Stan Zompí – University of Potsdam

(*)ABA effects and their theoretical implications (advanced)

Despite having long been regarded as theoretically uninteresting idiosyncrasies, suppletive stem alternations have recently been found to obey systematic *ABA generalizations: for certain triples of contexts ⟨C₁, C₂, C₃⟩, C₁ and C₃ never share the same stem allomorph to the exclusion of C₂ (Bobaljik 2012, Moskal 2018, Smith et al. 2019, and much subsequent work). Classical Distributed Morphology (DM) accounts, relying on Underspecification and the Elsewhere Principle, explain such generalizations by positing featural subset relations among the contexts involved (usually C₁ ⊂ C₂ ⊂ C₃, though we will see that what’s needed is weaker).

In this course, we’ll begin by surveying these classical DM accounts and the independent evidence we occasionally find for featural subset relations among contexts. Once we’ve familiarized ourselves with how these accounts handle simple *ABA generalizations in a single “paradigm column” (e.g. ⟨NOM.SG.FEM, ACC.SG.FEM, DAT.SG.FEM⟩), we’ll turn to the key empirical challenge: the same accounts stumble when we try to extend them to paradigms crossing two or more inflectional dimensions (e.g. case × gender).
After experimenting with some unsuccessful fixes (such as impoverishment or disjunctive contexts), we will explore more radical departures from the classical DM model—in particular, recent implementations of Nanosyntax (e.g. Caha 2023) and violable constraint–based optimization models (e.g. Zompì 2023, Zompì & Sun 2024).