An age-old conundrum in the field of speech research is the way in which sounds are represented in the human mind. Many phonologists argue that a mental representation consists of bundles of abstract categories that constitute the minimum amount of information needed to distinguish each sound from every other sound within a particular language. However, for many phoneticians, sound representations are more detailed, and may even be characterised by the different acoustic distributions that an individual has been exposed to over their lifespan. Major (2008, p. 68) highlights that such discussions are not unique to first language acquisition but also extend to the context of bilingualism. Foreign language speech research, for example, has sought to examine whether phonetic or phonological properties of the native (L1) sound system most influence the acquisition of a foreign (L2) sound system (ibid.). However, few L2 speech studies to date have attempted to bridge the gap between phonetics and phonology by considering implications for both types of theory.This three-paper thesis addresses the phonetics-phonology distinction through a series of studies that focus on the production of L2 French vowels by L1 Anglophones and the factors that correlate with L2 pronunciation development. As will become apparent, the nuanced relationships between L2 pronunciation and the factors under examination cannot—it will be argued—be fully understood without focusing on multiple levels of linguistic representation. As such, the idea that a wholly phonetic or phonological approach to L2 speech learning can provide a comprehensive account of L2 speech learning will be critically evaluated.More specifically, Articles 1 and 2 analyse two potential predictive factors of learners’ L2 vowel production: the degree to which these learners successfully process the same L2 vowels in perception, and the quantity and quality of L2 input to which learners have been exposed. On the other hand, Article 3 analyses a potential outcome of L2 vowel development: the adaptation of the L1 vowel system in order to incorporate new phonetic detail. Both the effects of L2 processing and L2 input on L2 pronunciation have received renewed attention in recent years, primarily because findings for the perception-production relationship in L2 speech learning have been inconsistent (e.g. de Leeuw et al., 2019), and because not all methods for measuring exposure have revealed an effect of L2 input on the development of L2 pronunciation (e.g. Flege and Wayland, 2019). Similarly, the exact relationship between L2 development and L1 speech plasticity remains unclear: although L1 phonetic changes have been observed among language learners due to L2 contact, a process known as ‘phonetic drift’ (Chang, 2012, 2013), whether these changes are environmentally-induced (i.e. due to new ambient linguistic input) or determined by the extent to which similar L2 sounds are actually being acquired, also merits further investigation. By analysing the correlation between L2 pronunciation and L2 perceptual processing, L2 input and L1 speech malleability, this research has broader implications concerning the representations that guide speech production and the effectiveness of different L2 speech models in accounting for language learners’ trajectories in L2 pronunciation development. We will argue that, taken together, results are consistent with the notion that both L1 phonetic and phonological representations determine the pronunciation of L2 vowels, but that the global properties of the entire L1 system and L2 system also interact. These findings are discussed in relation to current theories of L2 speech learning and important considerations for future models are proposed.