O modelo de dependências universais: assentando bases teóricas e revisando diretrizes metodológicas
Universal Dependencies (UD) are a framework for morphosyntactic annotation of natural language texts. This article assumes that this framework implies three types of adequacy: computational adequacy (related to natural language processing), typological adequacy (related to the overall proposal of cr...
| Autor: | |
|---|---|
| Tipo de recurso: | artículo |
| Estado: | Versión publicada |
| Fecha de publicación: | 2024 |
| País: | Brasil |
| Institución: | Universidade Federal do Paraná (UFPR) |
| Repositorio: | Revista da ABRALIN (Online) |
| Idioma: | portugués |
| OAI Identifier: | oai:ojs.revista.abralin.org:article/2176 |
| Acceso en línea: | https://revista.abralin.org/index.php/abralin/article/view/2176 |
| Access Level: | acceso abierto |
| Palabra clave: | Dependência universais Tipologia Conceitos comparativos Conceitos descritivos Universal Dependencies Typology Comparative concepts Descriptive concepts |
| Sumario: | Universal Dependencies (UD) are a framework for morphosyntactic annotation of natural language texts. This article assumes that this framework implies three types of adequacy: computational adequacy (related to natural language processing), typological adequacy (related to the overall proposal of crosslinguistic comparison), descriptive adequacy (related to the annotation of individual languages). The discussion on this paper focuses the last two types of adequacy. The proposal made here is that the underlying theoretical assumptions of UD’s framework need to be clearly explicated, because these assumptions determine the establishment of categories based on which annotations for individual languages are made. In this line, I examine the definitions proposed in the framework in order to determine what are the underlying theoretical notions, and I confront general (universal) guidelines for annotation and language-particular guidelines. The overall assessment in this paper is that, on the one hand, from the perspective of proposing general guidelines for annotation, UD’s framework should be built as a set of “comparative concepts” (HASPELMATH, 2010); on the other hand, from the perspective of proposing guidelines for the annotation of individual languages, the framework should be built as a set of “descriptive categories”. This dual view of the UD’s framework can ensure theoretical and methodological equivalence, and its implementation can optimize linguistic analysis and description tasks assisted by computational models. |
|---|