Humans can make an on-the-fly rapid inference when presented with a novel word, using prior knowledge and context. While this is an everyday occurrence necessary for effective communication, the mechanisms of rapid inference and learning during reading are not well understood. One reason for this complexity is that the processes of inference are shrouded by those associated with processing any familiar word. To isolate potential mechanisms of inference alone, we designed a zero-shot learning task where a novel word was presented in a highly constrained context allowing for easy inference of its meaning1. We recorded intracranial depth EEG (iEEG) from 15 human subjects undergoing clinical epilepsy monitoring while they performed the task. We first identified neural correlates of lexical processing independent of context on a fine temporal scale. Thereafter, using word predictions generated by an LLM we found separate processes associated with context-based integration of familiar and novel words. Thus, using temporally and spatially precise iEEG recordings, we identify a hierarchical and distributed system involved in language comprehension and learning. These results provide a theoretical framework to further unravel the mechanisms of rapid linguistic inference through modeling and better tasks. These results could also inform building better artificial networks and neuromorphic chips for few shot learning.