Translated from German. Kafka's clinical, matter-of-fact tone still scored zero — VEYL saw right through the register to the specific human beneath.
VEYL scored the full novella at 0% — no ambiguity, clearly human. Going in, we predicted this might land mid-range: Kafka's clinical, matter-of-fact narration (describing surreal events in businesslike sentences) is the kind of register AI language models are trained to produce. Add the translation-from-German factor, which tends to smooth out idiomatic quirks, and 25-40% seemed plausible. But VEYL's feature stack picked up on what actually mattered: the surreal specificity (an insect with "stiff arched segments," "pitifully thin" legs), the strange character voice (Gregor cataloging his own body), and the unusual emotional stance (a man treating his own transformation as a scheduling problem). Translated prose can still carry the fingerprint of a specific human author — and Kafka's does.
A short excerpt — VEYL scored the first ~60,000 words of the full text.
Curious what your own writing scores?
VEYL scans free up to 15,000 words per month. No credit card, no extras — just paste the text.
Try VEYL free →