What this shows
Quantitative analysis of prosodic features across 4.3 million verse lines in the Finnic runosong corpus. Runosongs (regilaul/runolaulut) traditionally follow the Kalevala meter: trochaic tetrameter with characteristic alliteration and synonymous parallelism between consecutive lines.
Verse Length
- Word count and character count distributions per verse line.
- Character count serves as an approximate syllable proxy (Finnish/Estonian avg ~4 chars/syllable).
Alliteration
- Strong: two or more content words sharing the same first letter. Weak: content words starting with different vowels (all vowels alliterate). Combined rate weights weak at 30%. Function words (pronouns, particles) are excluded.
- Shows which letters alliterate most and the most common alliterating word pairs.
Lexical Repetition
- Measured as Jaccard word overlap between consecutive verse lines: |A∩B| / |A∪B|.
- Score 0.0 = no shared words, 1.0 = identical lines (likely refrains).
- Note: This measures lexical overlap, not true runosong parallelism. In Kalevala-meter poetry, parallelism is synonymous substitution — consecutive lines express the same idea using different poetic synonyms (e.g. neiu/piiga, mets/padrik). True parallelism detection would require semantic analysis.