An AI scores the GOP Debate

Presidential debates are important. For candidates, it allows a chance to connect with voters and differentiate themselves from a crowded field. For voters, it allows an opportunity to hear from candidates they may not be familiar with and begin to form impressions. Which candidate was the clearest and most confident?

Debates are filled with nuances that may not be at once obvious to viewers. Nuances may include word choice, rate of speech, or clarity.  Voice patterns supply clues as to confidence and authenticity. Artificial Intelligence is adept at revealing those nuances. Many people may be better at processing voice tone, but pitch can be more nuanced.

We use two AI approaches to analyze the August 23 GOP Debate. First, we employ Natural Language Processing, or the analysis of the words and how they are delivered. Second, we utilize Audio Analysis, or the study of voice patterns.  When coupled, these two AI approaches provide subtle but important insights by combining linguistic and vocal analysis. For instance, do both the words spoken and the voice patterns display confidence, or does the voice tone show doubt even if stronger words were used?  For simplicity, six linguistic and six vocal criteria are measured:

Language                     Vocal

AI defines confidence as: 

1. High use of certainty words — shows decisiveness (“I will cut taxes”)
2. Positive sentiment and emotive language — conveys optimism
3. Lower language complexity — aims for clarity
4. More certainty words — (“I will absolutely stand firm on this policy”)
5. Dynamic voice pitch variation — engaging tone
6. Vocal Variation — projects authority
7. Low Timing and Intensity — steady, stable voice
8. Higher Acoustic Features — animated, passionate delivery

The Most Confident Candidates

Haley and Christie appeared most assured and decisive overall based on both linguistic and acoustic metrics. Hutchinson, Pence, and Burgum displayed more signs of uncertainty through their language use and vocal delivery.

The Confident

Nikki Haley scored highest on sentiment showing more positive emotional language. She also had higher use of certainty words (15). She also showed the greatest vocal dynamism, which is associated with a smooth, steady, stable voice.

Chris Christie: Forceful tone but steady and moderate delivery.  More certain words than uncertain ones indicating decisiveness.  He spoke with a low pitch and the lowest tone, attributes that are associated with confidence and command.

The Unsure

Asa Hutchinson: Scored the highest positivity demonstrating positive emotional language.   His low vocal pitch and tone is often perceived as more mature and authoritative.  Most candidates, however, spoke at an 8–9^th grade level Asa spoke at a 12^th grade level.  His message may have been lost in his high language complexity indicating less straightforward communication.  This was a problem as he also spoke the fewest words. Asa also used the most uncertain language.  While his voice tone expressed maturity, it was not backed up by the linguistic approach he took.

Mike Pence:  Lower acoustic features indicating a flatter, less dynamic vocal pitch. His low vocal pitch, low vocal timing variation but higher intensity shows some vocal instability.  This may have resulted from his back and forth with Vivek Ramaswamy.

Doug Burgum: Low vocal pitch suggests smaller vocal range. High vocal timing variation indicates shaky vocal delivery.

Center Stage

Many may consider Vivek Ramaswamy to have done well in the debate. He certainly got the most words in, (More than Tim Scott and Asa Hutchinson combined) but he showed a mixed confidence level. He did use strong modal words like “must” and “will” when listing his principles and beliefs. But some tentative phrases like “I think” and “probably” when discussing nuances.  Vocally he showed the greatest vocal variation and the highest pitch of the male speakers.  While this may be engaging and demonstrate passion, its overuse can detract from the message or come across as insincere.

At center stage, Ron DeSantis really missed an opportunity.  His performance was very middling on the scoring metrics.  He was in the middle of the pack on words spoken and used an equal number of certainty and uncertainty words.  Ron’s vocal timing and intensity were also in the middle of the pack, while his pitch was a bit high.  High pitch is often associated with nervousness or uncertainty.

Combining Natural Language Processing and Audio Analysis allows a comprehensive examination of linguistic and vocal cues sheds light on the contenders’ demeanor. This comprehensive approach underscores the significance of linguistic certainty, positivity, clarity, and vocal dynamism as hallmarks of confidence.

David Pope is the Chief Data Scientist at Speech Craft Analytics.

Image: Dnet