Sentiment Intensity & Polarity
Two complementary measures of affect, both from SentiMerge-weighted scores per token (Emerson & Declerck 2014, which harmonises several German sentiment lexica by lemma and part of speech). Intensity (SI) uses absolute values — how much emotionally charged language a song mobilises, regardless of direction — and is normalised by token count, so songs of any length are comparable. Polarity (SP) uses signed values — whether net affect skews positive or negative. The analysis extends Schneider (2026). Toggle the views and select an archive below.
Linguistic Insights
Intensity falls; the positive–negative balance holds
Affective intensity declines significantly over the corpus period (p < 2.2e−16), but year explains only about 7% of variance between songs — a real but modest effect. Signed polarity stays close to zero throughout (overall mean SP ≈ -0.098). What changes is the quantity of emotionally charged language, not its direction: recent songs are neither more positive nor more negative, just sparser in affective vocabulary.
What the measure can and cannot see
SI counts only tokens the SentiMerge lexicon recognises; words it does not cover — anglicisms, slang, coinages — score zero. Because Deutschrap, which is lexically distinctive and heavily English-influenced, comes to dominate the recent corpus (see Genres per Year →), part of the measured decline in intensity may reflect lexicon coverage and a shift in genre composition rather than a change in how emotional the songs are. The trend is best read as a property of the corpus as sampled and lexicalised, not a direct measure of songwriters' emotional expression.
Mean Sentiment Intensity (SI) per year · dots = fewer than 5 songs in that year
These analyses are based on corpus data as of July 29, 2026.