October 09, 2005

Can humanists count?

A few days ago, I wrote about a strange conjunction from A(gence) F(rance) P(resse). Arnold Zwicky responded with a subtle psycho-grammatical deconstruction. Jean Véronis suggested perhaps it was careless translation of a French-language original (see the update for details). My reaction was to wonder whether some modern humanist scholars, like the Pirahã, are unwilling to count things.

I'm not talking about Arnold and Jean, who are empirically scrupulous and numerically expert. To follow my train of thought, we need to start with the excess of demonstratives in the French version of the AFP texts:

Les gagnants sont alors discrètement contactés avant la cérémonie pour leur laisser la possibilité de décliner cette offre. Mais en fait ils sont peu nombreux à résister à cette récompense et sont même beaucoup à venir recevoir leur prix en personne et à leurs propres frais.

The winners are discretely contacted beforehand to give them an opportunity to decline. It is a testament to the growing prestige of the event that very few turn down the offer and agree to attend at their own expense.

Now, I don't think that there's anything about French and English that forces this difference. At least, the English version could have used object-phrases like "this offer" and "this award", and French is sometimes happy enough with null complements, or with a form of le instead of a form of ce. However, all this reminded me of Jean-Claude Sergeant and his characterization of the English language, which seems to predict the opposite distribution of demonstratives:

L'anglais est une langue paradoxale. Assez rigidement structurée sur le plan syntaxique, on pourrait la qualifier de "no nonsense language", c'est-à-dire de langue où l'à-peu-près n'a pas sa place, au risque de voir cette appréciation immédiatement contredite par l'emploi surabondant de reprises anaphoriques – this, that et leurs formes plurielles – et par les multiples possibilités de raffinement lexical que lui permet le jeu des postpositions adverbiales. On approchera un peu mieux encore la vérité de la langue en rappelant que le terme "reasonable" renvoie davantage à ce qui relève du bon sens que de la raison.

English is a paradoxical language. Quite rigidly structured on the syntactic level, we could call it a "no nonsense language", that is to say a language where there is no place for more-or-less, at the risk of seeing this insight immediately contradicted by the overabundant use of anaphoric references -- this, that and their plural forms -- and by the multiple possibilities of lexical refinement permitted by the play of adverbial postpositions. We will come a bit closer to the truth of the language by recalling that the term "reasonable" refers more to common sense than to rationality.

In previous posts, I've spot-checked some of Sergeant's other claims about differences between French and English. In each case, a few minutes' research over morning coffee was enough to call his generalizations into question. It doesn't seem to be true that "we can hardly, in English, accumulate at the start of the phrase the elements of complementation as this is practiced in the French press", nor that "English will avoid all ambiguity as to the identity of agents intervening in a phrase", at least if we use web search to count instances similar to Prof. Sergeant's own examples of English/French comparison.

So the extra demonstratives in the French AFP text reminded me of a Sergeant claim that I haven't checked yet: that English is characterized by "overabundant use of anaphoric references -- this, that and their plural forms". And I thought I'd do another coffee-break experiment.

I picked three stories in each language from this morning's news about the earthquake in Pakistan. The French-language stories were from AFP, Reuters and Le Monde; the English-language stories were from AFP, Reuters and BBC Online.

The French stories turned up 16 demonstratives (8 ce, 7 cette, 1 cela) in 2251 total words, for a rate of 7.1 demonstratives per thousand words. The English stories turned up 9 demonstratives (3 that, 5 this and 1 these) in 2270 total words, for a rate of 3.9 demonstratives per thousand words. (Note that I had to check all the instances of that by hand, to distinguish the demonstratives from the complementizers. This prevented me from trying to evaluate the idea on a larger scale over breakfast.)

I certinly don't claim that this little experiment -- based on a too-small sample of a single genre -- is definitive. However, I'll assert that it calls into question the idea that "this, that and their plural forms" are "overabundant" in English prose, at least in comparison to appropriately matched French prose.

More broadly, I'm starting to wonder: is there a subculture of humanistic scholars, embedded within modern western civilization, that has lost or rejected the idea that there is a connection between a generalization and our expectation about counts of its instances? The attitude of the Pirahã towards counting seems exotic and even improbable; but the attitude of scholars like M. le Professeur Sergeant puzzles me more.

Posted by Mark Liberman at October 9, 2005 03:15 PM