Two distinct measurement schemes have emerged for the new technique of two-dimensional terahertz spectroscopy (2DTS), complicating the literature. Here, we argue that the 'conventional' measurement scheme derived from nuclear magnetic resonance and its optical-frequency analogues should be favored over the 'alternative' measurement scheme implemented in the majority of 2DTS literature. It is shown that the conventional scheme avoids issues such as overlapping nonlinearities and facilitates physical interpretation of spectra, in contrast to the alternative scheme.