Whether morpheme-based processing extends to relatively unproductive derived words remains a matter of debate. Although whole-word storage and access has been proposed for some derived words, such as Japanese de-adjectival nominals with the unproductive ("-mi") suffix (e.g., Hagiwara et al. in "Language" 75:739-763, 1999), Clahsen and Ikemoto ("Ment Lex" 7:147-182, 2012) found masked priming from de-adjectival nominals with productive ("-sa") and unproductive (-mi) suffixes to their adjectivally-inflected base morpheme. Using masked and unmasked priming, we examine whether adjectivally-inflected base morpheme primes facilitate the processing of Japanese de-adjectival nominal targets with a productive or unproductive affix, including an orthographic-overlap condition and semantic relatedness measure that Clahsen and Ikemoto (2012) did not include. Our results replicate and extend Clahsen and Ikemoto (2012), revealing significant, statistically-equivalent morphological priming effects for "-sa" and "-mi" affixed targets, independent of orthographic and semantic relatednesss, suggesting that the processing of derived words with the unproductive "-mi" affix makes recourse to morpheme-level representations.