If you have ever played a 2-player game of Bananagrams, you know how much different it is that a regular game with 4 or 5 players. A 2-player game feels like a marathon and requires a lot of endurance. Playing longer games is a good way to build up and learn to maintain your focus.
For my casual play, the adaptation that I often use is to divide the regular pool of tiles in two, put half of them back in the banana, and play with just the other half. That way, each player winds up with the same number of letters in their grid as in a regular 4-player game (which is, for some reason, my standard). But you can wind up with asymmetric halves, so it's best to switch between the two halves, or else mix them together and split them periodically.
Conversely, 8-player games are like a 100 yard dash: You have to go as fast as you can, and just as coming out of the starting blocks as quickly as possible is critical for winning a short race, a small advantage in the initially drawn set of tiles can make all the difference when playing Bananagrams. Generally this results in more random outcomes than longer matches, but I like to view many-player matches as a challenge to my ability to play fast.
Of course, a big factor in why 8-player games go so fast is that they split the same 144 tiles up such that each player only has to make an 18-tile grid. Contrast that with the 36-tile grids in a typical 4-player game. A solution to this is mixing together two sets of Bananagrams (which explains the introduction of the Bananagrams Jumbo and Double Bananagrams sets which (as I understand it) both consist of two bananas sold together, giving a total of 288 tiles.) I've never tried playing a many-player game with two sets of Bananagrams. I would definitely like to see how different that format feels. I suspect that it would still be a bit faster than a 4-player, 1-banana game.
When it comes to the most fiercely competitive situations, like tournaments, some multi-player games such as Scrabble seem to converge to 2-player games. The argument is that this minimizes the influence of chance and allows the players' skills to maximally determine the outcome of the game. This does not hold for other games, like Settlers of Catan which does not work well with only two players (requiring a minimum of three), in part because player interaction is an important component of gameplay. Consequently it, and other so-called Eurogames, are probably not suitable for Bananagrams-style acceleration (a.k.a., Banananification).
Bananagrams is a wonderfully flexible game that can scale to small or large groups, and the feel of the game can be altered by increasing or decreasing the number of tiles per player. There must be many other ways that the game can be changed a little, giving it a different texture while maintaining its core Bananagrams nature. I hope to find and explore such ways in future posts.