ACL Rolling Review opens August cycle amid NLP submission surge

The ACL Rolling Review's August 2026 cycle opens submissions on August 3, and it lands on a system that is visibly bending under NLP's growth. Per the ARR dates page, reviews are due September 7, author response runs September 14 to 24, meta-reviews release October 8, and the cycle ends October 11. Nothing about the calendar is dramatic. What is dramatic is what it is being asked to hold.

The May 2026 cycle drew 17,087 submissions with 1,424 qualified area chairs and 10,636 reviewers, up from 10,518 in the January cycle. In a Medium post asking whether ARR is actually broken, David Jurgens, an Assistant Professor at the University of Michigan School of Information and Program Chair for ACL 2026, reports that around 40% of May submissions had no qualified reviewer authors, versus 18% in January. Outsider papers were accepted to ACL 2026 Main or Findings at 25%, against 39% for papers with qualified authors, and all-student outsider papers at 10%.

ARR's response to the scale problem has been structural rather than editorial. The changelog and dates show a switch from 8-week to 10-week cycles in May 2025 to give reviewers and authors more room, and a mandatory reviewer expectation for all qualified authors. A tool called Revas, developed with academic labs to flag common review issues, is being trialled from March 2026. The Call for Papers still bars generative AI from authorship and requires disclosure of LLM-assisted polishing or low-novelty text generation, with authors on the hook for correctness.

The honest caveat is that a submissions page is not a state-of-the-field report. The 17,087 May figure and Jurgens's outsider-author analysis come from a single Program Chair's post, not an ARR audit, and we do not know how many of those May papers survived to commitment at EMNLP or AACL, or how the August cycle will actually staff its reviewer pool. Take the specifics as reported, not settled.

For teams planning submissions, the takeaway is less about the deadline and more about the reciprocal-service math. In a queue this size, being able to field qualified reviewers from your own author list is becoming a de facto acceptance-rate lever, and the groups best placed to benefit are the industry labs and larger university groups that can absorb the load without slowing their own work.