2nd CALL FOR PAPERS
A Special Issue of the Computational Linguistics Journal
Modality and Negation
Computational linguistics has seen achievements in handling language at different levels of linguistic abstraction, from tokenization to semantic role labeling. Given a sentence, systems can more or less reliably determine who does what to whom when and where. However, texts do not always express factual information; on the contrary, language is often used to express uncertainty, opinion, evaluation, or doubt. Accordingly, computational linguistics has started to take into account the subjective aspects of language. There is now research that focuses also on determining who states that someone does something somewhere at a certain point in time (perspective) and based on what evidence (evidentiality), how certain someone is about stating something (certainty), the truth value of the facts being stated (negation), or the subjective evaluation of these facts (positive/negative opinion).
Linguistic phenomena such as modality and negation allow the expression of subjective aspects of meaning. Modality and negation are two information-level concepts that are well described from a philosophical perspective. Modality (Palmer 1986) is related to the attitude of the speaker towards her statements in terms of degree of certainty, reliability, subjectivity, sources of information, and perspective. It is related to other concepts like hedging (Hyland 1998), evidentiality (Aikhenvald 2004), uncertainty (Rubin et al. 2005), and factuality (Saurí 2008). Negation (Tottie 1991, Horn 2001) is used to position information as a counterfact, a fact that does not hold in the world. Both modality and negation are complex linguistic phenomena that are challenging both from a theoretical and a computational point of view. Their complexity is due to the fact that both phenomena interact with each other (de Haan 1997) and with other aspects of the linguistic context, such as mood, tense, and lexis. While modality and negation tend to be lexically marked, the class of markers is relatively large and heterogeneous. For example, while negation words such as "not" are clear indicators of negation, other terms such as modals, adverbs, conjunctions and multi-word expressions can also express negation and subjectivity. Moreover, processing modality and negation involves disambiguating the markers and determining their scope.
The treatment of modality and negation is very relevant for all NLP applications that involve deep text understanding. This includes applications that need to discriminate between factual and non-factual information (uncertain facts, opinions, attitudes, emotions, and beliefs), such as information extraction, opinion mining, sentiment analysis, text mining, and question answering, as well as other applications that process the meaning of texts, such as recognizing textual entailment, paraphrasing, and summarization. Hence, the adequate modeling of these phenomena is of crucial importance to the NLP community as a whole. While the area is still relatively new compared to areas like machine translation, parsing or semantic role labeling, it is now growing quickly.
For this special issue we solicit full-length article submissions describing innovative and challenging research on aspects of the computational modelling and processing of modality and negation. We specifically invite submissions that take into account linguistic aspects of the phenomena and bring a theoretical basis to research on computing the factuality and certainty of the events in a statement, finding the source and evidence for the statement of a fact, and determining whether a statement has a truth value. We encourage submissions that have a substantial analysis component, in the form of an analysis of the task and data and/or an error analysis of the proposed method. Submissions can address aspects of either modality or negation or both, provided that they lead to an enhanced understanding of the phenomena, as opposed to a straightforward engineering solution.
Possible topics include, but are not limited to:
- Linguistically informed modelling of modality and negation for NLP
- Analysis of the relevant information/knowledge involved in processing modality and negation
- The computational complexity of processing modality and negation
- Novel machine learning approaches for learning modality and negation
- Processing modality and negation across domains and genres
- The interaction of modality and negation for determining the factuality of events
- The influence of the linguistic context on the processing of modality and negation
- Evaluation of systems: metrics and application-based evaluation
- Submission of full articles: 31 March 2011
- Preliminary decisions to authors: 15 July 2011
- Submission of revised articles: 15 September 2011
- Final decisions to authors: 1 November 2011
- Final versions due from authors: 25 November 2011
Articles submitted to this special issue must adhere to the Style Guidelines of the Computational Linguistics Journal (http://cljournal.org/style.html). The submission guidelines can be found in the Computational Linguistics web site (http://cljournal.org/submissions.html). As in regular submissions to the journal, paper submissions should be made through the CL electronic submission system (http://cljournal.org/submissions/index.php/cljournal).
CLiPS - University of Antwerp, Belgium
Computational Linguistics and Phonetics - Saarland University, Germany