Software Spots the Spin in Political Speeches
New Scientist (09/17/08)No. 2674, P. 22; Hutson, Stu

Queen’s University researcher David Skillicorn has created an algorithm
that evaluates word usage within the text of a conversation or speech to
ascertain whether a person is being truthful. The program counts usage of
first person nouns, seeks out phrases that offer qualifications or
clarifications of more general statements, and looks for increased rates of
action verbs and negatively charged words, which signal higher levels of
spin.

Skillicorn used the algorithm to study speeches of 2008 presidential
contenders John McCain, Barack Obama, and Hillary Clinton, and determined
that the level of spin in their addresses reflected the occasion.

Voice analysis is another technique for determining spin, and Vox Institute
founder Branka Zei Pollermann uses auditory analysis software to build a
voice profile by mapping seven parameters of a person’s speech and then
comparing the profile with the speaker’s facial expressions by using
researcher Paul Ekman’s Facial Action Coding System as a guide.

Pollermann’s analysis of McCain’s speeches demonstrates that the
candidate’s flat tone and mismatched facial expressions could work against
him, while Obama, who exhibits greater pitch modulation and closer
correlation between speech and facial expressions, is a more politically
astute speaker.

Meanwhile, University of Tokyo researcher Yoshimasa Ohmoto
and colleagues are working on a facial recognition system for robots and
artificial intelligence agents that studies basic eye, nose, and mouth
movements to determine whether a person is lying.