Prospects of Inflectional Morphology in Harmonic Serialism

这个项目研究了一种循环的、基于优化的方法来处理屈折形态学的前景，这种方法依赖于调和序列主义，一种标准并行优化理论的衍生替代方法。为此，该项目追求两个相互关联的目标：第一个目标是证实和声序列主义是一种可行的屈折形态学方法，覆盖与其他模型（如分布式形态学或范式功能形态学）大致相同的基础。鉴于和声序列主义在音系学和句法学方面都取得了成功，建立该模型作为（屈折）形态学的工作理论开辟了一个单一的，统一的方法来处理所有基于形式的语法成分的可能性，一个有可能调和两个广泛采用但似乎不兼容的方法，最低限度的句法和优化理论音系学在一个连贯的，高度限制性的框架。在我最近的专著《调和序列中的屈折形态学》中提出的方法（春分，谢菲尔德）主要是为了提供一个基本的概念证明，通过显示如何一些众所周知的现象，（词缀顺序，扩展指数，析取阻塞，非局部词干同形，和 * 阿坝模式）可以解释，有时直接，但更经常是全新的方式，which哪一个would seem似乎be experienced经验or conceptually概念superior优越in at least至少some of the cases案件considered考虑.然而，这显然只是第一步：为了达到第一个目标，需要对更多语言的范式进行系统的研究，并推导出指数序、选言阻塞和扩展指数效应;第二个更深远的目标是试图表明，该方法可以为现有形态学理论的一些难以解决的问题提供新的、令人信服的解决方案，包括：指数下降，依赖性，范式缺口，形态运动，不连续出血，和学习算法的规格不足。谐波序列主义方法的属性，使它适合于这些现象的新观点，而其他成熟的方法则不是，它是独一无二的，它结合了一个衍生的，循环的方法来构建结构与优化理论的方法来优化。前一个属性意味着，在屈折形态学的决定可能是短视的（并可能最终导致不透明，在Kiparsky的意义上），和派生的中间阶段可能是至关重要的，以确定最终输出形式的属性。后一个属性使得它可以容纳违反和排名约束的证据，如修复或最后手段，未标记的出现，阴谋，约束简单性和通过重新排名的参数化。在实践中，这个项目的两个主要目标是紧密相关的：通过关注第二个目标，第一个目标也将自动得到解决。

This project investigates the prospects of a cyclic, optimization-based approach to inflectional morphology that relies on Harmonic Serialism, a derivational alternative to Standard Parallel Optimality Theory.To this end, the project pursues two interrelated goals: The first goal is to substantiate Harmonic Serialism as a viable approach to inflectional morphology, covering roughly the same ground as other models like Distributed Morphology or Paradigm Function Morphology. Given that Harmonic Serialism has been successfully pursued in both phonology and syntax, establishing the model as a working theory of (inflectional) morphology opens up the possibility of a single, unified approach to all form-based components of grammar, one which has the potential to reconcile the two widely adopted but seemingly incompatible approaches of minimalist syntax andoptimality-theoretic phonology in a coherent, highly restrictive framework. The approach laid out in my recent monograph ``Inflectional Morphology in Harmonic Serialism'' (Equinox, Sheffield) is primarily designed to provide a basic proof of concept, by showing how somewell-known phenomena (affix order, extended exponence, disjunctive blocking, non-local stem allomorphy, and *ABA patterns) can be accounted for, in sometimes straightforward but more often radically new ways, which would seem to be empirically or conceptually superiorin at least some of the cases considered. However, this is clearly only the very first step: To reach this first goal, many more paradigms from many more languages need to be systematicallyinvestigated, and exponent order, disjunctive blocking and Extended exponence effects derived.The second, more far-reaching goal is to try to show that the Approach can offer new and convincing solutions to some recalcitrant Problems for existing morphological theories, in the areas of impoverishment, exponent drop, deponency, paradigm gaps, morphological movement,discontinuous bleeding, and learning algorithms for underspecification. The property of the harmonic serialist Approach that makes it amenable to new perspectives on these phenomena whereother, well-established approaches are not is that it is unique in combining a derivational, cyclic approach to structure-building with an optimality-theoretic approach to optimization. The former property implies that decisions in inflectional morphology can be myopic (andmay ultimately give rise to opacity, in the sense of Kiparsky), and that intermediate stages of derivations may be crucial for determining the properties of eventual output forms. The latter property makes it possible to accomodate evidence for violable and ranked constraints, like repair or last resort, emergence of the unmarked, conspiracies, constraint simplicity, and parametrization by reranking.In practice, two main goals of this project are tightly interrelated: By focussing on the second goal, the first goal will automatically also be addressed.