Micro-cues of language evolution: A Multifactorial model of V2 loss in Central Romance

语言为什么会变化？像拉丁语这样功能齐全的语言是如何演变成法语或意大利语的？语序等核心特征的演变如何阐明使语言变化成为可能的认知机制？ 这些基本问题通过本项目提出的新的重建方法和多因素模型得到了回答。超越了目前对语言变化的常规研究，即从不同地区聚合不同的文本类型，以返回不规则和细长的变化曲线，我们的新校准方法（CaM）通过从给定地区选择特定的文本类型来控制变化，以识别稳定的进化证据。通过排除变异的非结构性因素，记录了更规则的进化曲线，并且可以制定和测试因果模型。CaM是通过对动词第二（V2）语序的损失，早期法语和意大利语的特点进行系统的调查。一个特定的文本类型是调查从诺曼底和威尼斯地区的法语和意大利语从最早的可用文本到16世纪50年代。在扩展的双语注释语料库，将公开提供的考试，支持一种新的多因素模型（MuM）的语言变化。这个想法是，相关配置提供的多个线索使抽象的语法特征，如词序是可学习的（Lightfoot 2009）：因此，这些配置的变化会影响相关词序的流行和分布。我们根据文献和正在进行的工作选择的线索是断言粒子si和宾语动词语序。这些线索的行为的变化预测预测V2词的损失作为衡量的定量和定性演变的每一种现象。验证模型的预测是通过识别事件，注释其形态句法属性和使用相关的统计方法，适用于整个时期的使用率。MuM和CaM的结果提供了i.在两种密切相关的语言中，一个主要语法变化的比较行为和年表，ii.促进这些变化的线索的身份、轨迹和影响程度，iii.对词序变化的一般因果模型的有效性的评估。从两种语言的特定类型和地区收集三个世纪以来的文本注释语料库，并识别和权衡所涉及的一些线索，这是一项高风险的工作，它带来了优化研究方法的高收益，并大大促进了对语言进化的理解。从两个网站的互补专业知识中建立协同作用，以开发一个更细粒度的概念框架，这个项目让我们更接近理解语法变化是如何可能的。

Why do languages change? How does a fully functional language like Latin evolve to become e.g. French or Italian? And how does the evolution of central characteristics such as word order illuminate the cognitive mechanisms that make language change possible? These fundamental questions are answered through the novel reconstructive method and multifactorial model proposed by this project. Going beyond the current routine research into language change that aggregates different text types from diverse regions to return irregular and elongated curves of change, our new Calibrated Method (CaM) controls for variation by selecting specific text types from given regions to identify stable testimonies of evolution. By excluding non-structural factors of variation, more regular curves of evolution are documented, and causal models can be formulated and tested. CaM is piloted through the systematic investigation of the loss of verb second (V2) word order that characterizes early French and Italian. One specific text type is investigated from the Normandy and Venetian regions for French and Italian from the earliest available texts to the 1550s. The examination in the extended bilingual annotated corpus, to be made publicly available, supports a novel multifactorial model (MuM) of language change. The idea is that multiple cues provided by related configurations make abstract grammatical characteristics like word order learnable (Lightfoot 2009): it follows that changes in these configurations impact the currency and distribution of the relevant word order. The cues that we select on the basis of the literature and of ongoing work are asseveration particle si and Object-Verb word order. Changes in behavior of these cues are predicted to prefigure the loss of V2 word as measured by quantitative and qualitative evolution of each phenomenon. Verification of the predictions of the model is achieved by identification of occurrences, annotation of their morphosyntactic properties and use of correlational statistical methods applied to rates of use across the period. The results of the MuM and CaM provide i. the comparative behavior and chronology of a major grammatical change in two closely related languages, ii. the identity, trajectory and degree of influence of the cues that promote these changes, iii. an assessment of the validity of a general causal model of word order change. The high risks in assembling an annotated corpus of texts from specific types and regions for two languages over three centuries, and identifying and weighting some of the cues involved, bring about the high gains of optimizing research methods and significantly furthering the understanding of language evolution. Building synergy from the complementary expertise of both sites to develop a finer-grained conceptual framework, this project gets us one step closer to understanding how grammatical change is possible.