Resonance Explained
Resonance is one of Discorra’s most powerful measures.
It tells you how much two corpora “speak the same language” — where they converge, and where they diverge.
1) What is resonance?
Think of resonance as the shared vibration between two voices.
When corpora use many of the same terms in similar frequencies, they resonate.
When they differ strongly, the resonance score falls.
Resonance measures the degree of overlap between frequency-weighted tokens in two datasets.
2) How is it calculated?
At a high level:
- Frequency weighting
We take each corpus and normalize word frequencies (per 10k tokens).
- Keyness comparison
Z-scores highlight words that are unusually strong in one corpus relative to the other.
- Resonance score
The overlap of high-weight words across corpora is aggregated into a 0–1 scale.
Raw counts are misleading. Z-scores let us normalize for dataset size and highlight relative strength of tokens, not just volume.
3) How to interpret resonance
- ✓
High resonance (0.7–1.0): Corpora share many of the same terms. This suggests alignment in language, tone, or topic.
- ✓
Moderate resonance (0.4–0.7): Some overlap, but also distinct vocabulary. Useful when comparing adjacent domains.
- ✓
Low resonance (0.0–0.4): Strong divergence. Corpora speak very differently; ideal for contrastive insights.
Two corpora can have high resonance but opposite sentiment. Resonance only measures vocabulary overlap, not positivity/negativity.
4) Why it matters
Resonance helps you:
- Spot where brand voice overlaps with competitors
- Identify white space (terms competitors use that you don’t)
- Track shifts in customer discourse over time
- Measure campaign alignment across channels
A brand may find high resonance with customers on “convenience” language, but low resonance on “sustainability.” That’s a signal to adjust messaging.