Part-of-Speech Distribution

Proportional distribution of 15 POS categories across all tagged word tokens, based on STTS (Stuttgart-Tübingen Tagset) annotations. Raw STTS tags are grouped into major categories; punctuation tokens are excluded. Use the selector to compare the full corpus against individual archives.

Linguistic Insights

Nominal versus verbal style as a register indicator

The ratio of nouns to verbs is one of the most robust indicators distinguishing written from spoken-style registers: informational, edited prose is noun-heavy, while spontaneous and spoken-style language is verb-heavy. Song lyrics as a whole sit closer to the verbal end than edited prose, though genre-internal variation is wide: Ulla Meinecke's and Dota's archives lean most verbal, while HipHop Songs — driven by dense name-dropping and brand references — tilts furthest toward the nominal end, reflecting rap's descriptive, reference-heavy lyricism rather than its spoken-word delivery. The adjectives point the same way: those that pack description into the noun phrase are outnumbered nearly two to one by predicative and adverbial ones, reversing the proportion found in edited written German. Corpus-wide values for both ratios and related measures are listed in General Statistics → Ratio Parameters.

What these categories split — and what they merge

The inventory charted here is finer than STTS in one place and coarser in another. Finer: interjections (ITJ: yeah, ey, oh, na) and onomatopoeia (ONO: bam, boom, brr, tatütata) form a layer almost absent from prose — little propositional content, but central to the sound and delivery of song, the one expressing emotion and direct contact, the other adding sonic iconicity and rhythmic texture. Coarser: ADJ merges the STTS tags ADJA (attributive) and ADJD (predicative and adverbial), so the corpus's most pronounced adjectival asymmetry cannot surface in this chart at all. Both distinctions have fuzzy edges — brr may be read as interjection or onomatopoeia, and ADJD shades into ADV for items like wirklich or einfach. Interjection rates and the attributive/predicative finding are discussed in General Statistics → Notable Observations.

Full Corpus (all archives)

Run generate_word_pos.R to populate this chart.

Centre: total tagged word tokens · Segments: proportions of 15 POS categories (STTS-based, punctuation excluded)

Category Key

NN — Common Nouns NE — Proper Nouns VV — Full Verbs VA — Auxiliary & Modal Verbs ADJ — Adjectives ADV — Adverbs ART — Articles PRON — Pronouns PREP — Prepositions KON — Coord. Conjunctions KOUS — Subord. Conjunctions PTK — Particles CARD — Cardinal Numbers ITJ/ONO — Interjections & Onomatopoeia Other — Other / unclassified

These analyses are based on corpus data as of July 29, 2026.