This word has caused more trouble than most when trying to decide on a pronunciation. The BBC’s pre-war Advisory Committee on Spoken English managed to change its mind twice in the course of ten years before giving up on a recommendation.
In 1928, the first edition of Broadcast English I – Recommendations to announcers regarding certain words of doubtful pronunciation gave ‘gárraazh’ (there was no IPA transcription).
In the extraordinary publication of the Society for Pure English – Tract no. XXXII, “The BBC’s Recommendations for pronouncing doubtful words, Reissued with Criticisms Edited by Robert Bridges’ (and remember that Bridges was the chairman of the BBC Committee!), Bridges writes “One can feel no sentiment about the pronunciation of garage, except to deplore that there should be another word added to the some 200 which used to be -age and are now commonly pronounced -edge or -idge; for instance, Jones records in his dictionary (1917) that cultivated Southern English people, in their ordinary conversation, pronounce parsonage as pahsnidg (pa:snid?). Anything that can check the spread of this disease is useful, and it is to be hoped that the B.B.C. announcers will set the example of a more agreeable solution than the phoneticians have predetermined.” Did Bridges really want us to say /?p??s?n???/? One must also wonder at the sight of the chairman of a committee issuing a critique of a report that his committee has published!
The second edition of Broadcast English I, published in 1931, after Bridges’ death, and when Bernard Shaw was chairman of the Committee, gave only ‘gárredge’. Shaw is reputed to have said that as sausage ends in -edge, there was no reason for garage to be any different.
The third edition (1935) went back to ‘gárraazh’ and added /?gær???/ (IPA had now been added to all the words). The only reason I can find for this is that the ‘public’ objected to what it considered a lowering of the BBC’s standards. An initial problem that faced the committee was that the 1st edition of the OED does not contain the word, but the supplement, published in 1933, giving the earliest citation from 1906, has both pronunciations. Henry Wyld, author of the Universal Dictionary of the English Language (1932), joined the committee in 1934, when it was reorganized, with four linguists (Daniel Jones, Arthur Lloyd James, Wyld and Harold Orton) making preliminary recommendations to the full committee. Wyld had also given both pronunciations in his dictionary, so it can be assumed that he had no objection to either of them.
It seems that no one considered the pronunciation /g??r??(d)?/, which is thought of as American, in any of these discussions. Both the Webster New International Dictionary and Funk and Wagnall’s Standard Dictionary indicate initial stress as either British or ‘especially British’.
I can’t believe that in 2011 anyone any longer cares much about which of these pronunciations they hear, but it would not be surprising if the claim made by Robert Walshe of the British Library, in this morning’s Today programme on Radio 4 – that /-?d?/ is increasing in frequency at the expense of /-??(d)?/ – was justified, as most common words ending in -age are pronounced (in British English at least) in this way.