Assessment of real-world speech input has proven valuable for understanding child language acquisition, and has potential for understanding plasticity later in life (e.g., second language learning, hearing aids and cochlear implants, language disorders). That being said, real environments are noisy, making it difficult to conduct acoustical analyses on typical recordings with one or two microphones, and existing wearable recorders are mostly proprietary or non-programmable. We've recently developed a programmable microphone array recorder (4-16 microphones), with the intention of making it open source. The aim is to use microphone arrays and beamforming to better isolate speakers from background noise. We are also in the process of adding dry-electrode EEG in order to enable measures of attention, allowing us to assess what a listener is actively processing rather than measuring all ambiant sound. At present, our hardware and basic device-level programming for audio recording are in place, and we are beginning to trial the devices and develop post-recording processing pipelines. We've tested an early version of the device on adult native Japanese speakers in London, and have found that those who have more exposure to English in their daily life are better at understanding British-accented speech in noise in laboratory tests.