Scholars have discussed Indonesian clitics theoretically from various perspectives. Regardless, their application in the corpus annotation field remains challenging and understudied. In this paper, I will qualitatively compare three corpus annotation tools that can be used to analyse Indonesian clitics: MorphInd, TreeTagger and NooJ. I find that these three computational tools have a number of fundamental differences. TreeTagger requires a separate tool to delineate a clitic from its host, before annotation. Conversely, NooJ and MorphInd integrate the clitic delineation and clitic annotation process, as shown in SANTI-POS. In terms of resources, MorphInd resource development seems to be restricted to its developers. Unlike MorphInd, TreeTagger and NooJ resources can easily be developed as they are available to the public. A survey confirms that NooJ is preferred due to its user-friendly installation and visual and transparent analysis. Going beyond annotation, I demonstrate how NooJ allows users to identify a clitic’s surface form, as well as its free form, in the Task Annotation Structure (TAS).

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

The Annotation of Indonesian Clitics by Nooj and Other Computational Tools

  • Prihantoro

摘要

Scholars have discussed Indonesian clitics theoretically from various perspectives. Regardless, their application in the corpus annotation field remains challenging and understudied. In this paper, I will qualitatively compare three corpus annotation tools that can be used to analyse Indonesian clitics: MorphInd, TreeTagger and NooJ. I find that these three computational tools have a number of fundamental differences. TreeTagger requires a separate tool to delineate a clitic from its host, before annotation. Conversely, NooJ and MorphInd integrate the clitic delineation and clitic annotation process, as shown in SANTI-POS. In terms of resources, MorphInd resource development seems to be restricted to its developers. Unlike MorphInd, TreeTagger and NooJ resources can easily be developed as they are available to the public. A survey confirms that NooJ is preferred due to its user-friendly installation and visual and transparent analysis. Going beyond annotation, I demonstrate how NooJ allows users to identify a clitic’s surface form, as well as its free form, in the Task Annotation Structure (TAS).