Association for Computational Linguistics (ACL)

What it is

The Association for Computational Linguistics (ACL) is a scientific and professional organization for people working on natural language processing, the field concerned with getting computers to handle human language. Its annual conference is one of the primary high-impact venues for NLP research, alongside EMNLP.^[1] The name refers both to the society and to its main meeting, so a piece of work is often identified simply by the conference it appeared in.

Origins

The society was founded in 1962 under a different name, the Association for Machine Translation and Computational Linguistics (AMTCL), with an initial membership of about 100 people.^[1] Its founding meeting took place on 13 June 1962, and the attendees were Victor Yngve, David Hayes, Harry Josselson, Ida Rhodes, Paul Garvin, Winfred Lehman, Richard See, Anthony Oettinger and Sydney Lamb.^[2] Victor Yngve of the University of Chicago was the first program chair and the first president of the association.^[3] In 1968 the organization changed its name to the Association for Computational Linguistics.^[1]

The original title joined machine translation and computational linguistics in a single name, which says something about how the two were understood at the time: translating between languages was among the most concrete and best-funded problems in the young field, and the society was built around the people working on it as well as on the broader questions of how to represent and process language by machine.

Meetings and proceedings

The first annual meeting was held on 25–26 August 1963 at the Denver Hilton, scheduled to precede the ACM national conference so that attendees could go to both.^[3] Publication of the annual meeting's proceedings, the Proceedings of the ACL, began in 1979.^[1] For a long time the meeting stayed in North America; the first ACL meeting held outside the region took place in Madrid in 1997.^[3] In 2020 the annual meeting received more submissions from China than from the United States for the first time, a marker of how widely the research community had spread.^[1]

Journals

The society's journal has changed name several times, and the lineage runs through all of them: it began with Mechanical Translation and Computational Linguistics, which the association took over in 1965, continued as the American Journal of Computational Linguistics (published 1974–1978 and again 1980–1983), and became Computational Linguistics from 1984. Since 1988 that journal has been published for the ACL by MIT Press.^[1]

Chapters and wider activities

The association has regional chapters: European (EACL), North American (NAACL) and Asian (AACL).^[1] Its activities extend well beyond the flagship conference and include EMNLP, IJCNLP, CoNLL, SemEval and the Workshop on Statistical Machine Translation (WMT), and it sponsors the journals Computational Linguistics and TACL.^[1] These events cover different parts of the field — main-conference research, shared evaluation tasks, parsing and learning-oriented work, machine translation — and give researchers recurring places to compare methods on common data.

Place in the field

A learned society matters in a field like natural language processing because it does the unglamorous connective work: it holds the meetings where results are presented and argued over, it keeps an archive of published papers, it maintains journals with editorial boards, and it gives a scattered international community a shared calendar and a shared set of standards. The ACL's history tracks the field's own shifts — from a small group centred on machine translation to a large, geographically distributed research area whose submissions now come from every continent, with regional chapters handling the local volume. Its conference and its journal remain the two places where work in the area is most often measured against the rest of the literature.