The NYTimes Measures the Words They Used at the Conventions

In today's New York Times there is an interesting graphic comparison of the frequency of words used by speakers at the recent American political conventions. (You will need to subscribe.)

Speeches are an interesting sort of public relations output; we can measure the words used, but speech can be so much more expressive than the printed word.

Some of the results we probably would have expected, for instance that the Democrats used the word "change" three times as much as the Republicans did, and that the Republicans used the word "god" twice as much as the Democrats. Or that Democrats (eager to focus on the issues) used "jobs," "economy," and "healthcare" twice as much as the Republicans, and that the Republicans (more eager, perhaps, to focus on the candidates than the issues) used "character" six times as much as the Democrats, and "vice president" twice as much.

Other results are more difficult to interpret. For instance, the Democrats used their opponent's name three times as much as the Republicans used their opponent's name. Interesting stuff.

We asked Peter Kowalski, KDPaine and Partners' Director of Research Strategy, for his comments:

[This kind of research] does bring to the surface different attempts at agenda setting, and efforts in framing the election / providing orienting cues to the public about how to think about the party/ election.

How is this applicable to interested, savvy PR people? If you work for a non-profit, analysis of this type can help to identify how your opponents have framed their side of the debate in the past, in order to better plan effective countermeasures and to test effective message strategies.

If you work for a business, analysis of competitor press releases, analyst quotes, and industry coverage can help to provide some competitive intelligence, and perhaps predict how the public will think about an industry in otherwise uncertain times.

Outside of formative research, this sort of word or phrase counting can be used in an evaluative sense when one is trying to shift how the media frames an issue, the attributes of an issue, or how a product, brand or company is defined (e.g. when Facebook made the shift from a site for college students to a social utility for everyone).

There is definitely some cool work that employs this. What's most important, from a research perspective, is to define well in advance what conclusions can be drawn from this data. For example, from this work we can't understand how each side is framing our current economic issues, or their solutions. All we know is that they are mentioning the economy.

Thank you, Peter. --WTP

Comments

[CML]

Is there an easy and free way to systematically collect this type of information? (i.e. word or phrase counting)

[Katie Paine]

CML, there really isn't anything that is both easy and free. Radian6 is easy, but not free. Technorati, Google and IceRocket are all free, but not nearly as easy.