You can read more analysis at the link above, but what do you see in what we have excerpted?
To begin with, an initial keyword search turned up a significant list of words that were statistically more likely to appear in one candidate’s corpus over the other.
Keywords more likely to appear in Clinton’s corpus include women, families, economy, together, work, American (and America), future, rights, create, men, growth, equal, fair, deserve, global, challenges, threats, opportunity, Israeli, help, stronger, students, income, invest, raising, everyone, God, children, abortion, responsibility, and vigorous.
Keywords in Trump’s corpus that were statistically more likely to appear when compared to Clinton’s were going, very, Hillary, I, they, are, it, I’m, don’t, great, China, really, billion, nobody, Mexico, immigration, disaster, Islamic, nice, mean, folks, me, somebody, people, ok, tremendous, problem, thought, totally, radical, anybody, got, amazing, border, rigged, foreign, money, worst, literally, trillions, terrorism, corrupt, deleted, destroyed, killing, oppressive, bigger, nobody, is, sad,and tired.
On a grammatical level, this cursory list suggests that Clinton is more likely to use keywords that are substantive and concrete. She mentions men, women, children, Americans, students, and teachers, while Trump uses more pronouns like I, they, me, anybody, somebody, nobody, people and folks.
Clinton also uses stronger verbs like deserve, create, and invest, while Trump usesgoing, are, got, and is.
Clinton talks about growth, responsibility, challenges and threats, while Trump uses the words great, bigger, problems, and worst.
When it comes to Trump’s rhetoric, what is perhaps most striking is the frequency with which he references himself. He says I, me, or my 850 times in these seven speeches. (He says I 700 times, me94 times, my 56 times, mine 5 times, and myself 2 times out of the total 25,722 words in the corpus.) What this means is that 3.3% of his words are self-references, which is a remarkably high figure by the standards of any typical corpus.
By way of comparison, Clinton says I 360 times, me 36 times, and my 52 times out of the total 23,089 words, bringing the total percentage of explicit self-references to 1.9%.