Cyclicity generalized

December 10, 2025
Online (zoom)
Starts at 18:00 (Moscow time)

 

Donca Steriade

Massachusetts Institute of Technology

 

Supplementary reading

 

Abstract

This talk presents evidence for a modification of the phonological cycle, which allows it to extend naturally to the analysis of new, almost-but-not-quite cyclic patterns of stem syncretism.

The cycle is a mechanism that generates phonological similarities between certain pairs of lexically related expressions. In its original conception (Chomsky & Halle 1968), this is an inheritance device designed for the case in which a complex form (the outer cycle or Derivative) inherits its properties from an expression it immediately contains (the inner cycle, or Base). The formalization of inheritance is implemented in different ways with rules vs. constraints, but the restriction of cyclic inheritance to pairs of nested forms is shared by all frameworks. I refer to this property as C(yclic)-Containment: the Base always contains the Derivative as its immediate subconstituent. 

In addition to C-Containment, cyclic inheritance has the properties of Directionality (aka Base Priority: Benua 1997); that of being able to transmit predictable properties, rule-governed properties from Base to Derivative (Steriade 2000); and that of Recursiveness, the fact that the Derivative in one pair can become the Base in another (Chomsky, Halle and Lukoff 1956).

The talk outlines evidence that has accumulated over the last 20 years showing that all these properties of the cycle, except C-Containment, are also present in other phenomena that promote stem syncretism:

i. Pseudo-cyclic inheritance: stem identity between derivatives and related words not contained within them (Steriade 2008, Steriade and Yanovich 2015, Steriade 2025).

ii. Paradigm uniformity: stem similarity between different cells in an inflectional paradigm. (Albright 2011, Kraska Szlenk 2007, Steriade 2000)

iii. Morphomes: stem identity between lexically related forms that are expected to be distinct but to exceed a certain degree of similarity (Aronoff 1994; Steriade 2016; 2023)

Like the cycle, these patterns are directional, recursive and can refer to, and transmit, predictable properties. But they are not bound by C-Containment, which is restricted to the cycle proper.

This question I address is how to analyze the cycle and the related patterns in (i-iii) in ways that do justice to the properties they all share, while restricting C-Containment to the cycle.

I will argue that the solution is to rethink C-Containment, which restricts cyclic inheritance to nested pairs of expressions, as a violable constraint. This allows us to provide parallel analyses for all stem identity and stem distinctness phenomena, including those in (i-iii). In addition, this allows unified analyses of standard cyclic and pseudo-cyclic inheritance (item (i)) within a single system: English stress will be shown to be a case in point.