News and Tribune

June 7, 2013

STAWAR: The governor’s speech

Local columnist

In his recently published book,  “The Secret Life of Pronouns,” University of Texas  social psychologist James W. Pennebaker   describes how the frequency of the words, phrases and punctuation marks we use, can reveal our inner feelings, self-concept,  social intelligence, and even mental health.  

 Pennebaker’s work focuses on what he calls “function words.” These include pronouns, articles, prepositions and other seemingly insignificant words which are often more revealing than the intended meaning of what we say. Function words account for only a fraction of our vocabularies, but comprise about 60 percent of the language we use. 

People are extraordinarily inaccurate in estimating their own and other people’s word frequencies so in order to objectively study these patterns, Pennebaker and his associates developed a software application called the Linguistic Inquiry and Word Count (LIWC). The LIWC calculates  how frequently people use various categories of words. Researchers  have examined a wide range of texts, using the LIWC, including e-mails, speeches, tweets, historical  documents, college essays, and even suicide notes. You can use a lite version of the LIWC to analyze tweets  at www.analyzewords. com, by simply entering  a Twitter handle.

Much of what the LIWC reveals is counter-intuitive. For example, President Obama was criticized by several commentators  for overusing the pronoun “I” in his speeches. After  a careful  analysis of press conference  transcripts, Pennebaker,  concluded that, Obama actually had the  lowest  use of “I” words  of any modern president. While Obama’s speech contained  2.6 percent “I” words, George W. Bush averaged 3.4 percent, Clinton had 4 percent, and George  H. Bush’s speech yielded  5.2 percent. Pennebaker notes that Obama’s lack of “I” words, however,  does not reflect humility, but rather signals self-confidence, emotional distance, and possible rigidity. According to Pennebaker, commentators were wrong about, not only how many self-references the President made, but also about what his pronoun use signified.     

Another of Pennebaker’s discoveries is that people readily accommodate other people’s linguistic styles. He calls this “Language Style Matching (LSM).” You can check out how well your style matches that of others at  

To further put the LIWC program through its paces, I tried it out on three speeches made by Indiana’s new governor, Mike Pence. I used a 2010 speech on the American presidency he gave to Hillsdale College in Michigan, along with the governor’s  2013 inaugural speech and State of the State address.

My contrary wife Diane pointed out that professional speechwriters often write these addresses, therefore, any analysis must take that into account. So in deference to her argument, my dubious interpretations, refer only to the text of the speech and not necessarily the speaker.

All three of Pence’s speeches demonstrated very similar language style. According to Pennebaker’s LSM application, the language style matching scores ranged from “above average” to “very much above average.”  

The 70 LIWC factor scores for all three speeches  were compared to the 2007 LIWC norms. Words with the highest and lowest frequencies were identified. The governor’s  speeches show a very high use of  “we” words, which is consistent with being confident, self-assured, but also  perhaps a bit distant. In regard to content words, the highest frequencies were seen for words relating to money, achievement and work. None of this is very surprising, given the governor’s stated focus on jobs, taxes and the economy. 

Words related to inhibition, such as “block”, “constrain” and “stop” were also relatively high in frequency, which may reflect   his traditionally conservative philosophy in regard to government. For example, on his first day in office the governor put a halt to all new state regulations to encourage job creation.

Religious words were next highest in frequency, followed by words related to death. The religious references are also understandable. The governor once defined himself  succinctly as, “a Christian, a conservative, and a Republican, in that specific order.” Surprisingly, however, death-associated words were found at  twice the expected frequency.      

Within the actual text the words “die”, “bury”, “grave”, “murder”, “corpse”, “blood-soaked” and “souls” were found. All of these words, however, came exclusively from the Hillsdale speech, in which the governor made many historical references that dealt with themes of warfare, strife and sacrifice and they account for high frequency of death associated words.

Overall there was also a high frequency of people-related words, which most political speeches are likely to have. The governor’s speeches contained a lot of large words and quite a few articles, which is typical for higher status educated males.  

At the other end of the spectrum, the governor’s speeches had very few words that signaled assent, such as “yes”, “agree”  and “OK.” This is consistent with the large number of inhibition words. You can’t agree all that much, when you’re inhibiting  things. There was a very infrequent use of “I” words, which fits in with the  high “we” word usage. There were no sexual words present, which is probably a very prudent policy for political speeches, as candidates in the last election learned.

There were few words associated with friends (such as “friend”, “buddy” and “neighbor”), which would be expected in political speech. The lack suggests some distancing and perhaps excessive formality.

Feeling words were also relatively low, suggesting an emphasis on rational rather than emotionalism. There was a low frequency of words relating to the past, which I found unusual, considering the historical features of the Hillsdale speech. This may reflect a purposeful attempt to maintain attention on listeners’ current concerns.  

Words associated with the body were also used infrequently, which may be related to the overall dearth of feeling words.  The data also suggest openness to auditory information, given the low frequency of both “seeing” and “feeling” words, while  “hearing” words were closer to the norm. 

Finally, the governor’s speeches used relatively few words associated with “home.” Like the lack of “friend” words this might also reflect formality instead of folksiness , and perhaps an emphasis on work and business, as opposed to,  domestic themes.  

When comparing the governor’s speeches to President Obama’s latest inaugural  and state of the union speeches, surprisingly the inaugural addresses matched language style  at the “far above average” level and the “State” speeches matched at the “above average” level.  Even the Hillsdale speech, which criticized President Obama’s word choice, ironically   matched  the president’s inaugural and State of the Union addresses the most, at the “far above average” level.    

Comparing the combined Pence speeches with the combined Obama speeches, there was significant overlap. Both had very few “I” words, “sexual” words, “body” words, “feeling” words, or words associated with the past. Both sets of speeches had high frequencies of “we” words, big words, inhibition words, and words related to “money”, “people” and “achievement.” The governor’s speeches, however, also contained a high frequency of words associated with “work”, “religion” and “death” while the president’s speeches  were characterized by a high number of  words relating to “certainty”, “causality” and the future.  

Pennebaker says that linguistic style is as unique as a fingerprint. Although Gov. Pence’s and President Obama’s speeches both contained strong signature features, I was struck by the many similarities. Of course, despite big differences in political philosophy, comparable contexts (such as inaugurations or outlining legislative agendas) may engender comparable styles. 

Just to be fair I also did a LIWC analysis of this column. Oddly enough this column also ends up looking very similar in style to  the speeches discussed. This may be because it cites so many of the words found in them. Also keep in mind the analysis may reveal as much about my editor, Diane, as it does about me.