Microstructure of white matter association tracts in the temporal lobe predicts the timing of electrophysiological activity and behaviour in picture naming

White matter (WM) microstructure has been mainly associated with accuracy of picture naming. However, little is known about how WM properties relate to the timing of picture naming and its underlying electrophysiological activity. Here, we linked diffusion-weighted, electrophysiological, and behavioural data, using time-based measures to conceptually connect WM deterioration to alterations in conduction speed and precision. Our sample included individuals of varying age, with or without stroke. We correlated fractional anisotropy (FA) of bilateral WM tracts traversing the temporal lobe with mean response times (RTs) and response time variability (coefficient of variation, CV) in picture naming, and with the onset latency of neuronal activity, represented by the electrophysiological P2 component time-locked to picture onset. We selected tracts showing associations with P2 latency and naming performance separately, on which we conducted path analyses examining the relationship between structure, neuronal activity and behaviour. Depending on the tract, higher FA was associated with faster and/or more consistent naming, suggesting that ventral tracts differentially contribute to naming speed and consistency. Importantly, path modelling for the right inferior fronto-occipital fasciculus (IFOF) revealed that FA predicted P2 latency, which in turn predicted RTs, possibly reflecting a structure-activity-behaviour relationship in lexical-semantic processing. P2 source reconstruction demonstrated anatomical proximity to posterior segments of the right IFOF, supporting these findings. Overall, this study provides a proof of concept for modelling structure-activity-behaviour relationships in language research, while highlighting the need for elaboration of such relations at a more mechanistic level.