The interdisciplinary nature of linguistics often leads to similar research questions being investigated using diverse corpora. In this paper, special attention is given to two key concerns in the corpora design: spontaneity and complexity. It is shown that spontaneity and complexity are not necessarily incompatible but often become the centre point of early methodological choices. These choices are here termed "non-negotiables" and it is demonstrated how these non-negotiables lead to obligations, constraints and concessions in the data collection process which shape the corpus. It is argued that the existence of different non-negotiables, influenced by different theoretical approaches, lead directly to the creation of different corpora. These different corpora then allow complementary lights to be shed on similar objects of study. Certain central theoretical concerns-spontaneity, complexity, generalisability and data co-construction-are also discussed. The paper concludes that an improvement in the description and diffusion of these decision processes would promote increased and improved interdisciplinary collaboration.