The political circus surrounding the U.S. election has already moved on to something more interesting but I wanted to take a look at last week’s presidential debates from a lexicological standpoint. Full disclosure: I didn’t actually watch the entire debate in real time because I value my sanity. However, I was able to download a good transcript from the folks over at The Rev. For the sake of brevity, I’ll only cover a few high-level thoughts.
First, using a text inspector tool, let’s take a look at the basic structure of the debate:
Speaking time between the two candidates was about even, but this estimate is based on the start of individual utterances and doesn’t count interruptions (of which there were many). I was actually surprised to see how much time the moderator, Chris Wallace, spent talking … although he did need to intervene quite often and most of this probably consisted of the phrase “Mr. President, please.” Trump had more sentences over the course of the debate but his type count — the number of unique words (or types) — was lower.
The average sentence length, the average number syllables per sentence, and average syllables per word, were lower for Trump and higher for Biden and Wallace (whom I’m including just for comparison). Biden fell in the middle in all cases. This seems to go against Trump’s normally convoluted speaking style and may be a factor of the transcription. Still, it suggests a more dynamic way of speaking — shorter, punchier statements — which could be beneficial at public rallies.
The Flesch Reading Ease score — a measure of ease of understanding … higher is easier — suggests that both candidates were speaking at about a 6th grade level during the debate. These scores are considered standard conversational English. Chris Wallace’s language was slightly more difficult to follow but not by much. The Flesch-Kincaid Grade test translates these scores to a U.S. grade level — again with Trump slightly easier to follow than Biden. The Gunning Fog index places the candidate’s speech in the middle school/junior high school range.
Parsing the transcript for individual words reveals a few interesting findings. Among the unique words uttered by each candidate (and not uttered by the other candidate) a few standouts include the following:
It’s interesting to see how Trump personalizes his communication with both his opponent and the moderator (e.g. Joe, Chris) while Biden does not. This hints at a speaking style honed in the business world where usage of a first name can signal an informal setting. Other unique words to Trump encompass a specific reference to actions (e.g. “closed” the country during the virus) or challenges (e.g. “name” one group that does X).
Biden is much more likely to use words like “America” and “American” to engage the audience in discussions of the country itself (e.g. “the American people”). I find it kind of weird that Trump doesn’t use this word at all. Biden also uses a few unique filler phrases (e.g. “here’s” the deal) and got caught up in arguments with Trump around the accuracy of certain facts (e.g. “discredited” reports).
Looking at word usage in situations where both candidates uttered the word at least once, we can see some interesting highlights of common pronouns:
In the chart, highlighted words are those that were used over twice as often by one candidate. Biden used the word “he” to refer to Trump throughout the debate, perhaps reinforcing his desire to connect to the audience. Trump made the debate more confrontational by speaking to Biden in the second person (e.g. “you”) or using his first name as noted above. Trump also has his own filler phrase (e.g. let “me” tell you) and signature words (e.g. “they”) that he uses to create generalized situations or to emphasize a point.
The she/her disparity refers to a segment discussing the qualifications of Judge Amy Coney Barrett. The difference between the possessive pronoun (“her” views) and the subjective pronoun (“she” thinks) is interesting just because of the implied difference between the focus on the individual and a more general discussion of a role in the country’s judiciary.
For filler words:
In addition to a few key filler phrases, Trump and Biden both have several crutch words they use frequently. Trump uses these words to emphasize the end of sentences (e.g. “okay”) and as a universal adjective/adverb (e.g. “very”). Biden’s filler words serve a similar adjective/adverb role (e.g. “totally” discredited).
Other notable words used by Trump included the word “forest” and the number “47.” Both of these represent attempts to inject new political talking points into the debate: “forest” being the latest argument against the realities of climate change (the fires in California being driven by poor forest management instead) and “47” being the number of years that Biden has been in office (expertise is bad!).
Finally, Biden used the word “fact” nearly ten times as often as Trump during the debate. My gut tells me that this approach won’t work on the typical Trump supporter but I admire the Vice President’s attempt to introduce some traditional forensic techniques (reality!) into the discussion.