Towards a General Theory of Multi-Valuation

理论语言学试图从看似随意的跨语言变异中揭示出对心理语法的深刻理解。这个项目将调查当协议的目标收到多个值时，协议是如何解决的。文献表明，这种情况的解决在不同的语言中是不同的。到目前为止，还没有人解释为什么语言在这个领域会有如此大的差异。我们在这个项目中的目标是，通过系统地审视协议的目标和发生多重估值的建筑，来解释如何解决多重估值实例的差异。如果元素接收到同一特征的多个值，则已知有两个主要选项可用：它可以将情况解析为语义上合适的值，例如Sg+Sg=Pl，或者它可以更喜欢壁橱协议控制器的值。这两种情况都在不同的语言中得到了证实，通常语言内部的差异取决于协议目标的类型(动词对名词对定语成分)。然而，这种变化并不是随机的。一些语言(例如俄语)显示所有协议目标的已解决复数协议，而其他协议目标未显示任何目标的解决方案(例如。斯洛文尼亚语)。此外，一些语言可以针对不同的目标使用不同的模式。例如，在英语中，动词上的两个单数特征被解析为复数，但对于定语成分，如指示词，解析一致是不可能的。然而，未经证实的是，这种语言中的定语成分在这种情况下表现出决心一致，但在动词上却不是。其次，我们将调查当元素收到冲突的值时，语言所采取的各种修复策略。众所周知，这样的语境是有标记的，需要进行修复才能使判决合法。具体地说，我们将专注于对壁橱控制器的协议解析，旨在为最近从连词一致性研究中提出的两步协议模型提供新的证据。最后，我们将研究扩展到合取和析取两种协议模式之间的不对称性。连体受试者能够控制已解决的一致，而非连体受试者则触发与两个分离词中最接近的一个的一致。然而，连词和析取都表现出多重赋值的效果特征，暗示着一个共同的句法结构。鉴于连词和析取的共同句法，我们无法解释它们不同的一致策略。我们的研究旨在表明解释的影响在这里起着作用，并且在特定的语义语境下，如否定，析取一致确实可以在析取中产生，这将暗示连词和析取之间不对称的原因是语义的，而不是句法的。

Theoretical linguistics strives to reveal deep understanding of the mental grammar from the seemingly random cross-linguistic variation. This project will investigate how agreement is resolved when the target of agreement receives multiple values. It has been shown in the literature that the resolution of such a situation is subject to variation across languages. As yet there is no explanation for why languages vary in the way that they do in this realm. Our aim in this project is to account for the variation in how instances of multi-valuation is resolved by systematically looking across both the targets of agreement, and constructions where multi-valuation occurs. If an element receives multiple values for the same feature, two main options are known to be available: it can either resolve the situation to a semantically suitable value, for instance Sg + Sg = Pl, or it can prefer the value of the closet agreement controller. Both situations have been attested across languages, often with language internal variation depending on the type of agreement target (verb vs. noun vs. attributive elements). Yet, this variation is not random. Some languages (eg. Russian) show resolved plural agreement across all agreement targets, whilst others show no resolution for any target (eg. Slovenian). Furthermore some languages can have differing patterns for different targets. In English, for instance, two singular features on a verb are resolved to plural, but resolved agreement is impossible for attributive elements like demonstratives. What is unattested however, is a language where attributive elements show resolved agreement in this situation, but not on verbs. Secondly, we will investigate the various repair strategies that languages take when an element receives conflicting values. Such contexts are known to be marked, and require repairs to allow the sentence to be licit. Specifically, we will focus on the resolution of agreement to the closet controller, aiming to provide novel evidence for the two-step model of agreement that has been advanced recently from studies of agreement with conjunctions. Finally, we extend our investigation to the asymmetry between the agreement patterns in conjunction and disjunction. Conjoined subjects are able to control resolved agreement, whereas disjoined subjects triggers agreement with the closest of the two disjuncts. Both conjunction and disjunction however show the effects characteristic of multi-valuation, hinting at a common syntactic structure. Given their common syntax, the different agreement strategies of conjunction and disjunction are unaccounted for. Our investigation aims to show that influences of interpretation are at play here, and that given certain semantic contexts, such as negation, resolved agreement can indeed arise in disjunctions, which would hint at the cause of the asymmetry between conjunction and disjunction is semantic, rather than syntactic in nature.