IMPRINT vocabulary

Baseline divergence

Baseline divergence is the largest component of the IMPRINT Drift Score, at 40 percent. It measures how far a calibration's vocabulary richness and average sentence length sit from the same measures in your baseline, as a proportion of the baseline value.

Written by Suman Debnath, creator of IMPRINTLast updated 5 September 2026
IMPRINT’s own term. This is product vocabulary, not established science. It is defined here so the measurement can be checked rather than taken on trust — the arithmetic is on the methodology page.

In more detail

Divergence is relative rather than absolute: the difference between the calibration value and the baseline average is divided by the baseline average, so the signal means the same thing for a writer of long, dense sentences and one of short, plain ones.

The two sub-signals are weighted equally and combined into a 0–100 scale capped at 100. Vocabulary richness is type-token ratio — unique words divided by total words. Average sentence length is words divided by sentence count.

Two properties are worth knowing before you read the number. The measure is absolute, so writing more richly than your baseline registers as divergence exactly as writing less richly does. And type-token ratio falls as text gets longer, so a longer calibration than baseline can register divergence created by length rather than by any change in thinking. Both are stated in the methodology's limitations.