naacl-topic-selection

NAACL Topic Selection

Every venue in the *ACL family reviews through the same ARR machinery, at

the same rigor, with heavily overlapping reviewer pools. So "is this good

enough for NAACL?" is usually the wrong question. The right ones: *does the

NAACL calendar fit this project's timing, and is the room that gathers at

NAACL the room this paper wants*. This skill decides both before a word is

written.

What the NAACL room is

Since the 2024 renaming — an all-member vote (September 2024) changed

"North American" to "Nations of the Americas" — the chapter's stated

identity spans the hemisphere. That identity shows up materially, not just

nominally: NAACL 2024 in Mexico City ran a "Languages of Latin America"

theme track, NAACL 2025 ran "NLP in a Multicultural World," and workshops

like AmericasNLP (Indigenous languages of the Americas) co-locate here.

None of this narrows scope — general NLP work is the bulk of every program

— but it defines what this venue's audience uniquely rewards:

• work on Spanish, Portuguese, French/Creoles, and Indigenous languages of

the Americas, especially with community involvement;

• dialectal, code-switching, and migration-linked language phenomena of

the hemisphere;

• multicultural and cross-variety evaluation methodology;
• plus the full breadth of general NLP that any *ACL flagship takes.

Routing among near-identical siblings

| Situation | Better target | Reasoning |

|---|---|---|

| Reviews will be ready when a NAACL commitment window is open | NAACL | Timing is the first-order variable in this family |

| No NAACL edition in the next ~12 months (skipped year) | ACL / EMNLP / EACL / AACL | Reviews age; commit them somewhere real |

| Corpus/resource paper where the artifact is the contribution | LREC-COLING | Resource-first reviewing culture |

| Result needs >8 content pages of argument | TACL | Journal format, *ACL audience, presentation slot at a conference |

| Contribution is a modeling/optimization advance, language incidental | ML venue (NeurIPS/ICML/ICLR) | The delta is judged against ML lineage, not NLP lineage |

| Americas-language work, early stage or niche | AmericasNLP or another co-located workshop | Right audience, faster feedback, non-archival options |

| Paper speaks to hemisphere-specific phenomena or communities | NAACL over equal-timing siblings | The one case where venue identity outranks timing |

The three-question fit test

1. Timing: which ARR cycle would this enter, and does a NAACL

commitment window plausibly follow it? (If the next edition is

unannounced, this answer is "unknown" — plan per naacl-workflow.)

1. Audience: name five groups you want reading this paper. Do they

attend NAACL specifically — or would any *ACL flagship reach them?

1. Theme leverage: does the current edition's theme track (they change

every edition; verify the live CFP) give this paper a lane where it

competes on its actual strengths?

Decision rule of thumb:
  timing_fits && (audience_specific || theme_fits)  -> NAACL
  timing_fits && generic *ACL audience              -> NAACL ok,
                                     siblings equally fine — pick by date
  !timing_fits                                      -> nearest sibling now;
                                     NAACL next time it exists

Two routing vignettes

Vignette A. A lab finishes a strong instruction-tuning analysis in

June of a skipped year. Audience: general NLP. Timing: the next NAACL is

ten-plus months out and unannounced; an EMNLP-feeding cycle closes in

weeks. Routing: the sibling venue, without sentiment — question two showed

nothing NAACL-specific about the audience, so waiting buys nothing and

costs baseline freshness.

Vignette B. A collaboration with a Paraguayan university has built the

first treebank for a Guaraní variety, with community annotators. Audience:

exactly the researchers, community members, and workshop ecosystem that

convene at NAACL and AmericasNLP. Routing: hold for the NAACL window even

at some cost in timing; also weigh LREC-COLING (resource-first reviewing)

and decide by whether the paper's core is the resource or the *analysis

the resource enables*. This is the identity-outranks-timing case, and it

is rarer than teams think.

Shaping the project once NAACL is chosen

• If the work touches any language of the Americas, invest early in native

data and community contact — the exact dimensions this audience probes.

• Build per-language and per-variety reporting into the experimental design

from day one rather than retrofitting it.

• Draft the contribution sentence with explicit language scope; if that

sentence sounds thin, the issue is the evidence plan, not the venue.

• Check the edition's tracks (industry, demos, student research) — a paper

straining to be a main-conference long paper is sometimes a strong track

paper.

Anti-patterns

• Choosing NAACL for prestige-ranking reasons: within this family the

tiers are a fiction; the programs differ by region and season, not bar.

• Sitting on finished work for a year "for NAACL" while its baselines rot.
• Framing a translated-benchmark study as an Americas-languages paper; the

audience that rewards the frame is the audience that detects the shortcut.

Output format

[Fit verdict] NAACL / sibling (which) / workshop / journal / ML venue
[Timing check] <cycle -> commitment window status>
[Audience check] <who convenes here that this paper needs>
[Theme leverage] <current theme track fit, if verified>
[Project shaping] <two concrete design commitments>