Poking at this dissertation (Continuum and Computational Modeling of Flexoelectricity by Sheng Mao):

https://repository.upenn.edu/server/api/core/bitstreams/3bc2...

> Flexoelectricity, ever since its discovery, has been regarded as an alternative of piezoelectricity at small scales. In fact, as early as the 1960s, Koehler et al. (1962), Turch´anyi, G. et al. (1973), Whitworth (1975) found that edge dislocations in centrosymmetric materials, such as sodium chloride, carry charge. Later, Perenko & Whitworth (1983) extended the observation to another kind of centrosymmetric material, ice. Piezoelectricity vanishes in these materials, therefore cannot be the source. Instead, a “pseudo-piezoelectric” effect was postulated by Evtushenko et al. (1987) for an explanation, which was later shown to be a result of flexoelectricity Mao & Purohit (2015).

Emphasis added.

So I think this was known but not fully understood by the time Perenko & Whitworth published Electric currents associated with dislocation motion in ice in 1983?

https://scholar.google.com/scholar?cluster=69134483967663101...

> In this paper we describe [an] experiment in which a small current is observed due to the movement of dislocations during plastic deformation [of ice].

The same authors of the paper TFA discusses published a preprint in 2022, which could also be what you're thinking of: https://arxiv.org/abs/2212.00323