This raises an interesting point about the nature of citation. In the case of this paper, the preprint or typescript or whatever it is, the author's recension (like a "director's cut"), is much more easily available than the printed book version. It's right there on the internet at
unil.ch and also amongst Bronkhorst's papers at
academia.edu, just the click of a mouse away. So I think there is a good argument for citing these recensions of the paper rather than the book. And this is especially the case if the author's recension is what you've actually been reading. One should of course cite this like a web page, with URL and date of consultation.
The quest for a page number in an inaccessible printed book is based on an idea that the book is more real, in some sense, more valid or stable than the author's recension. But I think that belief can probably be challenged quite strongly.