How many spoken languages? How many computer languages?

« previous post | next post »

Jeff Shaumeyer wrote recently on my Facebook wall to report that

In another facebook conversation a friend said "I read that there have now been more programming languages than spoken languages of all time." Is this even remotely possible?

Mike Geis immediately fixed on one problem with the claim, the problem of counting languages, whether you're counting human languages (spoken or signed) or computer languages. While we were contemplating these well-known issues (sources that attempt to put a number on human languages give a range — things like "5,000 to 10,000" — and the number of languages listed in the Ethnologue go up with each edition; the 15th edition has 6,912 entries, but a new edition will be out soon, and it's bound to have more), Jeff posted that

the friend discovered that he dramatically misremembered the result he was paraphrasing.

(whew!) but returned to the original claim, saying,

Just by orders of magnitude I found it incredible that more computer languages/dialects could have been created in the last hundred years than the total of spoken languages/dialects that had ever been.

Here's the second problem: "of all time" in Jeff's first message, "that had ever been" in the second.

We have records of virtually all the computer languages that have been devised. (Yes, records can be lost, but surely not many of them have been.) But for human languages, we're not even sure what the inventory is now, and the documentation of of human languages in detail is really a very recent enterprise in human history. The further back you go, the more languages you find that are known only by name or by a few words, and in any case you become dependent on written records, which drastically undersample the inventory of human languages.

It's hard to know: if you go back to the Dawn of Language (whenever that was — that's something else we don't really know much about), how many languages have there been? Surely more than the 7,000 or so we know about now, but how many more? An order of magnitude? Two? (For comparison, according to this Wikipedia page, "As of May 2006 The Encyclopedia of Computer Languages [actually, the on-line Roster of Programming Languages] by Murdoch University, Australia lists 8,512 computer languages." Goodness knows how they counted.)

As for the comparison between computer languages and human languages, you can imagine the scene: a bunch of computer people (maybe in a bar) undertake to list as many computer languages as they can and as many human languages as they can. Computer languages are almost sure to come out ahead, and not just because computer people generally don't have experience with a great many human languages. Computer people go through a lifetime of shifting from one computer language to another (and, sigh, one operating system to another, and, sigh, one text formating scheme to another, etc.), so they are impressed by the variety of resources; and they spend a fair amount of time comparing computer languages. Most ordinary (non-linguist) speakers of languages, including computer people, don't have comparable experiences with human languages.

So there would be an attention effect here: computer people are obliged to attend, over their lifetimes, to a considerable number of computer languages, so they think there are a huge number of them.

Back in the real world, I've been a mite suspicious of the roughly 8,500 figure in the Roster of Programming Languages. And for good reason: now that I look at the site, I see that there are 17 entries for LISP/Lisp, 31 for FORTRAN/Fortran, and 41 for ALGOL/Algol. So grouping "dialects" together, the way the Ethnologue does for human languages, would probably bring the figure down from thousands to hundreds.


Comments are closed.