Beyond Semantics: a theory of language in which semantics has no role in explaining how language functions

在心灵和语言的理论家中有一种广泛的信念：为了理解语言、思想和现实之间的关系，我们需要一种意义和内容的理论;也就是说，一种规范的、形式的意义科学，它是我们关于意义的普通观念的延伸和理论深化。形式的、规范的科学可以采取真值条件理论、推理实践理论、断言理论等形式。深层问题困扰着作为研究纲领的语义学。两个中心问题是：（a）说谎者悖论：矛盾似乎出现时，语言谈论自己。最简单的说谎者悖论是一个句子，它自己说它不是真的。目前在语义学中有许多关于如何处理这些悖论的相互竞争的方法，包括存在真正矛盾的想法。(b)模糊性：自然语言的一个普遍特征是谓词，如“pink”、“heap”、“human”等，都是模糊的。问题是，模糊谓词适用的事物范围具有模糊边界，但这种模糊边界的性质很难分析。模糊谓词产生了Sorites悖论，暗示模糊语言是不连贯的。理论家们认为，尽管存在表象（认识论）或语义上的不确定性（超价值论），谓词确实是尖锐的。对于这些问题的正确解决办法没有普遍的共识。我认为这些悖论不会被语义学解决，语义学是问题而不是解决方案。原因是语义学与谓词（如“true”、“heap”等）具有扩展（也就是说，它们为真的一组事物）的想法紧密相连。扩展是语义学的核心理论工具。它们之所以是语义学的核心，是因为句子的真值的定义，包括那些关于真值的定义，都是根据扩展给出的。例如，一个简单的原子句，“O是F”，对于某个事物O和F，当且仅当O在F的扩张中时为真。但是，真值谓词和模糊谓词具有扩展的信念产生了真值和模糊性的悖论。提出谓词缺乏扩展似乎是语义学无法理解的，因为它们拥有意义，在语义学中，需要一个关于谓词如何有助于真值条件的理论。我提供了一个替代框架，全球表现主义（GE），在其中我们停止分析语言的语义理论。GE是一种关于语言活动和潜在认知结构的理论。它引申了价值表现主义的一些核心思想，根据这些思想，价值主张表达情感状态。并不是所有的断言都表达情感状态，而是所有的断言都表达了非代表性的状态。虽然每个断言都有表示性内容，但它不是从底层的表达状态继承的。GE提供的不是内容理论，而是一种表达主义的理论，即所有句子都谈论内容和真值条件。GE描述了真值谓词的功能，但真值本身并没有理论上的定义，因此谓词扩展没有理论上的作用。GE通过控制句子的非语义组合性特征，为说谎者提供了一种不确定性解决方案，它摆脱了威胁任何解决方案的多种形式的报复悖论。Sorites悖论的解决方案也是一个不确定性解决方案：该悖论源于这样的原则，即模糊谓词可以容忍它们所应用的事物的微小变化。通用工程意味着，人们不能说事物中微小的精确变化如何影响谓词的应用。

There is a wide-spread belief amongst theorists of mind and language: that in order to understand the relation between language, thought, and reality we need a theory of meaning and content; that is, a normative, formal science of meaning, which is an extension and theoretical deepening of our ordinary ideas about meaning. That formal, normative science may take the form of a theory of truth-conditions, a theory of inferential practice, a theory of assertion, and so on. Deep problems afflict semantics as a research program. Two central problems are:(a) The Liar-paradoxes: contradictions seem to arise when language talks about itself. The simplest Liar paradox is a sentence that says of itself that it is not true. There are currently within semantics many competing approaches about how to treat these paradoxes, including the idea that there are true contradictions.(b) Vagueness: a pervasive feature of natural languages is that predicates, like 'pink', 'heap', 'human', etc, are vague. The problem is that the range of things to which vague predicates apply have fuzzy boundaries, but the nature of this fuzziness is very difficult to analyze. Vague-predicates generate the Sorites paradox that implies that vague language is incoherent. Theorists have argued that predicates are really sharp, despite appearances (epistemicism) or semantically undecided (supervaluationism). There is no general consensus about the right solutions to these problems. I contend that these paradoxes are not going to be solved by semantics, that semantics is the problem rather than the solution. The reason is that semantics is deeply tied to the idea that predicates, like 'true', 'heap', etc, have extensions, that is, a set of things of which they are true. Extensions are central theoretical tools in semantics. The reason they are central to semantics is that the definition of the truth of sentences, including those about truth, is given in terms of extensions. For example, a simple atomic sentence, 'O is F', for some thing O, and F, is true if and only if O is in the extension of F. But the belief that the truth-predicate and vague predicates have extensions generates the paradoxes of truth and vagueness. Proposing that predicates lack extensions seems unintelligible for semantics, since their possessing meaning, requires, within semantics, a theory about how predicates contribute to truth-conditions. I offer an alternative framework, Global Expressivism (GE), in which we stop analyzing language in terms of semantic theory. GE is a theory about language-activity and underlying cognitive structures. It extrapolates some core ideas of value expressivism, according to which, assertions of value express affective states. It's not that all assertions express affective states, but all express kinds of states that are non-representational. While each assertion has representational content, it is not inherited from the underlying expressed state. Instead of a theory of content, GE provides an expressivist theory of talk of content and talk of truth-conditions for all sentences. GE describes the truth-predicate's function, but truth itself is not defined theoretically, so predicate extensions have no theoretical role. GE entails a form of indeterminacy solution to the liar, which escapes from the many forms of revenge paradox threatening any solution, through the non-semantic compositionality features governing sentences. The solution to the Sorites paradox is also an indeterminacy solution: the paradox arises from principles that say that vague predicates are tolerant of small changes in things they apply to. GE entails that one cannot say how small precise changes in things affect application of predicates.