1 Introduction

Beijing Foreign Studies University has launched the GLOBE Corpus Project, aiming to build a balanced corpus for 101 languages. The arGLOBE Corpus is the Arabic sub library of the project, with a designed scale of 1 million words and texts mainly from 2010 to 2022.

Overview of Corpus Construction for Arabic Language

In the study of traditional Arabic grammar, corpus linguistics has a long history, such as the use of authentic corpora by Sibelius in the Book of Sibelius. The construction of the Modern Arabic Corpus started relatively late, but progress has been made. The existing electronic corpora include general corpora, dialect corpora, diachronic corpora, learner corpora, specialized corpora, and parallel corpora. However, the construction of a balanced corpus of Arabic written language still needs to be promoted, and the comparability between existing corpora is insufficient. The arGLOBE corpus adopts the Brown corpus sampling mode, aiming to construct a comparable Arabic written language balanced corpus and promote multilingual comparative research.

Creation of 3 arGLOBE corpora

The arGLOBE corpus is a contemporary Arabic written language balanced corpus constructed based on the Brown corpus sampling scheme, containing approximately one million words and mainly sourced from texts between 2010 and 2022. This language corpus provides three versions: raw language materials, part of speech encoding, and word form restoration, using the Stanford NLP toolkit's stanza ar package for natural language processing. The arGLOBE corpus also provides a word frequency table, including words, binary phrases, and ternary phrases, suitable for teaching and research. The corpus has been uploaded to the Beiwai CQPweb multilingual corpus platform, which supports multiple online search functions.

4 Research Prospects

The arGLOBE corpus provides rich resources for Arabic language research, supporting language ontology and interdisciplinary research. Word frequency tables and phrase tables are helpful for dictionary compilation and teaching, and part of speech tagging information can be used for structural complexity analysis. Comparable with the Brown family corpus, it facilitates language comparison and multilingual typology research. Different subgenre corpora can be extracted for discourse analysis and domain variation research. Researchers can build their own small-scale corpus and use arGLOBE as a reference for keyword extraction. The 1.0 version of arGLOBE corpus provides new opportunities for Arabic language research, but compared with Chinese and English corpora, there is room for improvement in the development of Arabic corpus. More types of corpora need to be constructed to reflect the language characteristics of different eras and conduct longitudinal research. Can develop a monitoring corpus, continuously collect corpus, and comprehensively reflect language use. Specialized Arabic language corpora such as learner corpora, literary corpora, and dialect oral corpora are also worth paying attention to, promoting linguistic analysis, foreign language teaching, stylistic research, and sociolinguistic research.

5 Conclusion

The study of Arabic corpus is an emerging field, and the construction of arGLOBE corpus is an important progress in this field. This corpus adopts the Brown corpus sampling standard and provides retrieval function through the "BFSU CQPweb Multilingual Corpus Platform". Researchers can use arGLOBE to conduct corpus based research and promote the development of Arabic language studies. At the same time, the "BFSU Global Corpus Cluster" advocates for joint construction and sharing, encourages more peers to participate in corpus construction, and promotes Arabic language teaching and research.