Cant

Do you ever get the feeling you're writing in an elaborate academic argot that's not intended to be understood?  Consider the following excerpt from my Statistical NLP homework:

To build my language model, I used a training corpus consisting of approximately seven million words of overheard pillow talk.  After filtering out all the b*d w*rds and tagging the rest for semanticality distortion using the TC algorithm (Tango and Cash, 1989), I proceeded to calculate (by hand, on a napkin) the ex post facto frequentive probability matrices, and from these I constructed a tetragrammaton model with max factors and several fortified fallback positions.  I then improved the model using Sleater-Kinney smoothing to straighten the curves, flatten the hills, and bring the surface to a beautiful glossy finish.  Finally, I evaluated the model's perspicacity on several corpora: the novels of Kilgore Trout, the complete lyrics of all the Cocteau Twins albums, and the contents of my refrigerator (see Photo 1 below).  The results are collected in Table 1.3.2.6.1a, in which the perspicacity values are reported as half-double-decaf-log probabilities with a twist of lemon.  Note that I have treated all negative results as experimental errors (as is standardly assumed) and that all zeroes have been replaced with smiley faces.

I don't really understand it, but I'm pretty sure I'm gonna get an A.

[Now playing in my head: "Evil" by Interpol]