Some molecular glass-formers can crystallize in the glassy state, some of which are van der Waals molecules and some are pharmaceuticals. The molecular mechanism responsible for this glass-to-crystal mode of crystallization is of interest to the glass transition research community as well as to the pharmaceutical industry because the effect is detrimental to stability of amorphous form of the drugs stored below the glass transition temperature. Two prominent models have been proposed for the molecular mechanism. In the homogeneous nucleation-based crystallization model, the molecular mechanism is the secondary relaxation, and the other model assumes that the molecular process responsible for crystal growth in the glassy state is from the local molecular motions. Crystal growth requires motion of the entire molecule, and in the glassy state the only such local molecular motion is engendered by the secondary relaxation of the Johari–Goldstein (JG) kind. While the JG secondary relaxation is the crux in the two models of glass-to-crystal growth, it has not been found in the glassy state of the pharmaceuticals studied so far. The examples include 5-methyl-2-[(2-nitrophenyl)amino]-3-thiophenecarbonitrile (ROY), indomethacin (IMC) and nifedipine (NIF). In the absence of any evidence of the JG secondary relaxation, the conundrum is that the two models of glass-to-crystal growth cannot be validated. It turns out these pharmaceuticals all have structural α-relaxations with narrow frequency dispersion. Empirically, glass-formers with narrow α-dispersion have JG secondary relaxation with weak relaxation strength, not well separated from the α-relaxation, and hence cannot be resolved. Theoretically, the narrow width of the α-dispersion is due to weak intermolecular coupling. In this article we enhance the intermolecular coupling of NIF by mixing with octaacetylmaltose to enhance the intermolecular coupling of NIF. In this way we have successfully resolved the JG secondary relaxation in the dielectric loss spectra of the NIF component in the glassy state, and validated the two models of glass-to-crystal growth.