Physical properties and evolutionary state of the Lyman alpha emitting starburst galaxy IRAS 08339+6517

Context. Though Ly alpha emission is one of the most used tracers of massive star formation at high redshift, it is strongly affected by neutral gas radiation transfer effects. A correct understanding of these effects is required to properly quantify the star formation rate along the history of the Universe. Aims. We aim to parameterize the escape of Ly alpha photons as a function of the galaxy properties, in order to properly calibrate the Ly alpha luminosity as a tracer of star formation intensity at any age of the Universe. Methods. We have embarked on a program to study the properties of the Ly alpha emission (spectral profile, spatial distribution, relation to Balmer lines intensity,...) in a number of starburst galaxies in the Local Universe. The study is based on Hubble Space Telescope spectroscopic and imaging observations at various wavelengths, X-ray data, and ground-based spectroscopy, complemented with the use of evolutionary population synthesis models. Results. We present here the results obtained for one of those sources, IRAS 08339+6517, a strong Ly alpha emitter in the Local Universe, which is undergoing an intense episode of massive star formation. We have characterized the properties of the starburst, which transformed 1.4 x 10(8) M-circle dot of gas into stars around 5-6 Myr ago. The mechanical energy released by the central super stellar cluster (SSC), located in the core of the starburst, has created a cavity devoid of gas and dust around it, leaving a clean path through which the UV continuum of the SSC is observed, with almost no extinction. While the average extinction affecting the stellar continuum is significantly larger out of the cavity, with E(B - V) = 0.15 on average, we have not found any evidence for regions with very large extinctions, which could be hiding some young, massive stars not contributing to the global UV continuum. The observed soft and hard X-ray emissions are consistent with this scenario, being originated by the interstellar medium heated by the release of mechanical energy in the first case, and by a large number of active high-mass X-ray binaries (HMXBs) in the second. In addition to the central compact emission blob, we have identified a diffuse Ly alpha emission component smoothly distributed over the whole central area of IRAS 08339+6517. This diffuse emission is spatially decoupled from the UV continuum, the H alpha emission, or the H alpha/H beta ratio. Both locally and globally, the Ly alpha/H alpha ratio is lower than the Case B predictions, even after reddening correction, with an overall Ly alpha escape fraction of only 4%. Conclusions. We conclude that in IRAS 08339+6517 the Ly alpha photons resonantly scattered by an outflowing shell of neutral gas are being smoothly redistributed over the whole central area of the galaxy. Their increased probability of being destroyed by dust would explain the low Ly alpha escape fraction measured. In any case, in the regions where the diffuse Ly alpha emission shows the largest Ly alpha/H alpha ratios, no additional sources of Ly alpha emission are required, like ionization by hot plasma as proposed for Haro 2, another galaxy in our sample. These results stress again the importance of a proper correction of scattering and transfer effects when using Ly alpha to derive the star formation rate in high-redshift galaxies