Pervasive Context-sensitivity in Natural Language: the problems and the solutions

当某人说出一个句子，比如(1)“Jill is a sailor”，为了理解他们的话语需要什么？一个标准的答案是，你所需要知道的就是句子中单词的意思以及它们的组合方式（即对句子结构的理解）。然而，近年来，这个标准答案受到了来自一些人的严重压力，这些人认为，语言意义的知识需要对话语语境的丰富知识。例如，假设吉尔是一名律师，业余时间喜欢航海。如果我们正在谈论朋友的爱好，我说(1)，似乎我说的是真的；在这种情况下，我说吉尔是个水手是对的。然而，现在想象一下，我们正在谈论我们朋友的职业。在这种情况下，如果我说出(1)，似乎我说了一些假的东西；在这种情况下，说吉尔是个水手是不对的。这似乎表明，掌握语言意义需要对词语在语境中的使用方式具有极高的敏感性。也就是说，像这样的例子似乎证明了我们语言中普遍存在的上下文敏感性。然而，普遍的上下文敏感性是有问题的，因为它破坏了语言理解可以由一个特殊的、封装的、只处理语言的心理“模块”来支撑的观点。相反，它似乎表明，理解语言需要丰富的一般和文化知识，例如，在一个上下文中，什么是相关的，别人在想什么。如果认为语言理解是一项非常丰富的事业是正确的，那么这反过来又会对诸如为有语言障碍的患者（例如失语症患者或自闭症患者）设计干预措施以及语言理解的计算机建模等事情产生严重影响（因为识别相关的上下文特征在人工智能领域是出了名的困难）。因此，该项目的目的是创建一个研究人员的国际网络，以调查普遍的上下文敏感性的证据，并探索在这一领域出现的理论立场。例如，该项目将探讨对(1)等案例的反应是否显示出文化差异，以及任何一种答案都能告诉我们关于这场辩论的什么。我们将在这一领域开放讨论，让认知心理学的研究人员参与进来，更彻底地探索这一领域中每一种哲学立场所做出的经验预测，以及如何通过实验来验证这些预测。总体而言，该项目将提供一个概括性的概述问题和解决方案，这些问题和解决方案出现在最近关于普遍的上下文敏感性的激烈辩论中，并确定在艺术和人文研究界内外推动辩论的最佳方式。拟议中的网络将主要涉及英国和中国的研究人员，利用两国现有的专业知识，但建立新的合作，我们希望能产生令人兴奋的发现。该网络将包括两次面对面会议（一次在中国，一次在英国），以及一系列进一步的活动（包括视频会议，远程博士指导，以及通过专门的开放访问网站进行在线合作）。该网络中心将扩展到包括艺术和人文社区以外的研究人员、早期职业研究人员和非学术利益相关者。通过发起这种包容性的对话，希望将取得原创和重要的成果，并在这一关键领域取得重大进展。最后，为该项目开发的连接和基础设施将为未来的中英合作提供一个跳板。

When someone utters a sentence, say (1) "Jill is a sailor", what is required in order to understand their utterance? A standard answer is that pretty much all you need to know is the meanings of the words in the sentence together with knowledge of the way they are put together (i.e. an understanding of sentence structure). However, in recent years this standard answer has come under serious pressure from those who argue that knowledge of linguistic meaning requires a wealth of knowledge about the context of utterance. For instance, imagine that Jill is a solicitor who likes to sail in her spare time. If we are having a conversation about our friends' hobbies and I utter (1) it seems that I say something true; in this context, I'm right when I say that Jill is a sailor. However, now imagine that we are having a conversation about our friends' professions. In this context if I utter (1) it seems that I say something false; in this context it is not right to say of Jill that she is a sailor. What this seems to show is that grasp of linguistic meaning requires an extremely high level of sensitivity to the way in which words are being used in context. That is to say, cases like these seem to demonstrate pervasive context-sensitivity in our language. However, pervasive context-sensitivity is problematic for it undermines the idea that language understanding could be underpinned by a special, encapsulated mental 'module' which deals just with language. Instead, it seems to show that understanding language requires a wealth of general and cultural knowledge, for instance about what is relevant in a context and what someone else is thinking. If it is right to think that language understanding is this very rich kind of enterprise, then this in turn has serious repercussions for such things as designing interventions for patients who have linguistic impairments (e.g. patients with aphasia, or those on the autistic spectrum) and for computer modelling of language understanding (since identification of relevant contextual features is something which is notoriously difficult within artificial intelligence circles).The aim of this project, then, is to create an international network of researchers to investigate the evidence for pervasive context-sensitivity, and to explore the theoretical positions which have emerged in this area. For instance, the project will explore whether reactions to cases like (1) show cultural variation, and what an answer either way tells us about the debate. We will look to open up discussion in this area to the input of researchers in cognitive psychology, exploring more thoroughly the empirical predictions made by each philosophical stance in this area and how these predictions might be tested experimentally. Overall, the project will provide a synoptic overview of the problems and solutions which have emerged in the recent, lively debate about pervasive context-sensitivity and identify the best way to take the debate forward, both within the arts and humanities research community and beyond.The proposed network will primarily involve researchers in the UK and China, making use of extant expertise in both countries but forging new collaborations which we hope will yield exciting discoveries. The network will involve two face-to-face meetings (one in China and one in the UK), together with a range of further activities (including video conferencing, remote PhD supervision, and on-line collaboration via a dedicated, open access website). The network hub will extend to include researchers from beyond the arts and humanities community, early career researchers, and non-academic stakeholders. By instigating an inclusive dialogue of this kind it is hoped that original and important results will be delivered and understanding in this key area will be significantly advanced. Finally, the connections and infrastructure developed for this project will provide a springboard for future UK/China collaboration.