Protoform reconstruction is the task of inferring what morphemes or words
appeared like in the ancestral languages of a set of daughter languages. Meloni
et al. (2021) achieved the state-of-the-art on Latin protoform reconstruction
with an RNN-based encoder-decoder with attention model. We update their model
with the state-of-the-art seq2seq model: the Transformer. Our model outperforms
their model on a suite of different metrics on two different datasets: their
Romance data of 8,000 cognates spanning 5 languages and a Chinese dataset (Hou
2004) of 800+ cognates spanning 39 varieties. We also probe our model for
potential phylogenetic signal contained in the model. Our code is publicly
available at this https URL

使用 RNN-based encoder-decoder with attention model 和 Transformer 模型进行原型重建，分析了语言谱系启示信号。

变形原型重构

Transformed Protoform Reconstruction

Phylogenetic comparative methods are new in our field and are shrouded, for
most linguists, in at least a little mystery. Yet the path that led to their
discovery in comparative biology is so similar to the methodological history of
balanced sampling, that it is only an accident of history that they were not
discovered by a typologist. Here we clarify the essential logic behind
phylogenetic comparative methods and their fundamental relatedness to a deep
intellectual tradition focussed on sampling. Then we introduce concepts,
methods and tools which will enable typologists to use these methods in
everyday typological research. The key commonality of phylogenetic comparative
methods and balanced sampling is that they attempt to deal with statistical
non-independence due to genealogy. Whereas sampling can never achieve
independence and requires most comparative data to be discarded, phylogenetic
comparative methods achieve independence while retaining and using all data. We
discuss the essential notions of phylogenetic signal; uncertainty about trees;
typological averages and proportions that are sensitive to genealogy;
comparison across language families; and the effects of areality. Extensive
supplementary materials illustrate computational tools for practical analysis
and we illustrate the methods discussed with a typological case study of the
laminal contrast in Pama-Nyungan.

本文介绍了在语言学领域中新兴的系统比较方法，即系统比较学方法，并解释了其背后的基本逻辑和其与采样的深刻联系。文章还介绍了相关概念、方法和工具，帮助语言学家在日常的语言类型学研究中使用这些方法。最后，通过一个 Pama-Nyungan 的语言类型学案例研究，阐述了本文讨论的方法。

采样的挑战及系统比较方法的应用：以 Pama-Nyungan 语言层面对比为例

Challenges of sampling and how phylogenetic comparative methods help: With a case study of the Pama-Nyungan laminal contrast

We study phylogenetic signal present in syntactic information by considering
the syntactic structures data from Longobardi (2017b), Collins (2010), Ceolin
et al. (2020) and Koopman (2011). Focusing first on the general Markov models,
we explore how well the the syntactic structures data conform to the hypothesis
required by these models. We do this by comparing derived phylogenetic trees
against trees agreed on by the linguistics community. We then interpret the
methods of Ceolin et al. (2020) as an infinite sites evolutionary model and
compare the consistency of the data with this alternative. The ideas and
methods discussed in the present paper are more generally applicable than to
the specific setting of syntactic structures, and can be used in other
contexts, when analyzing consistency of data with against hypothesized
evolutionary models.

研究了来自 Longobardi、Collins、Ceolin 和 Koopman 的句法结构数据中存在的系统发育信号，通过比较生成的系统发育树和语言学社区所认可的树，探讨了这些数据符合马尔科夫模型假设的程度，同时将 Ceolin 等人的方法解释为一个无限位点的演化模型，并与数据的一致性进行比较。本文讨论的思想和方法不仅适用于句法结构的具体设置，还可以用于其他上下文，分析数据与预测的演化模型的一致性。