Searches the sermon transcript for patterns that indicate the pastor is referencing the original biblical languages. Matches within 20 characters of each other are de-duplicated so the same reference isn't counted twice.
the Greek/Hebrew/Aramaic word/term/phrase/root/verb/nounin (the) Greek/Hebrew/AramaicGreek/Hebrew/Aramaic: or Greek/Hebrew/Aramaic,translated from/oforiginal Greek/Hebrew/language/textDe-duplication: matches within 20 characters of each other count as one reference.
Two-pronged approach that scans for historical figures and event/term patterns. De-duplicated within 30 characters to avoid double-counting when a name appears near a term (e.g. "Luther's Reformation" counts as 1, not 2).
De-duplication: matches within 30 characters of each other count as one reference.
Each row on the admin page calculates its total independently:
Row Total = min(Max, |Count × Bonus|) × sign(Count × Bonus)
The Total Bonus is the sum of all row totals. This value is added to the PSR score to produce the Total Score:
Total Score = PSR + Total Bonus (clamped 0–100)
The first 3 rows are free-text word searches. Type any word and the system counts exact whole-word matches (case-insensitive) in the transcript. The first row defaults to "Jesus". Special characters in the search word are escaped so punctuation is matched literally.
These rows pull their scores directly from the sermon's AI-generated category scores (the same values shown on the sermon detail page). The count field shows the category score (read-only) and the bonus slider lets you apply a manual adjustment. These are useful when you believe the AI over- or under-scored a particular category.