Though less elegant than Jamais Vu, I believe what you (and we all) experience is Temporary Cognitive Overload. Terribly oversimplifying, I'll use imprecise language in favor of concepts, and deal only with reading. I bold not to shout but to present bites. (TL:DR just read bold.)
Cognitive load theory and schema theory go hand in hand in learning (theory). Schemas are frameworks of information (like a steel-framed skyscraper in your mind); they start as very basic ("This is a cell") and become more complex and facile ("NADH-Q oxidoreductase, Q-cytochrome c oxidoreductase, and cytochrome c oxidase are mitochondrial transmembranous enzyme complexes responsible for oxidative phosphorylation, blah balh".) They allow (and form) Long Term Memory. We need a framework ("cell") into which we can stick a fact before we can remember it for more than a very few minutes. The more we know about something (the better our schemas are), the more easily we learn. Working Memory (WM) allows us to process what we read and place it into a schema so that we can remember it. We have limited Working Memory (processing ability) available to us at any given time. Efficient processing => Long Term Memory (LTM).
Inefficient Processing => What Did I Just Read? (I know I read it, I know it was in a language I understand, I understood it, but I can't remember what the hell it said.) IP /= schema identification /= LTM.
Efficient processing => OK, That Makes Sense; What's Next? (This relates to things I know; how does it relate to things I'm about to be exposed to?) EP => schema identification => LTM.
Cognitive Load takes up processing speed (reducing WM). If cognitive load is great enough, all WM is used up, and we will be unable to identify/form a schema. There are several types of Cognitive load: intrinsic (how complex the information is), extrinsic/ineffective (a bunch of things including distractions, emotionally demanding states [e.g. stress], and especially the way in which material is presented, e.g. inducing splitting of attention, etc.) and germane (what's left over to actually form schemas). They are (kind of) additive. Good schemas reduce cognitive load (increasing WM).
If you are reading at your limit of WM, one final additional 'load' will make you unable to remember what you have just/will immediately read. Because Cognitive Overload does not disappear immediately upon reduction of load (e.g. you are alarmed that you have not remembered what you read, so you decrease attention splits, and commit to rereading with intent), you need a few moments to experience reduced load before you regain WM. That's why your second reading, if immediate, might not sink in, whereas if you got up, sipped water, and sat down again, you might have enough recovery time to regain WM. If you're not suddenly anxious, which would decrease WM. Etc.
Just a quick aside, if 100 people write which or deer or else or other easy word you're familiar with 30 times in one minute, ~70% will begin to doubt that it is a real word. This is Jamais Vu on a cognitive overload scale!
Answer: it's called Temporary Cognitive Overload.
Schema theory in about 40 easy steps.
There are no quick and dirty cognitive theory papers. Wiki will have to do.