Extreme language in presidential debates: Reagan, Trump and everyone in between

Tyler Schnoebelen
13 min readOct 19, 2016

--

If you follow politics in America even a little bit, you know that Republicans talk a lot about taxes and that Donald Trump loves the word tremendous. But how do these rank relative to each other and to what Democrats (and Hillary Clinton, in particular) tend to talk about? Well, one finding is that over the years, Republican candidates have been even more preoccupied with Hillary Clinton than they have been with Ronald Reagan. Another finding is that the debates for the current election have been ~157% more negative than all previous debates.

This post is partly descriptive: I’m taking all of the debates since 1976 (plus the Nixon-Kennedy debates of 1960) and finding the words and phrases that most distinguish the two major parties’ candidates. You can read nifty work by Burt Monroe, Michael Colaresi and Kevin Quinn if you’re statistically minded or descriptions at The Language Log for easier-to-understand versions. And you may want to check out my earlier posts about the second debate, the vice presidential debate, and/or the first debate. Those use machine learning to find the most Trumpian and Clintonesque moments of the debates.

How negative was Walter Mondale? How extreme was Ronald Reagan?

But this post is also driven by a hypothesis about the kinds of partisan framings that separate the two parties. It’s impossible to step into this without mentioning George Lakoff and his notions of the divide of “the strict father” vs. “the nurturant parent” (I think we can all agree that nurturant is a horrible word).

One way of thinking about the political spectrum is how much you believe things are clear-cut versus ambiguous. I’m going to work to show how this line of thinking best explains what we see in the data. I’ll look at negativity, in particular, but demonstrate how different assumptions change whether the negativity of pre-2016 debates were led by Democrats or Republicans.

The most Republican/Democratic phrases

Let’s kick off with Republican vs. Democratic words. I’m only going to use candidates that have at least 5,000 words. So that means that Geraldine Ferraro and Jack Kemp make it into the data but Scott Walker and Howard Dean do not.

The key words and phrases are not that extreme, although they do pick out a lot of things that elicit love or hate. Certainly, the graph seems to pass the sniff test: it’s government vs. health (care) as the defining party words. Over the years, the Republicans have talked about the federal government 440 times to the Democrats’ 100. Meanwhile, the Democrats have, in their primary debates, presidential and vice presidential debates said universal health care 124 times. The Republicans have said that phrase in debates: never.

The Democrats use a lot more senator, mainly Senator Clinton and Senator Obama in the 2008 primaries and Senator Sanders in this years’ primaries. That’s more of a term of address than a topic per se. But the other names like Bush are probably best thought of as topics. The Democrats have, over the years, said Bush 530 times. The Republicans, though they have had George HW Bush, George W Bush, and Jeb Bush, have only used the word 336 times. The Democrats were not using Bush affectionately.

Hillary Clinton has been referred to by Republicans 366 times since Gary Bauer first invoked her in a Republican debate in Michigan in 2000; Democrats have used her names in debates 40 times total. By contrast, Ronald Reagan was first mentioned in a 1976 debate between Jimmy Carter and Gerald Ford. That first use was Carter’s. Democrats have actually only used Reagan’s first-name-last-name 28 times. Republicans have invoked it 317 times. So yes, whether you use log-odds or straight counts, Republicans have invoked Hillary Clinton more than Ronald Reagan. (And the Democrats have been even more obsessed with Bush.) In other words, topics and people can carry a lot of affect. The words and phrases that distinguish political parties show where the ideological divides are most sharp.

In addition to topics, we do see a few other tendencies that are worth tracking down (later): it’s, they, tell you, to try to, what I, and I think that. But for the moment, I’m going to set those to the side with the proviso that people are often misguided in what they think pronouns indicate about a speaker. Some important correctives on the meaning of I can be found in James Pennebaker’s work.

Let’s switch to the current candidates and distinguish them from one another:

You’ll notice that Trump’s key phrases and words are not really topics or positions except for deals. Among Clinton’s phrases, clearly universal health care and the affordable care (act) are topics and the republicans should probably be considered one, too. Senator Sanders is more typically referential-while-she’s-in-the-same-room-as-he-is.

But instead of topics, the phrases that distinguish them are much more stylistic in nature. Trump’s hyperbolic tremendous and a disaster, his interruptive excuse me and intensifying very, very. We also see his much-written about tendency to think about winners/losers and his slogan about making America great again.

Stylistically, Clinton uses the discourse kind of you know and she often starts her conversational turns with Well, and So I. She has a preoccupation with action both individual and collective (everything I can, we’ve got to). It is hard for me to avoid a rabbit hole of well, but I’ll refer you to this link and mention that when Supreme Court justices speak after someone they are about to disagree with, well is very commonly how they begin.

The charts above are primarily descriptive in nature. That is, they are guided by the relative frequency of words and normalized so they more accurately show what would happen if people actually spoke the same number of words.

Several of Donald Trump’s key phrases were extremes. How unusual is this in the history of US debates? As you’ll start to see in the graphs below, Trump’s negativity is particularly strong and hard to contest. But where we trace it back to historically depends on some important assumptions.

The most negative years and the most negative candidates (okay, it’s “2016” and “Donald Trump”)

I take a bunch of lists of negative English words: Janyce Wiebe and colleagues’ work on subjectivity, Minqing Hu and Bing Liu’s work, the General Inquirer, and AFINN. At a first pass, let’s just look at, per year, how many of the words are negative. I did some light editing to remove the most popular errors like mean and vice which are almost always used as part of I mean and vice president in the debates so shouldn’t count as negative terms.

Okay, that’s almost impossible to read, though basically what it’s showing you is that 2016 is a banner year for negativity. The different word lists disagree about when exactly the other peaks are, but in general, the 1960 Kennedy-Nixon debate looks positively pleasant.

You’ll also notice the peaks in 1976, 1984, 1992, and 2004. These years all share in common the fact that one of the parties has an incumbent. In 1976, Carter was up against sitting president Ford, in 1984, Mondale was going against sitting president Reagan, in 1992, Bill Clinton was up against sitting president George HW Bush, and in 2004, Kerry was up against sitting president George W Bush.

Those first four peaks are not merely challengers against incumbents, but specifically Democratic challengers against Republican incumbents. Could it be the Democrats bringing in all the negativity? Here’s the breakdown of negativity per candidate — I’ve restricted myself to candidates in the peak negativity years (but annotated for when they were talking in other years):

Using word counts (so very frequent words count as many times as the candidates use them)

In other words, this has been a tremendously negative year. Lindsey Graham and Donald Trump lead the way, but you’ll see Bernie Sanders and Martin O’Malley up there, too. Notice that Hillary Clinton is towards the bottom, which makes sense given her focus on positivity — her slogan is Stronger Together, after all.

In terms of the past, Democrat Walter Mondale was more negative than Republican Bob Dole (they faced off in 1976) and Ronald Reagan (they faced off in 1984). Geraldine Ferraro, the Democratic VP pick in 1984, was among the least negative. She faced off with Republican George HW Bush in 1984. He was also one of the least negative. Ross Perot, an independent, and Bill Clinton, a Democrat faced him in 1992 and were both more negative, though Bill Clinton doesn’t break the top ten.

George W Bush was also middle-of-the-pack in negativity. John Kerry, his Democratic opponent in 2004, was fairly high up the negativity list.

But we are using word lists and those are context-insensitive. They have words like catastrophe, abusive, evil, and liar, but these are relatively rare words. These are the more typical words (by both per person word counts and — for the computational among you, tf-idf scores).

  • Graham and Pataki: destroy
  • Trump: bad
  • Sanders and Pence: criminal
  • Cruz and Rubio: illegal
  • O’Malley: crime
  • Kerry: lost and terror
  • Mondale: worse

One way of letting rare — but likely much more truly negative — words pack a greater punch is to shift from adding up total occurrences of negative words and instead counting each individual word form once.

Using word “types”, so each word form only counts one time — giving less weight to the most frequent words (like ‘bad’)

When we do this, you can see that Trump and Graham are still at the top of the negativity charts. Cruz, Sanders, Pataki and Rubio are also still at the top. Because Pence used so much criminal but not a lot of other negative words, he drops very far down the list.

Meanwhile, Reagan jumps up higher than Mondale or Carter. Ford, Dole, and Ferraro stay at the bottom as does Hillary Clinton. Bill Clinton drops below George HW Bush and Ross Perot. George W Bush is still down towards the bottom but John Kerry drops waaaay down when we use word forms instead of total counts.

So notice how different assumptions about “what counts” change the story of “until this year, negativity in debates has been driven by Democratic challenges to Republican incumbents”. If we had different word lists or counted words differently, we’d get rather different results.

A party of extremes and a party of neutrals

Part of the idea behind Wiebe and her colleagues’ subjectivity project is that words are not just negative-positive-neutral but are more-or-less subjective.

The last row of the following chart is about words that aren’t considered subjective. Democrats and Republicans have about as many word forms though Republicans have a lot more counts-per-word-form.

In terms of strongly subjective language, it is the Republicans who lead both in Negative and Positive. The Democrats — consistent with Lakoff’s idea of nurturing parent metaphor instead of the strict father metaphor — have more neutral terms. Trump and Romney are very different Republicans. And humans. Let’s look at the differences they have in strongly subjective negative terms.

Trump has 65 word forms that are strongly subjective negatives and that he’s used 3+ times. They include bad, disaster, rid, hell, mess, badly, horrible, worst, unbelievable, disgrace, abuse, hate, liar, catastrophe, incompetent, disgraceful, irredeemable and ashamed.

We have twice as much data for Mitt Romney so it isn’t surprising that he has more negative, strongly subjective words than Trump: 82 all together. But Romney’s words have a remarkably different flavor: critical, rid, bad, burden, trouble, enormous, unacceptable, disagree, reject, mistake, veto, concerned, terror. Also 6 heck’s and 3 darn’s. He does use words like inexcusable, twisted, treason, and chaos but the extremity of his negativity seems less.

A history of evil

As we saw in the earlier colorful bar charts, changing whether we count word token counts or word forms gives us a different answer to “what was the earliest antecedent to negativity? Was it Walter Mondale or Ronald Reagan? We’re a bit delayed from looking at examples, so let’s head towards them. To get started, let me show you the negative words Mondale and Reagan used in any of the debates they took part in:

  • Both: worse, crime, illegal, lost, worst, destroy, bad, crisis, guilty, guilt, humiliation, kill, terrible, dead, despair, warning
  • Mondale: abuse, madness, desperate, desperately, losing, scandal, terribly
  • Reagan: bankrupt, cheat, fraud, ridiculous, abhorred, abuses, badly, bloody, damage, destructive, dire, evil, horrendous, loss, mad

Notice the relationships between Reagan and Mondale’s words. Reagan has the more emotionally immediate mad while Mondale has the longer, less immediate madness. Reagan deploys mad in an analogy where he’s painting Carter as a mad witch doctor (!):

I sometimes think he’s like the witch doctor that gets mad when a good doctor comes along with a cure that’ll work.

Not only is madness less emotionally direct than the monosyllabic mad, but Mondale is talking about having automated systems that could trigger a nuclear war (hey, btw, ethics is probably the most important thing you can be talking about in terms of artificial intelligence). The subject of the sentence is not any particular person but a general we and it is placed in a the format of a question that is, of course, a rhetorical question not really meant to be answered.

Why don’t we stop this madness now and draw a line and keep the heavens free from war?

Mondale also talks about abuse while Reagan talks about abuses. Presumably multiple instances > a single. Mondale has three uses of abuse and Reagan has one of abuses. Let’s see how things add up. Mondale, first as a VP candidate against Bob Dole where he uses abuse twice, then in a presidential debate with Ronald Reagan:

Even today this administration is fighting all the Watergate reforms, opposed the appointment of a special prosecutor, opposed the reforms that were cried out for adoption following the revelations of the abuse of the CIA and the FBI. And with a record like that, and with all of the abuse of public faith and trust that we’ve been through, surely that, too, is another reason for a new generation of leadership.

When the Republican platform says that from here on out, we’re going to have a religious test for judges before they’re selected for the Federal court, and then Jerry Falwell announces that that means they get at least two Justices of the Supreme Court, I think that’s an abuse of faith in our country.

Reagan in debates with Mondale uses abuses not as a plural noun but as a verb. The lead up to this is actually the rather graphic description of an attack that lead to the law. My gut is that talking about the abuse of a pregnant woman and the death of her fetus is more gut-wrenching and emotional than the kinds of governmental abuses that Mondale talked about. But this is just a gut sense — Mondale was much closer to the deeply disturbing shocks of Watergate that many of us are inured to.

the California State Legislature unanimously passed a law that was signed by the then-Democratic Governor — signed a law that said that any man who so abuses a pregnant woman that he causes the death of her unborn child shall be charged with murder.

In some ways, regardless of these words, Reagan is also using things like abhorred, dire, horrendous and evil. In particular, evil is difficult to argue with. If something is truly evil, it is rarely ethical to just standby.

Evil has been used in the debates 53 times — 70% of the time by Republicans, 20% by Democrats. Ben Carson and John McCain used evil six times each. Martin O’Malley, Rick Santorum, and Barack Obama have used it six times. Tom Brokaw, Rand Paul, Carly Fiorina, Newt Gingrich, and Ron Paul used it twice each. Here’s the first use in a debate (well, in a debate I have a transcript for). It’s from Marvin Kalb, asking Reagan a question in the 1984 debate with Walter Mondale:

Mr. President, you have often described the Soviet Union as a powerful, evil empire intent on world domination. But this year you have said, and I quote. “If they want to keep their Mickey Mouse system, that’s okay with me.” Which is it, Mr. President? Do you want to contain them within their present borders and perhaps try to reestablish detente — or what goes for detente — or do you really want to roll back their empire?

Reagan then replied:

I have said on a number of occasions exactly what I believe about the Soviet Union. I retract nothing that I have said. I believe that many of the things they have done are evil in any concept of morality that we have. But I also recognize that as the two great superpowers in the world, we have to live with each other. And I told Mr. Gromyko we don’t like their system. They don’t like ours. And we’re not going to change their system, and they sure better not try to change ours. But between us, we can either destroy the world or we can save it. And I suggested that, certainly, it was to their common interest, along with ours, to avoid a conflict and to attempt to save the world and remove the nuclear weapons. And I think that perhaps we established a little better understanding.

In most philosophies, a true evil must be opposed. That’s what makes politically performative evil dangerous: to act against something that is NOT evil as if it IS evil leads you down perilous and polarizing paths.

So perhaps we trace Donald Trump’s extreme rhetoric — though he never said evil in the debates, even when others in the Republican primaries were — back to Ronald Reagan. The point, I think, is to see the problems and the possibilities in Reagan’s response.

Reagan walks a tight rope of moral accusation and pragmatic actions. Can you have a “better understanding” with something evil? Should you? And more to the point, do you let your fear of evil lead you and others into evil of your own?

--

--

Tyler Schnoebelen

Linguistics and language, data science and artificial intelligence, UX and design, travel, San Francisco. Want to talk about emoticons, emoji or AI?