<p>Tactile sensors that can detect material softness, or elastic modulus, are critical for intelligent robots and embodied neuroprosthetics. Artificial electronic skins (eSkins) that mimic the cutaneous stretchability and mechanosensory apparatus can facilitate human-like touch perception in these applications. Existing devices primarily rely on the piezoresistive mechanism to detect changes in pressure and lateral strain upon contact with a target object. However, resistors are highly susceptible to differences in conductive nanomaterial morphology which compromises sample-to-sample repeatability and introduces large cyclic hysteresis. Furthermore, their fabrication and interconnection complexity hinder the development of scalable, high-density arrays. Here, we overcome these limitations with a stretchable, fully capacitive sensing array that uses rigid islands in a novel architecture to obtain multimodal information. In this configuration, normal pressure is transduced to the capacitive pixels with rigid islands while free-standing pixels increase in capacitance as a function of out-of-plane deformation induced by the softness of the touched object. Electrode dimensions and layout are investigated to determine their effect on the accurate differentiation of material moduli (72 kPa to 1.36 MPa). The sensor output trend is maintained even after a five-fold miniaturization of the array sensing area from a 25 mm to a 5 mm square. Finally, the device is integrated with a dynamic robotic gripper for real-time material classification. Our unique eSkin sensor provides sophisticated feedback while minimizing data acquisition and analysis complexity, which is advantageous for efficient training of future machine learning algorithms.</p>