ACL Rolling Review (ARR) invites the submission of long and short papers on substantial, original, and unpublished research in all aspects of Computational Linguistics and Natural Language Processing. The purpose of ARR is to improve the efficiency, turnaround, and process consistency of ACL reviewing, while keeping the diversity (topical, geographic and otherwise) and iterative improvement through collaborative work with conference program chairs. ARR will use Open Review as its platform (but reviews will not be open in ARR). The reviewing and acceptance of papers for publication will be done in two steps:
- Step 1 – Centralized Rolling Review: Authors submit papers to a unified review pool with deadlines every two months. Review is handled by an area chair (AC), and revision and resubmission of papers are allowed.
- Step 2 – Commitment to a Publication Venue: A publication venue is a conference or workshop that accepts reviews from ARR. When an opportunity to commit to a publication venue comes around, authors may submit papers with fully completed reviews (including meta reviews). Program chairs decide the process for committing ARR reviewed papers, as well as the criteria and process for deciding to accept a subset of these submissions into their event.
SCOPE OF SUBMISSIONS
All topics in Computational Linguistics / Natural Language Processing are welcome at ARR. At submission, papers must select a topic area to assist with AC and reviewer matching. The current set of areas are (in alphabetical order):
- Computational Social Science and Cultural Analytics
- Dialogue and Interactive Systems
- Discourse and Pragmatics
- Efficient/Low-Resource Methods for NLP
- Ethics, Bias, and Fairness
- Generation
- Human-centered NLP
- Information Extraction
- Information Retrieval and Text Mining
- Interpretability and Analysis of Models for NLP
- Language Modeling
- Linguistic Theories, Cognitive Modeling, and Psycholinguistics
- Machine Learning for NLP
- Machine Translation
- Multilingualism and Cross-Lingual NLP
- Multimodality and Language Grounding to Vision, Robotics and Beyond
- NLP Applications
- Phonology, Morphology, and Word Segmentation
- Question Answering
- Resources and Evaluation
- Semantics: Lexical and Sentence-Level
- Sentiment Analysis, Stylistic Analysis, and Argument Mining
- Speech Recognition, Text-to-Speech and Spoken Language Understanding
- Summarization
- Syntax: Tagging, Chunking and Parsing
- Theme tracks (proposed by participating publication venues and specific to those venues)
For more information about how to choose the right area, see this page.
ARR welcomes diverse types of contributions in the above areas. When making a submission at ARR, the authors can indicate one or more of the following contribution types that their submission is making:
- Approaches to low-resource settings
- Approaches low compute settings-efficiency
- Data resources
- Data analysis
- Model analysis & interpretability
- NLP engineering experiment
- Publicly available software and/or pre-trained models
- Position papers
- Reproduction study
- Surveys (all papers are expected to include reviews of related literature. This category is meant for the papers that go beyond that, e.g. in scope or in establishing new interdisciplinary connections)
- Theory
Both positive and negative results for experimental studies are welcome, and have the same challenge of justifying to the program committee why this particular result is interesting and important. A negative results paper might report on non-reproducibility or non-generalizability of previously published results, their misattribution (‘right for the wrong reasons’), or present an idea that seemed great (for well-justified reasons), but didn’t work.
ARR welcomes scholars from other communities who make a substantive contribution to the computational processing of language. Here are some points to keep in mind for a successful submission:
- Interdisciplinary submissions typically aim to introduce concepts/methods/theories from one community to the CL/NLP community.
- For machine learning application papers on areas other than language, if the authors feel that their work fits the NLP applications track (e.g. because it finds a new application for a language model), they are welcome to try to convince the reviewers and area chairs that their work should be presented in this venue.
- A good rule-of-thumb to consider whether your work is connected to a given community is to consider whether your list of references has a substantial proportion of papers from that community.
PAPER SUBMISSION INFORMATION
Long Papers
Long papers must describe substantial, original, completed and unpublished work. Wherever appropriate, concrete evaluation and analysis should be included. Long papers may consist of:
- up to eight (8) pages of content
- unlimited extra space after the conclusion for limitations (required, see below) and optionally ethical considerations
- plus unlimited pages of references
Submissions that exceed the length requirements, or are missing a limitations section, will be desk rejected.
Short Papers
Short paper submissions must describe original and unpublished work. Please note that a short paper is not a shortened long paper. Instead, short papers should have a small, focused contribution that can be made in a few pages. Some common kinds of short papers are (non-)reproduction reports, negative results, opinion pieces, interesting application nuggets (but all these types of contributions could also be presented in long papers, given sufficient substance).
Short papers may consist of:
- up to four (4) pages of content
- unlimited extra space after the conclusion for limitations (required, see below) and optionally ethical considerations
- plus unlimited pages of references
Submissions that exceed the length requirements, or are missing a limitations section, will be desk rejected.
Instructions for Two-Way Anonymized Review
Papers must not include authors’ names and affiliations. Furthermore, self-references that reveal the authors’ identities, e.g., “We previously showed (Smith, 1991)…” must be avoided. Instead, use citations such as “Smith previously showed (Smith, 1991)…” Papers should not refer, for further detail, to documents that are not available to the reviewers.
Supplementary materials, including any links to repositories, should also be anonymized. This includes author responses during the review process. Links to file hosting services that can track downloads, such as Dropbox, are not allowed.
Submissions that violate these requirements will be desk rejected.
ARR is subject to the ACL Policies for Review and Citation, which were updated in early 2024. Beginning with the February 15, 2024 ARR deadlines, there is no anonymity period or limitation on posting or discussing non-anonymous preprints while the work is under peer review. However, the new policy incentivizes anonymous submissions by special paper awards and priority in acceptance decisions for borderline papers.
Authorship
The author list for submissions should include all (and only) individuals who made substantial contributions to the work presented. Each author listed on a submission to ARR will be notified of submissions and reviews. Please refer to the new ACL policy publication ethics for authorship criteria.
Please notice that once the paper has been submitted, no changes to the list of authors are allowed. There are additional authorship rules for resubmissions (see below).
Citation and Comparison
Authors are expected to cite all refereed publications relevant to their submission but may be excused for not knowing about all unpublished work (especially work that has been recently posted and/or is not widely cited).
In cases where a preprint has been superseded by a refereed publication, the refereed publication should be cited in addition to or instead of the preprint version.
Papers (whether refereed or not) appearing less than 3 months before the submission deadline are considered contemporaneous to a submission, and authors are therefore not obliged to make detailed comparisons that require additional experimentation and/or in-depth analysis.
For more information, see the ACL Policies for Review and Citation.
Multiple Submission Policy
There are several cases to consider:
- ARR + Other Venue: ARR precludes multiple submissions. ARR will not consider any paper that is under review in a journal or another conference at the time of submission, and submitted papers must not be submitted elsewhere during the ARR review period. This policy covers all journals and refereed and archival conferences and workshops without exception (e.g., TACL, Computational Linguistics, IJCAI, SIGIR, AAAI, ICASSP, ICML, NeurIPS, etc). In addition, we will not consider any paper that overlaps significantly in content or results with papers that will be (or have been) published elsewhere, without exception.
- ARR + Commitment: The commitment process is treated as being under review for a conference. That means (a) while in an active review cycle at ARR, a paper cannot be committed to a conference, and (b) between commitment to a conference and a decision or withdrawal, a paper cannot be submitted to ARR. For example, if you got reviews in the June 2023 cycle, you cannot both commit to EMNLP (August 22nd deadline) and submit a revised version to the August 2023 ARR cycle. Note that this is a change of policy relative to the early days of ARR, when this type of dual submission was permitted for ACL and NAACL 2022.
- Commitment + Commitment/Other Venue: Whether you can commit/submit to two venues simultaneously depends on the dual submission policies of those venues. Typically, it is not permitted.
- ARR + ARR: For the sake of clarity, this policy also covers ARR itself; authors may not resubmit to ARR work that is already under review to ARR. As of 2023, ARR does not have overlapping cycles, and so this is a moot point.
Submissions that violate requirements 1, 2, or 4 will be desk rejected.
Resubmission Policy
Authors may resubmit to ACL Rolling Review.
- When resubmitting, the authors must provide a link to the previous submission. Any paper previously submitted that does not acknowledge its earlier version will be desk rejected. To count as a new paper rather than a revision, the submission must be on a sufficiently different topic that the prior reviews no longer apply (e.g., further improving the proposed methods or extending the set of experiments or analysis does not qualify, nor does extending a short paper to a long one). Please note that resubmission does not mean that papers are stuck with old reviewers: revised papers can still be sent to new reviewers and meta-reviewers on authors’ request (see below), with the authors indicating why they prefer that option.
- To reduce the workload for the ACs and reviewers, and to focus the new reviews on the changes, the authors must provide a summary of changes and revisions in response to reviews from the previous round. For the latter, there should be an overview of changes and a point-by-point response to each weakness and suggestion from all reviewers (including meta-review where relevant), either describing the revision or arguing against it. It is recommended to refer to the location of revisions using section and line numbers. Some authors find it useful to color-code the changed parts in the text of the submission; this is allowed, but not mandated. A color-coded version of the paper may be included with the explanation of revisions in the same PDF.
- Resubmissions must be modified versions of the original submission that address issues raised by the reviewers and meta-review; the authors may not simply resubmit the exact same paper, nor may they submit a completely new paper as if it were a resubmission. Resubmissions that ignore relevant feedback provided during the previous review round can be desk rejected by the EiC team.
- If the authors want to add an author as part of a resubmission, they may do so with a (brief) justification; except in extremely rare circumstances, authors may not be removed.
- Resubmitted papers will go back to the original reviewers and area chair, where possible, unless the authors request new reviewers or AC (with justification, e.g. violations of reviewer guidelines). Note that new reviewers are instructed to form their own view of the paper first: they only see the previous reviews and author explanation of revisions after they submit their initial reviews. See the information for authors (https://aclrollingreview.org/authors#step4) for further discussion.
Submissions that had previously received very low evaluation from reviewers and a meta-review score of 1 generally need a wholesale revision that can be expected to take a lot of time. We reserve the right to desk reject such resubmissions if they are sent back to ARR within 6 months of the original submission without such a revision.
Authors considering a resubmission should also refer to the withdrawal policy below.
Withdrawal Policy
Authors may withdraw their submission at any time. However, a submission that is withdrawn more than 48 hours after the submission deadline may not be resubmitted until the second subsequent ARR cycle, so withdrawing after this time requires contacting the ACs.
Once the submission receives the ARR meta-review, it is no longer considered under review, and can be submitted or committed elsewhere with no restrictions. There is no need to withdraw a meta-reviewed paper to resubmit to ARR or commit to a conference that accepts ARR reviews. When resubmitting a revised version, please follow the resubmission policy above.
Once a paper receives even one review at ARR, it cannot be simply withdrawn and submitted as a new publication. If you intend to resubmit to ARR, do not withdraw the previous version. Simply resubmit following the above resubmission policy.
Ethics Policy
Authors are required to honour the ethical code set out in the ACL Code of Ethics.
The consideration of the ethical impact of our research, use of data, and potential applications of our work has always been an important consideration, and as artificial intelligence is becoming more mainstream, these issues are increasingly pertinent. We ask that all authors read the code, and ensure that their work is conformant to this code. Authors are encouraged to devote a section of their paper to concerns about the ethical impact of the work and to a discussion of broader impacts of the work, which will be taken into account in the review process. This discussion does not count towards the page limit, as long as it is placed in the end of the paper. In addition, we provide a responsible NLP research checklist, which authors must complete as part of their paper submission. The content of checklists is used in the review process, and incorrect or misleading information in the checklist can result in desk rejection. Furthermore, ACL’25 will resume the publication of responsible NLP checklists for accepted papers for improving the overall transparency of the scientific process.
We reserve the right to reject papers on ethical grounds, where the authors are judged to have operated counter to the code of ethics, or have inadequately addressed legitimate ethical concerns with their work. Indeed, the ARR review form includes a section addressing these issues and papers flagged for ethical concerns by reviewers or ACs will be further reviewed by the Ethics Advisory Committee (EAC).
AI Writing/Coding Assistance Policy
Generally, generative AI tools do not qualify for authorship. Their use for writing or coding, as well as its scope, must be disclosed in the Acknowledgements section, as well as flagged in the Responsible NLP Checklist. For coding, details may also be included in the README files.
ARR follows the new ACL Policy on AI Writing Assistance, which recognizes the following cases are recognized as appropriate use of such tools:
- a. Assistance purely with the language of the paper. This covers models used for paraphrasing or polishing the author’s original content, rather than for suggesting new content—similar to tools like grammar checkers, spell checkers, dictionaries, and synonym tools. The use of tools that only assist with proofreading, like grammar or spell checkers, does not need to be disclosed.
- b. Short-form input assistance. This covers predictive keyboards or tools that offer suggestions during typing, that might be powered by generative language models. The use of such tools does not need to be disclosed.
- c. Literature search. This covers search assistants, e.g., to identify relevant literature. The usual requirements for citation accuracy and thoroughness of literature reviews apply.
- d. Low-novelty text. This covers the automatic generation of text about pre-existing ideas. Authors should specify where such automatically generated text was used, and convince the reviewers that the generation was checked to be accurate and is accompanied by relevant and appropriate citations. If the generation copies text from existing work, the publication ethics policy applies to that text. Authors need (for example) to acknowledge all relevant citations: both the source of the text used and the source of the idea(s).
- e. New ideas. This covers when generative model output reads to the authors as new research ideas that would deserve co-authorship or acknowledgment from a human colleague (e.g., topics to discuss, framing of the problem), which the authors then develop themselves. As with all new ideas, the authors should conduct a literature search to determine relevant prior work and cite to ensure proper credit. The authors should disclose if models were used in this manner.
- f. New ideas + new text. ACL does not consider a generative model to be an entity that can fulfill the requirements of co-authorship.
In all cases, all authors are fully responsible for the correctness of their methods, results, and writing. They should check for potential plagiarism, both of text and code.
Limitations (required section)
Authors are required to discuss the limitations of their work in a dedicated section titled “Limitations”. This section should be included at the end of the paper, before the references, and it will not count toward the page limit. This includes both, long and short papers. Papers without a limitations section will be desk rejected. Note, prior to the December 2023 cycle, this was optional.
Please note that this section should not introduce new methods, analysis, or results. We reserve the right to desk reject the submissions that use this section to introduce more content that should have been part of the main paper. It can only discuss the limitations of the work presented in the main content of the paper.
Reviewing Requirements
Submitting to ARR comes with a reviewing requirement. At least one author must participate in the review process, unless an exception applies (e.g., every author is either new to the community, has insufficient experience, or is already serving in another capacity). Papers that do not meet this requirement, and are not covered by an exception, may be desk rejected. There are of course legitimate reasons for one to suddenly be unavailable to review, and we will respect these (e.g. already serving in ARR in another capacity).
Authors who will review are nominated at submission time. They must:
- Have an updated OpenReview profile, including affiliation, semantic scholar link, dblp link, ACL anthology link, and an email address where they can receive OpenReview messages. See here for how to maximize the likelihood of well-matched review assignments.
- Have published at least three papers in the main computational linguistics and NLP venues in the last 5 years. (we consider publication history recorded at ACL anthology, which covers AACL/IJCNLP, ACL, CL, COLING, CONLL, EACL, EMNLP, NAACL, LREC, and TACL). Both short and long papers count, as do Findings papers.)
- Once added as a reviewer, the new reviewers must complete their reviewer registration form for the cycle and set their availability to at least 4 papers within 48 hours.
- The nominated authors who already have ARR reviewer profiles must also set their load for the current cycle to at least 4 papers per submission, prior to submission. In case of the same research group submitting multiple papers, it is the authors’ responsibility to coordinate within the group and ensure that either different senior authors are nominated, or the senior author undertakes a proportionately higher review load and will mentor junior subreviewers.
- All reviewers must familiarize themselves with the latest version of the ARR reviewer guidelines and perform the reviews on time and in line with ARR requirements.
The number of papers assigned to each reviewer in a given cycle depends in part on the paper matching process. To mitigate the issue of overwhelming volume of assignments, ARR has implemented a load balance check across cycles. Please note that even with four assigned papers, contributing four reviews doesn’t fully reciprocate the volunteer effort that the authors receive for their submission to complete a full review cycle (it involves not just three reviews and a meta-review from the AC, potentially with emergency substitutions, but also effort from the SACs and PCs during paper commitment).
See this blog post for further information about this requirement (as of April 2024).
Paper Submission Process, Criteria and Templates
Submission is electronic, using the OpenReview.net platform. All long, short, and theme papers must follow the ACL Author Guidelines. Here are the paper submission form fields for your reference.
ARR provides a submission checklist. The checklist is intended as a reminder to help authors improve the quality of their papers, and decrease the risk of a desk rejection.
Paper submissions must use the official ACL style templates, which are available here (Latex and Word). Please follow the paper formatting guidelines general to “*ACL” conferences available here. Authors may not modify these style files or use templates designed for other conferences.
Submissions that do not conform to the required styles, including paper size, margin width, and font size restrictions, will be rejected without review.
Here is the current version of the review form, and here is the current version of the AC meta-review form. These forms will be re-assessed and updated periodically.
Desk Rejection
Papers can be desk-rejected for a variety of reasons, including format and anonymity violations, dual submissions, and self-plagiarism (significant overlap in content with other submissions or publications by the same authors). We provide authors with a checklist to assist with avoiding some of the common issues (the list is not exhaustive).
Starting from December 2024, ARR will also enforce desk rejections for incorrect or misleading filing of the responsible NLP checklist, and for violating the resubmissions policy (i.e. submissions that are not accompanied with either revisions in response to prior reviews, or justifications for not doing so, as described in the section on resubmissions).
Note that while we do our best for the bulk of desk rejections to happen before the start of the review process, sometimes serious problems are missed at initial checks and discovered late. Hence, desk-rejections can happen at any time in the review cycle.
Optional Supplementary Materials: Appendices, Software, and Data
ARR encourages the submission of these supplementary materials to improve the reproducibility of results and to enable authors to provide additional information that does not fit in the paper. Supplementary materials may include appendices, software, or data. For example, pre-processing decisions, model parameters, feature templates, lengthy proofs or derivations, pseudocode, sample system inputs/outputs, and other details that are necessary for the exact replication of the work described in the paper can be put into appendices.
However, reviewers are not required to consider material in appendices. If the pseudo-code, or derivations, or model specifications are an important part of the contribution, or if they are important for the reviewers to assess the technical correctness of the work, they should be a part of the main paper rather than appendices.
Appendices should come after the references in the submitted pdf. They do not count towards the page limit. Software should be submitted as a single .tgz or .zip archive, and data as a separate single .tgz or .zip archive. Supplementary materials must be fully anonymized to preserve the two-way anonymized reviewing policy.
If supplementary software is provided through a link to an online repository, it should be properly anonymized (e.g., Anonymous GitHub). Links to cloud services like Google Drive, Dropbox etc. are not acceptable, as they might track the download activity.