Area Descriptions

When submitting a paper to a conference, it is sometimes difficult to determine which track a given paper should be submitted to. In order to help with such decisions, we provide more detailed descriptions for the NAACL 2021 tracks below.

We are aware that the areas overlap and that there may not be a perfect fit for every submission. At submission time, please choose the area that you think fits best (and optionally, a secondary area.) The area chairs and reviewers in each area will have broad expertise to cover these cases, including work in new and emerging areas. The PC co-chairs will also rebalance papers after submission, so your final track may change from what you initially selected.

Computational Social Science and Cultural Analytics. This area includes work that tries to understand and analyze human behavior through the lens of language (including social media text), as well as the work at the intersection of NLP with literature and culture.
Dialogue and Interactive systems. This area includes work on conversational AI and systems that participate in or analyze written and spoken conversations between people and machines.
Discourse and Pragmatics. This area includes work that studies texts longer than one sentence, but does not fall into related areas (e.g. Dialog or Summarization). This includes topics such as coreference resolution, discourse parsing, and argument mining.
Ethics, Bias, and Fairness. This area includes work that analyzes, detects and mitigates stereotypical bias or offensive wording in language data as well as work discussing ethical concerns about NLP applications.
Green NLP. This area includes work whose primary focus is to measure or reduce the amount of compute required to build effective models.
Language Generation. This area includes work on models, data, and analysis for building systems that produce natural language text, but does not fall into related areas (e.g. Dialog or Summarization).
Information Extraction. This area includes entity and relation extraction, as well as work that converts text to a more structured form such as a database or knowledge base, inferring ontologies, and populating and reasoning with knowledge bases more generally.
Information Retrieval and Text Mining. This area includes work on Information Retrieval focused on text, such as document crawling, indexing, search, retrieval, , ranking, classification, and clustering.
Interpretability and Analysis of Models for NLP. This area includes work that aims to understand or explain models, their components, and their behavior, using various methods and approaches (black-box analyses, intrinsic interpretability, challenge datasets, etc.).
Language Grounding to Vision, Robotics and Beyond. This area includes work that bridges between language and aspects of the physical world such as images, video, robotics, time and space.
Language Resources and Evaluation. This area includes work on new tasks, datasets, and evaluation metrics. Some resource and evaluation papers may be more suitable for the task-specific tracks (e.g. dialog, MT, or QA), and the authors should use their best judgment to decide where to submit.
Linguistic Theories, Cognitive Modeling and Psycholinguistics. This area includes work that studies or models human language processing, typology and acquisition, and work that is more theoretic in nature, e.g. without implemented systems.
Machine Learning for NLP: Classification and Structured Prediction Models. This area includes work on models that make discrete structured decisions (e.g. multi-class, graphs, or trees), whether latent, fully, or otherwise supervised.
Machine Learning for NLP: Language Modeling and Sequence to Sequence Models. This area includes models that predict sequences of text, which can include pre-training and are often self-supervised based on naturally written text.
Machine Translation. This area includes work that aims to translate one human language to another, generally construed, including but not limited to work on multilingual, unsupervised, and document-level machine translation.
Multilinguality. This area includes work where the focus is on models (or data) that specifically work on multiple languages at the same time (such as multilingual BERT or XLM Roberta-based methods). This includes work directed at improving performance holistically across many languages, and not on improving performance on one language at a time.
NLP Applications. This area includes work that is primarily application-oriented. Work appropriate for this track includes applied work in different areas such as education, language learning, health, and others. Also welcome are efforts that study the challenge of deploying NLP systems and in-depth, real-world evaluations of systems in practice.
Phonology, Morphology and Word Segmentation. This area includes work that studies properties of subword units and how they combine to form meaning units.
Question Answering. This area includes work focused on automatically answering questions posed in natural language in a variety of styles, including but not limited to factoid/non-factoid, textual/knowledge-base QA, open-domain, reading comprehension, or common sense QA.
Semantics: Lexical Semantics. This area includes work that focuses primarily on the meaning of individual words and phrases, including word and entity embeddings.
Semantics: Sentence-level Semantics and Textual Inference. This area includes work that focuses on sentence-level semantics, sentence embeddings, compositionality, and reasoning, including semantic parsing and natural language inference.
Sentiment Analysis and Stylistic Analysis. This area includes work that focuses on non-literal and non-factual aspects of the meaning or style of natural language text.
Speech. This area is appropriate for all work that includes speech signals, including both recognition and synthesis.
Summarization. This area includes work that focuses on producing a smaller natural language summary of a focused aspect of a larger text collection.
Syntax: Tagging, Chunking, and Parsing. This area includes work on data, algorithms and models for analysis of syntactic and related structure in text.