This paper investigates the nature of second language (L2) morphological variability (e.g., Hawkins and Chan 1997; Tsimpli and Dimitrakopoulou 2007; Hazdenar and Schwartz 1997; Lardiere 1998a, 1998b, 2005, 2007, 2009; McCarthy 2007, 2008), more specifically the feature calculus in SLA. The processing of uGender on past participles in L2 French serves as a test of the hypothesis that a universal parser acts as a learning mechanism—that is, the source of re-assembly of feature matrices. If learners are checking features and relying on morphological underspecification, a cost of feature checking or matching under insertion is expected (Dekydtspotter and Renaud 2009). Four groups of participants—French natives (n=11), second-semester (n=25), fourth-semester (n=12), and advanced (n=12) American learners of French—read 28 experimental items in a self-paced moving window experiment. Items consisted of a context and a subsequent segmented sentence with matched or mismatched past participle gender (masculine vs. feminine). Grammaticality judgments and reading times (RTs) on the past participle were analyzed. L2 learners' processing profile suggests that L2 grammar acquisition is mediated by a universal parser, whereas judgments (i.e., for those learners at a lower proficiency level) appear to lag behind. The processing data is suggestive of feature valuation, morphological defaults, and spell-out desiderata, which support a feature re-assembly account (e.g., Lardiere 2009).