The enhancement of the emission rate of charge carriers from deep-level defects in electric field is routinely used to determine the charge state of the defects. However, only a limited number of defects can be satisfactorily described by the Poole-Frenkel theory. An electric field dependence different from that expected fr om the Poole-Frenkel theory has been repeatedly reported in the literature, and no unambiguous identification of the charge state of the defect could be made. In this article, the electric field dependencies of emission of carriers from DX centers in A1(x)Ga(1-x)As:Te, Cu pairs in silicon, and Ge:Hg have been studied applying static and terahertz electric fields, and analyzed by using the models of Poole-Frenkel and phonon assisted tunneling. It is shown that phonon assisted tunneling and Poole-Frenkel emission are two competitive mechanisms of enhancement of emission of carriers, and their relative contribution is determined by the charge state of the defect and by the electric-field strength. At high-electric field strengths carrier emission is dominated by tunneling independently of the charge state of the impurity. For neutral impurities, where Poole-Frenkel lowering of the emission barrier does not occur, the phonon assisted tunneling model describes well the experimental data also in the low-field region. For charged impurities the transition from phonon assisted tunneling at high fields to Poole-Frenkel effect at low fields can be traced back. It is suggested that the Poole-Frenkel and tunneling models can be distinguished by plotting logarithm of the emission rate against the square root or against the square of the electric field, respectively. This analysis enables one to unambiguously determine the charge state of a deep-level defect.