Abg Corpus

The present paper presents the task of compiling a linguistic corpus of Brazilian Portuguese, which was undertaken by the authors. It is called ABG Corpus, and this article is also about the computational tools developed for the task. Our main goal is to reunite a large amount of texts, both from spoken and written language to, in the best way possible, represent the Brazilian language in a way that we could use it as a database for our researches, Guide (2016) and Benevides (2017). The ABG corpus has 3.616.625 word tokens and 92.602 types of words, being that 1.938.805 of those tokens are from spoken language corpora and 1.676.820 tokens come from written corpora. Based on the corpus linguistics framework and through the use of computational tools developed using Python, this article shows and provides access to the ABG Corpus, the computational tools (stress marker, phonological structure identifier, syllabifier), as well as some phonological information (stress and syllable related), already present on the corpus. We end by inviting the community to further expand our findings and explore this new tool.