The vocabulary
Theological terms ranked by rate (per 1,000 words) in the selected station. Below the dashed line, every word is rarer than promise in the full corpus.
Theological terms · per 1k · occurrences · songs
All words
| word | per 1k | occurrences | songs | % of songs |
|---|
The absences
Words from the same theological register, and how seldom they occur in the selected station.
The near-zero board (rate per 1k · raw count)
Frequent term · scarce counterpart (rate ratio)
Word lookup
Any word's rate and rank in the selected station.
Word explorer
Type a word to see every song that sings it (in the selected station) and how each song handles trouble. Searching wait against breakthrough and watching them land in different buckets is the point.
Suffering
This music engages suffering at roughly the rate the Psalms do, but it rarely names what the suffering is, and about half the time it answers with a promise that the trouble will lift rather than with worship offered while it stays.
Every song in the selected station placed by what it does with trouble.
Resolution spectrum
The two responses to unreversed trouble
Named or unnamed
Patterns
Counted by occurrence: every time a word is sung, repeats included. Each pie totals only its own slices, not the whole corpus.
Songs
Search and filter the selected station. Click any song for its full classification.
Method
The corpus is 504 songs in rotation on three Christian radio stations, K-LOVE, Air1, and KSBJ, chosen to separate ownership from format. The station toggle scopes most tabs to the songs actually on K-LOVE or Air1; "All 504" is the whole corpus.
Two analytical layers sit behind the site. Local word counting is deterministic and needs no model: every lyric is tokenised, stopwords removed, variants folded, and each word tallied for occurrences and songs. A per-song classification by a language model then records what each song does with trouble. Because the three station subsets are different sizes, per 1,000 words is the only comparable rate, and it is the headline number throughout.
Source and contact
All code and data are public: github.com/ChristianRadioLyrics/christianradiolyrics
Questions or corrections to Zach at: christianradiolyrics@outlook.com
Limits, stated plainly
- The lament classification is a single model pass with no blind human validation and no measured error rate.
- Airplay weighting was available but set aside; monitoring data could not be verified for completeness, so every song counts once.
- Lyrics are crowd-sourced and may contain transcription errors.
- The Psalter comparison (roughly 65 of the 150 psalms are laments) is a common scholarly estimate, not a fixed count.
- This is one corpus at one moment. Nothing here supports a claim about change over time.