Bimodal corpus-based language comparison. A study of connectives of contigency in European languages

这个项目涉及表达条件、因果或让步关系的连接词，如“if”、“because”和“although”。本文利用双模态翻译语料库（EPTIC-J）研究了相关连接词在欧洲二十种民族语言中的使用和分布。EPTIC-J语料库包含了欧洲议会的演讲翻译，以及他们的同声传译和欧洲联盟其他语言的书面翻译。这种语料库设计使我们能够研究连接词在原语言中的分布以及在（离线/书面以及同步/口语）翻译中反映的跨语言对应关系。权变连接词在解释上是多层次的，因为它们具有真值条件解释，同时也在话语层面和论证层面引发交际效果。讨论的主要问题是：（1）在不同的解释层次上，意义的各个方面如何映射到语言形式（从属成分的分布和分句的线性顺序），以及（2）什么程度的认知负荷与特定的连接词及其所表达的关系有关。我们只使用语料库的书面部分来解决问题（i）。对于问题（ii），我们依赖于口译数据，使用口译中的翻译选择和耳声跨度值（EVS/décalage）作为认知负荷的指标。通过应用一种新的方法（双峰比较语料库分析）来研究一个从不同角度（类型学，句子语义学，话语/会话和论证）研究的主题，并通过将口语数据整合到图片中，该项目旨在重新考虑现有的索赔和概括，并开辟新的视角。

This project deals with connectives that express a conditional, causal or concessive relation, such as `if', `because' and `although'. It makes use of a bimodal translation corpus (EPTIC-J) to study the use and distribution of relevant connectives in twenty European national languages. The EPTIC-J corpus contains transcriptions of speeches made in the European Parliament as well as their simultaneous interpretations and written translations into the other languages of the European Union. This corpus design allows us to study connectives in terms of their distributions in original language as well as crosslinguistic correspondences reflected in (offline/written as well as simultaneous/spoken) translation. Connectives of contingency are interpretively multi-layered, insofar as they have a truth-conditional interpretation while also triggering communicative effects at the discourse level and the level of argumentation. The main questions addressed are (i) how aspects of meaning at various levels of interpretation map to linguistic form (distribution of subordinators and linear order of clauses), and (ii) what degrees of cognitive load are associated with specific connectives and the relations that they express. We address question (i) by using the written part of the corpus only. For question (ii) we rely on the interpreting data, using translation choices in interpreting and ear-voice span values (EVS/décalage) as indicators of cognitive load. By applying a novel methodology (bimodal comparative corpus analysis) to the study of a topic that has been investigated from various perspectives (typology, sentence semantics, discourse/conversation and argumentation), and by integrating spoken data into the picture the project intends to reconsider existing claims and generalizations in a new light as well as open up new perspectives.