English has borrowed words from French from at least as long ago as Anglo-Norman times. French invariably has stress on the ultima (the final syllable), but British English regularly retracts that stress by one syllable closer to the beginning of the word when it borrows a given word from French. This is what accounts for stresses like gárage and bístro in British English loans from French, as it does for rappróchement (heard today from a speaker of British English on the BBC World News Service, for example). Cf. American English garáge, which preserves the stress of the French original.
How to account for this retraction? Here, as many times before in these posts, Y-H-B (who was once called “the markedness man” by the late Robert Austerlitz, quondam professor of linguistics at Columbia) invokes markedness to explain the phenomenon. Final position of stress in French as perceived by an English speaker is interpreted as being marked because of the fundamental distinction in English between nominal and verbal stress, as in prógress (noun) vs. progréss (verb), etc., etc. Since verbs necessarily make reference to time, whereas nouns do not do so necessarily, verb as a category is marked and noun unmarked. Hence when British English speakers––but not Americans–– hears final stress in a French word that is being borrowed, they nativize it by automatically unmarking it by retracting the stress one syllable closer to the beginning of the word. Voilà!