Forty Years of ACL Meetings, 1963-2002

After a foundational meeting on June 13, 1962, The Association for Machine Translation and Computational Linguistics (AMTCL) held its first Annual Meeting at the Denver Hilton on Aug. 25-26, 1963, on the two days preceeding the national conference of the Association for Computing Machinery. The proceedings of this first AMTCL meeting were published later in the journal Mechanical Translation 7(2). The association changed its name in 1968 to the Association for Computational Linguistics (ACL), but without affecting the continuity of the organization. So ACL meetings have been held annually since 1963, for what now amounts to forty years, and we now have the Proceedings of the Fortieth Annual Meeting of the ACL.

Victor Yngve of the University of Chicago was the first program chair and first Association president. Other officers of the founding board were David Hayes, Vice-President, and Harry Josselson, Secretary-Treasurer, who were joined by Ida Rhodes, Paul Garvin, and Winfred Lehman as council members. Richard See, Anthony Oettinger and Sydney Lamb constituted the nominating committee. The published purpose of the organization was to "encourage high standards by sponsoring meetings, publication and other exchange of information."

It is remarkable that this group of scientists had the foresight and faith in computational linguistics (CL) to found the ACL and conduct annual meetings as early as 1963, but, in fact the founding meeting was preceeded by ten years of discussion on studying language computationally, dating back to a first meeting on CL convened by Yehoshua Bar-Hillel at the M.I.T. faculty club on June 17-20, 1952. Victor Yngve relates how that first meeting led to the founding of the ACL in his contribution to a twentieth anniversary celebration in the 1982 ACL Proceedings (pp.92ff). One of the fantastic results of Steven Bird's ACL Anthology effort is that all of this sort of material will now be easily available. The ACL's history will no longer languish on older members' shelves.

In many ways the annual meetings are the heart and soul of the ACL, and the Proceedings therefore the best record of the remarkable scientific activities and accomplishments of its members. There are many reasons for this. First, the association's journal (begun in 1974) publishes a smaller number of longer papers, and therefore illuminates a narrower band of the research results. But second and more importantly, ACL has a (controversial) tradition of insisting on an unusual level of quality in its contributed papers. The typical ACL call for papers solicits contributions "on substantial, original, and unpublished research on all aspects of computational linguistics" and stresses that "[papers] should emphasize completed work rather than intended work."

The stringent formulation is reflected in an acceptance rate many journals envy, typically around 25% during those years for which statistics are available (most years since 1983). As a consequence, researchers in CL typically view ACL Proceedings as archival, i.e., the sort of publication which need not be duplicated in a journal. Unfortunately, this has not been reflected in general availability until the current Anthology, another reason to celebrate its advent. The selectivity of the conference has been a matter of pride for many members, but it is also controversial, first because the acceptance of a contributed paper is a condition for many members for having their conference travel costs reimbursed, so that selectivity tends to depress conference attendance rates, and, less importantly, because external evaluations of research in academia -- which are unaware of the ACL's selectivity -- may discount contributions in conference proceedings entirely as evidence of scientific contribution. A solution to the selectivity dilemma has emerged over the past five years, and this has involved adding specialized satellite workshops to the main program of the conference, whose publication is separate. There has likewise been experimentation with demonstration and poster sessions, and again, these have normally been distinguished in publication. In this way publication in the Proceedings continues to carry professional prestige, even while larger numbers are attracted to the annual conference. This electronic publication is welcome among other reasons for the uniform accessibility it gives to the large mass of ACL "grey literature" alongside the Proceedings.

The concerns of the research have narrowed over the years, even while some very general issues recur frequently. While early researchers had a sense of probing not only language, but also the nature of symbolic computation, artificial intelligence, and even human intelligence considered philosophically and psychologically, later volumes restrict themselves much more soberly to language and a linguistic perspective. On the other hand, a very consistent thread throughout the forty years of ACL proceedings is the conviction of its researchers that CL is of enormous practical use -- even very theoretical papers often express their wish to help realize the dream of the machine that communicates in natural language, with all the practical benefits that will bring with it. It is gratifying to see papers from more recent years celebrate the ways in which language technology is being applied practically. The general issue of the proper balance of practical and applied work may not often be addressed explicitly, but one certainly gets the sense that it has been present all along, with significant swings occasionally in the one or the other direction.

Since 1997 ACL has taken great pains to involve all of the world's computational linguistics in its conferences, and this has increased interest and attendance again. Mitch Marcus, Eva Hajicova, Phil Cohen and Wolfgang Wahlster played especially important roles in this. The increased interest motivates in yet another way efforts to improve the easy availability of scientific literature, certainly for all those who are now discovering the ACL's forty years of research on language and computation. The ACL should not hide its light under a bushel!

John Nerbonne, ACL President 2002
Last Revision: June 4, 2002