Source and coverage

Data comes from the Formula Dream F1 Team Radio Archive, a fan-curated and unofficial collection of team radio clips. The full archive holds 39,309 radio clips totaling 797,666 words, spanning eight seasons from 2018 through 2026. The 2022 season is absent from the archive, and 2025 and 2026 coverage is incomplete. Of the full clip count, 37,820 (96.2%) have transcripts; the remaining 1,489 audio-only clips were excluded from all rate calculations.

Preparation

Year and date fields were coerced to integer and datetime types, with parsing errors set to null. Empty text fields were filled with blank strings to allow safe regex matching. Word counts used a regex pattern matching letters, apostrophes and hyphens, correctly handling contractions like “can’t” and hyphenated terms like “pit-wall” while excluding numbers and punctuation.

Profanity detection

Two detection methods were combined. The first, targeting masked or bleeped words, used a regex pattern to catch censored tokens such as “f***” or “s**t,” identifying 1,123 masked tokens that were then mapped to word families through pattern matching. The second used an explicit 40-word lexicon spanning eight languages, covering terms such as “f***,” “s***,” “damn,” “idiot,” “cazzo,” “mierda” and “putain,” which identified 384 explicit swears.

Classification

Detected swears were sorted into four severity tiers: strong (e.g. “f***,” “s***,” “c***”), mild (e.g. “damn,” “hell,” “piss”), insult (e.g. “idiot,” “bastard,” “b****”) and blasphemy (e.g. “jesus,” “christ,” “god”). A combined “vulgar” count — strong, mild and insult tiers — totaled 1,507 instances; blasphemy was tracked separately. Each swear was attributed to either the driver or the pit wall by splitting transcripts into driver and pit-wall segments, which found drivers swearing 3.4 times more often than their race engineers.

Aggregation

Individual swear occurrences were compiled into a long-form table of 1,526 rows with metadata including clip ID, driver, team, token, word family, severity tier and position. Messages were then aggregated to a message level, producing a 1,171-row table flagging each message’s swear count, strong-swear count, masked-token count, and whether it qualified as a “sweary message.” Further aggregation tables broke totals down by driver, team, season, session type and race, computing swears per 1,000 words and the percentage of messages flagged as sweary for each group.

Other reported findings

The archive’s underlying analysis also reports a temporal trend — swearing peaking in the 2020–21 seasons at 2.2 swears per 1,000 words before falling to 1.52 per 1,000 words in 2024 — and a session-type split showing Qualifying running far swearier (3.03 per 1,000 words) than first practice (0.99 per 1,000 words). The per-message data behind these two figures was not available for this analysis, so they are reported here as the archive’s own methodology notes rather than independently verified figures.

What this page uses

The tables above are drawn from two aggregation outputs: a per-driver summary covering all 35 drivers who appear in the driver-attributed radio archive, and a word-family summary covering all detected swear tokens. Per-driver totals used here (37,320 messages, 785,741 words, 1,490 swears) run slightly below the archive-wide totals quoted above, since some transcribed clips are pit-wall-only or could not be attributed to a specific driver and sit outside the by-driver breakdown.

Source: Formula Dream F1 Team Radio Archive (fan-curated, unofficial), 2018–2026 seasons.