In this paper, we report the results of our acceptability judgement survey with 120 subjects to argue for a population split among Japanese speakers regarding the grammatical accessibility of particle stranding ellipsis in the language. This point is supported by the bimodal distribution of the number of speakers who either accepted or rejected this elliptic pattern. The split in question, we argue, reflects ongoing competition between two, or possibly more than two, partially different grammars available to Japanese learners, a natural consequence if the grammars concerned are consistent with the incoming PLD and the narrowly restricted hypothesis space curved out by innate predispositions/UG. We further note that some Japanese speakers may even fluctuate between the two grammars, thereby accounting for their varied, unstable judgments concerning this construction.