Contemporary Christian radio · 504 songs · ~178,000 words · one moment, 2026

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.

A descriptive word study of the songs in rotation on three Christian stations. It reports what was measured. The conclusions are yours.

The vocabulary

The theological terms of the corpus, ranked by how often they are sung. Below the dashed line, every word appears less often than the word promise.

Theological terms · occurrences · songs (of 504)

    All words in the corpus

    word occurrences songs % of songs per 1k

    The absences

    Words from the same theological register, and how seldom they occur in the corpus.

    The near-zero board

    Frequent term · scarce counterpart

    Word lookup

    Any word in the corpus — how often it is sung, in how many songs, and where it ranks among all 3,823.

    Suffering

    Every song placed by what it does with trouble — not whether trouble appears, but where the song leaves it.

    Resolution spectrum · 504 songs, left to right

    The two responses to unreversed trouble

    Named or unnamed

    Songs

    Search and filter all 504. Click any song for its full classification. Filters combine across groups; within a group, any selected value matches.

    Method

    The corpus is 504 songs in rotation on three Christian radio stations — K-LOVE, Air1, and KSBJ — chosen to separate ownership from format rather than to sample any single brand. The song list was cross-checked against five independent 2026 chart sources; lyrics were drawn from an open, licensed lyrics database.

    Two analytical layers sit behind the site. The first is local word counting: every lyric tokenised on one machine, stopwords removed, and each remaining word tallied for total occurrences and for how many songs contain it. This layer is deterministic and needs no trust in any model — it is the basis of tabs 01–03.

    The second is a per-song classification by a language model, which read each song and recorded what it does with trouble and how it addresses its listener. This layer underlies tabs 04–05.

    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 — third-party 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.

    Source and contact

    All code and data behind this site are public: github.com/ChristianRadioLyrics/christianradiolyrics

    Questions or corrections: christianradiolyrics@outlook.com

    What this measures is what these songs say about trouble, and what they promise about how it ends. Nothing else.