We used EEG hyper-scanning to investigate speech processing of real-time continuous speech during conversation between non-native adults. In this study, seven dyads (n=14) of adult non-native English speakers completed a ‘spot-the-difference' picture game in both their first language (L1) and English (L2). Participants were seated in separate rooms and conversed using stand-mounted microphones while listening to each other through EEG insert headphones. EEG was recorded simultaneously from both participants using 32-channel systems with bilateral mastoids. Conversations were recorded and speech amplitude envelopes of both takers were extracted and time-aligned to the EEG. We applied a multivariate temporal response function (mTRF) approach to estimate neural tracking at the acoustic level (e.g., speech envelope) for both perceived and produced speech in the speakers’ first and second languages. Preliminary results demonstrated that we could obtain reliable auditory neural tracking using this paradigm. Further work will involve analysing the transcripts to derive lexical measures (e.g. lexical predictability and frequency) and correlating them with EEG in L1 and L2 conversations. We will extend these findings by testing additional real-world conditions, such as speech in noise.