This thesis investigates spoken verb + noun collocations in L1 and L2 English candidates undertaking a language exam. It further introduces the new Trinity Lancaster Corpus of L1 Spoken English (TLC-L1), a corpus developed to align in context to the TLC-L2 (Gablasova et al., 2019). By considering differing proficiency levels of L2 speakers, the research takes a pseudolongitudinal approach to investigate collocational development with core findings including the influence of topic and register on the use of collocations in a language exam, which are also reflected in the TLC-L1 corpus, as well as evidence of a nonlinear developmental trajectory of English language learners and their use of verb + noun collocations. Finally, the thesis brings three major contributions and implications to the field: (1) methodological with the development and application of a new corpus, (2) theoretical through the analysis of collocations in the under-investigated mode of speech and (3) pedagogical with suggestions and examples of corpus-informed language teaching materials.