Inside of a piezo buzzer is a little unimorph/bimorph disk which converts voltage to mechanical movement. To function as a buzzer, you just feed it an ac signal at its resonant frequency. However, John Alexander, the author of this project: STM Project, patented a design that uses the unimorph disk inside of a cheap buzzer as a fine-adjustment mechanism (US5866902).
Basically, if you apply a DC voltage to the across the ceramic part of the disk and the brass part, the disk will flex a very tiny amount (around 100nm/V).
My question is this: Piezo electric elements are given a polarity by applying a strong electric field that aligns the dipole moments inside of the piezo material. Thus, I cannot a priori know if the piezo disk that I remove from a buzzer will extend or contract when I apply a positive voltage bias. Keeping in mind that an application of 15V would yield a flex of around 1um, what is a clever way to test a disk for its polarity?