The proper treatment of NON-DERIVED ENVIRONMENT BLOCKING (NDEB), also known as the Derived Environment Constraint, has long been the subject of debate by phonologists. Past approaches include the Strict Cycle Condition (Mascaró 1976), the Elsewhere Condition (Kiparsky 1982) and underspecification (Kiparsky 1993). However, since the introduction of Optimality Theory (McCarthy & Prince 1993, 1994, Prince & Smolensky 1993), phonologists have tried to model NDEB in terms of parameterised constraints (e.g. Burzio 1997) or constraint conjunction (e.g. Łubowicz 1998).

In this paper I present a case of NDEB found in the stress patterns of Tohono O'odham words. Secondary stress is assigned to all odd-numbered syllables in derived words, but is blocked on word-final odd- numbered syllables in underived words. I claim that all the presented facts about Tohono O'odham stress can be accounted for in terms of co- phonologies (cf. Orgun 1996, Inkelas et al. 1997, Inkelas 1998). By showing the intricate interaction between, on the one hand, stress assignment to latent vowels and, on the other, their behaviour with respect to perfective truncation, I argue that Tohono O'odham stress can be viewed as being assigned ‘cyclically’ and also as exhibiting the effect of bracket erasure. These facts, as I will show, are captured naturally by the co-phonology model. This co-phonology analysis is contrasted with the mono-stratal, non-constituency-based optimality-theoretic account argued for in Fitzgerald (1996, 1997). It is demonstrated that the co- phonology analysis yields a simpler and more explanatory account of the Tohono O'odham facts than Fitzgerald's account.

I begin this paper with an illustration of the stress patterns of both underived and derived forms in §2. I will then provide a co-phonology account for NDEB in §3. An alternative analysis is considered in §3.5. In §4 I illustrate the interaction between stress, latent vowels and perfective truncation, and finally the formal analysis is presented in §5.