NILC-Metrix: assessing the complexity of written and spoken language in Brazilian Portuguese

Leal, Sidney Evaldo; Duran, Magali Sanches; Scarton, Carolina Evaristo; Hartmann, Nathan Siegle; Aluísio, Sandra Maria

doi:10.1007/s10579-023-09693-w

Cited by 4 publications

(2 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…A estas métricas, foram acrescidas duas medidas de complexidade da escrita discutidas em Leal et al (2023): a métrica de complexidade lexical Brunet e a métrica de complexidade sintática da distância do grafo. A escolha dessas duas métricas foi motivada pelos resultados apresentados em Leal et al (2018) para a tarefa de identificação da complexidade textual, e pela facilidade de replicação das mesmas para o português.…”

Section: Avaliação Intrínsecaunclassified

See 1 more Smart Citation

Transferência de estilo textual arbitrário em português

Botton da Costa,

Paraboni

2023

Linguamática

View full text Add to dashboard Cite

Na Geração automática de língua natural, modelos de transferência de estilo textual arbitrário objetivam a reescrita de um texto usando qualquer novo conjunto de características estilísticas desejado. Em se tratando do idioma português, entretanto, observa-se que os recursos linguístico-computacionais necessários para o desenvolvimento de modelos deste tipo ainda são consideravelmente escassos em comparação à língua inglesa. Assim, como um primeiro passo em direção ao desenvolvimento de métodos avançados deste tipo, o presente trabalho investiga a questão da transferência de estilo textual arbitrário com o uso de paráfrases em português, combinadas ao uso de modelos neurais construídos a partir de arquiteturas do tipo sequência-para-sequência e por refinamento de grandes modelos de língua. Além dos modelos de reescrita textuais propriamente ditos, o estudo apresenta também recursos inéditos para a tarefa na forma de um córpus de paráfrases e de um modelo de \textit{embeddings} validado nas tarefas de similaridade e simplificação sentencial, com resultados comparáveis ao estado da arte.

show abstract

Section: Avaliação Intrínsecaunclassified

“…A escolha dessas duas métricas foi motivada pelos resultados apresentados em Leal et al (2018) para a tarefa de identificação da complexidade textual, e pela facilidade de replicação das mesmas para o português. Em ambos os casos, optamos por reimplementar o código apresentado em Leal et al (2023) com uso do pacote spaCy em Python.…”

Section: Avaliação Intrínsecaunclassified

Transferência de estilo textual arbitrário em português

Botton da Costa,

Paraboni

2023

Linguamática

View full text Add to dashboard Cite

show abstract

Mean dependency length — a new metric for requirements quality

Barbosa,

de Oliveira,

Cerqueira

et al. 2024

INCOSE International Symp

View full text Add to dashboard Cite

This paper proposes the mean dependency length (MDL) as a metric for measuring natural language requirements quality. Dependency length is a linguistic feature based on dependency grammar, which natural language researchers have traditionally used to evaluate syntactic complexity in other contexts. In this study, aided by MATLAB‐based algorithms, the authors assessed MDL over a requirements set composed of 249 original statements, rephrased into five pattern systems. Null hypothesis and effect size testings revealed that MDL is sensitive to the application of pattern rules and to the differences among the patterns, both in an absolute approach and in comparison with other metrics. Furthermore, it was also demonstrated that MDL is aligned with users' values, especially for understandability issues, and can be measured automatically. Finally, the work concluded that MDL is a convenient metric for assessing the quality of natural language requirements.

show abstract

Exploring Spoken Discourse and Its Neural Correlates in Women With Alzheimer's Disease With Low Levels of Education and Socioeconomic Status

Malcorra,

García,

Marcotte

et al. 2024

Am J Speech Lang Pathol

View full text Add to dashboard Cite

Purpose: Early impairments in spoken discourse abilities have been identified in Alzheimer's disease (AD). However, the impact of AD on spoken discourse and the associated neuroanatomical correlates have mainly been studied in populations with higher levels of education, although preliminary evidence seems to indicate that socioeconomic status (SES) and level of education have an impact on spoken discourse. The purpose of this study was to analyze microstructural variables in spoken discourse in people with AD with low-to-middle SES and low level of education and to study their association with gray matter (GM) density. Method: Nine women with AD and 10 matched (age, SES, and education) women without brain injury (WWBI) underwent a neuropsychological assessment, which included two spoken discourse tasks, and structural magnetic resonance imaging. Microstructural variables were extracted from the discourse samples using NILC-Metrix software. Brain density, measured by voxel-based morphometry, was compared between groups and then correlated with the differentiating microstructural variables. Results: The AD group produced a lower diversity of verbal time moods and fewer words and sentences than WWBI but a greater diversity of pronouns, prepositions, and lexical richness. At the neural level, the AD group presented a lower GM density bilaterally in the hippocampus, the inferior temporal gyrus, and the anterior cingulate gyrus. Number of words and sentences produced were associated with GM density in the left parahippocampal gyrus, whereas the diversity of verbal moods was associated with the basal ganglia and the anterior cingulate gyrus bilaterally. Conclusions: The present findings are mainly consistent with previous studies conducted in groups with higher levels of SES and education, but they suggest that atrophy in the left inferior temporal gyrus could be critical in AD in populations with lower levels of SES and education. This research provides evidence on the importance of pursuing further studies including people with various SES and education levels. What Is Already Known on This Subject: Spoken discourse has been shown to be affected in Alzheimer disease, but most studies have been conducted on individuals with middle-to-high SES and high educational levels. What This Study Adds: The study reports on microstructural measures of spoken discourse in groups of women in the early stage of AD and healthy women, with low-to-middle SES and lower levels of education. Clinical Implications of This Study: This study highlights the importance of taking into consideration the SES and education level in spoken discourse analysis and in investigating the neural correlates of AD. Supplemental Material: https://doi.org/10.23641/asha.24905046

show abstract

NILC-Metrix: assessing the complexity of written and spoken language in Brazilian Portuguese

Abstract: This is a repository copy of NILC-Metrix : assessing the complexity of written and spoken language in Brazilian Portuguese.

Cited by 4 publications

References 42 publications

Transferência de estilo textual arbitrário em português

Transferência de estilo textual arbitrário em português

Mean dependency length — a new metric for requirements quality

Exploring Spoken Discourse and Its Neural Correlates in Women With Alzheimer's Disease With Low Levels of Education and Socioeconomic Status

Contact Info

Product

Resources

About