How we work
We only publish research built on data you can download and recompute. These rules apply to every study.
Code computes every number
Every number in the text and on the charts is produced by a program from the published data. An author — human or AI — cannot type a figure by hand: the site build checks this and refuses to publish an article that contains one.
Open data and methodology
Each study ships with CSV tables holding the same numbers as the charts, plus the word lists used for counting. Charts can be embedded — the code is under each one.
We never republish lyrics
Lyrics belong to their authors. We use them for counting only and never publish them in full or in part: only numbers leave our pipeline — how many songs contain a word or a theme.
How we count words
- Words are lemmatized: “loving”, “loved” and “love” count as one word.
- A word counts once per song: a chorus sung eight times doesn’t make its words the chart’s top words.
- Function words — pronouns, prepositions, conjunctions, particles, interjections — are not counted. The most common light verbs (get, go, know, want) are counted but excluded from word rankings.
- A song’s language is detected from its words, not from the artist’s country.
- There is no lemmatizer for Kazakh, so Kazakh words are compared as written and themes are matched by word stems.
Words behind each theme
Ambiguous words are excluded: “sir” does not count as a mention of god, and “to look” does not count as eyes.
love
- English
- love, lover
- Russian
- любовь, любить, любимый
- Portuguese
- amor, amar
- German
- liebe, lieben
- Spanish
- amor, amar
- Kazakh (word stems)
- махаббат, сүй
night
- English
- night, tonight
- Russian
- ночь, ночной
- Portuguese
- noite
- German
- nacht
- Spanish
- noche
- Kazakh (word stems)
- түн
heart
- English
- heart
- Russian
- сердце
- Portuguese
- coração
- German
- herz
- Spanish
- corazón
- Kazakh (word stems)
- жүрек
money
- English
- money, cash, dollar
- Russian
- деньги, бабки, кэш, лавэ, доллар
- Portuguese
- dinheiro, grana
- German
- geld, cash
- Spanish
- dinero
- Kazakh (word stems)
- ақша
dance
- English
- dance, dancing
- Russian
- танцевать, танец, танцы, танцпол
- Portuguese
- dançar, dança
- German
- tanzen, tanz
- Spanish
- bailar, baile
- Kazakh (word stems)
- биле
tears
- English
- cry, crying
- Russian
- слеза, плакать
- Portuguese
- chorar, lágrima
- German
- weinen, träne
- Spanish
- llorar, lágrima
- Kazakh (word stems)
- жыла
god
- English
- god, jesus
- Russian
- бог, господь
- Portuguese
- deus, jesus
- German
- gott
- Spanish
- dios
- Kazakh (word stems)
- құдай, алла
eyes
- English
- eye
- Russian
- глаз, взгляд
- Portuguese
- olho
- German
- auge, blick
- Spanish
- ojo, mirada
- Kazakh (word stems)
- көз
Limitations
Every source has blind spots, and each study names them. For example, the Shazam chart shows which songs people identify, not how much they stream them. And not every song has lyrics in the open database: songs without lyrics are left out of word counts, and their share is reported.
Corrections
When we find a mistake, we fix the article and say so at its end: what was wrong, what changed and when. Earlier versions of the data stay in the repository history.