In looking at any picture, there are an infinite number of dimensions by which to analyze it. Recently, research at the University of Texas at Austin has come up with a technique to mathematically determine the dimensions that most people see when looking at a specific image. The method, tentatively called Meaning Extraction, is actually fairly simple.
Initially, several hundred people are asked to describe the picture -- much the same way you did. Afterwards, the most common words that people used were determined. For each person's essay, we then counted the number of times that they used each of these common words. We then used a method called factor analysis (principle axis, varimax rotation, based on words used by at least 5% of the essays -- approximately 120 words). This method produces groups of words that naturally hang together.
One cluster of words, for example, includes ka, right, left, side, oza, read, blue, facing, quencher, and ribbon. A second factor included yellow, banner, writing, blue, white, letters, green, corner, paragraph. Yet another was made up of wall, shadow, light, upper, behind, sitting, portion, level, background, and table. Note how each of these factors is drawing on a particular meaning cluster.
We are now applying the Meaning Extraction method to other pictures, people's descriptions of their feelings of emotions, Shakepeare's plays, the ways women talk about their breast cancer, etc. Indeed, the method doesn't require our ability to actually read or even understand the text. For example, we are able to do the same analyses even if the text is in Spanish, German, or some other language. The computer simply clusters words together on a mathematical basis.
So where did we come up with the "psychological meaning" of these clusters? OK, we just made that part up. Not randomly, of course, but drawing on what we are finding in some of our other projects. If you have any thoughts or suggestions about the feedback you received, please email us.
Much of the research we are conducting with language is based on a computer text analysis program called Linguistic Inquiry and Word Count, or LIWC, developed by James W. Pennebaker, Roger J. Booth, and Martha E. Francis. To learn more about LIWC, go to the LIWC website.
If you would like to read some of the professional papers dealing with language and text analysis, check out our reprint website.
Any other questions or concerns should be sent to:
James W. Pennebaker
Department of Psychology
The University of Texas at Austin
Austin, TX 78712 USA
Pennebaker@mail.utexas.edu