A MaxEnt learner for super-additive counting cumulativity

Whereas most previous studies on (super-)gang effects examined cases where two weaker constraints jointly beat another stronger constraint (Albright 2012; Shih 2017; Breiss & Albright 2022), this paper addresses gang that arise from multiple violations of a single constraint, which Jäger Rosenbach (2006) referred to as counting cumulativity. The super-additive version cumulativity is the focus paper; not only overpower violation but also surpass mere multiplication severity its violation. I report natural language examples morphophonological alternation in compound suppressed by existence marked segments manner: laryngeally consonants Korean tensification and nasals Japanese Rendaku. Using these test cases, argues types super-additivity cannot be entirely captured traditional MaxEnt grammar; instead, modified model proposed, degree penalty scaled up number violations, through power function. This provides computational implementation proposed learns necessary parameters given quantitative data. A series learning simulations show learner able detect find appropriate exponent values for those, correctly capturing probability distributions input