Three-letter Initialisms

Language Log 2022-12-11

I spent last week at the ASA conference in Nashville. The "ASA", as members all call it, is the Acoustical Society of America — but asa.org is "American Students' Assistance", while asa.com is the "American Sailing Association", and Wikipedia offers 75 other options for the ASA initialism.

This is general problem. Another organization that I belong to is the LSA, as the members of the Linguistic Society of America call it — but lsa.org is the "Louisiana Sheriffs' Association", and lsa.com is apparently malware associated with Windows' Local Security Authority, so I won't link to it. The Linguistic Society of America, having been pre-empted by the Louisiana Sheriffs, used to be online at the URL lsadc.org, but is now linguisticsociety.org. Wikipedia offers 48 other options for LSA. They include "Latent Semantic Analysis", which is the earliest of the word-embeddings at the root of the tree whose recent fruits include ChatGPT — which recently evoked some confusion on this blog as to the meaning of the GPT initialism.

There are obviously 26^3 = 17576 three-letter initialisms, and nearly all of them are spoken for, multiple times over.

Web (and Wikipedia) search suggests that nearly all possible 3-letter sequences are Out There — though some, like AZU, are words or names, which makes it harder to find many uses as initialisms or acronyms. Someone with more patience than I have could estimate how they're divided up among various types of initialisms: companies, organizations, concepts, algorithms, methods, diseases, drugs, and so on.

To help explore this issue, I wrote a little program to generate random 3-letter sequences. Here's 1001 outputs.

There's also the question of how many such initialisms a reasonably literate person knows. Obviously this will vary a lot from person to person, but I'm guessing it's generally over 100.