Whether this might be challenging or not, I'm keen to study AN nodes for future uses as in motion graphics. Any assistance is appreciated. Thanks in advance. I am not too fussy on font types or choices of words.
The first effect is easily achieved by using an inverse delay falloff with a high back interpolation. The initial transformations of the objects are stored in an ID key, and they are added to a group then sorted by their x location:
The second effect can be achieved by bending the text when the falloff becomes bigger than 1, that is, the back of the interpolation. To get those values, we subtract one and take the maximum of zero to clamp it to positive numbers. Then this output is multiplied by some negative scalar to define the amount of bending. Finally, the Object Attribute Output node is used to write those bend factors, assuming all object have a simple deform modifier with the default name and bend type: