Frequency-domain molecular absorption spectroscopies compare the energy spectral density of a light beam before and after interaction with resonant molecules in a limited spectral range, yielding the wavelength-resolved sample-specific attenuation. However, in these schemes, temporal integration hides the dynamics of the interaction between the electric field of the excitation and the sample. Here, we employ attosecond-precision electric-field-resolved spectroscopy to trace the complete energy transfer dynamics between mid-infrared (MIR) optical waveforms and vibrating methylsulfonylmethane (DMSO2) molecules in aqueous solution, with sub-cycle resolution. Resonant absorption and stimulated emission are observed on a few-femtosecond scale and optical-phase-dependent coherent transients manifest on the femtosecond-to-picosecond scale. The oscillations of the coherently-excited molecules dephase within several picoseconds owing to their varying environments in the solution.