When we are standing up, gravity pulls blood down which is detected by "baroreceptors" in the circulatory system. The circulatory system compensates by increasing the heard rate, so that blood pressure in the upper body is maintained. When the baroreceptors register such a blood pressure drop, the locus coeruleus neurons become more active. The locus coeruleus responds to stress by increasing release of norepinephrine (a.k.a., noradrenaline). This has a number of effects, including increasing heart rate and blood flow and boosting motivation, thinking, and alertness. But researchers have found that increases in norepinephrine coincide with a slight slowing of certain kinds of thinking.
The thesis is that "insight problems", the kind where your brain kind of wanders and does background processing, and then suddenly presents a solution in an "A-ha!" moment, happen faster when you have lower norepinephrine levels. A simple "A-ha"-type problem to test, and one which is very popular in psychological research, is anagramming.
In one experiment [M.P. Walker, C. Liston, J.A. Hobson, R. Stickgold, "Cognitive flexibility across the sleep-wake cycle: REM-sleep enhancement of anagram problem solving", Cognitive Brain Research, Volume 14, p.317-324 (2002)], researchers woke people up at different stages of sleep and asked them to solve anagrams. They found that subjects who had just been in REM sleep were better at anagramming than subjects who had been in some other sleep stage. And REM sleep has a lower locus coeruleus activity level.
In another experiment, [D. M. Lipnicki and D. G. Byrne, Cognitive Brain Research, Volume 24, p.719-722 (2005)], researchers had subjects solve 5-letter anagrams (like unscrambling DEFSU to get FUSED) and perform arithmetic problems (73 - 58 + 19 - 26 = ?), while standing up and then while lying down. When standing up, average anagramming time was 29.4s +/- 6.3s. When lying down, average anagramming time was 26.3s +/- 5.4s. The difference in anagramming times (3.1 seconds) is smaller than the standard deviation of the data, but the statistical analysis suggests that the error bars are not that important, allowing us to conclude that the average person in the study anagrammed 10% faster while lying down.
A similar test found that standing is correlated with people solving simple arithmetic problems a few percent faster, but it's such a small difference that it doesn't seem statistically significant.
Summarizing: When you stand up, your body tends to become more alert and less relaxed, and you may anagram slightly slower as a consequence.
Caveats: This study was performed on randomly selected subjects with no proclivity for word games. When given 32 5-letter anagrams to solve, and 45 seconds to solve each in, the average number of solutions was 8 or 9 (whether sitting or standing). A typical Bananagrams player would do significantly better, I would be willing to bet.
I wonder if the cause and effect might flow the other way: if lower blood pressure allows one to anagram faster, can focusing on anagramming cause blood pressure to be temporarily lowered? Or will it just cause one to lie down?
Anagramming is only one of the skills used in playing word games, of course. The kind of concentration necessary to play Bananagrams fast most likely requires the mind to not be in a relaxed state. (And what I'd really like to see is a study on playing Bananagrams itself). Still, whenever the professional Bananagrams league launches, one of the first drugs they should consider regulating is beta-andrenoceptor antagonist propranolol.
D. M. Lipnicki and D. G. Byrne, "Thinking on your back: Solving anagrams faster when supine than when standing", Cognitive Brain Research, Volume 24, p.719-722 (2005).
The full paper can actually (to my complete surprise) be bought through Amazon, or if you have access to it through your library, you can get the paper from here.