Asymmetries in Presupposition Projection and the Nature of Meaning

语言话语随着时间的推移而逐渐被处理，导致给定表达前后的时间不对称。本项目解决了观察到的不对称是否仅仅是语言话语随时间展开的副产品，或者它们是否在语言知识和表征中发挥直接作用的基本问题。解决这个问题是至关重要的，因为它对我们概念化句子意义的方式以及它们如何与上下文信息相互作用有着重要的潜在影响。这与语言知识和其他认知能力如何相互作用的一般问题有关；这是研究语言和人类思维的核心问题。预设解释的不对称性是语言意义的一个特殊方面，它通常与语言和语言外语境相互作用，为研究这一问题提供了一个理想的案例研究。为了说明这一点，(1)中句子中的预设触发词“stop”引入了Mary曾经来上课的预设，也就是说，(1)通常在这已经被认为是理所当然的语境中使用：(1)Mary不再来上课。重要的是，传统上认为复合句中的预设是不对称计算的，例如触发的预设(例如：“停止”)需要在前面的话语语境中得到支持。(2)中的对比说明了这一点。在触发之前引入支持预设的材料，如（2a），会产生恰当的话语，而相反的配置（2b）则不会（用“#”表示）。玛丽过去常来上课，但现在不来了。玛丽不再来上课了，而她过去常常来上课。然而，这种对比可能（至少部分）是由于独立的因素，例如冗余约束，并且可能反映违反的处理约束，而不是语法上的硬连线。这种效果在不同的连接词中也会有所不同，例如在断语或条件句中。最后，像(2)这样的基本数据点没有说明这种影响是由于线性顺序还是底层层次结构造成的，而底层层次结构被认为对意义的计算至关重要。因此，预设投影中的不对称问题远未解决，而且是多方面的，需要仔细的实验研究。目前的项目将从实验和理论上探索这些问题，将基于形式语义学和语言哲学的关于预设的细粒度理论预测与语言处理模型相结合。这个项目的产出将有助于认知科学中几个学科的一些正在进行的辩论，例如语言学、哲学、心理学和神经语言学，从而使我们更普遍地理解人类的思想。

Linguistic utterances are processed incrementally as they unfold in time, resulting in a temporal asymmetry between the before and after of a given expression. The present project addresses the fundamental question of whether observed asymmetries are merely a by-product of linguistic utterances unfolding in time, or whether they play a direct role in linguistic knowledge and representations. Addressing this question is critical because it has important potential consequences on the very way we conceptualize the meaning of sentences and how they interact with contextual information. This bears on the general issue of how linguistic knowledge and other cognitive faculties interact; a core issue in the study of language and the human mind. Asymmetries in the interpretation of presuppositions - a particular aspect of linguistic meaning, which characteristically interacts with both the linguistic and extra-linguistic context - provide an ideal case study for investigating this issue. To illustrate, the presupposition trigger 'stop' in the sentence in (1) introduces the presupposition that Mary used to come to class, i.e., (1) is typically uttered in contexts where this is already taken for granted: (1) Mary stopped coming to class. Importantly, presuppositions in complex sentences are traditionally thought to be asymmetrically computed, such that the presupposition of a trigger (e.g. 'stop') requires support in the preceding discourse context. This is illustrated by the contrast in (2). Introducing the material supporting the presupposition before the trigger, as in (2a), makes for a felicitous utterance, while the reverse configuration (2b) does not (indicated by '#'). (2) a. Mary used to come to class and she stopped (coming to class). b. #Mary stopped coming to class and she used to (come to class). However, such contrasts may (at least in part) be due to independent factors, e.g. redundancy constraints, and could reflect violable processing constraints rather than being grammatically hard-wired. The effect also may vary across connectives, e.g., in disjunctions or conditionals. Finally, basic data points such as (2) leave open whether the effect is due to linear order or the underlying hierarchical structure assumed to be crucial for the computation of meaning. Thus, the issue of asymmetry in presupposition projection is far from resolved, and so multi-faceted that careful experimental investigation is called for. The present project will explore these issues experimentally and theoretically, combining fine-grained theoretical predictions about presuppositions, rooted in formal semantics and philosophy of language, with models of language processing. The outputs of this project will contribute to a number of ongoing debates in several disciplines in the cognitive sciences, e.g., linguistics, philosophy, psychology, and neurolinguistics, thereby informing our understanding of the human mind more generally.