In standard Optimality Theoretic analyses of weight-sensitive stress, no consideration is given to the
possibility that languages may reposition the edges of the prosodic word in order to avoid violations of
constraints related to syllable weight. I propose here that this type of constraint interaction does occur.
Past work has noted that in some cases, the prosodic word seems to slightly misalign with the
morphosyntactic word, due to phononological pressures. If this is true, we might expect – given that word-level
stress placement (among other phenomena involving the prosodic word) is often sensitive to syllable
weight – that in some cases, prosodic word edges can shift in service of syllable-weight-related constraints.
I propose that this underexplored prediction is borne out in Karuk (isolate, California) and Majhi Punjabi
(Indo-Aryan). Previous accounts of the stress systems of these languages have resorted to novel footing
mechanisms (Karuk: Sandy 2017; Majhi Punjabi: Dhillon 2010). With adjustable prosodic word edges, such
special mechanisms can be avoided.
Section 2 analyzes the placement of stress on heavy-adjacent syllables in Karuk. I propose that the
optimal prosodic word ends in a heavy syllable, and that the optimal position for stress is the penultimate
syllable of the prosodic word. This leads to placement of stress on the syllable preceding the rightmost
heavy syllable. Section 3 analyzes a pattern of mixed stress attraction/repulsion by superheavy
syllables in Majhi Punjabi. I propose that a constraint penalizes unstressed superheavy syllables within the
prosodic word. Violations of this constraint are avoided either by placing stress on a superheavy syllable
(stress attraction) or by excluding a superheavy syllable from the prosodic word (stress repulsion).
Section 4 discusses some facts that suggest that prosodic constituents are created early, in the morphology-phonology
interface module (following e.g. Lee and Selkirk 2022). Section 5 concludes.