--- In
cybalist@yahoogroups.com, "Brian M. Scott" <BMScott@...>
wrote:
> At 8:15:17 PM on Saturday, October 1, 2005, Grzegorz
> Jagodzinski wrote:
[...]
>> All laws are descriptive, contrary to theories whose aim
>> is to answer the question "why". However, laws also
>> *require* things to happen so-and-so, in order to satisfy
>> what the laws say. As Newton's law requires apples to fall
>> onto the ground, so Zipf's law requires frequent words to
>> be shortened (if they are too long). Both things *must*
>> happen.
> Don't be ridiculous. 'The length of a word tends to bear an
> inverse relationship to its relative frequency' doesn't
> require anything of any specific word; it's a vague,
> qualitative description of a lexicon.
Okay, now I have more time. Newton's law does not require
apples to fall; it says that unless prevented, they do fall.
This is a description of what is observed; a counterexample
would merely show that the description is inadequate. Indeed,
thanks to Einstein & Co. we now know that it *is* inadequate,
though not in this particular way. Similarly, this observation
of Zipf's requires nothing; it merely describes the data. And
if you read the paper 'Word Length, Sentence Length, and
Frequency -- Zipf Revisited', by Bengt Sigurd, Mats Eeg-Olofsson,
and Joost van de Weijer, you know that it's definitely an
incomplete description of the data, since the actual relationship
looks more like a gamma distribution with a peak.
Finally, we know that there are also processes that work in the
opposite direction, extending some common words when they get too
short; consider, for instance, the history of French <aujourd'hui>.
Brian