Korpus zur Erforschung von Registerphänomenen bei Martin Luther (RegiL)

The following documentation describes the design, architecture and corpus building pipeline of the version 0.1 of the RegiL-corpus ("Korpus zur Erforschung von Registerphänomenen bei Martin Luther"). RegiL v. 0.1. consists of seven subcorpora, 72 corpus documents and 179,996 tokens (text layer) and is converted into the search and visualisation format ANNIS (Krause 2019). The mostly fully automatic annotation (layers "pos" and "lemma") is based on a flat manual normalisation in two steps (layers "text" and "norm"). The corpus was built with the aim of representing different registers in Luther's writings. Text categories such as text type or genre were used as an approximation for a variety of registers. The corpus was created in project B04, Emergence and transfer of register patterns: Situational-functional parameters of intraindividual variation in the writings of Martin Luther and Johannes Bretke of the CRC1412 Register (funded by Deutsche Forschungsgemeinschaft DFG, 416591334).下文将阐述RegiL语料库（RegiL-corpus，原德语全称为"Korpus zur Erforschung von Registerphänomenen bei Martin Luther"，即马丁·路德作品语域现象研究语料库）0.1版本的设计思路、架构体系与语料构建流程。RegiL 0.1版本包含7个子语料库、72份语料文档与179996个文本层Token，且已转换为支持检索与可视化的ANNIS格式（Krause, 2019）。该语料库的标注体系以两步扁平化人工归一化（对应文本层与归一化层）为基础，同时辅以近乎全自动化的词性标注（pos层）与词元（lemma层）标注。本语料库的构建目标为完整呈现马丁·路德作品中的各类语域，研究中以文本类型、体裁等文本分类维度作为多样语域的近似表征。本语料库由德国研究基金会（Deutsche Forschungsgemeinschaft, DFG，资助编号416591334）资助的CRC 1412「语域」项目B04课题——"语域模式的涌现与迁移：马丁·路德与约翰内斯·布雷特克作品中个体内变异的情境-功能参数"——研制完成。