More political text analytics
« previous post | next post »
I spent a few minutes this morning getting transcripts for all 12 Republican and all 9 Democratic debates, and over the next few days I'll do some additional Breakfast Experiments™ on the results. One trivial thing is a complete type-token plot, from texts constructed by concatenating all the transcript pieces attributed to each remaining candidate across all the debates:
Unsurprisingly, the trends we saw earlier have continued: Cruz has the highest rate of vocabulary display, and Trump the lowest.
As I've noted before, Trump's low rate of vocabulary display is partly due to his repetitive rhetorical style. Thus his last unscripted contribution to the 3/10/16 debate in Coral Gables started this way:
It depends on what comes up. You never know. It depends on what comes up. Look, look, we had a great president, Ronald Reagan. We had Tip O'Neill, speaker. And what do we do, we take these two men that are very, very different men, they got along, they had relationships, and they got things, and very beautifully. Nobody is complaining about the deals that Ronald Reagan made. And he made it with Tip O'Neill. We need to have people get together and work good deals out, good deals out from our standpoint.
Compare a corresponding passage from Senator Cruz:
Donald, you are welcome to be president of the Smithsonian. You know, there are some in Washington who are having fevered dreams of a brokered convention. They are unhappy with how the people are voting and they want to parachute in their favored Washington candidate to be the nominee. I think that would be an absolute disaster and we need to respect the will of the voters. It's one of the reasons why in the course of this election — listen, everyone up here has worked very hard. But Donald is right, there are only two of us that have a path to winning the nomination, Donald and myself.
There are also differences in the distribution of word frequencies, but that will have to wait for another morning. Meanwhile, here are links to some earlier LLOG coverage of relevant linguistic aspects of the current presidential primary campaigns:
"R2D2", 3/27/2016
"Trump the Thing Explainer?", 3/19/2016
"Trump reviews", 3/17/2016
"Trump's Gettysburg Address", 3/12/2016
"The most Kasichoid, Cruzian, Trumpish, and Rubiositous words", 3/11/2016
"The Trump insult haiku", 2/29/2016
"Bigly", 2/26/2016
"Trump's future conditional head-scratcher", 2/19/2016
"Trump's rhetorical style", 12/26/2015
"Donald Trump's repetitive rhetoric", 12/5/2015
"More Flesch-Kincaid grade-level nonsense", 10/23/2015
"Vocabulary display in the CNN debate", 9/18/2015
"Political pitches", 9/17/2015
"Make America rather formidable again", 9/10/2015
"Political vocabulary display", 9/10/2015
"The most Trumpish (and Bushish) words", 9/5/2015
"More BS from George F. Will", 8/28/2015
"Back to the Bushisms industry?", 8/17/2015
"Phenomenal to the women", 8/11/2015
"Did a blind squirrel happen to find a nut?", 8/8/2015
"Trump's eloquence", 8/5/2015