Portuguese Corpora Research Articles

Intelligent Support for Low Literacy Adults: The European Portuguese iRead4Skills Corpus

This paper presents the Portuguese dataset of the iRead4Skills project (Dataset 1: corpora by complexity level for FR, PT, and SP – v.2.0), a representative sample of written European Portuguese for automatic complexity assessment that addresses a gap in existing resources for Portuguese. The corpus was created within the framework of the iRead4Skills project, which encompasses Portuguese, French, and Spanish. The project aims to develop an intelligent system to evaluate text complexity while recommending appropriate reading materials to native adult learners with low literacy skills. The corpus compilation involved a manual selection of text samples across various textual genres and document types, covering a wide range of existing written materials and focusing on the reading needs and reading habits of the target audience—low literacy adults enrolled in vocational education and training centres or adult learning (AL) centres. The collected texts were categorised into the three distinct levels of complexity targeted and defined by the project: very easy, easy, and plain levels. Texts of higher complexity were also included, resulting in the creation of four distinct sub-corpora. The resulting Portuguese dataset consists of 2,186 texts and 942,818 tokens and serves as the foundational source for training and testing the project’s complexity analysis systems. This paper presents a comprehensive overview of the compilation process of the corpus, encompassing its methodological design and the challenges faced. Although some existing Portuguese corpora were used for complexity studies and tool development, these primarily consist of texts classified according to CERF levels and retrieved from didactic materials designed for L2 teaching/learning or texts produced by L2 learners. The corpus presented in this paper introduces a new resource that addresses a significant gap in materials needed to inform and support studies and applications related to text complexity. The resulting dataset provides a novel and important language resource for European Portuguese, with several applications including research on linguistic complexity, development of automatic text complexity and readability assessment systems, and educational purposes.

«A praia está perto»: Zum Gebrauch des Verbsestarin Ausdrücken der statischen Lage im Portugiesischen

AbstractThe article examines the use ofestarin Portuguese expressions of permanent location, which runs counter to normative grammar, according to whichserandficarshould appear in those cases. This use ofestaris not mentioned in grammatical descriptions, either. Yet, it occurs in contemporary Portuguese (both Portugal and Brazil). The same goes for the Portuguese- and Spanish-lexified creoles, some of which usetá/stá(< estar),to express location. However, almost all rely on a single verb in these contexts. The research question was studied empirically in a corpus (N=1,037) based on Google searches with seven exact phrases includingestar,ser, andficar, subsequently complemented with an analysis of two other Portuguese corpora. The results show thatestarappears in expressions of permanent location, both in Portugal (22,5%) and Brazil (17,5%). However, there is a statistically significant difference,ficarbeing more frequent in Brazil andserin Portugal. Moreover, the higher frequency ofestarin Portugal may indirectly explain the popularity officarin Brazil. In addition,estarappears in written nonfiction, pointing to expressions of the typeestar localizado/situadoas its possible origin, also taking into account that no instances of its use were found in older documents.

Extending automatic transcripts in a unified data representation towards a prosodic-based metadata annotation and evaluation

This paper describes a framework that extends automatic speech transcripts in order to accommodate relevant information coming from manual transcripts, the speech signal itself, and other resources, like lexica. The proposed framework automatically collects, relates, computes, and stores all relevant information together in a self-contained data source, making it possible to easily provide a wide range of interconnected information suitable for speech analysis, training, and evaluating a number of automatic speech processing tasks. The main goal of this framework is to integrate different linguistic and paralinguistic layers of knowledge for a more complete view of their representation and interactions in several domains and languages. The processing chain is composed of two main stages, where the first consists of integrating the relevant manual annotations in the speech recognition data, and the second consists of further enriching the previous output in order to accommodate prosodic information. The described framework has been used for the identification and analysis of structural metadata in automatic speech transcripts. Initially put to use for automatic detection of punctuation marks and for capitalization recovery from speech data, it has also been recently used for studying the characterization of disfluencies in speech. It was already applied to several domains of Portuguese corpora, and also to English and Spanish Broadcast News corpora.

A perspectivação conceptual em autocitação factiva e fictiva

Este artigo analisa o fenômeno da Autocitação Factiva e Fictiva sob a ótica perspectivação conceptual ou construal, tendo em vista trabalhos de Talmy (1996, 2000), Langacker (1999, 2008), Verhagen (2005, 2007) e Rocha (2006a e b, 2010). Com base em dados extraídos de corpora em Português Brasileiro e Europeu para fins exclusivamente qualitativos, postula-se que o discurso direto, em forma de autocitação, dispõe de instâncias factivas de seus tipos comunicativos como diametralmente opostas, mas contíguas, às suas instâncias fictivas, gerenciadas por operações de construal ora mais subjetificantes, ora mais objetificantes.