[UPDATE 6/6/14: see end of post]
I should really say another antecedent, given that the Stanford Encyclopedia of Philosophy identifies three other versions of this argument. I take the following from Stanislaw Lem’s Summa Technologiae:
Physicist and science fiction writer Anatoly Dneprov has described an experiment in his novella, whose aim was to debunk a thesis about “infusing with spirituality” a language-to-language translation machine by replacing the machine’s elements such as transistors and other switches with people who have been spatially distributed in a particular way. Performing the simple functions of signal transfer, this “machine” made of people translated a sentence from Portuguese into Russian, while its designer asked all the people who constituted the “elements” of that machine what this sentence meant. No one knew it, of course, because the language-to-language translation was carried out by the system as a dynamic whole. The designer (in the novella) concluded that “the machine was not intelligent”… [p. 324]
Lem’s book was published first in Polish in 1964, and Anatoly Dneprov died in 1975, so this comfortably predate’s John Searle’s 1980 version of the argument. I haven’t been able to identify what novella Lem is referring to here, as his notes and bibliography have no mention of Dneprov’s work. It would be great if anybody out there did know, and I could make a note of it here.
Update 6/6/14: Internet success! As can be seen from the comments below, the Philosophy Department of Moscow State University’s Center for Consciousness Studies has posted a translation of this story. Go! Read! Enjoy. As an aside, I also like how this story is well within the bounds of the then orthodoxy of dialectical materialism.