Where to Submit Natural Language Processing Work
Published · paper counts are current, not as-published
Natural language processing runs more venues than any other field indexed here, and they are harder to plan around than that number suggests. The useful facts are not about subject matter - they are about which venues actually run in a given year, and how much of the field two of them absorb.
The Calendar Has No Rule You Can Rely On
9 venues are indexed here, and only 2 of them keep a schedule you can state in a sentence. The rest run irregularly: gaps of one year and two alternate without settling, and a venue can skip a year and return. Computer vision is the opposite case - there 2 venues split the calendar cleanly by parity, so you always know which is open. Here you have to look it up, which is what the table is for.
| Venue | Papers | Indexed editions | Years | Runs |
|---|---|---|---|---|
| ACL | 23,254 | 27 | 2000-2026 | every year |
| EMNLP | 20,347 | 22 | 2004-2025 | every year |
| NAACL | 8,406 | 17 | 2000-2025 | irregularly |
| COLING | 8,129 | 13 | 2000-2025 | irregularly |
| EACL | 4,564 | 10 | 2003-2026 | irregularly |
| SEMEVAL | 3,671 | 17 | 2007-2026 | irregularly |
| IJCNLP | 1,715 | 7 | 2005-2025 | irregularly |
| CONLL | 1,322 | 11 | 2010-2024 | irregularly |
| AACL | 757 | 4 | 2020-2025 | irregularly |
The Field's Memory Runs Deepest Here
That matters for more than trivia. A venue with a long indexed run is the only place you can watch vocabulary turn over rather than infer it: terms that carried a field in one decade and vanished in the next are visible as a series rather than as an anecdote. The research trends page ranks that movement year by year across the whole corpus.
Less Concentrated Than It Looks
The two largest venues take 60% of the field's indexed papers. That is a lot, but it is less than computer vision, where the top two take 75%. The difference is the long tail: this field keeps 7 further venues alive, several of them regional, where computer vision keeps 2.
What This Does Not Show
- Not deadlines, and not acceptance rates. The corpus records published papers, not submission dates or review outcomes - which are the things a submission plan actually turns on.
- "Irregular" is a description, not a verdict. It means the gaps between indexed editions are uneven. A venue may keep a schedule this corpus cannot see, and one that moved between venue codes would read as a gap.
- Editions indexed here, not a venue's full history. A venue older than its first year above ran before this index reached it.
- A field is not its indexed venues. Language work also appears at the general machine-learning venues and in journals outside this set.
For the same comparison in computer vision, see where to submit computer-vision work. The numbers here are read from the database when you load the page.