Very interesting article!
One thing I wonder, did you take into consideration the ambiguity of words and names in Trump’s and Clinton’s speeches?
For example when Trump mentions Clinton in a sentence, but he is speaking about Hillary’s husband Bill. From the context, a person can easily distinguish, but if you only look at the words, the dimension of meaning is lost. Unless of course you assume that whenever Clinton is mentioned, it refers to Hillary Clinton, but then your analysis will be wrong in some cases.
There are many other interesting points to analyze once you actually consider the meaning of names. One example: how often do the candidates mention leaders of others countries, and which (again taking into consideration ambiguity).
If this sounds useful to you, you can actually achieve this kind of analysis with our entity linking methods. Our Natural Language Understanding API, which identifies all names in the speech, links it to the correct person or location (represented by the correct Wikipedia article), and categorizes them (i.e. Hillary Clinton is a democratic politician), is one of them :)