Skip to main content

26. After acceptance

The email says Accept — or Accept to Findings, which is also a win (Chapter 2). Congratulations; now the work is not quite done. Acceptance starts a short, deadline-driven last mile: a camera-ready version with its own rules, a talk or poster to prepare, and the small task of getting people to actually read what you wrote. This chapter is that last mile. It is mostly logistics, but logistics you can get wrong, and the camera-ready deadline is as immovable as the submission one.

The camera-ready​

The final version is not the submission with the author names pasted on. Several things change, and a few are easy to forget.

  • The extra page. Accepted papers get one more content page — "up to nine (9) pages of content" for a long paper (five for a short), plus unlimited acknowledgments and references (ACLPUB). It exists for the changes reviewers asked for, not for padding. Use it to honor what you promised in the rebuttal.
  • De-anonymize. Add the author names and affiliations under the title — full names, "do not abbreviate given names to initials" — and turn every anonymized self-citation back into a normal one ("as we showed in Okeke 2024," not "(Anonymous, 2024)"). Restore the code and data links you withheld for review.
  • Add the acknowledgments. A new, unnumbered Acknowledgments section goes "immediately before the references"; it does not count toward the page limit and was absent from the review version. Thank your funders with grant numbers, your compute providers, and the people who helped — but do not try to thank reviewers by guessing who they were.
  • Sign the license. You will sign the ACL copyright/license agreement on behalf of all authors through the conference system. Authors retain most rights; it is routine.
  • Final polish. Turn off page numbers (the final version is unnumbered), make every figure and table legible, fix the typos reviewers flagged, and confirm you are still within the limit. Then send the code and data you said you would.

Honor what you promised​

The extra page is there for a reason: the changes you committed to in the rebuttal. If you wrote "we will add the significance test," "we will discuss Müller et al.," or "we will release the code," the camera-ready is where you make good — area chairs and conscientious reviewers do sometimes check, and a promise quietly dropped is noticed. It is also simply how trust in the process survives.

Poster or oral?​

Most accepted papers present as posters; a smaller set are given oral or spotlight slots. Neither is a consolation prize — posters often get you better conversations than a talk does. Conferences also ask for materials uploaded ahead of time: a poster PDF, slides, and frequently a pre-recorded video for the virtual audience. Check your conference's presenter instructions and the deadline, which comes weeks before the event.

Giving the talk​

*ACL talks are short — ACL 2025, for instance, gave each presentation eight minutes plus three for questions (ACL 2025 FAQ). You cannot present the paper in eight minutes, so do not try. Tell the story:

  • One arc. The problem, why it matters, your one idea, the single result that supports it, and the takeaway. Cut everything else.
  • Slides that read from the back row. Big fonts, one point per slide, few words, colorblind-safe figures (Chapter 16 on figure craft applies here too). Describe each figure aloud for anyone who cannot see it well.
  • Practice and time it. Run it out loud until it fits with room to spare. Anticipate the obvious questions; "I don't know — good question" is a perfectly respectable answer.

Making a poster that works​

A poster is a visual argument, not your paper shrunk to fit. The classic failure is eight pages of 10-point text taped to a board that nobody reads.

  • A title and a one-sentence takeaway large enough to read from across the aisle — many people will only get that far, so make it count.
  • A few figures, very little text. The poster supports your spoken explanation; it is not a substitute for it.
  • Rehearse the one-minute and the five-minute version, and bring a way for people to find the paper later — a QR code to the Anthology page is plenty.

At the conference​

The hallway is the point. The talks are recorded; the conversations are not. Introduce yourself, ask people about their work, and have a thirty-second and a two-minute description of yours ready. Being a first-year is not a handicap — most people are generous, and "I'm a first-year PhD student working on X, could you tell me about Y?" opens almost any door.

Getting the work seen, honestly​

A paper nobody reads helped nobody. So, without overselling:

  • Put it where people look. Post it to arXiv if you have not, and let the ACL Anthology page be the canonical link.
  • Say it plainly. A short, honest thread or post — what problem, what you found, one figure, a link — does more than breathless hype. The "we beat SOTA" reflex from the jargon chapter reads as thin; let the result speak.
  • Release what you promised, and make it easy to use — a documented repository, a model card, a dataset card. Reusable artifacts are cited far more than unreusable ones (Chapter 22).
  • Keep your record current. Update your website and your Semantic Scholar / Google Scholar profile so the paper is findable under your name.

Common mistakes​

  • Treating the camera-ready as a formality. Forgetting the acknowledgments, leaving a self-citation anonymized, or skipping the de-anonymization is a sloppy final impression on a paper you worked months on.
  • Padding the extra page instead of spending it on the reviewer-requested changes.
  • Dropping a promised change or release. It is noticed, and it erodes the trust the rebuttal earned.
  • A talk that reads the paper aloud, or a poster that is the paper at 8 point. Tell a story; show an argument.
  • Overselling. Hype ages badly and the field has a long memory; modest and accurate wins respect.

Further reading​