We present time-resolved optical spectroscopy of the dwarf nova GW Librae during its rare April 2007 super-outburst and compare these with quiescent epochs . The data provide the first opportunity to track the evolution of the principal spectral features . In the early stages of the outburst , the optically thick disc dominates the optical and the line components show clear orbital radial velocity excursions . In the course of several weeks , optically thin regions become more prominent as strong emission lines replace the broad disc absorption . Post-outburst spectroscopy covering the I-band illustrates the advantages of Ca ii relative to the commonly used Balmer lines when attempting to constrain binary parameters . Due to the lower ionisation energy combined with smaller thermal and shear broadening of these lines , a sharp emission component is seen to be moving in between the accretion disc peaks in the Ca ii line . No such component is visible in the Balmer lines . We interpret this as an emission component originating on the hitherto unseen mass donor star . This emission component has a mean velocity of \sim - 15 \pm 5 km s ^ { -1 } which is associated with the systemic velocity \gamma , and a velocity semi-amplitude of K _ { \mathrm { em } } = 82.2 \pm 4.9 km s ^ { -1 } . Doppler tomography reveals an asymmetric accretion disc , with the S-wave mapping to a sharp spot in the tomogram with a velocity consistent to what is obtained with line profile fitting . A centre of symmetry analysis of the disc component suggests a very small value for the WD orbital velocity K _ { 1 } as is also inferred from double Gaussian fits to the spectral lines . While our conservative dynamical limits place a hard upper limit on the binary mass ratio of q < 0.23 , we favour a significantly lower value near q \sim 0.06 . Pulsation modeling suggests a WD mass \sim 1 M _ { \odot } . This , paired with a low mass donor , near the empirical sequence of an evolved CV close to the period bounce , appears to be consistent with all the observational constraints to date .