I made some adjustments to data processing based on your valuable comments to my previous post on this subject. For example, one of you mentioned counting the ‘$’ symbol in addition to the word ‘dollar(s)’ under the keyword ‘dollar’, as both would be spoken the same way. I also converted numbers into words, i.e. 1 = one, 32 = thirty two.
Most words here are shown counts for exactly as they were spoken, while others (like verbs) are grouped by all forms. For instance, “destroy” includes destroy, destroys, destroyed, and destroying; “lie” includes lie, lies (also as nouns), lied, and lying; “win” includes win, wins, won, and winning.
Since both candidates theoretically had equal speaking time, I chose to show the actual word frequency counts rather than percentages of total words spoken by each candidate, rankings, or other forms of data manipulation. The idea is that a word spoken a certain number of times by a candidate and heard by the audience is, in my opinion, just as meaningful as percentages, and it’s easier to interpret.
Side notes:
1. The transcripts may contain spelling mistakes and/or variations of words, so the counts may not be exact and could vary. For example, “healthcare” might sometimes appear as two words, “health care.” I believe the margin of error should be similar for both candidates, but I didn’t manually check every word or read the entire transcript. 2. These are just raw counts. To infer true meaning, context is necessary, so please interpret with care.
At first glance the biggest thing I can guess is that Trump just talked a lot more
2 comments
* **Data source:** June 2024 debate – [CNN debate transcript](https://edition.cnn.com/2024/06/27/politics/read-biden-trump-debate-rush-transcript/index.html), September 2024 debate – [ABC debate transcript](https://abcnews.go.com/Politics/harris-trump-presidential-debate-transcript/story?id=113560542)
* **Data processing:** Python
* **Data visualization:** Excel
I made some adjustments to data processing based on your valuable comments to my previous post on this subject. For example, one of you mentioned counting the ‘$’ symbol in addition to the word ‘dollar(s)’ under the keyword ‘dollar’, as both would be spoken the same way. I also converted numbers into words, i.e. 1 = one, 32 = thirty two.
Most words here are shown counts for exactly as they were spoken, while others (like verbs) are grouped by all forms. For instance, “destroy” includes destroy, destroys, destroyed, and destroying; “lie” includes lie, lies (also as nouns), lied, and lying; “win” includes win, wins, won, and winning.
Since both candidates theoretically had equal speaking time, I chose to show the actual word frequency counts rather than percentages of total words spoken by each candidate, rankings, or other forms of data manipulation. The idea is that a word spoken a certain number of times by a candidate and heard by the audience is, in my opinion, just as meaningful as percentages, and it’s easier to interpret.
Side notes:
1. The transcripts may contain spelling mistakes and/or variations of words, so the counts may not be exact and could vary. For example, “healthcare” might sometimes appear as two words, “health care.” I believe the margin of error should be similar for both candidates, but I didn’t manually check every word or read the entire transcript.
2. These are just raw counts. To infer true meaning, context is necessary, so please interpret with care.
At first glance the biggest thing I can guess is that Trump just talked a lot more