Prediction during language comprehension has been found for high-cloze words in constraining sentences (e.g. Federmeier et al., 2007) and for parallelism (e.g. either…or) in sentence structure (Staub & Clifton, Jr., 2006; Warren, Dickey, & Lei, 2015), with both leading to faster reading times when the predicted word/syntactic structure is encountered versus an unexpected alternative.  The present pair of studies investigates if individuals adapt these particular lexical and syntactic prior probabilities depending on the context in which comprehension occurs.  In a self-paced reading paradigm, participants read sentences from Federmeier et al. and Staub & Clifton within 1) an informative/predictable context in which filler sentences informed the reader that the prior probabilities would be helpful during comprehension, and 2) an uninformative/unpredictable context in which filler sentences did not support utilization of the prior probabilities for comprehension.  Reaction time at the critical and post-critical regions was compared across contexts, as well as compared across the two studies in which the order of the contexts was switched.  Preliminary analyses reveal that, while lexical and syntactic prediction are robust effects during language comprehension, evidence of adaptation of these prior probabilities is more elusive.  Implications of these results will be discussed and a future study will be proposed.