Bursts from internal shocks: is it really synchrotron emission?

Within the standard internal shock scenario, synchrotron emission would produce a spectrum with slope F(v) proportional to v^(-1/2), as immediate consequence of the cooling timescale being shorter than the integration time. This is in disagreement with the harder observed spectra, indicating that a different mechanism is responsible for the burst emission. Furthermore in this scenario pair production is expected when photons produced by inverse Compton emission are taken into account.