Physical interactions between bubbles and crystals affect gas migration and may play a major role in eruption dynamics of crystal-rich magmas. Strombolian eruptions represent an end member for bubble-crystal interactions, in which large bubbles (significantly larger than the crystal size) rise through a crystal-rich near-surface magma. Indeed, volcanoes that produce Strombolian eruptions often generate ejecta with > 30 vol% (often > 45 vol%) average crystallinity. At Stromboli Volcano, Italy, average crystallinity can reach 55 vol%, which is approaching the eruptibility limit for magmas. At such high crystallinities the solids interact mechanically with each other and with bubbles. This complex rheology complicates the two-phase (liquid-gas) slug flow model often applied to Strombolian eruptions. To examine the effect of crystals on bubble rise, we performed analogue experiments in which large bubbles rise in a vertical tube filled with silicone oil and polypropylene particles. The particles have a slightly lower density than the oil, and therefore form a layer of oil + particles at the upper surface. We varied surface pressure, particle volume fraction, length of the particle-bearing cap, and bubble size to examine the ways in which these parameters influence Strombolian-type eruptions. We show that in experiments, suspended solids begin to affect bubble rise dynamics at particle volume fractions as low as 30 vol% (or, when divided by the random close packing value, a normalized particle fraction φ=0.64). Bubbles in experiments with higher particle contents deform as they rise and burst through a small aperture, generating surface fountains that begin abruptly and decay slowly, and longer-lasting acoustic signals of lower amplitude than in particle-poor experiments. Particle fractions > 38 vol% (φ>0.80) generated strong deformations on fast-expanding bubbles that applied a high stress on the cap, but they trapped bubbles that were less overpressured. Qualitatively, the gas release behavior observed in particle-rich experiments is consistent with observations of Strombolian eruptions. Moreover, we estimate that the observed crystallinity of pyroclasts at Stromboli volcano represents φ>0.8. From this we suggest a “weak plug” model for Strombolian eruptions that evolves towards a low-viscosity equivalent of Vulcanian-style plug failure with a more crystalline, stronger, and less permeable plug. Importantly, this model allows the rise of several bubbles in the conduit at the same time and suggests that longer-lasting, more pulsatory and complex eruptions may reveal an increase in near-surface crystallinity, shedding some light on changing conduit conditions that could help determine the different gas rise regimes involved in passive degassing, puffing, and different expressions of Strombolian explosions.