I have already given the basic explanation of the method, so this is a heads up for some basic troubleshooting. Kenneth Burke's students had two major challenges with this method:
1. "Alice and the EAT ME cake" syndrome: This is where you just connect all words into a huge index which then becomes the whole book, since of course in a text all words are connected in some way. The solution is to focus on key terms, which generally are to be found at key points in a plot, turning points in an argument, or may also be featured at the beginning and the end. You can limit yourself to only operational synonyms, but even with that "there must be some waste motion here" where one feels a bit lost until one starts discovering the patterns of the text. Yet a lot of Burke's students attested that there does come a moment where the text and its structure unfolds to the critic in ways they may never have experienced before.
2. The problem of bias: One has to be open to "let the text speak" to a certain extent, rather than just seeking the bare minimum of support one needs to confirm a preconceived notion. One can speculate, but those speculations should be based on solid textual evidence. If one is going beyond that, one should let the reader of the analysis know it. Start by just observing, patiently collecting more and more equations, until one has built up a large but manageable index of equations. Burke required all his students to "make clear all elements of inference or interpretation." One can start with equations or key terms first. Either one can work well. Starting with key terms goes faster, starting with equations is more thorough.