Cluster methods can also be distinguished from one another based on their primary function. For instance, some methods measure the level of connectivity between several words of concepts, while other methods determine the extent to which specific words or concepts pull other words and concepts towards them. This “pull” might be consider a form of semantic centripetal force. Methods also might be used to identify distribution of words or concepts, or the density of a pack of words and concepts.
As one example, a doctoral student of mine used a powerful digitally based analytic tool to identify themes found in a series of movies that tend to cluster together. She was able to move beyond the obvious thematic overlaps in the movies to much subtler connections found between the use of words (dialogue) found in the movie scripts. My own identification of six distinctive cultures in postsecondary education (Bergquist, 1993; Bergquist and Pawlak, 2008) is based on my analysis of word clusters to be found in the interviews I conducted over many years with members of collegiate institutions throughout North America. I engaged a similar analysis in exploring the sub-cultures of those engaged in professional coaching (Bergquist and Brock, 2008).
It is in the use of semantic patterning that something tends to emerge from nothingness. We find only randomness in the words we hear, read and use everyday. Nothingness is to be found in what we don’t say; yet, a profound something is to be found in the sequencing of silence with words we use and the words we receive from other people. There is no such thing as a “failure to communicate” (to use a phrase from the movie, Cool Hand Luke). We are conveying something when we choose not to speak or “sit back” and are not allowed to speak.
Furthermore, even among the words that we speak and write, there are “cracks” that appear to be “saying” nothing, but actually “say” a great deal. With a focus on semantic patterning, these “cracks” are found to reveal powerful relationships that exists between words, and between words and nonwords (gestures, tones of voice, silence, etc.). My doctoral student, for instance, found a strong gender bias underlying the way characters in her movies dealt with martial conflicts and tragedy in their lives. This bias was not found in the scripts or explicit intensions of the screen writers, directors or producers. Yet, these biases were in plain, painful sight when the cluster analysis was conducted and semantic patterning was revealed.
I reach for the caveat once again. Words exist as utterances, but do not actually exist as “things” that somehow relate to one another. One can’t “correlate” the relationship between words, unless numbers are assigned to the words. There certainly is no way to describe or somehow graphically-portray the relationship between words and concepts. As already mentioned, the most important feature to be found in the clustering of words exists in the cracks that appears between the words—these are the differences between the words, the surprising connection between and parallels to be drawn between the words, or the closeness (density) that exists between a group of words. As in the case of paradigms and models, semantic patterning and cluster analysis might actually consist of nothing—at least nothing we can hold dear to our hearts.
