Klarinet Archive - Posting 000456.txt from 2004/06

From: Joseph Wakeling <joseph.wakeling@-----.net>
Subj: Re: [kl] "Tunes create context like language"
Date: Mon, 28 Jun 2004 10:32:42 -0400

Ormondtoby Montoya wrote:

>>American social scientist George Kingsley
>>Zipf discovered that if he ranked words in
>>literary texts according to the number of times
>>they appeared, a word's rank was roughly
>>proportional to the inverse of its frequency. In
>>other words, a graph of one plotted against
>>the other appeared as a straight line.
>>
>>
>
>The above seems like a meaningless statement to me, similar to: "I'm
>tall, so this proves that I'm tall."
>
>

No. It's obvious that the higher the frequency, the lower the rank, and
vice versa ("lowest" rank being 1, 2, ... i.e. most frequent, second
most, etc.). But it's not at all obvious how the frequency should
decrease as rank increases. Zipf's Law is a precise mathematical
function relating the two, and it's interesting because it indicates a
wider "spread" in the language than if the words were just distributed
randomly. Basically it's a sign of structure in language---the most
likely interpretation being that words are used relative to words that
have already been used.

-- Joe

---------------------------------------------------------------------
Klarinet is a service of Woodwind.Org, Inc. http://www.woodwind.org