More initial thoughts:
It's laborious to assign weights to words in dictionary, or, perhaps, only a portion of it may be assigned integer values based on raw data collected. Either ways, the curve is automatically obtained. A thesaurus may be useful to grade relative weights based on word intensity level.
Opening and closing angles for sentences shouldn't be a hassle as we can quantify all the 26 characters. Of course, the opening alphabet also need to be combined with one or more succeeding alphabets of the first word in a sentence in order to determine the opening angle of the sentence projectile, the sentence curve. Similarly for the last word where the last alphabet may be taken together with the preceding alphabets to get landing angle.
So, total possible permutations for opening angles
= (26P1) * (26P1) * (26P1) * (26P1)
= (26!/25!) * (26!/25!) * (26!/25!) * (26!/25!)
= 26 * 26 * 26 * 26 = 456,796 opening angles
Similarly, 456,976 closing angles.
The solutions would include stings like 'aaaa' and 'bbbb' which we don't ignore. The first would be a scream and the second could be a character stammering or pausing.
I have ignored case sensitivity as that matters only when a sentence is shouted.
This quantification is humanely possible by a medium-sized research team.
I have limited to three succeeding and preceding alphabets going by the social media philosophy that there are degrees of connections till the third degree.
We can also give weights to punctuation marks. Eg: comma is a pause, semicolon is a stronger pause, and a period the strongest pause. They mostly impact the +x axis, the sentence throw and it's length. We can say that they are vector displacements. As though you pressed spacebar on keyboard.
Thus the computer graphs be generated.
Oh, and since each word in a sentence has its own takeoff and landing angles, the sentence curve is made of mini projectiles, their shapes also determined by the word buoyancies along the +y and -y axis.
When bestselling novels are fed into this software, it may provide useful information as to how important scenes shape up. I lay emphasis on scenes as they are the building blocks of a novel.
So, retrospectively, we can see the shapes of important scenes, which primarily influence the plot.
The graphs are two dimensional but can be made three dimensional if page numbers are fed along the +z axis, into the screen. Thus we know a desired plot shape is obtained on average at say, page 180, and so on.