ACL Policy on Publication Ethics - Revision history

Aoife Cahill: Switch to ARR policy on dual submission rather than the outdated wiki from ACL

2026-03-18T13:55:26Z

Switch to ARR policy on dual submission rather than the outdated wiki from ACL

← Older revision		Revision as of 13:55, 18 March 2026
Line 210:		Line 210:
	b. If the previously published or submitted material is used as a basis for a new submission, clearly indicate how the new submission differs from the previously published work(s).		b. If the previously published or submitted material is used as a basis for a new submission, clearly indicate how the new submission differs from the previously published work(s).

	Due to the volume of submissions received with duplicated content, the overlap threshold is now 10%. This means that there must be no more than a 10% (e.g., in terms of number of tokens) overlap between: (a) the text that a manuscript, at any point in the publication process, presents as original; (b) any other works submitted to ACL venues or published anywhere; and (c) concurrent submissions under review at ACL venues. Author’s own pre-prints are excluded from this limitation. See also the [[~~Double_Submission_Policy_for_Conferences~~\|ACL policy on double submission]].		Due to the volume of submissions received with duplicated content, the overlap threshold is now 10%. This means that there must be no more than a 10% (e.g., in terms of number of tokens) overlap between: (a) the text that a manuscript, at any point in the publication process, presents as original; (b) any other works submitted to ACL venues or published anywhere; and (c) concurrent submissions under review at ACL venues. Author’s own pre-prints are excluded from this limitation. See also the [[https://aclrollingreview.org/cfp\|ACL policy on double submission]].

	==== Electronic Posting ====		==== Electronic Posting ====
Line 218:		Line 218:
	==== Dual Submission ====		==== Dual Submission ====

	Authors should follow the [[~~Double_Submission_Policy_for_Conferences~~\|ACL policy on double submission]]. Articles submitted for consideration should not have been published previously and should not be concurrently under consideration for publication elsewhere. Authors must disclose all prior publication(s) and current submissions when submitting an article. At the venue chairs’ discretion, some forms of submission articles may be exempted from this rule.		Authors should follow the [[https://aclrollingreview.org/cfp\|ACL policy on double submission]]. Articles submitted for consideration should not have been published previously and should not be concurrently under consideration for publication elsewhere. Authors must disclose all prior publication(s) and current submissions when submitting an article. At the venue chairs’ discretion, some forms of submission articles may be exempted from this rule.

	==== Author Conduct ====		==== Author Conduct ====
Line 323:		Line 323:
	h. Reviewers and editors must not use confidential information from a reviewed manuscript for personal advantage before the information is publicly available, including leveraging research ideas or data.		h. Reviewers and editors must not use confidential information from a reviewed manuscript for personal advantage before the information is publicly available, including leveraging research ideas or data.

	i. Secondary reviewers, if any, should first be officially added to the reviewing system before they are invited to provide a review, and they must similarly treat the manuscript as a confidential document.		i. Secondary reviewers, if any, should first be officially added to the reviewing system before they are invited to provide a review, and they must similarly treat the manuscript as a confidential document. Please follow ACL's policy on secondary reviewers at [https://aclrollingreview.org/reviewerguidelines#q-can-i-use-a-secondary-reviewer? \| ARR Reviewer Guidelines]

	j. Reviewers and editors must report observed discrepancies or signs of potential author, reviewer, or editor misconduct, such as plagiarism, data fabrication, data misuse, or duplicate submission to the designated channels.		j. Reviewers and editors must report observed discrepancies or signs of potential author, reviewer, or editor misconduct, such as plagiarism, data fabrication, data misuse, or duplicate submission to the designated channels.

Nmadnani at 18:13, 14 March 2026

2026-03-14T18:13:40Z

← Older revision		Revision as of 18:13, 14 March 2026
Line 1:		Line 1:
	~~<!-- <p style="font-size:2em">ACL Policy on Publication Ethics</p> -->~~
	<hr/>		<hr/>

Kokilj: /* Definitions */ added and ordered alphabetically

2025-09-26T12:54:38Z

Definitions: added and ordered alphabetically

← Older revision		Revision as of 12:54, 26 September 2025
Line 28:		Line 28:
	== Definitions ==		== Definitions ==

	'''~~Reviewer~~'''		'''ACL Venues'''

			The superset of ACL submission, reviewing, and publication venues, including: ACL Rolling Review; ACL main conference; ACL Regional body main conferences; ACL SIG conferences; workshops at these conferences; TACL; CL.

			'''ARR commitment'''

			The six-month period during which an ARR-reviewed paper can be committed to a venue without re-review.

			'''Author'''

			Any person named in the “Authors” field of a submission.

			'''Camera-ready'''

	~~Reviewer refers~~ to ~~people writing peer reviews of manuscripts submitted~~ for ~~consideration~~ to ~~ACL venues~~.		The last date by which the accepted paper, formatted to venue specifications, must be delivered for publication. “Post camera-ready” refers to any time after this date.

	'''~~Editor~~'''		'''Conflict of interest'''

	~~Editor refers to people involved in making~~ a ~~decision about a submitted manuscript’s rejection, acceptance, scheduling, or format, but not writing direct (i.e. non~~-~~meta) reviews. Editors include action editors, area chairs, senior area chairs, programme chairs, and similar roles~~.		The ACL definitions of a conflict of interest are detailed at the [[ACL Conference Conflict-of-interest policy]] page.

	'''Convener'''		'''Convener'''
Line 40:		Line 52:
	A person who arranges meetings of groups or committees, for example conferences or workshops.		A person who arranges meetings of groups or committees, for example conferences or workshops.

	'''~~Author~~'''		'''Corresponding author'''

	~~Any person named in~~ the ~~“Authors” field of a submission~~.		A designated author responsible for all communication with the venue. A manuscript must keep at least one and no more than two corresponding authors at all times.

	'''~~Works~~'''		'''Editor'''

	~~Author-created material~~ submitted ~~for review~~ or ~~created for presentation~~. ~~This includes manuscripts as well as appendices, presentations, code~~, ~~datasets~~, ~~videos~~, ~~images~~, ~~etc~~.		Editor refers to people involved in making a decision about a submitted manuscript’s rejection, acceptance, scheduling, or format, but not writing direct (i.e. non-meta) reviews. Editors include action editors, area chairs, senior area chairs, programme chairs, and similar roles.

	'''~~Conflict of interest~~'''		'''Final decision'''

	The ACL ~~definitions of a conflict of interest are detailed at~~ the ~~[[ACL Conference Conflict~~-of-~~interest policy]] page~~.		The programme-chair or journal-editor outcome on a manuscript (accept, reject, assign to venue). For papers routed through ACL Rolling Review (ARR), the final-decision date is the venue-assignment decision.

	'''~~ACL Venues~~'''		'''First author'''

	The ~~superset of ACL~~ submission~~, reviewing, and publication venues, including: ACL Rolling Review; ACL main conference; ACL Regional body main conferences; ACL SIG conferences; workshops at these conferences; TACL; CL~~.		The author listed first on the manuscript; typically the primary contributor. Under this policy, the first author position cannot change after the submission deadline.

	'''Generative tools'''		'''Generative tools'''

	Here we include text and image generation tools, commonly referred to as generative AI tools, trained on massive datasets to generate multimedia outputs. These can be privacy-preserving or non-privacy preserving. An example of a non-privacy preserving tool is a text or image generation tool, that stores input data for a commercial purpose, and is typically hosted on remote servers managed by third-party companies. An example of a privacy-preserving tool is a text and image generation tool that doesn’t store input data.		Here we include text and image generation tools, commonly referred to as generative AI tools, trained on massive datasets to generate multimedia outputs. These can be privacy-preserving or non-privacy preserving. An example of a non-privacy preserving tool is a text or image generation tool, that stores input data for a commercial purpose, and is typically hosted on remote servers managed by third-party companies. An example of a privacy-preserving tool is a text and image generation tool that doesn’t store input data.

			'''Reviewer'''

			Reviewer refers to people writing peer reviews of manuscripts submitted for consideration to ACL venues.

			'''Review process (for conference submissions)'''

			The interval beginning at the submission deadline and ending at the final decision (acceptance or rejection by the venue).

			'''Resubmission'''

			A manuscript submitted again as a new entry after (a) rejection, (b) a “major revisions” outcome, or (c) expiry of an ARR commitment window. Resubmissions reset authorship alteration rules as described in Section 4.

	'''Scientific integrity'''		'''Scientific integrity'''

	Adherence to scientific ethical principles and professional standards essential for the responsible practice of research.		Adherence to scientific ethical principles and professional standards essential for the responsible practice of research.

			'''Submission deadline'''

			The official cut-off time at which a manuscript (or re-submission) must be uploaded to be considered for review.

			'''Top-level domain (TLD)'''

			The final segment of an Internet domain name (e.g., .edu, .ac.uk, .org). Email-address changes that keep the same TLD (3-g) are treated differently from those that introduce a new TLD (3-h).

			'''Works'''

			Author-created material submitted for review or created for presentation. This includes manuscripts as well as appendices, presentations, code, datasets, videos, images, etc.

	== Scope ==		== Scope ==

Kokilj: approval dates

2025-09-26T12:52:06Z

approval dates

← Older revision		Revision as of 12:52, 26 September 2025
Line 2:		Line 2:
	<hr/>		<hr/>

	''Approved by the ACL Exec, 2024-06-15, 2025-04-25; Authored by Cahill, A., Derczynski, L., & Jaidka, K.''		''Approved by the ACL Exec, 2024-06-15, 2025-04-25, 2025-07-19, 2025-09-26; Authored by Cahill, A., Derczynski, L., & Jaidka, K.''

	<hr/>		<hr/>

Kokilj: /* Scientific Integrity */ corrected in-text citations

2025-07-20T03:06:36Z

Scientific Integrity: corrected in-text citations

← Older revision		Revision as of 03:06, 20 July 2025
Line 102:		Line 102:
	d. Selective reporting: This includes omitting data or results in such a way that the research, intentionally or not, becomes misleading; applying inappropriate or misleading statistical techniques to enhance the significance or importance of research findings; or drastically changing the reporting of the conditions or results of experiments conducted.		d. Selective reporting: This includes omitting data or results in such a way that the research, intentionally or not, becomes misleading; applying inappropriate or misleading statistical techniques to enhance the significance or importance of research findings; or drastically changing the reporting of the conditions or results of experiments conducted.

	e. Neglecting ethical standards: Engaging in research involving human subjects without adherence to established ethical standards and applicable legal or institutional requirements. This includes failing to obtain informed consent from participants, debriefing participants in studies involving partial disclosure, not ensuring the confidentiality and privacy of the data collected where required, and conducting human subjects research without oversight by an Institutional Review Board (IRB) or an equivalent ethics review body where available. In cases where no such body exists, authors must justify the ethical safeguards implemented, following internationally recognized ethical standards such as the Belmont Report [1], the Declaration of Helsinki [2], or the OECD AI Principles [3].		e. Neglecting ethical standards: Engaging in research involving human subjects without adherence to established ethical standards and applicable legal or institutional requirements. This includes failing to obtain informed consent from participants, debriefing participants in studies involving partial disclosure, not ensuring the confidentiality and privacy of the data collected where required, and conducting human subjects research without oversight by an Institutional Review Board (IRB) or an equivalent ethics review body where available. In cases where no such body exists, authors must justify the ethical safeguards implemented, following internationally recognized ethical standards such as [https://www.hhs.gov/ohrp/regulations-and-policy/belmont-report/index.html the Belmont Report], [https://www.wma.net/policies-post/wma-declaration-of-helsinki/ the Declaration of Helsinki], or [https://oecd.ai/en/ai-principles the OECD AI Principles].

	g. Flawed evaluation practices: Continuing to use evaluation measures that are known to be flawed, unverifiable, or misleading without acknowledging their limitations undermines the reliability of research findings and can be considered a breach of scientific integrity. While we recognize that community norms and common practices influence evaluation choices, authors are responsible for critically assessing the evaluation methods they adopt. Where known flaws exist, authors should disclose them and, when possible, take steps to improve evaluation practices.		g. Flawed evaluation practices: Continuing to use evaluation measures that are known to be flawed, unverifiable, or misleading without acknowledging their limitations undermines the reliability of research findings and can be considered a breach of scientific integrity. While we recognize that community norms and common practices influence evaluation choices, authors are responsible for critically assessing the evaluation methods they adopt. Where known flaws exist, authors should disclose them and, when possible, take steps to improve evaluation practices.

Kokilj: Added references for scientific integrity

2025-07-20T03:04:09Z

Added references for scientific integrity

← Older revision		Revision as of 03:04, 20 July 2025
Line 360:		Line 360:

	[https://aclanthology.org/2022.findings-emnlp.35/ Kirk, H. R., Birhane, A., Vidgen, B., & Derczynski, L. (2022). Handling and Presenting Harmful Text in NLP Research. In Findings of EMNLP. Association for Computational Linguistics.]		[https://aclanthology.org/2022.findings-emnlp.35/ Kirk, H. R., Birhane, A., Vidgen, B., & Derczynski, L. (2022). Handling and Presenting Harmful Text in NLP Research. In Findings of EMNLP. Association for Computational Linguistics.]

			[https://www.hhs.gov/ohrp/regulations-and-policy/belmont-report/index.html National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research. (1979). The Belmont Report: Ethical principles and guidelines for the protection of human subjects of research. U.S. Department of Health, Education, and Welfare.]

			[https://oecd.ai/en/ai-principles Organisation for Economic Co-operation and Development. (2019). OECD principles on artificial intelligence. OECD.]

	[https://www.usenix.org/conference/usenixsecurity23/instructions-presenters USENIX: Instructions for Presenters]		[https://www.usenix.org/conference/usenixsecurity23/instructions-presenters USENIX: Instructions for Presenters]

			[https://www.wma.net/policies-post/wma-declaration-of-helsinki/ World Medical Association. (2013). World Medical Association Declaration of Helsinki: Ethical principles for medical research involving human subjects. JAMA, 310(20), 2191–2194.]

Kokilj: /* Scientific Integrity */

2025-07-20T03:01:06Z

Scientific Integrity

← Older revision		Revision as of 03:01, 20 July 2025
Line 112:		Line 112:
	j. Inappropriate use of AI: Authors and reviewers are encouraged to use generative tools responsibly and transparently, in accordance with the policies on [[#Generative Assistance in Authorship]] and [[#Generative Assistance in Peer Review]]. Their use in ways that are inconsistent with the [[ACL Policy on Publication Ethics]] is prohibited.		j. Inappropriate use of AI: Authors and reviewers are encouraged to use generative tools responsibly and transparently, in accordance with the policies on [[#Generative Assistance in Authorship]] and [[#Generative Assistance in Peer Review]]. Their use in ways that are inconsistent with the [[ACL Policy on Publication Ethics]] is prohibited.

	As outlined in the [ACL Ethics Review ~~and Enforcement Processes~~], breaches of the ACL Policy may lead to actions consistent with ACL’s post-publication ethics procedures, such as public correction statements or, in rare cases, retraction.		As outlined in the [[Process for ACL Publication Ethics Review]], breaches of the ACL Policy may lead to actions consistent with ACL’s post-publication ethics procedures, such as public correction statements or, in rare cases, retraction.

	==== Guidelines for Generative Assistance in Authorship ====		==== Guidelines for Generative Assistance in Authorship ====

Kokilj: /* Authorship */ added Scientific Integrity section

2025-07-20T02:57:42Z

Authorship: added Scientific Integrity section

@@ Line 89: / Line 89: @@
 Authors must follow the [https://www.aclweb.org/adminwiki/index.php/Authorship_Changes_Policy_for_ACL_Conference_Papers Authorship Changes Policy for ACL Conference Papers] as well as venue-specific policies.
 ==== Guidelines for Generative Assistance in Authorship ====
@@ Line 185: / Line 209: @@
 f. This policy may not be applied to works published before this policy comes into effect (April 25<sup>th</sup> 2025), even if a work in scope appears to have contravened it.
 === Content of Submitted, Published, or Presented Material ===

Kokilj: /* Content Guidelines */

2025-07-20T02:51:08Z

Content Guidelines

← Older revision		Revision as of 02:51, 20 July 2025
Line 209:		Line 209:

	g. For content about broader impact, see below.		g. For content about broader impact, see below.

			h. Scientific Integrity: Content that constitutes a breach of scientific integrity is unacceptable. For a description of what constitutes breaches in scientific integrity and how they are handled, refer to the [[#Scientific Integrity]] section under [[#Authorship]].

	Examples of unallowable content include changes to the content of the paper during review or after acceptance that deviate from the original scope of the paper or constitute unacceptable content; conjecture, unless relevant to the presented research supported by citations; illegal content; potentially harmful content, unless preceded by a warning and gap (see policy on [[#Harmful content]]); toxic content/inflammatory writing/hate speech unless it is a necessary scientific example (see policy on [[#Harmful content]]); hidden or obscured prompts or cues inserted into the paper to manipulate its automated processing (see [[#Guidelines for Generative Assistance in Authorship]]).		Examples of unallowable content include changes to the content of the paper during review or after acceptance that deviate from the original scope of the paper or constitute unacceptable content; conjecture, unless relevant to the presented research supported by citations; illegal content; potentially harmful content, unless preceded by a warning and gap (see policy on [[#Harmful content]]); toxic content/inflammatory writing/hate speech unless it is a necessary scientific example (see policy on [[#Harmful content]]); hidden or obscured prompts or cues inserted into the paper to manipulate its automated processing (see [[#Guidelines for Generative Assistance in Authorship]]).

Kokilj: /* Content Guidelines */ example and in-text citation

2025-07-20T02:49:44Z

Content Guidelines: example and in-text citation

← Older revision		Revision as of 02:49, 20 July 2025
Line 210:		Line 210:
	g. For content about broader impact, see below.		g. For content about broader impact, see below.

	Examples of unallowable content include changes to the content of the paper during review or after acceptance that deviate from the original scope of the paper or constitute unacceptable content; conjecture, unless relevant to the presented research supported by citations; illegal content; potentially harmful content, unless preceded by a warning and gap (see policy on [[#Harmful content]]); toxic content/inflammatory writing/hate speech unless it is a necessary scientific example (see policy on [[#Harmful content]]).		Examples of unallowable content include changes to the content of the paper during review or after acceptance that deviate from the original scope of the paper or constitute unacceptable content; conjecture, unless relevant to the presented research supported by citations; illegal content; potentially harmful content, unless preceded by a warning and gap (see policy on [[#Harmful content]]); toxic content/inflammatory writing/hate speech unless it is a necessary scientific example (see policy on [[#Harmful content]]); hidden or obscured prompts or cues inserted into the paper to manipulate its automated processing (see [[#Guidelines for Generative Assistance in Authorship]]).

	==== Societal Impact ====		==== Societal Impact ====