A person talks about a familiar face, and their speech works like a mirror of memory. In people with temporal lobe epilepsy, that mirror sometimes trembles: words come more slowly, phrases break off. Doctors hear such complaints, but standard tests, where you have to name a picture, often show that everything is fine.
Researchers recorded stories about famous people from 51 patients and 30 healthy individuals. A neural network turned the sound into text and counted words and pauses. It turned out that the patients' vocabulary is no worse: they use just as many rare words. But their phrases are shorter (5.54 seconds versus 6.59), and pauses are more frequent (10.81 per minute versus 9.29).
Previously, only an experienced doctor could notice such subtle changes in a long conversation. Now they can be measured from a recording of a conversation. A patient talks about familiar faces at home, and a program monitors the pace of speech. The mirror of memory can be hung in every smartphone.
Word frequency was measured on the Zipf scale: a word with frequency 3 occurs about once per million words, and one with frequency 4 about once per ten thousand. Rare words turned out to be a marker of vocabulary.
Not published on arXiv. Here are its page, the PDF and the formalization: you can check it yourself, not only read our account.