This special issue is devoted to any aspect related to processing negation from a computational perspective. The call for papers was broad and welcomed contributions describing theoretical insights, annotation schemes and corpora, empirical studies on processing and representing the meaning of negation, and applications that benefit from processing negation. We received 15 submissions and 6 articles were accepted for publication after a thorough review process.

The special issue opens with a survey describing recent advances in processing negation (Blanco and Morante Reference Blanco and Morante2020). Six articles describing new research follow. These six articles can be grouped based on two criteria: the languages they work with and whether they process negation in order to improve some application. Three articles work with English texts (Schulder, Wiegand, and Ruppenhofer Reference Schulder, Wiegand and Ruppenhofer2020; Barnes, Velldal, and Øvrelid Reference Barnes, Velldal and Øvrelid2020; Sykes et al. Reference Sykes, Grivas, Grover, Tobin, Sudlow, Whiteley, McIntosh, Whalley and Alex2020), and three articles work with texts in other languages: Spanish (Jiménez-Zafra et al. Reference Jiménez-Zafra, Cruz-Díaz, Taboada and Martín Valdivia2020; Taul et al. Reference Taulé, Nofre, González and Martí2020), and French and Brazilian Portuguese (Dalloux et al. Reference Dalloux, Claveau, Grabar, Oliveira, Moro, Gumiel and Carvalho2020). Regarding applications, three articles present work on processing negation for sentiment analysis (Jiménez-Zafra et al. Reference Jiménez-Zafra, Cruz-Díaz, Taboada and Martín Valdivia2020; Schulder et al. Reference Schulder, Wiegand and Ruppenhofer2020; Barnes et al. Reference Barnes, Velldal and Øvrelid2020), two work in the biomedical domain (Dalloux et al. Reference Dalloux, Claveau, Grabar, Oliveira, Moro, Gumiel and Carvalho2020; Sykes et al. Reference Sykes, Grivas, Grover, Tobin, Sudlow, Whiteley, McIntosh, Whalley and Alex2020), and one presents a corpus with focus on negation annotations (Taul et al. Reference Taulé, Nofre, González and Martí2020). In the remaining of this introduction, we briefly summarize the articles in this special issue.

Schulder et al. (Reference Schulder, Wiegand and Ruppenhofer2020) present a methodology to generate lexica for sentiment polarity shifters such as alleviate and failure. They present a bootstrapping approach combining classifiers with human annotations, and the resulting lexica include polarity shifters that are nouns, verbs, or adjectives. Barnes et al. (Reference Barnes, Velldal and Øvrelid2020) show that modeling negation explicitly is beneficial for sentiment analysis. More specifically, they show that a multitask neural network architecture that learns both sentiment and scope of negation outperforms one that learns negation in an end-to-end manner. Finally, Jiménez-Zafra et al. (Reference Jiménez-Zafra, Cruz-Díaz, Taboada and Martín Valdivia2020) work with texts in Spanish and show that incorporating a scope detector into an existing model for sentiment analysis yields better results.

Taulé et al. (Reference Taulé, Nofre, González and Martí2020) define 10 criteria to identify the focus of negation in Spanish. The criteria have been defined based on a detailed and in-depth analysis of linguistic phenomena. They apply these criteria to the annotation of scope and focus in the NewsCom corpus, accounting for the first corpus annotated with focus in Spanish.

Dalloux et al. (Reference Dalloux, Claveau, Grabar, Oliveira, Moro, Gumiel and Carvalho2020) present new corpora for Brazilian Portuguese and French manually annotated with negation cues and their scopes in clinical documents. They also present automatic methods based on supervised machine learning approaches for the automatic detection of negation cues and their scopes, namely vector representations and neural networks. They find that LSTM-based neural architectures are more efficient than GRUs in the scope detection task. Sykes et al. (Reference Sykes, Grivas, Grover, Tobin, Sudlow, Whiteley, McIntosh, Whalley and Alex2020) present a new corpus of radiology reports annotated with negation. They also propose and experiment with a rule-based method and two neural network models for detecting negation in the new corpus. They conclude that their specialized rule-based method yields better results than the neural models, albeit by a small margin.