First I took the transcriptions of more than one hundred Obama speeches from the website [link] Then I counted all words using an adapted version of the Snowball stemmer for Processing. I also picked up an Obama photo using Google Image to find a good one.
Finally I use my image pattern searcher algorithm to fit all the words in the image. In this case the more times the word appears in the speeches the bigger it is (proportionally).
Here you have an example of the 20 more often words in the speech: can,785, american,683 , people,648, one,619, time,561, years,548, country,521, work,481, make,477, need,454, just,440, know,415, america,399, go,384, come,356, care,355, nation,352, now,345, like,343, right,332.