Vowels Distribution
Each song contributes five data points – one per base vowel (a, e, i, o, u, counted case-insensitively) – plotted against the song's total base-vowel count. The chart reveals whether the frequency hierarchy of vowels is stable across individual texts and how much variability exists from song to song. The German umlauts (ä, ö, ü) are not included here because their much lower frequency would place them in an indistinct band near the x-axis; for the complete vowel inventory see Vowel Proportions, and for the overall vowel-to-consonant ratio across archives see Consonant-to-Vowel Ratio.
Linguistic Insights
Vowel hierarchy robustness
The five trend lines – one per base vowel – run approximately linearly from the origin, stacked vertically in the order e > i > a > u > o. That this hierarchy, well-established for German prose, replicates in song lyrics is itself a finding: register-specific constraints such as rhyme and metre do not measurably distort relative vowel frequencies. The narrow scatter around each trend line further shows that the slope – i.e. relative vowel density – does not drift systematically with text length, confirming the hierarchy at the level of individual songs rather than merely in the corpus aggregate. Note that elision forms such as geh' or hab' drop the vowel letter and will shift it's position slightly downward.
Vowel-specific outliers as style markers
A song's residual distance from its vowel's trend line measures how strongly that text deviates from the expected vowel density. Songs with an unusually high i-share may reflect phonaesthetic strategies exploiting high-vowel brightness, or a lexical bias toward frequent function words (ich, mit, in); outliers toward o and u suggest deliberate use of dark vowel colouring, a technique studied under the heading of Vokalinstrumentierung. Systematic residual analysis thus offers a quantitative entry point into phonostylistics: which authors or genres show consistent light- or dark-vowel profiles?
Full Corpus (all archives)
x-axis: total vowel tokens per song (a + e + i + o + u, case-insensitive) · y-axis: count of the individual vowel · Songs above 1,500 (x) or 700 (y) are not shown (<1% of songs)
These analyses are based on corpus data as of July 29, 2026.