 'we airplanes'," due to the fact that avion means airplane. Hofstadter goes on to tell us the obvious: that "humans are so good at unconsciously selecting just the appropriate meaning of an ambiguous word according to its content that probably 99% of native French speakers have never even noticed that the word avions has two potential meanings."

But the author is not done with Systran yet. Because he submitted the classic French poem "Ma Mignonne" to Systran's software, he goes on to tell us, "Apparently Systran does a bang-up job of translating thousands of pages of a certain meticulously 'cured' English in a timely domain," and yet can't even come close to handling a simple sixty-word poem (treated purely as prose). The excuse that the poem is in Old French holds no water. Any speaker of current French can effortlessly understand "Ma Mignonne".

I would guess that machine translation does a lousy job of translating Renaissance Italian poetry as well. The next LISA meeting should really examine this issue!