Abstract: While highly restrictive context can predict a specific word (e.g. Kuperberg & Jaeger, 2016), it is infrequent in natural conditions. A graded prediction of certain features of the upcoming word might be a more robust mechanism (Luke & Christianson, 2016). This idea is not new and has been formulated across different disciplines in a number of ways. A few linguists point that in connected speech (individual sentences and connected texts) each word communicates/limits possibilities for the upcoming word on multiple levels: orthographically, phonologically, semantically, and syntactically (e.g., Jackendoff, 2003; Sag, Wasow, & Bender, 2003). In probabilistic models each word becomes a prior to the upcoming word (e.g., Frisson, 2009; Hale, 2001; Smith & Levy, 2013). In this talk I will show my work in this area, where I use two different paradigms: eye-tracking and EEG. First, I will present evidence from eye-tracking which indicates that morphosyntactic information is processed before the eyes fixate the target word - parafoveally (Stoops & Christianson, 2017; 2019). Second, I will present evidence that preceding context can induce the expectations regarding word order and morphosyntactic case inflections as reflected in the event-related potentials (ELAN and P300) and electrophysiology of neuronal oscillations in the delta, theta, and alpha bands as revealed by time-frequency analyses (Stoops, Ionin, & Garnsey, 2016; 2017; 2019). Finally, I will discuss the opportunities these findings present to specify morphosyntactic processing in the cognitive models of eye-movements and to map the electrophysiological findings to the neuroanatomy of the saccadic circuit.