Like other theories of generative grammar, Optimality Theory (Prince&Smolensky 1993/2004) is much concerned with explaining the limits of variability between human languages. Unlike many other theories, OT has an answer built right into its founding assumptions: prioritization as a mechanism inherently leads to variation, because conflicts may be resolved in various different ways, depending on the assignment of priorities; but at the same time the possibilities of variation are limited because the constraints themselves are held constant across languages. What varies, fundamentally, is not the constraints on language form, but the way that their inherent conflicts are resolved. Because of its particularly clear structure, and because of the explicitness of its commitment to universality, it is possible to ascertain with great exactness the entire range of predictions made by a postulated set of constraints. The manner in which this has generally been pursued involves (1) stating a set of constraints, (2) exhaustively permuting the relative priorities among the constraints, to determine the languages they predict, where by “language” is meant a set of forms, and (3) comparing these predicted languages against what is known about attested human languages. This has been useful and instructive, and has given much information about the empirical force of the approach. But this methodology leaves the internal structure of the individual grammars unexamined. We may learn whether or not a given array of predicted forms appears among known human languages; but we gain little or no insight into how the constraints relate to each other inside the grammar to create the patterns we observe. Gaining this internal knowledge is absolutely crucial if we wish to advance from small-scale models, with a few constraints and a few structures under consideration, to more ambitious theories which embrace a wide diversity of phenomena.. Our project develops methods to determine the internal structure of grammars, in this sense. A two-pronged effort is required. On the one hand, the formal and computational techniques needed to find the grammar-internal structure must exist, or must be brought into existence. On the other, a suitably rich and well-researched empirical domain must be examined as the focus of inquiry, so that the formal methods are given content. The formal side rests on work by the second proposer and his associates (Prince 2002, Prince 2006, Brasoveanu & Prince 2011), which makes it possible for the first time to determine with full exactness the detailed structure of constraint interactions in a grammar. On the basis of this foundational work, we have jointly developed further techniques that allow for rigorous comparison of interaction patterns across grammars. The empirical side of the project is focused on systems of stress patterning (‘stress prosody’) in human languages, specifically the phenomenon of word stress, which is concerned with the rhythmical properties of words, This is an extremely rich area, with hundreds of languages studied and analyzed over the last 30 years. Our analytical goal is to achieve a complete account of the most interesting stress system, which is based on Alber 2005, and to relate it exhaustively to the others in the family, demonstrating how variation in assumptions leads to variation in results. This requires an explicit procedure, currently under development, which extracts the higher-level properties of predicted systems from the fine-grained lower-level information present in the evaluation by the constraints. Though developed in the context of prosody, the procedure itself applies to systems of constraints of any kind, so that our results can be applied to any linguistic domain. As the project evolves, we will examine other typologies, including those currently collected by the research group working at the University of Verona on Romance and Germanic dialects. Importo previsto relativo alle misssioni: 6500 E.