1. I find it counter-intuitive / misleading to have quotations of previous acts of speaking to God coded as if they were in their original context.

I understand why this approach was taken, but wonder if an additional type of speech (e.g. referenced or quoted) that distinguished this from an actual speech act might solve the problem and make the data more useful.
2. In building Searches to see how closely other searches mimicked the interactive, I discovered that the documentation is incomplete as it does not tell us whether or not an element is unique to this data set or mirrors other coding. Examples of non-unique coding appear to be addressee, speaker, places, things, literary type. The example that left me confused was "Theme" ... is it or is it not unique to this dataset?