I think we can safely say that the ‘battle’ for second syllable stress on contribute and distribute is now lost. Almost all age groups now appear to me, with no valid statistical evidence whatever, to be putting the stress on the first syllable, and if John Wells’ survey was to be carried out again, I’m sure the percentages would be even higher for initial stress than they were in 2008 (the date of publication of the 3rd edition of his Longman Pronunciation Dictionary). Wells also allows the pronunciation contri’butory (3rd syllable stress), which I heard for the first time (consciously at least) a few weeks ago, from a BBC Radio reporter. I suspect he may have been flustered, and that this was a slip, but it obviously fitted the rhythm of his sentence. Can we expect to hear tri’butary at some time in the future?
It’s interesting that while the four-syllable words kilometre, controversy, trajectory, aristocrat and exigency have alternative pronunciations with first or second syllable stress (stress on the first syllable being the older one), and three-syllable words such as balcony, secretive, quandary, vagary, orchestra, contribute and distribute are now usually stressed on the first syllable rather than the second (orchestra appears to have been the earliest of these to shift – the OED 1st edition reports that Byron stressed the second syllable, and while a 1798 edition of Johnson’s dictionary stresses the first syllable, a later one edited by John Walker (1810) has second syllable stress), there are also words going the other way: integral and communal are now commonly heard in the UK with second syllable stress. Wells did no survey for integral, and shows 68% to 32% in favour of 1st syllable stress on communal, but I think things are changing. Integral, meanwhile, is simultaneously showing another, different, change: possibly by analogy with the word intricate, the /r/ is being shifted to the second syllable, leading to the pronunciation /ˈɪntrɪgəl/.
A problem with the Wells surveys from the first edition of his dictionary on is that those reporting their own usage may have given the pronunciation they felt was the ‘correct’ one, rather than the one they actually used – and in many cases, although they might believe they used one pronunciation, dispassionate observation by others might prove otherwise. I am sure John Wells is aware of this. Only close analysis of a large oral corpus could demonstrate the true position.
Jack Windsor Lewis has a section on this question on his website here.