Context : Aims : This paper investigates the peculiar behaviour of the light even ( alpha-elements ) and odd atomic number elements in red giants in the galactic bulge , both in terms of the chemical evolution of the bulge , and in terms of possible deep-mixing mechanisms in these evolved stars . Methods : Abundances of the four light elements O , Na , Mg and Al are measured in 13 core He-burning giants stars ( red clump stars ) and 40 red giant branch stars in four 25 \arcmin fields spanning the bulge from -3 to -12 ¡ of galactic latitude . Special care was taken in the abundance analysis , performing a differential analysis with respect to the metal-rich solar-neighbourhood giant \mu Leo which resembles best our bulge sample stars . This approach minimizes systematic effects which can arise in the analysis of cool metal-rich stars due to continuum definition issues and blending by molecular lines ( CN ) and cancels out possible model atmosphere deficiencies . Results : We show that the resulting abundance patterns point towards a chemical enrichment dominated by massive stars at all metallicities . Oxygen , magnesium and aluminium ratios with respect to iron are overabundant with respect to both galactic disks ( thin and thick ) for [ Fe/H ] > -0.5 . A formation timescale for the galactic bulge shorter than for both the thin and thick disks is therefore inferred . To isolate the massive-star contribution to the abundances of O , Mg , Al and Na , we use Mg as a proxy for metallicity ( instead of Fe ) , and further show that : ( i ) the bulge stars [ O/Mg ] ratio follows and extend to higher metallicities the decreasing trend of [ O/Mg ] found in the galactic disks . This is a challenge for predictions of O and Mg yields in massive stars which so far predicted no metallicity dependence in this ratio . ( ii ) the [ Na/Mg ] ratio trend with increasing [ Mg/H ] is found to increase in three distinct sequences in the thin disk , the thick disk and the bulge . The bulge trend is well represented by the predicted metallicity-dependent yields of massive stars , whereas the galactic disks have too high Na/Mg ratios at low metallicities , pointing to an additional source of Na from AGB stars . ( iii ) Contrary to the case of the [ Na/Mg ] ratio , there appears to be no systematic difference in the [ Al/Mg ] ratio between bulge and disk stars , and the theoretical yields by massive stars agree with the observed ratios , leaving no space for AGB contribution to Al . Conclusions :