According to the Canonical Binding Theory (Chomsky 1981), anaphors must be bound in their local domain and pronominals must be free. The discovery of “long-distance anaphors” (e.g. Thrainsson 1976, Giorgi 1984), which violate the locality condition, induced the search for independent criteria. Giorgi (1984:310) proposed a widely adopted criterion: “pronouns can have split antecedents and anaphors cannot”. Recent minimalist binding theories derive this property of anaphors from the way a dependency on the antecedent is established which makes it intrinsic to binding. However, this leads to an important problem, since some languages have elements that i) may be locally bound and, hence, look like anaphors; yet ii) allow split antecedents which is a property of pronouns (e.g. Japanese and Korean, Katada 1991, Kasai 2000). In this paper I analyze the data of another such language, namely Meadow Mari (Uralic), and show that such facts require a modular approach to binding (see Reuland 2011). I further argue that here the left periphery contains the relevant factor

中文翻译：

---

Meadow Mari 的反身性：约束和同意

根据规范绑定理论（Chomsky 1981），回指必须绑定在其本地域中，代词必须是自由的。违反局域性条件的“远距离照应”（例如 Thrainsson 1976，Giorgi 1984）的发现引发了对独立标准的搜索。Giorgi (1984:310) 提出了一个被广泛采用的标准：“代词可以有分裂的先行词，而回指不能”。最近的极简绑定理论从对先行词的依赖建立的方式推导出回喻的这种属性，这使得它成为绑定的内在因素。然而，这会导致一个重要的问题，因为某些语言具有以下元素： i) 可能是本地绑定的，因此看起来像回指；然而 ii) 允许拆分先行词，这是代词的属性（例如日语和韩语，Katada 1991，Kasai 2000）。在本文中，我分析了另一种此类语言的数据，即 Meadow Mari（乌拉尔语），并表明此类事实需要采用模块化方法进行绑定（参见 Reuland 2011）。我进一步认为，这里的左边缘包含相关因素