1999A&A...349..765M

We have analyzed the properties of the star formation episodes taking place in a sample of blue compact and irregular galaxies by comparing their multiwavelength observational properties with the predictions of evolutionary population synthesis models. This method has allowed us to constrain the age, star formation regime (instantaneous or extended) and Initial Mass Function (IMF) slope, as well as the shape and strength of the interstellar extinction in these regions. We find that star formation episodes are essentially short with a mean age of 3.5Myrs. Some galaxies may be undergoing their first global episode of star formation while for the rest of the sample older stars contribute to at most half the optical emission. The Wolf-Rayet star population (WR) is well reproduced by the models and provides the strongest argument in favor of a short duration of the star formation episode. Supernova rates are relatively large. The accumulation of supernova explosions within few Myr has contributed to a quick metal enrichment of the ISM and to its disruption by the release of huge amounts of mechanical energy. V-K colors agree well with the prediction that red supergiant stars are rare in low metallicity regions. A general agreement is found between the predicted and observed far infrared emissions suggesting that the fraction of hidden stars contributing to the ionisation is minimum, except in some specific objects. A saillant result of this study is that the IMF slope appears to be very universal, on average very close to that of the solar neighborhood and with no dependence on the metallicity, contrary to previous claims. We have also found no dependence whatsoever between the shape of the extinction law and the metallicity. It is likely that the strong radiation associated to the bursts destroys the dust component responsible for the 2175Å bump. Finally we confirm that extinction affecting the stellar continuum is in some cases significantly weaker than that derived from the Balmer emission lines. Such a discrepancy can lead to underestimations in the value of the Hβ equivalent width by a factor as large as 2, leading to an overestimation of the age of the burst. Similarly, the Wolf-Rayet bump to the Hβ luminosities ratio can also be affected by this differential reddening leading to an overestimation of the WR star population. As bursts get older they appear dustier, possibly as a result of dust ejection during the evolution of their most massive stars. Finally, we have found a serious general discrepancy between the predicted and the measured radio luminosities. While part of this discrepancy might be attributed to aperture mismatching in some cases, it points to the presence of additional radio sources not included in present evolutionary models.