Abstract A theoretical discussion of units in linguistic theory would be, in a sense, incomplete without a discussion of the systems, whether overt or implied, that the units are associated with. This paper traces conceptualizations of units and their accompanying systems in several disciplines. We identify some important problems with rule-based accounts (Parsons 1937) of social action and discuss the transition to non-rule-based theory afforded by ethnomethodology (e.g. Garfinkel 1963, 1967; Heritage 1984, 2011). We draw direct parallels between these issues and analogous developments in mathematical logic (Godel 1992) and philosophy of mind (Fodor 1968, 1983; Lucas 1961; Putnam 1960, 1967 etc.), and argue that these stem directly from fundamental properties of a class of all formal systems which permit self-reference. We argue that, since these issues are architectural in nature, linguistic theory which postulates that linguistic units are the outputs of a consistent, self-referential, rule-based formal systems (e.g. Hauser, Chomsky & Fitch 2002) will inevitably run into similar problems. This is further supported by examples from actual language use which, as a class, will elude any theoretical explanation grounded in such a system.

中文翻译：

---

语言单位及其系统

摘要 在某种意义上，如果不讨论与单元相关联的系统（无论是公开的还是隐含的），语言理论中对单位的理论讨论将是不完整的。本文追溯了多个学科中单元及其伴随系统的概念化。我们确定了社会行动的基于规则的解释（Parsons 1937）的一些重要问题，并讨论了民族方法学提供的向非基于规则的理论的转变（例如 Garfinkel 1963, 1967; Heritage 1984, 2011）。我们将这些问题与数理逻辑 (Godel 1992) 和心智哲学 (Fodor 1968, 1983; Lucas 1961; Putnam 1960, 1967 等) 的类似发展直接比较，并认为这些直接源于类的基本属性所有允许自我参照的正式系统。我们认为，由于这些问题本质上是建筑问题，假设语言单位是一致的、自我参照的、基于规则的形式系统的输出的语言理论（例如 Hauser, Chomsky & Fitch 2002）将不可避免地遇到类似的问题。来自实际语言使用的示例进一步支持了这一点，作为一个类，这些示例将回避基于此类系统的任何理论解释。